Update - Tuesday 10 March: In an update to this story, a Discord Spokesperson provided the following comment: "We recently became aware of an issue impacting some Arc Raiders players involving debugging features intended for developers building and testing Social SDK game integrations. This resulted in some players’ Discord information from the game being stored locally on their device, and viewing it would require access to the device or the files themselves. Embark has released a hotfix for the issue and we are providing guidance to developers and updating the Discord Social SDK with additional protections.”

A computer engineer has discovered a serious vulnerability within Discord's SDK that allows games to store Discord DMs between players in game logs without any security measures. System engineer Timothy Meadows published a blog post, revealing an incident where Arc Raiders was storing DMs between two gamers in plaintext to a local log file. Thankfully, at the time of writing, the problem has since been hotfixed by Embark Studios.

Timothy discovered that Arc Raiders' Discord SDK was using a completely unencrypted bearer token and logs "all events" including any private conversations to the user's local drive without any encryption. A bearer token stores the user's Discord credentials, and anyone who gets this token has full access to the Discord user's account, including private DMs, friends list, and account settings.

This is made worse by the fact that if Arc Raiders crashes and the user sends log files to Embark Studios (the game's development team), the company's employees will have that user's full account credentials and any DMs that were sent to the log files.

Arc Raider uses the Discord SDK to show your Discord friends list in-game and invite Discord friends to the game. For this limited functionality, Timothy states the game only requires a "limited OAuth scope for game activity display." This would solve the issue and stop Arc Raiders from recording DMs to log files and storing a user's full account credentials to the game's log files as well. Some engineers who've inspected Discord's API say the issue lies solely with Discord, however.

I dug into the ARC Raiders Discord token leak issue; this might not be ARC Raiders or Embark's fault. Discord's new Social SDK has a logging hook you can override, and as far as I can tell Discord is failing to scrub log events of sensitive information.API: discord.com/developers/d...

— @eidolon.photon.institute (@eidolon.photon.institute.bsky.social) 2026-03-10T11:17:16.528Z

Thankfully, Embark Studios has since patched the issue with a hotfix. The game company assured users that no private or personal data was sent outside of gamers' PCs, and the company itself has not reviewed or kept any personal information that might have been sent to them. Embark Studios has completely disabled Discord's SDK and is conducting an audit to ensure that there are no other problems with the SDK.

This isn't the first time Discord has to deal with security issues. The social app was hacked by a ransomware group late last year, demanding $3.5 million from Discord's developers, and allegedly stole 70,000 government ID photos.

To the surprise of us in the newsroom and likely numerous existing Discord users who have felt the application get heavier and less performant over time, one lone developer has successfully backported Discord with most of its key functions intact to Windows 95 and Windows 98 Second Edition. Boiling down Discord to its most basic form, it still makes sense that this is possible. After all, any fancy modern instant messenger is just an IRC client with a fancy hat.

IProgramInCpp did this development work and stands out as one of many examples of unexpected applications being made to work on Windows 95 and 98 — keep in mind that we've once seen the entire .NET Framework (and thus thousands of dependent apps) be backported to Windows 95.

According to the developer, running the aptly named Discord Messenger on Windows 95 will require much more of the end user than on Windows 98 Second Edition, where the client works out of the box. For older versions of Windows, you'll need to compile the program with MinGW.

Detailed installation instructions are available on the GitHub page for those hoping to get a Discord client running on their old PC. Still, fortunately, Windows 98 SE, Windows XP SP2, and other newer versions of Windows should be able to run this one pretty seamlessly. It even still works on modern Windows versions, though you have very little reason to use it on a modern system unless you hate that Discord overhead.

Of course, this port isn't necessarily feature-complete, and some features will seemingly never be added, whether due to technical limitations or attempts to avoid flagging Discord's spam detection. Technically, using your own Discord client breaks TOS, so users are advised to be cautious with this client, especially if a safer alternative account is an option.

Features implemented during writing include the most basic messaging functionality: Direct Messages, servers, images, attachment uploads, message edits, replies, and deletion. This includes typing indicators, embeds, viewing server member lists, and untrusted URL indicators.

Unimplemented features to be added include the friend list, dark mode, entering voice channel and joining voice channels, blocking/closing DMs, muting channels, changing display name & custom status, and other core functionality still currently missing. Of these features, it seems that voice call support— both connecting to servers and supporting the protocols in use— is pretty far off in the future, which is sensible considering how relatively cutting-edge that aspect of Discord is compared to the rest, which is again basically just an IRC client with a fancy hat (or a dozen).

Some features that are just unplanned boil down to this application's intended use as an alternative Discord client rather than your central client. Sending friend requests, creating DM channels, entering servers, and most log-in methods, including QR codes and email addresses, are unplanned since they will likely trigger Discord's anti-spam measures.

A few weeks ago we reported on Runway Gen-2, an online text-to-video tool, which generates 4-second clips based on your prompts. Up until now, the free service has only been available in a closed beta that operated via a series of private channels on Runway's Discord server. However, as of today, Gen-2 is available to everyone and you can use it via the company's website instead of Discord. 

To use Runway Gen-2, you'll need to create a free account and at Runway's site and then navigate to app.runwayml.com/ai-tools/gen-2 for the tool (or click the link in your control panel there). The company has had its Gen-1 tool, which modifies existing videos to make them look different (ex: turning people into Claymation), available to everyone for months, but Gen-2 takes simple text prompts such as "robot drinking a beer" and turns them into clips. 

As we noted in our prior story on Gen-2, the video clips have no sound at all and they often have very limited movement. Though you can download the clips in MP4 format, they often look more like animated GIFs. For viewing purposes, I converted all the clips embedded in this article into GIFs, which has caused the colors to dither, but it seems to have no effect on their smoothness.

You can also upload a photo and then give Gen-2 a prompt that tells it to use that image as inspiration. I uploaded a headshot of myself and asked it show "this guy drinking a beer." 

The person in the video looks quite a bit like me -- he's a middle-aged, bald white guy with a salt-and-pepper beard and glowing green glasses -- but he's not really drinking a beer; it's not quite clear what he's supposed to be sticking in his mouth.

After you've entered your prompt (and possibly photo), you click the Generate button and the tool tells you if you're in the queue for your request to be processed. It then gives you a percentage showing you how much of your video is completed. In my tests, it always took less than a minute between when I entered my prompt and when the video was completed. This is consistent with what I experienced when using the beta version of Gen-2, except, in that case, I'd send my prompt to a Discord bot which didn't show a progress percentage but instead would just send me back the video after a minute or so.

When the video is complete, you will see it on-screen and you can play it right there or click a button to download it as an MP4 file. It will also be stored in the assets section of your account.

Unlike with the beta version of Gen-2, it appears that there's a limit to how many videos you can generate for free. My free account showed a limit of 60 seconds of video, which at 4 seconds per clip, amounts to 15 clips.  

If you use up your credits or you want extra features such as an upscaled (higher resolution) or videos that don't contain Runway's watermark, you need to pay for a Standard account which is $15 a month or $144 a year when paid at once. For that price, you get 125 seconds of video per month and can pay extra for more. 

Since gaining access to the private beta a few weeks ago, I've spent a lot of time playing with Gen-2. I'm impressed with how good many of the clips are, but it's very inconsistent and people can look like they come straight from the uncanny valley, with strange expressions in their eyes or misshapen body parts.

For better or worse, the tool is pretty bad at recreating specific characters. I asked it, at various times, to create videos of the cartoon characters Peppa Pig, Paddington Bear and Blue from Blue's Clues and they look sort-of like those characters and are usually cartoony, but the likenesses aren't very accurate. One particular image of a photorealistic Peppa Pig is really creepy and seems like it was ripped from a horror film.

You can also forget about asking for something involving a corporate logo. I asked it several times to create videos that used the Tom's Hardware name or logo and got gibberish letters instead. 

However, one thing Runway Gen-2 is really good at is generating images of robots drinking alcoholic beverages or doing other "bar" things. Every time I asked it for robots drinking or smoking or dancing or pouring beer or washing their hands in the men's room at a bar, I got pretty decent output. I even got good clips of robots doing standup comedy or lounge singing. Asking for robots playing billiards or foosball or darts a bar was more of a mixed bag.

Other areas of success I found include families eating meals together and time-lapse views of cities or nature scenes.

Despite the limitations of Runway Gen-2, we have seen a number of  very creative people turn them into longer movies by stitching many clips together. Perhaps the most famous example of a Runway Gen-2 movie is the Pepperoni Hugspot Commercial pizza commercial, made by a creator who goes by the name Pizza Later. Pizza later worked with a friend to create a commercial for a fake flatulence drug called FlatuLess, which we've embedded below. 

Note that the music and voiceovers were created with other AI tools and all the clips were stitched together by humans using Adobe After Effects. The clips look like they are from an old commercial because Pizza Later used the Red Giant VHS filter on them.

If you want to try Runway Gen-2 and see what you can create with it, all you need to do is head to runwayml.com and sign up for a free account.

Gaming goes with the Raspberry Pi like peanut butter and jelly, lending to tons of cool projects over the years; But we’ve never seen a project quite like this one from maker and developer Konrad Beckmann. Using a Raspberry Pi Pico, he’s able to interface with a Nintendo 64 via the cartridge port, a project of his that we covered back in 2022. However the latest update has enabled him to successfully stream the latest Zelda game, The Legend of Zelda: Tears of the Kingdom, to a Nintendo 64 console.

We spoke with Backmann about the new update, and he explained the original plan was to make an RP2040-powered cartridge that’s capable of running large games directly on the Nintendo 64. This proved to be difficult, however, so at the moment, he’s shifted his focus to streaming content to the Nintendo 64 through the Pico module.

Because he’s using a Raspberry Pi Pico W, he’s able to implement wireless support into the project. The games are run on a separate PC and formatted for streaming video to the Pico with a resolution of 320 x 240px. This configuration makes it possible to play games like Tears of the Kingdom and more using the microcontroller.

In addition to streaming Zelda, he’s successfully set up a screen sharing session from his PC and also implemented support for streaming his PlayStation 5 using pyremoteplay. This required a solution in which the PS5 was streamed to his PC first, enabling a successful run of Bloodborne on the Nintendo 64.

Beckmann went on to elaborate that the RP2040-powered cartridge he uses in this particular project is the PicoCart64 v1 lite version. The software driving the project is Python-based and can be found over at GitHub. The code is totally open source for anyone who wants to check it out and understand more about how it goes together. A Discord server is also available for anyone that wants to ask questions and dig into the development of the project.

If you want to recreate this Raspberry Pi project or just see it in action, check out Konrad Beckmann’s Twitter profile for more links and details. Beckmann has been experimenting with using the Raspberry Pi Pico on the Nintendo 64 for a few months, so be sure to follow him for more cool projects as well as any future updates on this one.

Text-to-video is the next big thing in AI. We saw a couple of weeks ago how awesome (and a bit creepy) the AI-generated Pepperoni Hugspot commercial was. Pizza Later, the person who developed that video, told us that they used a tool called Runway Gen-2 to do the moving images in that project. With its text-to-video engine, they were able to give simple prompts such as "a happy man/woman/family eating a slice of pizza in a restaurant, tv commercial," and get photo-realistic content. 

