GTX 1630 on Par With GTX 1050 Ti, Benchmarks Show
GTX 1630 and 1050 Ti share identical performance metrics in gaming.
According to a Weibo post by Inno3D, Nvidia's new GeForce GTX 1630 is an incredibly slow gaming GPU, with new benchmarks showcasing performance parity with the entry-level GTX 1050 Ti from six years ago. If Inno3D's benchmarks are applicable to most titles, this makes the GTX 1630 one of the slowest GPUs on the market right off the bat, and perhaps little more than a refreshed GTX 1050 Ti with newer video encoders and decoders thanks to the newer Turing architecture.
The GTX 1050 Ti came out in 2016 as Nvidia's entry-level gaming GPU powered by the Pascal architecture. The 1050 Ti features 768 CUDA cores running on a 128-bit memory bus with 4GB of GDDR5 memory. Power consumption peaked at 75W with clock speeds starting at 1290MHz for the base clock, and boost frequencies hitting 1392MHz for the reference model.
But even in 2016, we were not completely satisfied with its performance. In our original review, when compared to the vanilla GTX 1050, the 1050 Ti showed lackluster performance for the $130 price point it originated at.
The GTX 1630 is Nvidia's latest entry-level GeForce GPU, marking the first time Nvidia has ever used an xx30 series badge for a GTX GPU. As expected, this GPU features less memory bandwidth and fewer cores than the GTX 1650, with a grand total of 512 CUDA cores running on a tiny 64-bit bus with 4GB of GDDR6 (12Gbps) memory. Thankfully clock speeds have been increased over the 1650 to compensate for the reduced specifications, featuring a 1,785 MHz boost clock vs 1,590 MHz on the 1650.
Benchmarks
Inno3D showed a benchmark chart comparing the GTX 1630, against the GTX 1050 Ti and the GTX 1050 in 12 titles. Some of these titles include popular AAA games such as Cyberpunk 2077, Control, Call of Duty Cold War, Assasins Creed Valhalla, Forza Horizon 4, and New World.
On average, the GTX 1050 Ti and GTX 1630 showed roughly equal performance, with the GTX 1050 being noticeably worse in most games. But there are a few exceptions where the GTX 1050 TI and even the 1050 were beating out the GTX 1630 in three of the twelve titles. Of course, game selection will absolutely affect the results.
This is some exceptionally slow performance for a brand new graphics card, even AMD's new but unexceptional Radeon RX 6400 and RX 6500 XT can outperform the GTX 1050 Ti.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
This makes Nvidia's GTX 1630 truly the slowest modern GPU we've seen in a very long time and a GPU that almost no one will buy for gaming with its rumored price of $165. A year ago, a case could have been made for the GTX 1630 during the global GPU shortage. But with GPU prices now at MSRP and the used GPU market becoming a gold mine for PC gamers, the GTX 1630 is already obsolete before it hits store shelves.
The only saving grace for the GTX 1630 is in its multi-media prowess with support for VP8 and H.265 codecs. It also features a much newer NVENC encoder for recording gameplay or just screen capturing from your monitor, which cannot be said of the RX 6400 which lacks H264 encoding entirely.
Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.
-
-Fran- Hm... Maybe I was wrong in my previous assumption and nVidia DID notice how people complained about the 6400 and 6500XT siblings by focusing on the lack of encoding features, so they're now just charging you for those instead as a "main" feature and not the abysmal performance xD!Reply
I mean, it would make sense from nVidia