I've just gotten access to the public beta of Runway Gen-2 and I'm really impressed by the realistic nature of its output. While the videos are short at just four seconds each, the quality of the imagery is impressive and it all works via sending short requests to a bot on Runway ML's Discord server.

By sending a few words of text to the @Gen-2 bot, I was able to get short, photo-realistic (or cartoon style) clips of everything from a family enjoying a sushi dinner to a robot with a serious drinking problem. The output was frequently not exactly what I asked for, but it was always interesting and was superior to the NeuralInternet Text-to-Video Playground I wrote about last week.

While anyone can join the server, you'll only see the list of Gen-2 chatrooms once you gain access to the beta program (which many are on the waitlist for). There are some rooms where you can chat and share projects with other users and then there are three rooms named Generate One, Generate Two and Generate Three where you can go to send prompts directly to the @Gen-2 bot. The moderators encourage you to keep issuing prompts to the same thread so as not to make a mess of each chatroom.

Prompting Runway Gen-2

A Runway Gen-2 prompt might look be something like "@Gen-2 A drunk humanoid robot that is looking at the camera and vomiting small screws out of its mouth." The bot will immediately respond back with your prompt and some parameters it's using (ex: "upscaling") which you can change by issuing a new prompt (more on that later). Then, a couple of minutes later, you'll get get a 4-second video based on your prompt. 

Here's what my drunk robot looked like. All of the videos are playable from within Discord and you can download them as MP4 files. I have separately converted all the video samples shown in this article into animated GIFs so we can view them more easily (and without pre-roll ads). 

You'll notice that the above clip wasn't exactly what I asked for. The robot is not vomiting up screws as I had intended. Instead, it is just looking menacingly at a cup of beer. My other attempts at this prompt were not exactly what I wanted either. When I left the word "drunk" out, I got a robot opening its mouth but not spitting anything out. 

Using Pictures With Runway Gen-2 Prompts

You can also feed images to the bot by either copying and pasting them into Discord along with the text prompt or putting the URL of the image into the prompt. Runway Gen-2 will not actually use the image you uploaded, however. It will only draw inspiration from the image in creating its own video. I uploaded images of myself many times and it gave me videos of people that looked a little like me, but were definitely not me.

For example, when I uploaded a picture of myself and gave it no further information, it showed a balding, middle-aged man with sunglasses who was not me, standing next to a river and some buildings. His mouth moved and the water moved.

The Runway Gen-2 bot is better at copying the emotion or topic of an image you provide. I showed it an image of myself having a look of disgust on my face and asked for "this guy looking at the camera and mouthing 'oh man'."

A lot of the users on the Discord server say that they achieved great results by generating a still image with another AI tool such as Midjourney or Stable Diffusion, then feeding that image to CLIP Interrogator 2.1 on Hugging Face, a tool which looks at an image and then gives you prompts that it thinks refer to that image. 

I tried that process, asking Stable Diffusion to make me an image of a boy on a sidewalk playing with toy robots in the 1980s. I then took the image into CLIP Interrogator and got some sample prompts for it which were pretty obvious such as "boy standing next to robot." Still, feeding the same image into prompt didn't quite give me what I wanted. I got a boy with two robots standing in front of a street, but it wasn't the same street or boy.

To Move or Not to Move

The time limitation itself usually means that there's not a lot of time for movement in each clip. But, on top of that, I found that many clips had very little movement in them. Often, it was just someone's head bobbing or some liquid flowing or smoke rising from a fire. 

A good way to get more movement is to put a prompt in which requests a time-lapse or panning of some kind. When I asked for a time-lapse of an Icelandic Volcano or a pan shot of a New York subway, I got some pretty nice results. When I prompted for a panning view of the Taipei skyline, I got clouds moving but no panning, and the city was definitely not Taipei.

Asking for running, chasing or riding may or may not get the job done. When I prompted for a "tortoise skateboarding," I got some kind of werid tortoise-like animal rolling down the street at rapid speed. But when I asked for Intel and AMD boxers fighting each other, I got a picture of two boxers that didn't move at all (and neither had Intel or AMD logos).

What Runway Gen-2 is Good and Bad At

Like other AI image generators, Runway Gen-2 doesn't do a great job of reproducing very specific, branded characters, products or places. When I asked it for Mario and Luigi boxing, I got two characters that look like knock-offs of Nintendo's characters. I asked many times for videos of Godzilla and got some giant lizards that not even the most casual fan would confuse with the King of Monsters. 

It was a little better with Minecraft references. When I asked for a creeper and an enderman eating pizza and again for a creeper eating at McDonald's, I got decent-looking creepers but an inaccurate enderman. Asking for a family of creepers eating pizza gave me a family of humanoids who look like they came from Minecraft. Anyone who has played Minecraft knows that creepers are green monsters with black spots. 

The tool is terrible with logos. I gave it the Tom's Hardware logo and asked it to use the logo in a commercial and it gave me back this this bizarre thing. 

When I asked it for an AMD Ryzen CPU on fire, I got something that looked vaguely like a PCU with a logo that you just have to see for yourself (below).

What Runway Gen-2 does really well is give you generic images of people and families doing things like eating. You may or may not get them to eat exactly what you want. When I asked for a family eating live worms, I got a family that looked more like it was eating salad. A family eating sushi in a 1970s pizza restaurant looked particularly realistic.

I feel compelled to point out that I almost always got white people when I asked for a person without specifying their ethnicity. The only time I got a non-white family (or person) without specifically asking for one is when I asked for the family eating sushi. This is a well-known problem with training data on many generative AI models.

Special Parameters

There are a handful of parameters you can add to the end of your prompt in Runway Gen-2 in order to change the output a bit. I didn't fool around with these a lot.

• --upscale delivers higher resolution
• --interpolate makes the video smoother
• --cfg [number] controls how creative the AI becomes. Higher values are closer to what you asked for. 
• --green_screen output ths the video with a green screen area you can use in editing
• --seed is a number that helps determine the outcome. By default, it's a random number every time, but if you use the same number again, you should get a similar result.

Stitching It All Together

If you search the Internet for examples of Runway Gen-2 videos, you may notice a lot of videos that are longer than 4 seconds and have sound. People make these videos by stitching together many different 4-second clips in a video editor and adding sound and music that they've gotten elsewhere.

One of the most famous of these Runway Gen-2 videos is the Pepperoni Hugspot pizza commercial which I mentioned above. But, in the Runway ML Discord, I see a lot of people posting  YouTube links to their creations. One of my favorites is "Spaghetti Terror" which was posted to Twitter by Andy McNamara. And Pizza Later's new lawyer commercial is a hoot.

Bottom Line

Runway Gen-2 is in private beta as I write this, but the company has said it intends to make it available to everyone soon, as it already has with its Gen-1 product. As a technology demo, it's truly impressive and I can see someone using its short clips in lieu of stock video or stock animated GIFs. 

Even if the time were extended to 60 seconds, it seems unlikely that this tool could replace professionally (or even amateurly) shot video anytime soon. Its inability to accurately reproduce very specific places and people is a huge downer, but it's also a limitation I've seen in every image-generating AI thus far. However, the technology is right there and, as the training data scales up, this could be even more impressive. 

ChatGPT can give some impressive results, and also sometimes some very poor advice. But while it's free to talk with ChatGPT in theory, often you end up with messages about the system being at capacity, or hitting your maximum number of chats for the day, with a prompt to subscribe to ChatGPT Plus. Also, all of your queries are taking place on ChatGPT's server, which means that you need Internet and that OpenAI can see what you're doing.

Fortunately, there are ways to run a ChatGPT-like LLM (Large Language Model) on your local PC, using the power of your GPU. The oobabooga text generation webui might be just what you're after, so we ran some tests to find out what it could — and couldn't! — do, which means we also have some benchmarks.

Getting the webui running wasn't quite as simple as we had hoped, in part due to how fast everything is moving within the LLM space. There are the basic instructions in the readme, the one-click installers, and then multiple guides for how to build and run the LLaMa 4-bit models. We encountered varying degrees of success/failure, but with some help from Nvidia and others, we finally got things working. And then the repository was updated and our instructions broke, but a workaround/fix was posted today. Again, it's moving fast!

It's like running Linux and only Linux, and then wondering how to play the latest games. Sometimes you can get it working, other times you're presented with error messages and compiler warnings that you have no idea how to solve. We'll provide our version of instructions below for those who want to give this a shot on their own PCs. You may also find some helpful people in the LMSys Discord, who were good about helping me with some of my questions.

It might seem obvious, but let's also just get this out of the way: You'll need a GPU with a lot of memory, and probably a lot of system memory as well, should you want to run a large language model on your own hardware — it's right there in the name. A lot of the work to get things running on a single GPU (or a CPU) has focused on reducing the memory requirements.

Using the base models with 16-bit data, for example, the best you can do with an RTX 4090, RTX 3090 Ti, RTX 3090, or Titan RTX — cards that all have 24GB of VRAM — is to run the model with seven billion parameters (LLaMa-7b). That's a start, but very few home users are likely to have such a graphics card, and it runs quite poorly. Thankfully, there are other options.

Loading the model with 8-bit precision cuts the RAM requirements in half, meaning you could run LLaMa-7b with many of the best graphics cards — anything with at least 10GB VRAM could potentially suffice. Even better, loading the model with 4-bit precision halves the VRAM requirements yet again, allowing for LLaMa-13b to work on 10GB VRAM. (You'll also need a decent amount of system memory, 32GB or more most likely — that's what we used, at least.)

Getting the models isn't too difficult at least, but they can be very large. LLaMa-13b for example consists of 36.3 GiB download for the main data, and then another 6.5 GiB for the pre-quantized 4-bit model. Do you have a graphics card with 24GB of VRAM and 64GB of system memory? Then the 30 billion parameter model is only a 75.7 GiB download, and another 15.7 GiB for the 4-bit stuff. There's even a 65 billion parameter model, in case you have an Nvidia A100 40GB PCIe card handy, along with 128GB of system memory (well, 128GB of memory plus swap space). Hopefully the people downloading these models don't have a data cap on their internet connection.

Testing Text Generation Web UI Performance

In theory, you can get the text generation web UI running on Nvidia's GPUs via CUDA, or AMD's graphics cards via ROCm. The latter requires running Linux, and after fighting with that stuff to do Stable Diffusion benchmarks earlier this year, I just gave it a pass for now. If you have working instructions on how to get it running (under Windows 11, though using WSL2 is allowed) and you want me to try them, hit me up and I'll give it a shot. But for now I'm sticking with Nvidia GPUs.

I encountered some fun errors when trying to run the llama-13b-4bit models on older Turing architecture cards like the RTX 2080 Ti and Titan RTX. Everything seemed to load just fine, and it would even spit out responses and give a tokens-per-second stat, but the output was garbage. Starting with a fresh environment while running a Turing GPU seems to have worked, fixed the problem, so we have three generations of Nvidia RTX GPUs.

While in theory we could try running these models on non-RTX GPUs and cards with less than 10GB of VRAM, we wanted to use the llama-13b model as that should give superior results to the 7b model. Looking at the Turing, Ampere, and Ada Lovelace architecture cards with at least 10GB of VRAM, that gives us 11 total GPUs to test. We felt that was better than restricting things to 24GB GPUs and using the llama-30b model.

For these tests, we used a Core i9-12900K running Windows 11. You can see the full specs in the boxout. We used reference Founders Edition models for most of the GPUs, though there's no FE for the 4070 Ti, 3080 12GB, or 3060, and we only have the Asus 3090 Ti.

In theory, there should be a pretty massive difference between the fastest and slowest GPUs in that list. In practice, at least using the code that we got working, other bottlenecks are definitely a factor. It's not clear whether we're hitting VRAM latency limits, CPU limitations, or something else — probably a combination of factors — but your CPU definitely plays a role. We tested an RTX 4090 on a Core i9-9900K and the 12900K, for example, and the latter was almost twice as fast.

It looks like some of the work at least ends up being primarily single-threaded CPU limited. That would explain the big improvement in going from 9900K to 12900K. Still, we'd love to see scaling well beyond what we were able to achieve with these initial tests.

Given the rate of change happening with the research, models, and interfaces, it's a safe bet that we'll see plenty of improvement in the coming days. So, don't take these performance metrics as anything more than a snapshot in time. We may revisit the testing at a future date, hopefully with additional tests on non-Nvidia GPUs.

We ran oobabooga's web UI with the following, for reference. More on how to do this below.

python server.py --gptq-bits 4 --model llama-13b

Text Generation Web UI Benchmarks (Windows)

Again, we want to preface the charts below with the following disclaimer: These results don't necessarily make a ton of sense if we think about the traditional scaling of GPU workloads. Normally you end up either GPU compute constrained, or limited by GPU memory bandwidth, or some combination of the two. There are definitely other factors at play with this particular AI workload, and we have some additional charts to help explain things a bit.

Running on Windows is likely a factor as well, but considering 95% of people are likely running Windows compared to Linux, this is more information on what to expect right now. We wanted tests that we could run without having to deal with Linux, and obviously these preliminary results are more of a snapshot in time of how things are running than a final verdict. Please take it as such.

We ran the test prompt 30 times on each GPU, with a maximum of 500 tokens. We discarded any results that had fewer than 400 tokens (because those do less work), and also discarded the first two runs (warming up the GPU and memory). Then we sorted the results by speed and took the average of the remaining ten fastest results.

Generally speaking, the speed of response on any given GPU was pretty consistent, within a 7% range at most on the tested GPUs, and often within a 3% range. That's on one PC, however; on a different PC with a Core i9-9900K and an RTX 4090, our performance was around 40 percent slower than on the 12900K.

Our prompt for the following charts was: "How much computational power does it take to simulate the human brain?"

Our fastest GPU was indeed the RTX 4090, but... it's not really that much faster than other options. Considering it has roughly twice the compute, twice the memory, and twice the memory bandwidth as the RTX 4070 Ti, you'd expect more than a 2% improvement in performance. That didn't happen, not even close.

The situation with RTX 30-series cards isn't all that different. The RTX 3090 Ti comes out as the fastest Ampere GPU for these AI Text Generation tests, but there's almost no difference between it and the slowest Ampere GPU, the RTX 3060, considering their specifications. A 10% advantage is hardly worth speaking of!

And then look at the two Turing cards, which actually landed higher up the charts than the Ampere GPUs. That simply shouldn't happen if we were dealing with GPU compute limited scenarios. Maybe the current software is simply better optimized for Turing, maybe it's something in Windows or the CUDA versions we used, or maybe it's something else. It's weird, is really all I can say.

These results shouldn't be taken as a sign that everyone interested in getting involved in AI LLMs should run out and buy RTX 3060 or RTX 4070 Ti cards, or particularly old Turing GPUs. We recommend the exact opposite, as the cards with 24GB of VRAM are able to handle more complex models, which can lead to better results. And even the most powerful consumer hardware still pales in comparison to data center hardware — Nvidia's A100 can be had with 40GB or 80GB of HBM2e, while the newer H100 defaults to 80GB. I certainly won't be shocked if eventually we see an H100 with 160GB of memory, though Nvidia hasn't said it's actually working on that.

As an example, the 4090 (and other 24GB cards) can all run the LLaMa-30b 4-bit model, whereas the 10–12 GB cards are at their limit with the 13b model. 165b models also exist, which would require at least 80GB of VRAM and probably more, plus gobs of system memory. And that's just for inference; training workloads require even more memory!

"Tokens" for reference is basically the same as "words," except it can include things that aren't strictly words, like parts of a URL or formula. So when we give a result of 25 tokens/s, that's like someone typing at about 1,500 words per minute. That's pretty darn fast, though obviously if you're attempting to run queries from multiple users that can quickly feel inadequate.

Here's a different look at the various GPUs, using only the theoretical FP16 compute performance. Now, we're actually using 4-bit integer inference on the Text Generation workloads, but integer operation compute (Teraops or TOPS) should scale similarly to the FP16 numbers. Also note that the Ada Lovelace cards have double the theoretical compute when using FP8 instead of FP16, but that isn't a factor here.

If there are inefficiencies in the current Text Generation code, those will probably get worked out in the coming months, at which point we could see more like double the performance from the 4090 compared to the 4070 Ti, which in turn would be roughly triple the performance of the RTX 3060. We'll have to wait and see how these projects develop over time.

These final two charts are merely to illustrate that the current results may not be indicative of what we can expect in the future. Running Stable-Diffusion for example, the RTX 4070 Ti hits 99–100 percent GPU utilization and consumes around 240W, while the RTX 4090 nearly doubles that — with double the performance as well.

With Oobabooga Text Generation, we see generally higher GPU utilization the lower down the product stack we go, which does make sense: More powerful GPUs won't need to work as hard if the bottleneck lies with the CPU or some other component. Power use on the other hand doesn't always align with what we'd expect. RTX 3060 being the lowest power use makes sense. The 4080 using less power than the (custom) 4070 Ti on the other hand, or Titan RTX consuming less power than the 2080 Ti, simply show that there's more going on behind the scenes.

Long term, we expect the various chatbots — or whatever you want to call these "lite" ChatGPT experiences — to improve significantly. They'll get faster, generate better results, and make better use of the available hardware. Now, let's talk about what sort of interactions you can have with text-generation-webui.

Chatting With Text Generation Web UI

The Text Generation project doesn't make any claims of being anything like ChatGPT, and well it shouldn't. ChatGPT will at least attempt to write poetry, stories, and other content. In its default mode, TextGen running the LLaMa-13b model feels more like asking a really slow Google to provide text summaries of a question. But the context can change the experience quite a lot.

Many of the responses to our query about simulating a human brain appear to be from forums, Usenet, Quora, or various other websites, even though they're not. This is sort of funny when you think about it. You ask the model a question, it decides it looks like a Quora question, and thus mimics a Quora answer — or at least that's our understanding. It still feels odd when it puts in things like "Jason, age 17" after some text, when apparently there's no Jason asking such a question.

Again, ChatGPT this is not. But you can run it in a different mode than the default. Passing "--cai-chat" for example gives you a modified interface and an example character to chat with, Chiharu Yamada. And if you like relatively short responses that sound a bit like they come from a teenager, the chat might pass muster. It just won't provide much in the way of deeper conversation, at least in my experience.

Perhaps you can give it a better character or prompt; there are examples out there. There are plenty of other LLMs as well; LLaMa was just our choice for getting these initial test results done. You could probably even configure the software to respond to people on the web, and since it's not actually "learning" — there's no training taking place on the existing models you run — you can rest assured that it won't suddenly turn into Microsoft's Tay Twitter bot after 4chan and the internet start interacting with it. Just don't expect it to write coherent essays for you.

Getting Text-Generation-Webui to Run (on Nvidia)

Given the instructions on the project's main page, you'd think getting this up and running would be pretty straightforward. I'm here to tell you that it's not, at least right now, especially if you want to use some of the more interesting models. But it can be done. The base instructions for example tell you to use Miniconda on Windows. If you follow the instructions, you'll likely end up with a CUDA error. Oops.

This more detailed set of instructions off Reddit should work, at least for loading in 8-bit mode. The main issue with CUDA gets covered in steps 7 and 8, where you download a CUDA DLL and copy it into a folder, then tweak a few lines of code. Download an appropriate model and you should hopefully be good to go. The 4-bit instructions totally failed for me the first times I tried them (update: they seem to work now, though they're using a different version of CUDA than our instructions). lxe has these alternative instructions, which also didn't quite work for me.

I got everything working eventually, with some help from Nvidia and others. The instructions I used are below... but then things stopped working on March 16, 2023, as the LLaMaTokenizer spelling was changed to "LlamaTokenizer" and the code failed. Thankfully that was a relatively easy fix. But what will break next, and then get fixed a day or two later? We can only guess, but as of March 18, 2023, these instructions worked on several different test PCs.

1. Install Miniconda for Windows using the default options. The top "Miniconda3 Windows 64-bit" link should be the right one to download.

2. Download and install Visual Studio 2019 Build Tools. Only select "Desktop Environment with C++" when installing. Version 16.11.25 from March 14, 2023, build 16.11.33423.256 should work.

3. Create a folder for where you're going to put the project files and models., e.g. C:\AIStuff.

4. Launch Miniconda3 prompt. You can find it by searching Windows for it or on the Start Menu.

5. Run this command, including the quotes around it. It sets the VC build environment so CL.exe can be found, requires Visual Studio Build Tools from step 2.

"C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Auxiliary\Build\vcvars64.bat"

6. Enter the following commands, one at a time. Enter "y" if prompted to proceed after any of these.

conda create -n llama4bitconda activate llama4bitconda install python=3.10conda install git

7. Switch to the folder (e.g. C:\AIStuff) where you want the project files.

cd C:\AIStuff

8. Clone the text generation UI with git. 

git clone https://github.com/oobabooga/text-generation-webui.git

9. Enter the text-generation-webui folder, create a repositories folder underneath it, and change to it.

cd text-generation-webuimd repositoriescd repositories

10. Git clone GPTQ-for-LLaMa.git and then move up one directory.

git clone https://github.com/qwopqwop200/GPTQ-for-LLaMa.gitcd ..

11. Enter the following command to install several required packages that are used to build and run the project. This can take a while to complete, sometimes it errors out. Run it again if necessary, it will pick up where it left off.

pip install -r requirements.txt

12. Use this command to install more required dependencies. We're using CUDA 11.7.0 here, though other versions may work as well.

conda install cuda pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia/label/cuda-11.7.0

13. Check to see if CUDA Torch is properly installed. This should return "True" on the next line. If this fails, repeat step 12; if it still fails and you have an Nvidia card, post a note in the comments.

python -c "import torch; print(torch.cuda.is_available())"

14. Install ninja and chardet. Press y if prompted.

conda install ninjapip install cchardet chardet

15. Change to the GPTQ-for-LLama directory.

cd repositories\GPTQ-for-LLaMa

16. Set up the environment for compiling the code.

set DISTUTILS_USE_SDK=1

17. Enter the following command. This generates a LOT of warnings and/or notes, though it still compiles okay. It can take a bit to complete.

python setup_cuda.py install

18. Return to the text-generation-webui folder.

cd C:\AIStuff\text-generation-webui

19. Download the model. This is a 12.5GB download and can take a bit, depending on your connection speed. We've specified the llama-7b-hf version, which should run on any RTX graphics card. If you have a card with at least 10GB of VRAM, you can use llama-13b-hf instead (and it's about three times as large at 36.3GB).

python download-model.py decapoda-research/llama-7b-hf

20. Rename the model folder. If you're doing the larger model, just replace 7b with 13b.

rename models\llama-7b-hf llama-7b

21. Download the 4-bit pre-quantized model from Hugging Face, "llama-7b-4bit.pt" and place it in the "models" folder (next to the "llama-7b" folder from the previous two steps, e.g. "C:\AIStuff\text-generation-webui\models"). There are 13b and 30b models as well, though the latter requires a 24GB graphics card and 64GB of system memory to work.

22. Edit the tokenizer_config.json file in the text-generation-webui\models\llama-7b folder and change LLaMATokenizer to LlamaTokenizer. The capitalization is what matters.

23. Enter the following command from within the C:\AIStuff\text-generation-webui folder. (Replace llama-7b with llama-13b if that's what you downloaded; many other models exist and may generate better, or at least different, results.)

python server.py --gptq-bits 4 --model llama-7b

You'll now get an IP address that you can visit in your web browser. The default is http://127.0.0.1:7860, though it will search for an open port if 7860 is in use (i.e. by Stable-Diffusion).

24. Navigate to the URL in a browser.

25. Try entering your prompts in the "input box" and click Generate.

26. Play around with the prompt and try other options, and try to have fun — you've earned it!

If something didn't work at this point, check the command prompt for error messages, or hit us up in the comments. Maybe just try exiting the Miniconda command prompt and restarting it, activate the environment, and change to the appropriate folder (steps 4, 6 (only the "conda activate llama4bit" part), 18, and 23).

Again, I'm also curious about what it will take to get this working on AMD and Intel GPUs. If you have working instructions for those, drop me a line and I'll see about testing them. Ideally, the solution should use Intel's matrix cores; for AMD, the AI cores overlap the shader cores but may still be faster overall.

A fix is on the way to solving the Discord bug plaguing Nvidia's GeForce gaming graphics cards. However, you may or may not already have the update since it apparently automatically downloads itself to your system without your knowledge. We've asked Nvidia for clarification on precisely who is affected, how the fix gets applied, and some other details and are still awaiting a reply, but here's what we know.

Nvidia announced the update for Discord a couple of days ago; however, the chipmaker's vague statement caused more confusion among GeForce graphics card owners. The post didn't include instructions other than that the patch will arrive as an OTA (over-the-air) update. Apparently, it practically requires no user intervention other than restarting their systems for it to take effect.

"NVIDIA GeForce users can now download an application profile update for Discord. This resolves a recent issue where some GeForce GPUs' memory clocks did not reach full speed while Discord was running in the background. The NVIDIA display drivers will automatically download and apply the updated application profile to your PC the next time you log into Windows. NVIDIA GeForce application profiles are a group of software settings used by the NVIDIA graphics driver to provide optimum performance when using a selected application," stated Nvidia in a recent blog post.

Initially, users were confused if the update would come as a Windows update or a driver update. Nvidia subsequently clarified that it's an application profile update that the Nvidia driver uses. However, the chipmaker didn't specify whether the GeForce Experience software is required for the automatic download and update to occur. Not everyone uses GeForce Experience because some gamers prefer to stick to the bare minimum and only install the GeForce driver.

There are still many unanswered questions, though. For example, Nvidia didn't confirm whether the mechanic is the same on the Studio Drivers and Game Ready Drivers. The company also didn't touch on the subject of Linux users. There's also little concrete information about which Nvidia GPUs and drivers specifically were impacted, and whether the same fix will occur on all of them.

The problem with automatic downloads occurring in the background is that you have no clue when or if the update has been installed. Affected users can leave Discord open and run a game to check whether the graphics card's memory is at full speed, but if the problem persists, it's also unclear what will trigger an update other than the suggestion to restart the PC. One Twitter user reported that Nvidia's automatic deployment doesn't purportedly work on Windows 11 systems or systems where automatic updates are disabled. In that case, you're better off applying the manual fix, which Nvidia has detailed here.

Our assumption right now is that, at the very least, you'll need GeForce Experience installed and running to get the application profile update to download and install — which means you'd also need to be logged in to GeForce Experience. We've reached out to Nvidia for clarification on some of the inquiries and will update with further information on the fix as it becomes available.

Update 02/03/2022 5:14 pm PT

Nvidia has divulged via its Nvidia Customer Care Twitter account that it has started deploying an updated application profile for Discord to fix the memory speed bug. However, it's not a Windows, Discord, or Nvidia driver update. Instead, it's a new application profile for the Nvidia display drivers, which the chipmaker can send over the air. Therefore, GeForce graphics card owners don't have to do anything other than restart their systems for the new profile to take effect:

You can find Nvidia's complete statement below:

"GeForce users can now download an app profile update for Discord. This resolves a recent issue where some GeForce GPUs memory clocks did not reach full speed w/ Discord running in the background. The update automatically downloads to your PC the next time you log into Windows."

Original Article

If you're an avid user of Discord and own an Nvidia graphics card, you may lose performance without knowing it. Nvidia has discovered a bug with the Discord application that will prevent the memory on even the best graphics cards from reaching its full speed.

The latest update to the Discord software results in a throttled memory clock on Nvidia graphics cards. When the software runs in the background, the graphics card will fail to hit the maximum memory clock, leading to a performance loss. It's an intriguing bug since it affects Nvidia graphics cards only and, more specifically, the memory subsystem. Unfortunately, the chipmaker didn't specify which models are impacted, so we can only assume that all graphics cards from the company are affected.

Nvidia didn't share how Discord disrupts the memory on the graphics card. However, Discord recently enabled AV1 support for Nvidia's latest GeForce RTX 40-series (Ada Lovelace) graphics card, so the bug could be potentially related to AV1 encoding support for video streaming. Nonetheless, it's a bug that has managed to slip through the developers.

Nvidia has stated it will push a fix for PC users with an over-the-air (OTA) update. However, there's a simple solution if you don't want to wait for the official fix. The chipmaker has shared the following method to solve the Discord application bug manually:

1. Click here to download the NVIDIA program GeForce 3D Profile Manager.

2. Open the GeForce 3D Profile Manager.

3. Click on the button Export SLI Profiles.

4. A file explorer window will pop up. Select a location to export and save the NVIDIA Profiles text file.

5. Open the text file saved in step 4 using Notepad (or any program that does not automatically apply formatting).

6. Perform a search for the section "Discord." Add a new line and type the following text as shown in the screenshot below:

Setting ID_0x50166c5e = 0x00000000

7. Save the edited NVIDIA Profiles as a txt file.

8. Go back to the GeForce 3D Profile Manager and click on the Import SLI Profiles button.

9. A file explorer window will appear. Select the updated "NVIDIA Profiles.txt" file and then click on the Open button.

10. Once completed, you may close the GeForce 3D Profile Manager application.

This week, the live-streamed broadcast on the Intel Insiders Community Discord channel highlighted a project with Intel Arc Graphics cards and LEGO. The show, a recording of which is now available to watch on Twitch, featured Zach Hill, an Arc Graphics tech marketing exec and LEGO fan. Hill has been working in his spare time for the last six months on creating a 1:1 or life-size model of the Intel Arc A750 graphics card using the huge library of LEGO bricks available in software.

The project aims to create a replica of the A750 as close as possible using available LEGO bricks. For this purpose, Hill uses BrickLink Studio, a free software package (LEGO Group owned) which allows 3D building and rendering of projects using the full gamut of LEGO bricks still produced. According to Hill’s estimation, the project, as you see it in the pictures throughout this article, is 80% finished. If you watch the video you can see and hear areas of the project which Hill still thinks require attention. One of the obvious non-LEGO issues, though, is that the sticker needs changing from Arc A770 to A750.

Hill and regular host Bob Duffy discussed whether the life-size graphics card could have LEDs or even RGB. A brief answer to this question is that there are some LEGO lighting kits available (previously released in a LEGO blacksmith and a lighthouse kit, for example), but in making the graphics card model a 1:1 size, there really wasn’t room to include them. This was also partly why Hill targeted the A750 for his LEGO design project – it doesn’t need LEDs to be realistic, just some silver accent LEGO bricks.

Non-LEGO fans will probably learn a lot about these creative building bricks from watching the video. Some of the things we picked up included a pair of new acronyms; AFOL (Adult Fans Of Lego), and SNOT (Studs Not On Top). The Arc A750 design uses quite a lot of bricks which are SNOTs.

On the topic of making these LEGO Intel Arc A750 graphics cards widely available after the project is completed, things seemed rather uncertain. It was highlighted that the LEGO Ideas Set program could get an official kit mass-produced with 10,000 votes behind it. However, Hill didn’t seem keen on this idea, perhaps because the project isn’t yet finished to his satisfaction. Also he seemed to direct those interested towards the future possibility of the BrickLink Studio design being made available and users sourcing / buying their own bricks using the catalog associated with the model.

If someone were to build the project in its present state, Hill reckons the 500 to 6000 bricks needed would cost about $120 or more. That's almost half the price of a real working Intel Arc A750 GPU.

Discord, a social platform that supports text messaging, voice calls, and video calls, is enabling support for hardware-accelerated AV1 video streaming for Nvidia's GeForce 40-series graphics boards, which are among the best graphics cards money can buy today. This is the first communication platform to support both hardware acceleration of AV1 encoding and decoding using the Ada Lovelace microarchitecture. 

"Discord is currently rolling out an update that enables AV1 streaming with GeForce RTX 40 Series GPUs," tweeted Gerardo Delgado, Nvidia's product line manager for content creators. "The update is rolling out starting this week and will slowly populate to all users. With AV1 you'll be able to stream up to 4K60 with nitro, at 8 Mbps!" 

The AV1 codec significantly reduces bandwidth requirements for high-resolution, high-framerate videos. By adding support for AV1 encoding even for just one family of GPUs, Discord will reduce the upload bandwidth needed for high-quality video streams for content creators and download bandwidth requirements for viewers with hardware that supports AV1 decoding. For those with a graphics processor that does not support AV1 decoding, the service will transcode video into something ubiquitous, such as H.264. 

Nowadays, several video streaming services like Netflix and YouTube stream videos in AV1 to hardware that supports AV1 decoding. Meanwhile, Discord is currently the only communication platform that supports AV1 decoding and encoding. This may not be surprising as Discord primarily aims at gamers, and this audience tends to own the latest and greatest hardware. 

Speaking of hardware, it is necessary to note that all the latest GPUs from AMD, Intel, and Nvidia support AV1 encoding and decoding. Therefore, it is likely that Discord and, eventually, other communication software will support hardware-accelerated AV1 encoding and decoding using other GPUs.

The best part about streaming is talking to your viewers. You can talk to them via game chat, Discord, or text chat, but text chat is the most common way to connect with viewers. This is where inside jokes are born, nicknames, and bonds. You can’t build those bonds if you’re not looking at the chat and responding to viewers.

Once your Twitch account is linked you’ll be able to view chat in OBS. There is a way to add chat into your overlay on OBS. This is what all the pros do, i.e. me! I love having a chat on my stream. It helps me when I go back and watch the VODs to see what the chat was like, how active it was, and if I missed anything. This can be done for Twitch, Facebook, and YouTube.

• Connecting with your viewers is easy via OBS. Popular streamers link their chats to their live shows.
• Linking Facebook, Discord and Twitch chat to OBS is a simple and effective means to interact with viewers.
• Text, emoji's and memes can be shared live via the chat.

Before tackling this project it would be prudent to take a look at our How To Record in OBS and How to Stream on Twitch guides. Those two guides will act as a foundation for building the best stream for your viewers.

Here is how to add chat to your stream using Facebook, Twitch (using StreamLabs and StreamElements and YouTube.

• Grab a great webcam from Amazon.

How To Add Facebook Chat to Your Stream

1. Open your Facebook Like Page.

2. Click on the Meta Business Suite on the left hand side of the page.

3. From the More menu, select Go Live.

4. Toggle the test broadcast button and then click Select to start a live broadcast.The test broadcast button enables you to check your configuration before going live.

5. From the Settings menu select Comments.

6. Pop out the comments module.

7. From the new Comments tab, copy the URL.

8. Open OBS.

9. For your chosen scene, go Sources and add a Browser source called Facebook Chat.

10. Replace the URL with the Chat URL.

11. Login to your Facebook account from the Browser window in OBS. Use the Interact button located above Sources and enter your login.

12. The chat window is now ready for use. Every time you start a live, refresh your URL comments Browser source. And you’re all set.

How To Add Twitch Chat To Your Stream

To add chat overlay to your Twitch stream you’re going to need to login into Streamlabs or StreamElements.

Using Streamlabs:

1. Go to the StreamLabs website.

2. Click on Widgets >> Chat Box.

3. Log in to your account.

4. Search for chatbox via the search bar and select Chat Box Widget Settings.

5. Copy the Widget URL. Remember, do not share this URL with anyone. It is linked to your account.

6. Open OBS and select the scene for use with the chat box. Here we are using the Browser scene for the chat box.

7. Under Sources, add a Browser source.

8. Rename the source to Twitch Chat Box and click OK.

9. Replace the URL with the Chat URL and click OK to save. The width and height of the chat box can also be tweaked to provide the best possible layout.

10. Click on Start Streaming to stream your show with an embedded chat window.

Using StreamElements

StreamElements Chat Box is a bit more involved because you need to select an overlay. Then you’ll have a separate customizable chat feature.

1. Open a browser to the StreamElements website and login.

2. Select Streaming Tools >> Overlays from the StreamElements page Menu.

3. Create a new overlay.

4. Set the overlay resolution to match that of your stream and click Start. In our case we chose 1080p.

5. Name the overlay “Chat Window” and click Save.

6. Click on Add Widget.

7. Go to Stream tools and select “Your Stream’s Chat”. This will place a black chat window on the screen. We can move the window around for the best layout.

8. Copy the URL link via the icon in the top right corner.

9. Open OBS and select the scene for use with the chat box. Here we are using the Browser scene for the chat box.

10. Under Sources, add a Browser source.

11. Rename the source to Twitch Chat Box and click OK.

12. Replace the URL with the Chat URL and click OK to save. The width and height of the chat box can also be tweaked to provide the best possible layout.

13. Click on Start Streaming to stream your show with an embedded chat window.

Using YouTube

YouTube’s chat integration with OBS is the easiest of them all.

1. Open a browser to YouTube.

2. Go to your YouTube channel, the icon is in the top right corner.

3. Select Go Live via the Camera icon’s dropdown menu.

4. Select Popout Chat in the three dots menu in the top right.

5. Copy the URL of the popup chat window.

6. Open OBS and select the scene for use with the chat box. Here we are using the Browser scene for the chat box.

7. Under Sources, add a Browser source.

8. Rename the source to YouTube Chat Box and click OK.

9. Replace the URL with the Chat URL and click OK to save. The width and height of the the chat box can also be tweaked to provide the best possible layout.

10. Click on Start Streaming to stream your show with an embedded chat window.

Do you have a pre-teen or early teen child? Are you struggling to think of what to get them for Christmas to complete or complement their PC setup? It’s that time of year again when we’re only a couple of weeks away from Santa leaving presents under the tree. If you’re still deciding what to get your loved ones, you don’t have much time left before there won’t be enough time to get your items delivered, so you’d better have a good think about what you can get your little monster for their gaming setup. 

My son is 13 going on 26 and plays a mix of games from Minecraft, and Roblox, to many a zombie game. He also likes to hang around with his friends on their own little Discord server while they play multiplayer games together and stream their gameplay to each other . Having watched a lot of YouTube videos, my son has been well indoctrinated into the RGB-everything persona when it comes to the decor in his room. Studio 54 never had such a light show, even during the disco days.

Headsets: Go for Quality, Wireless is a Plus

So, what are some affordable options for kids who play PC games? I’ve found that my son goes through peripherals fairly quickly, thanks to a mixture of abuse, and the fact that some devices just don’t meet the needs of an active and excited teenager.

Headsets are always the first to go. I started off by letting him have my hand-me-downs when he was younger, but as he matured, he started to take more of an interest in the products that he wanted to use. The cheaper, flimsy headsets would only last a few months, so I started upgrading him to sturdier name brands and quickly became a fan of the quality and price of headsets like the HyperX Cloud II, which you could often find discounted - especially on sales holidays such as Black Friday.

It’s definitely worth spending a few extra bucks to get a product that’s more durable and likely to last a while. One thing he’s wanted recently though is a wireless headset, especially for using with his VR headset, as there are already enough cable issues when using that. So this year I’ve opted for a HyperX Cloud Alpha wireless headset and hope it will meet his expectations.

Wireless gaming headsets are a particularly good idea for kids, because they can get excited and accidentally yank at the wires of corded ones. On the other hand, you need to make sure that the headsets stay charged and have decent battery life. It almost goes without saying that you want a headset that uses 2.4 GHz wireless technology, not just Bluetooth because Bluetooth quality isn’t as good. If you want to find a top-tier headset, check out our list of the best gaming headsets. There are also a couple of good headset deals right now:

• HyperX Cloud Alpha Wireless: now $165 at Amazon (was $199)
• HyperX Cloud II: now $49 at Amazon (was $99)
• Logitech G435 Lightspeed Wireless: now $49 at Amazon (was $79)

RGB Mouse Mats: A Really Cheap Gift

A simple and inexpensive present that I’ve gotten for my boy over the last few years is a new mouse mat. Sometimes it’s just a plain color, and other times it may be decorated with whatever currently trending interest he may have. A mouse pad is usually a little stocking filler as they don’t usually cost much, but of course, you can go crazy and spend a small country’s GDP on one with flashing lights and wireless charging if you’re so inclined.

As it so happens, Tom’s Hardware maintains a list of the best RGB mouse pads and we test them to make sure they really shine. There are a couple of solid sales on pads now too:

• SteelSeries QcK Gaming Surface XL: now $34 at Best Buy (was $59)
• Corsair MM700 RGB Mouse Pad: now $49 at Best Buy (was $59)
• HyperX Fury S: was $14 at HyperX (was $29)

Keyboards: Compact and Colorful

When it comes to keyboards and mice, it’s all down to preference, and the prices of these range from quite low to obscene. Think of factors like size, space, mechanical or membrane, RGB, and price. For my son, it has to have the RGB; for him it’s more important than the functionality of the product, but I’m going to make sure it’s capable of both and also a good value for money.

He has a small desk in his room and doesn’t ever use the number pad on his current keeb, so I’ve opted to get him a smaller TKL (Ten-Key-Less) board which still keeps the functionality of the F-keys but makes a little more room on the desk. I opted for a Logitech G Pro TKL which I managed to find on sale for around $50, this keyboard comes with mechanical switches and all the RGB he could wish for.

We maintain an up-to-date list of the best gaming keyboards. There are also a few compelling keyboards on sale right now:

• Logitech G PRO Mechanical Gaming Keyboard: now $89 at Amazon (was $129)
• Corsair K70 RGB MK.2 Low Profile: now $99 at Best Buy (was $149)
• ROCCAT Pyro Wired Mechanical Keyboard: now $49 at Best Buy (was $79)

Mouse: Save By Staying Wired

Also for a mouse, I found a Logitech G Pro Wired for $30 that conveniently matches the keyboard and is nicely priced. He’s not worried about having a wireless keyboard and mouse and going wired for these two peripherals usually saves you a lot of money, and does away with the pain of your mouse or keyboard running out of juice at the most inconvenient time.

We have a list of the best gaming mice that’s based on our extensive testing and research. However, there are also some great gaming mouse deals right now:

• HyperX Pulsefire Haste: now $24 at HyperX (was $49)
• Logitech G PRO Wireless: now $49 at Amazon (was $129)
• SteelSeries Rival 3: now $21 at Best Buy (was $29)

Monitor: A Dramatic Upgrade, But Not Cheap

The last thing I bought my son for Christmas is a new monitor, as he was using a very old second-hand monitor that I had passed down the line many years back, just like the PC he’s using - made from all my old bits and pieces. The first thing to take into consideration when thinking about getting a monitor is the use case, and then, think about what hardware you have powering it. A monitor is one of those pieces of kit that can last multiple PC builds, so worth making the right choice.

My son uses his computer for school work and gaming and has an older Nvidia 6GB GTX 1060 powering the graphics. Space on his desk is also important, so I wouldn’t want anything larger than a 27-inch screen. Being a little gamer, and enjoying the FPS games, something with a decent refresh rate of 144 Hz plus would be nice, but with the GPU he has, he’s not able to play games at a higher-than-1080p resolution. So I’m considering a monitor that’s limited to a 1080p (1920 x 1080) resolution.

Going for a 1080p monitor also gives you a lot of options and keeps the price low as high-speed, 1080p displays are available for less than $200. If you’re willing to spend closer to $250 or $300, you can get a 2K (2560 x 1440) screen that will make text and images sharper. Either way, if you upgrade the graphics card in the future, then you can think about another monitor with a higher resolution and perhaps convert the one you buy today into a second monitor.

We have a very detailed list of the best gaming monitors you can buy, based on our testing. However, these screens are on good sales at the moment:

• Dell S2721HGF (27-inch, curved 144 Hz, 1080p): now $179 at Dell (was $259)
• Dell G2722HS (27-inch, 165 Hz, 1080p): now $149 at Dell (was $279)
• Alienware AW2521HFL (25-inch, 240 Hz 1080p): now $199 at Best Buy (was $249)

Bottom Line

The main takeaway from buying any bits of computer, peripherals, or monitors is to think about the purpose you want them to fill, the amount you’re willing to spend, and if it’s for the holidays, whether your child actually wants them.

Be sensible and shop around for good deals and prices and don’t make any rush purchases. Hopefully, you’ve already got your shopping done, but if not, check out our lists of the best deals on PC hardware and tech right now and the best monitor deals.

We’ve seen makers integrate hundreds of applications with various Raspberry Pi projects but not too many use the instant messaging platform Discord. So today, we’re sharing a project from Reddit user Android2771, who has created an easy way to look at the latest post of a given Discord server using a simple Pi-powered e-ink display setup.

E-ink displays are limited as far as refresh rates go but excel at providing low-power display options for projects using text and images. In addition, some of them provide multiple color options, but in this case, Android 2771 uses a small, 2.7-inch e-ink display with just black and white support. These screens are standard in e-readers but work well for homemade Pi projects like this digital calendar and can quickly adapt to a wider variety of applications using tools like PaperPi.

This project relies on a Raspberry Pi as the central controller. First, it’s responsible for checking Discord for the most recent text posts and images. Then, it processes the data into a format that fits the tiny e-ink display parameters. The result is an e-ink display that regularly updates with posts from any Discord server you like.

It doesn’t take too much hardware to recreate this project as it’s mainly just a Raspberry Pi with an e-ink screen using a network connection to retrieve the chat log data. In this project, Android2771 is using a Raspberry Pi 4, but there’s no reason an older model like a Raspberry Pi 3B+ wouldn’t work in its place. The e-ink screen used in this project is a 2.7-inch model from Waveshare, but you could use any e-ink screen, including larger ones with more color options.

In the original post, Android2771 explains that he used a custom Discord bot to interface with a local API in charge of the e-ink display output. Python Pillow was responsible for handling the image processing, turning them into black and white images with a small enough resolution to appear fully on the small e-ink screen.

Unfortunately, the code used in this project is not open source, but Android2771 left a trail for makers to follow who are interested in creating a similar system of their own. If you want to make this Raspberry Pi project yourself, check out the original thread shared to Reddit for sources, links, and plenty of details to get you started in the right direction.

Graphics card-snatching bots continue to plague nearly all e-tailers as they become more sophisticated, and one frustrated shopper has managed to track down a bot that seems to have found a way around AMD.com's protections, thus preventing them from purchasing a graphics card. 

As we've seen elsewhere, online shoppers have been fighting a losing battle against scalper bots in Europe. Redditor DifficultEstimate7 shared their story of how they and some friends have been trying to secure an AMD GPU from the weekly drop for over four months. Despite awaiting graphics card drops poised with mouse, keyboard, and credit card details at hand, they have been defeated by bots such as the Vuurvlieg AMD Companion script, time and again.

DifficultEstimate7 and friends are understandably frustrated by the AMD store's stock vanishing in front of their eyes. Unfortunately, even if they have managed to get through to the checkout screen, they have found that stock has evaporated before they can make a payment. AMD has implemented anti-bot protection for its store, but it is obviously being bypassed.

The Redditor believes he has found the source of the issue. The picture you see above was discovered by DifficultEstimate7 in a Discord channel where the AMD EU stock situation was being discussed. It purports to show that a script dubbed the 'Vuurvlieg AMD Companion' managed to secure 214 graphics cards.

AMD dropped a total of 350 graphics cards at the last drop, according to the source. That means this bot snatched up about 64% of the GPUs released, and there's no telling what other scripts or bots were fired off at the same time.

You might ask yourself why the redditor didn't purpose a script/bot to do the necessary work for them? Yes, they had already thought of this, but the apparently very successful 'Vuurvlieg AMD Companion' is donationware and isn't accepting new users. Perhaps the script's author got a very nice donation, and it would be too expensive for ordinary customers to be worthwhile. In some ways, this raises parallels to the Best Buy $199 'GPU paywall,' as it really isn't worth the investment for a single ordinary customer.

Being in the market for a GPU isn't a situation that many PC gamers, enthusiasts, or DIYers will have relished for many months. Rather than enjoying the pleasure of choice, the current situation has folks looking at what (if anything) they can get, what multiple of MSRP will they have to pay, or what will tide them over until GPU availability and pricing becomes more reasonable. Scalper bots continue to exacerbate the issue as they become more sophisticated, so web stores like AMD.com will have to respond in kind if we expect to see some normalcy anytime soon.  

The unofficial Minecraft-inspired NFT game "Blockverse" appears to have become one of the latest scams in the crypto world. At launch, and in a mere eight-hour period, Blockverse raised more than $1 Million from a community of users through the sale of 10,000 NFTs (Non-Fungible Tokens) priced at 0.05 ETH (~$127) each. Little did the users know that they'll lose their funds soon after in a rug pull - a sadly classic cryptoverse move.

Rug pulls (the term coined for when blockchain-based projects disappear with users' funds) are a constant of the daily blockchain life - in 2021, around 3,300 projects pulled rugs on their users, stealing approximately $7 billion worth of investor's money. Rug pulls work in much the same way that snake-oil sellers did back in the day: promising a product but not delivering it; cue the charlatan running away with his clichéd dollar-marked bag of money. Blockverse seemingly carried this out a mere two days later. The project deleted its digital footprint - website, Discord server, and game server all disappeared, and so did the only connection from users to their investment - their access-granting NFTs rendered worthless.

The possession of one of these 10,000 NFTs would give the limited player pool of users access to the game upon its launch. Being a digital token, these NFTs are also inherently tradeable, meaning that single accounts could hold more than one of these "access slips." Blockverse pulled off a significant pre-order event if it were a legitimate project, selling the equivalent of 10,000 licenses. Users paid $127 for a digital pre-order game. But a digital trail may have forced Blockverse to come back - partially.

The game's community members managed to track down a Coinbase address linked with the Blockverse founders. This address had funded most of the efforts with establishing the Blockverse; alongside it was a Cloudflare IP. So it would certainly be enough for law enforcement (or someone with less stellar intentions) to arrive at the founders' identities.

Three days after closing all of Blockverse's digital presence, and with the paper trail leads already disseminated through the community, its founders took to Twitter to explain the events from the following previous days. First, they claimed the project's legitimacy and explained that the reason for the flight was nothing more than an attempt at protecting their safety. But, according to them, several overwhelming and concerning complaints regarding some of the Blockverse's elements - gas fee prices for the NFT acquisition, lack of player capacity at 10,000 users, and lack of utility in the $DIAMOND tokens that only served as an access license.

"The FUD [fear, uncertainty, and doubt] quickly descended into harassment, threats, and doxxing," the creators wrote on Twitter. "The team noticed all this and panicked, deleting the discord server on impulse. Everything else was closed to prevent the continuation of harassment that had occurred so far. Even then, the plan was to reopen once everyone had time to calm down."

"There is absolutely no reason for us to leave permanently, as all the technical work and infrastructure for the project thus far has already been put in place. It was more work to take things down than to leave it up, but again, everything was done in the interest of protecting personal safety."

Despite this relative show of face (the project website is still down, for example), investors appear to be on the edge regarding the legitimacy and intentions of the project's founders. However, many have lost confidence in the project's founders. As a result, a group of community members is now negotiating that the original Blockverse developers cede control of the project. Interestingly, Blockverse's founders have sat at this bargaining table and agreed to hand over all relevant contracts and code while keeping the 500 ETH raised by the initial token sale - essentially paying themselves around $1 million for the initial development work. Blockverse has also apparently made a similar proposal to ArkDev, the founder of blockchain gaming platform NFT Worlds, although there's no public information of where that offer stands.

Even more surprisingly, the troubled project community members seem to be willing to cut their losses by settling with the founders. They're eager to let them keep 100 ETH (worth approximately $250,000 at current market prices) for the project's data while using the remaining funds to hire a development team. 

"What we really want are the assets and 200-400ETH to hire people to run the project," they said. "Most rug pulls recover little of anything, so the community is realistic in terms of a settlement. We are in agreement to letting them keep some. We just want enough to run the project."

A new type of malware, dubbed ‘Panda Stealer’ by researchers, is spreading through spam emails and malicious Discord links, and has its sights set firmly on your ever valuable cryptocurrency. According to Trend Micro, the phishing emails appear as business quote requests, with an XLSM file attached that’s loaded with malign macros. 

Panda Stealer appears as an innocent XLSM file with macros that once enabled download a "loader" which executes the main "stealer" application. Alternatively, an XLS file may be downloaded, containing a formula that hides a Powershell command that accesses paste.ee, a Pastebin alternative, to download a further PowerShell command. Once running, Panda Stealer tries to detect keys, addresses, and other data associated with cryptocurrency transactions and wallets holding funds including Dash, Bytecoin, Litecoin, and Ethereum. Right now we are unsure if the latest cryptocurrency, Chia is affected. It will also attempt to steal credentials from other applications such as NordVPN, Telegram, Discord, and Steam. It’s capable of taking screenshots of the infected computer, and sucking data from browsers like cookies, passwords, and cards.

Panda Stealer seems to be a variant of Collector Stealer, a cracked build of which is freely available online. While there’s no evidence yet of a particular criminal group behind Panda Stealer, Trend Micro was able to identify an IP address being used by the malware for command and control. It led to a rented Shock Hosting virtual server, and having been reported, the server has been suspended. 

This may not be enough to quell the threat, however, as VirusTotal found 264 similar files in its database, calling home to 140 C&C servers and from more than 10 download sites, some of them from Discord, which may be being used to share the malware between criminals.

PlayStation and Discord are working together to "bring the Discord and PlayStation experiences closer together on console and mobile starting early next year," SIE president and CEO Jim Ryan wrote in a blog post. For the first time, Discord will have some sort of functionality on a console; the popular chat and social app was popularized by PC gaming before spreading widely.

It is unclear if there will be a native Discord app on the PlayStation 5 or PlayStation 4, or if it will be plugged in to the PlayStation user interface somehow, nor is there information on how mobile messaging may work. Fans have long called for Discord on consoles to unify their gaming communications.

Ryan's post suggests more will be made clear in the months to come, so perhaps we'll learn how it works soon.

This comes as Sony Interactive Entertainment has made a minority investment in Discord's Series H round of funding. This deal was announced mere weeks after Discord stopped talking with Microsoft, which was in talks to buy the company for as much as $10 billion dollars. It appears that Discord is continuing going public through an IPO.

Nothing in the announcement suggested that this partnership is exclusive, and that Discord could not build functionality within the Nintendo or Xbox ecosystems in the future. Whether that happens, however, is anyone's guess.

Things are getting a bit recursive around here. Epic Games today announced the addition of several PC apps, including the Itch.io game platform, to the Epic Games Store.

Itch.io is mostly known for indie titles, many of which are free or cost very little, but it also hosts “game jams” that encourage developers to create something based on the parameters set by the event’s organizers.

Bringing the platform to the Epic Games Store essentially means that a marketplace primarily focused on games will feature another marketplace primarily focused on games. Hopefully someone goes further down the rabbit hole at some point.

Itch.io isn’t the only program heading to the Epic Games Store. The marketplace has also expanded to include the iHeartRadio client, the Krita painting app, the Brave browser and the KenShape model generator as part of this push outside gaming.

Epic Games said that social platforms Houseparty (which it owns) and Discord will also be added to the Epic Games Store "in the future." These additions don’t appear to be included with the store’s public roadmap, but that might change.

The company also indicated that games and apps won’t necessarily be confined to their own silos by announcing that "Fortnite Crew Members will receive three free months of Spotify service" if they don’t yet have a Premium membership. Spotify is another Windows app that is already available on Epic.

For anyone who isn’t in the know: Fortnite Crew is a subscription that offers access to each season’s battle pass, additional V-Bucks and other benefits in exchange for $11.99 per month. Now those benefits appear to be expanding beyond Fortnite.

All of these additions make it clear that Epic Games didn’t just want to compete with the likes of Valve and GOG when it released the Epic Games Launcher. It’s actually creating its own app store that just happened to launch with a focus on games.

That might lend some additional context to the company’s dispute with Apple over app distribution on iOS and iPadOS devices.

For now, however, it seems that Epic Games isn’t content with offering an alternative game marketplace for Windows 10 users. It has a long way to go before apps become a core part of the Epic Games Store, but this is a sttart. 

A recent Microsoft update for Windows 10 appears to be wreaking havoc on some computers since its release several days ago. Update KB5001330 for builds 2004 and 20H1 and update KB5001337 for builds 1909 and 1903 are causing serious issues, from annoying frame rate bugs all the way to BSODs, as spotted by ComputerBase.

The issues don't end there, apparently. For some people, the gaming-related issues vary wildly: some users experience unstable FPS with v-sync enabled and when sharing a screen on discord. For others, the issues are completely game-dependent.

Additionally, there have been numerous reports of other issues that aren't related to gaming, like DNS issues and boot-looping. For some, the update won't even install and will stay stuck at a certain percentage while installing.

Probably one of the worst issues related to this update is a nasty old bug that can potentially delete user files. Windows Latest reported a few days ago that the 'temporary user bug' is back again, causing your user login to disappear, along with your files located within that user profile (like the Documents folder). Luckily this is a very rare issue.

How To Delete the Windows 10 Update

If you are one of the unlucky people to have problems pertaining to this update, you can uninstall it manually with little to no effort.

All you need to do is head into the settings app, go to 'Update & Security,' click on the 'View Update History' section, and in that menu, there's an option to uninstall updates.

Once there, uninstall KB5001330 for 2004 and 20H1 builds and KB5001337 if you're running a build older than 2004.

Hopefully, Microsoft will get to the bottom of this problem by either withdrawing the update completely from its servers and/or rushing out another update to fix all these issues.

Generally, it takes Microsoft around 2 weeks to a month to permanently fix bugs like this, so we'd recommend pausing Windows Updates for that amount of time.

Streaming online doesn't have to cost a fortune with DIY systems like StreamPi. Using a Raspberry Pi, you can create a custom stream deck experience totally catered to your streaming needs.

It's a cross-platform, open-source application designed to integrate with common platforms used by streamers, and it's triggered with touchscreen buttons. The StreamPi website lists Samuel Quinones and Debayan Sutradhar as its primary developers.

The app is currently designed to work with both Linux and Windows machines. Users can program custom functions that work with apps like OBS, Spotify, and recently new additions like Twitter, Command Line, and more. There are also plans in the works to expand support with additional programs like Discord. The Raspberry Pi runs StreamPi and accepts input from the user, which triggers events using any of the supported applications.

It also features themes for the interface that can be changed and customized with a little programming. You can also see the device in action — the team behind the project posted a demo of a Raspberry Pi 4-powered system on Twitter.  

To explore this project in-depth, visit the official GitHub for StreamPi. Check out our list of best Raspberry Pi projects for more useful creations from the maker community.

Riot Games today announced the launch of Valorant its tactical first-person shooter, which was simply called Project A when it was teased in October 2019. The company also said that it plans to release the game some time this summer 2020.

Valorant was clearly inspired by tactical shooters like Counter-Strike: Global Offensive and Tom Clancy's Rainbow Six Siege. Riot said the game relies on precise gunplay, like CS:GO, but it also features unique character-based abilities like those found in Siege and Overwatch.

• Popular Right Now: Raspberry Pi Turns 8 Years Old
• Best Gaming Mice of 2020
• Best High Refresh Gaming Monitors

Riot said in a gameplay video that Valorant matches are round-based contests between attackers looking to plant a bomb and defenders hoping to stop them. Players have to purchase weapons and abilities at the start of each round.

The gameplay video makes it clear that Valorant will also feature voice chat--unlike the text-chat-only League of Legends–and reports indicate that it will feature a ping system like the one found in League (and the Apex Legends battle royale) as well.

Riot also shared details about how Valorant will run:

"Here’s what we think it takes for you to trust a game enough to invest: 128-tick servers, at least 30 frames per second on most min-spec computers (even dating back a decade), 60 to 144 FPS on modern gaming rigs, a global spread of datacenters aimed at <35ms for players in major cities around the world, a netcode we’ve been obsessing over for years, and a commitment to anti-cheat from day one."

More details about Valorant should be revealed soon. Riot has set up accounts for the game on Twitter, Facebook, Instagram and YouTube; it's also established a Valorant subreddit and Discord server. Links can be found on the site.

In December, AMD released its Adrenalin 2020 driver. Though this major driver release touted stability among other things, there was a notable amount of complaint across forums. The new 20.1.1 drivers list a total of 27 fixed issues including per game fixes and other updates. 

There are several known issues that still exist, including an issue with the popular RX 5700 graphics card that sees users intermittently experience a black screen while gaming or on the desktop (the workaround is to disable hardware acceleration in apps running in the background, like Web browsers or Discord). 

The full list of fixes and known issues, as per AMD, are below.

Fixed Issues:

• The audible beeps at game startup from Radeon Chill, Radeon Boost and Radeon Anti-Lag have been removed. These features now offer audible indicators only when activated or deactivated via hotkey.
• The Radeon ReLive on screen timer indicator during recordings has been disabled by default but can be enabled in Radeon Software settings.
• Controls for vertical sync may be hidden or disappear when Radeon Enhanced Sync is enabled.
• Radeon ReLive may experience freezing or pausing issues during recordings when a high resolution camera is connected and in use.
• CPU usage may sometimes remain high once Radeon Game Advisor has been invoked during a game.
• Some users may experience an error message “Another instance is running” during download of a software update through the Radeon Software home screen.
• The DuplicateDesktop process may sometimes cause high CPU usage while a game is running.
• Radeon Software may close or may experience a crash upon resuming from sleep.
• The toast message detailing the hotkey to open Radeon Software’s Overlay may still show up in some games after Radeon Software Overlay has been disabled.
• Lost Ark may experience stuttering intermittently during gameplay.
• Using a custom stream key with Radeon ReLive may fail to stream your content.
• The ‘Stream’ button may remain active when in the process of choosing a region to stream even when a region has not been selected or chosen.
• Some Radeon R9 200, Radeon R9 300 and Radeon R9 Fury series graphics products may experience instability with a limited number of DirectX®9 or DirectX®11 games when using a high refresh rate 120hz+ display. A workaround if you are experiencing this issue is to lower your displays refresh rate.
• Some mjpeg clips may experience a green tint on Radeon RX 5700 series graphics products when using Windows® Media Player or the Movies & TV application.
• MechWarrior 5: Mercenaries may experience a game crash and DXGI dialogue error when running the game using HDMI and Radeon FreeSync display configurations.
• Live streaming using the DouYu application with hardware acceleration enabled may cause video corruption on Radeon RX 5700 series graphics products.
• Trials Rising™ may experience excessive fog/smoke in some areas of the game.
• Missing text or corruption may be experienced in the right eye when playing the VR game Boneworks.
• Fixed result overflows that can be experienced with Radeon RX 5700 series when using SETI@Home.
• The ‘Shop AMD Products’ button may open the AMD.com homepage instead of the proper shopping web link.
• The scrolling arrow options may intermittently fail to work during Radeon Software installation.
• Up and Down arrow keys don’t work when using the search bar in Radeon Software.
• Enabling HDR enabled displays in Windows® may cause colors to become washed out.
• Radeon Software sidebar appears behind the Windows® taskbar when the taskbar is set to the same side of your display.
• Playing Tom Clancy's: The Division 2 with HDR enabled and performing a task switch may cause display color corruption that persists even once the game is exited.
• Mixed Reality Portal may experience color corruption or distortion near the edge of viewing areas on some headsets.
• Resident Evil 2 may experience screen flashing when launching the game using DirectX12 API.

Known Issues

• he Radeon Software Overlay hotkey notification may sometimes be displayed during video playback in web browsers or launching some video player applications.
• Integer Scaling option is not showing up or available on some Windows®7 system configurations.
• Factory Reset install may keep previously configured Radeon Software game profiles. This can cause mismatch between global graphics settings and per profile settings.
• Text overflow in some UI boxes or toast messages may be experienced in some language localizations.
• Radeon Software may open with an inconsistent size or may not keep its previously set size when opened.
• Some Vulkan gaming applications may crash when performing a task switch with Radeon Image Sharpening enabled.
• Integer Scaling may cause some video content to show flicker when the display resolution is set to less than native resolution.
• Some Radeon RX 5700-series graphics users may intermittently experience a black screen while gaming or on desktop. A potential temporary workaround is disabling hardware acceleration in applications running in the background such as web browsers or Discord.

If the fixes benefit your use case (or you simply want to update to the latest versions), the 20.1.1 drivers are available for download now on the AMD website. Per usual they come in two flavors, one for Windows 10 64-bit and one for Windows 7 64-bit.

Valve is facing more competition than ever in the PC gaming market as companies like Twitch, Discord, and Epic Games introduce their own stores to compete with Steam. PCGamer reported earlier this week that Valve has turned to new features like Steam Link Anywhere and the largely open source Steam Networking Sockets APIs to help make sure its platform will remain popular with gamers and developers alike.

Steam Link Anywhere seems to do what it says on the tin. In the Steam Link beta build released on March 14,  the only change was listed as "Added support for Steam Link Anywhere, now in early beta." Valve explained that "Steam Link Anywhere allows you to stream games to your Steam Link from any computer running Steam, as long as your computer has good upload speed and your Steam Link device has a good network connection."

This is just the latest iteration of Steam Link. Valve introduced the technology with dedicated hardware in November 2015, and following numerous price cuts; itdiscontinued that hardware in November 2018. In the meantime, the company released a Steam Link app for Android (and eventually iOS) that let people access their Steam libraries on their phones. The app's also available for Samsung smart TVs and Raspberry Pi 3 devices.

The current version of Steam Link is limited to home networks. Steam Link Anywhere appears to remove that limitation, provided the host computer's upload speeds and the target device's download speeds can support game streaming. That's a significant improvement to Steam Link that could make it even more popular with people who want to be able to access their Steam libraries no matter where they happen to be.

PCGamer also spotted new Steam Networking Sockets APIs. These APIs would allow developers to use Valve's networking technology, which is used in massively popular titles like Counter-Strike: Global Offensive and Dota 2, in their own games. Valve has reportedly said that "a large portion" of the API is now open source--which lets developers poke around to see exactly how it works--and can offer "faster and more secure connections."

This is the kind of development that many players won't think about, because networking really only comes to most people's attention when there's a problem with it, but could make things significantly easier for developers. Valve is set to explain the advantages of using its Steam Networking Sockets APIs over other multiplayer technologies at GDC 2019 on March 21.

Valve announced that it's retiring the Video section of Steam as part of a "refocus" on gaming-specific content in the marketplace.

Steam's Video section previously allowed people to sell documentaries, movies, and other video content that didn't have to be explicitly related to gaming. Yet it seems that Steam users were mostly interested in the platform as a game marketplace--which makes sense given that it's primarily a place where people can buy PC games--rather than a central hub of digital content.

"For the past few years, we have worked on expanding Steam beyond games and software by building a video platform that supports paid and free video content," Valve said. "In reviewing what Steam users actually watch, it became clear we should focus our effort on offering content that is either directly related to gaming or, is accessory content for games or software sold on Steam."

The company said that gaming-related videos will now be discovered "via the associated game or software store page, or through search, user tags, recommendations, etc." Other videos will be "retired" over the coming weeks. Steam users will still be able to view the videos they've already purchased, which is a relief, but it's clear that Valve wants to exit the non-gaming-video business.

This decision arrives during a time of transition for Steam. Over the last few years Valve has rethought many aspects of the store, from how it displays reviews to what screenshots developers can use to promote their latest titles, and overhauled its Steam Chat service. All that's in addition to the company's work on virtual reality, cross-play with other platforms, and other tech.

Steam also faces more competition than ever now that Discord's in the game-selling business, the Epic Games Store has debuted, and services like Twitch have promoted their own storefronts. Other companies, from Microsoft to Electronic Arts, have been planning Netflix-like subscription services that would obviate the need to buy games one at a time from platforms like Steam.

All that competition is bound to lead to changes in Steam. Removing non-gaming content, which apparently wasn't all that popular anyway, is an easy way for Valve to focus on other areas.

Discord announced earlier this year that it was entering the game-selling business. Part of its plan to compete with the likes of Steam and GOG was to help indie developers launch their games in exchange for timed exclusivity via the First on Discord program. Now the company has revealed the first games debuting with that program, and it turns out the "first on" part might not mean exactly what people think it does.

First on Discord's inaugural batch of games features seven titles: Minion Masters, Mad Machines, Bad North, At Sundown, Sinner: Sacrifice for Redemption, King of the Hat, and Last Year: The Nightmare. Aside from debuting on Discord, the games don't seem to have much in common. They range from a real-time card game and competitive shooter to a "brutal ball sport game" and "real-time tactics roguelite."

Yet several of these games can already be played. Bad North has already debuted on Xbox One, PlayStation 4, and Nintendo Switch. Minion Masters has an Early Access edition available on Steam. Several other titles have Steam pages and will be released on consoles later this year. At least for the launch, First on Discord appears to come with a few caveats and developers plan to take full advantage of the timed exclusivity.

Being limited to the Discord Store is the one aspect of the program that seems to be set in stone. Games are supposed to be limited to the company's storefront--at least on PC--for their full release. But even that limitation is vague--Discord said in its original announcement that games would be exclusive to its platform for "usually 90 days." That implies that devs will have wiggle room, but no other details are available.

The company also didn't say exactly when the OG pack of First on Discord games will debut. The Discord Store is currently available in a limited beta, and in its most recent announcement, Discord merely said these games will arrive "later this fall." It's not clear if they'll be released at the same time, if the Discord Store will be generally available before their launch, or when more information about First on Discord will be revealed.

Discord's announcement did highlight the synergy its primary service has with the new storefront. Each of the games listed has an official Discord server where people can discuss the game, communicate with the developers, and chit-chat with people who have a similar taste in games. Eventually many people who buy these games via the Discord Store are likely to play them with friends while using Discord proper.

It seems Discord isn't content to let Valve imitate many of its features in the new Steam Chat, because the gamer-focused communications platform is fighting back with the new Discord Store, which augments the Nitro subscription offering and establishes yet another storefront for video games. The new offering is still in beta--it's rolling out to 50,000 of the service's Canadian users today--but its ambitions are already clear.

This new storefront will take various forms. Discord emphasized the addition of "golden games" people might have missed to Nitro, the $5 monthly subscription that offers various perks in exchange for supporting the otherwise free platform's development, in a blog post about the new store. Companies are starting to bring the subscription model of movies, books and music to games; Discord is capitalizing on that shift.

Not that signing up for Nitro will be the only way to get new games via Discord. That's where the Discord Store comes in. The company said it wants to launch "a curated game store experience similar to one of those cozy neighborhood book shops with recommendations about the hottest and newest games from us to you." That emphasis on curation would help the Discord Store differentiate itself from larger platforms like Steam.

Developers will get 70 percent of the revenues drawn from sales in the Discord Store, which is the same amount they'd receive from Valve or Apple if they sold their game via Steam or the App Store. Some devs will also receive direct support from Discord thanks to the new "First On Discord" initiative, which will help publish indie games in exchange for timed exclusivity, and the company said most will stay on its platform for 90 days.

All of that could become hard to manage--especially when you consider that many games are launched from something like Steam, Origin, Battle.net, or similar apps. Discord is betting that won't be a problem, though, because it's also introducing a Game Library that can automatically scan your system for games and open them and their corresponding launcher right from the existing Discord app, no hassle required.

Discord said in its announcement that it now has more than 150 million users. That's a lot of people who rely on the service to voice chat with friends playing the same game, socialize with people who have similar interests, and keep up with the gaming industry's latest gossip.

It makes sense, then, for the company to sell games that interest those people. Offering access to select titles could increase the number of people who sign up for Nitro, and selling individual games could also let Discord monetize the many people who can't or won't pay a monthly fee. Twitch did essentially the same thing in 2017 when it launched Twitch Games Commerce to make it easy to buy games you see on stream.

Discord is making a lot of changes with the introduction of the Discord Store. But the way it's approached them, from looking to help people experience games they missed and offering a curated selection of games to making it easy to launch anything on your system, highlights exactly why so many people like it. Discord knows what gamers want, and so far, it's managed to give it to them in a way other companies haven't.

Sometimes it feels like you have to open a half-dozen applications before you start playing a game. You've gotta open Steam, launch a game, make sure your streaming app is working (everyone streams these days), and then start chatting with friends in something like TeamSpeak or Discord. Valve hopes to have removed that last step, at least, by updating Steam Chat with so many features it might as well be a new service.

These updates to Steam Chat were released in beta form back in June. Now they're available to all Steam users, whether it's via the Steam app or the web, and Valve said it's also begun work on a Steam Chat app for iPhone and Android devices. It almost seems like the company is quietly hoping to become the central hub of PC gamers' lives, rather than merely being a place where they buy a bunch of games they might never play.

That means Valve is going to have to convince people to use Steam Chat instead of TeamSpeak, Discord, and similar platforms. (With Discord likely being the frontrunner when it comes to constant interaction rather than just in-game communication.) To that end, the new Steam Chat includes new group chats that offer in-line media, the ability to invite people with a link, and the option to make multiple text and voice channels.

Let's get this out of the way: As we said in our coverage of the new Steam Chat beta, yes, these new group chats make it seem like someone at Valve opened Discord and just changed a few things around to fit in with Steam's existing design, and that's probably going to be many Discord users' immediate reactions to these new group chats.

But the updates to Steam Chats don't stop with group chats. There's also a new friends list that lets you see your buds organized by what game they're playing, whether or not they're playing together, and if they're a member of any of your group chats. A new "rich presence" tool lets devs show exactly what your friends are doing in their game, too, and improvements have been made to call quality and the platform's security.

Will this be enough to convince people to use Steam Chat instead of other platforms? Maybe. But it's hard to move between platforms, and many people have already filled Discord servers with messages, channels, and memes they would have to give up by starting over with Steam Chat. People also like to play games that aren't on Steam--surprise!--and they are probably used to using something like Discord for those games.

Valve released an open beta version of its new Steam Chat service that boasts a more feature-rich friends list, improved group messaging tools, and other updates that could help it compete with more popular chat apps. The company said that many of these changes would also let it make more updates to web-based Steam components in the future, which should help improve the core experience offered by the nigh-ubiquitous game marketplace.

The new Steam Chat's first improvement comes in the form of an upgraded friends list. Your friends are now grouped together based on what game they're playing, whether or not they're currently in a game with each other, and their presence in one of your group chats. For those of you who still remember Myspace, the new friends list will also let you designate some friends as "favorites," so get ready to compete for that coveted status. To top it off, this updated friends list will also share information about what's happening in their games thanks to a new "rich presence" option for developers.

Next comes the new group chats. Let's get this out of the way right off the bat: these new group chats are very similar to Discord's servers. You have the in-line media, the ability to create multiple text or voice channels in each group, and the ability to invite friends to a group with a link... it's almost like someone at Valve opened Discord and decided to make the UI look more like Steams. That isn't a bad thing--Discord has become popular with many gamers because it's one of the best communications platforms out there--but it's hard not to think Valve took a little too much inspiration here.

Finally, Valve said it also made some "improvements" to Steam's voice chat. The company didn't elaborate on the general improvements, but it did say that you can quickly see if your friends are in a voice chat right from your friends list, and that you should notice "clear, crisp voice quality before, during, and after your games." (Which, again, sounds a lot like what Discord's offered to its users for the last few years.) Perhaps even more importantly, here's what Valve said about some of the privacy and security updates it's made to Steam's voice chat with this new release:

"Steam voice chat was rewritten from the ground up with a new WebRTC-based backend. As a result, voice chat uses high-quality Opus encoding, voice traffic is encrypted, and all traffic is sent through Steam servers rather than directly to peers. This keeps your IP address private, which masks your physical location and also prevents network attacks."

The new Steam Chat is available now via the Steam client and web browsers. Valve is collecting feedback via this section of its forums. The company said that the "new UI framework" and "important architectural improvements under the hood" will allow it to improve aspects of the core Steam experience once the new Steam Chat makes its debut. We don't know what the company plans to update, but if the overhauls are as extensive as those made to Steam Chat, the platform-slash-marketplace could soon be pretty different from what we've grown accustomed to.

Discord continues to attract gamers with its voice, video, and text features. Different channels or “servers” are full of people who play the same titles and use the app to discuss new content, ask questions, or look for other players to join their in-game party. Now the app has a new feature, Rich Presence, that allows you to easily join or spectate those in the middle of a gaming session.

Each server displays the list of people in it, but the roster also shows the game each person is currently playing. With the new feature, you can click on a person’s profile  to see more details about their session. For instance, you can see their location in the game, such as a specific level or lobby. Some titles will even give people the option to join in co-op gameplay or spectate the player’s actions. There’s also an ability that will allow you to send party invites via Discord chat. Prospective players can see the game you want to play as well as the number of open slots available.

At launch, Rich Presence is already supported by a few games, including major titles such as Call of Duty: WWII, Divinity: Original Sin 2, and Payday 2. You can try out the new Discord features with these games today. Take a look at the list below to see the full roster of titles that currently support Rich Presence.

Call of Duty: WWII (Sledgehammer Games)Divinity: Original Sin 2 (Larian Studios)Payday 2 (Overkill)Battlerite (Stunlock Studios)Duelyst by (Counterplay Games)SpeedRunners (DoubleDutch Games)Tooth & Tail (Pocketwatch Games)Foxhole (Clapfoot Inc.)Unturned (Smartly Dressed Games)Holodrive (Bitcake Studio)Brawlhalla (Blue Mammoth Games)Squad (Offworld Industries)GRIP (Caged Element)We Need to Go Deeper (Deli Interactive)Move or Die (Those Awesome Guys)Hellion (Zero Gravity)Osu! (PPY)Ballistic Overkill (Aquiris Game Studio)

Game developers who want to add Rich Presence integration in their games can visit the app’s developers page, where they can download the SDK and learn about the different commands available for the program.

Rich Presence is the latest addition to an app that continues to grow in popularity. Discord initially started out as a text and voice chat service. Last month, however, the developers enhanced the app when they added support for video chat and screensharing. Users can host a video session, which holds up to 10 people, and also use the picture-in-picture feature so they can browse other servers and messages while on the call. For more information about Rich Presence, you can visit the app's website or watch a short video that showcases its functionality.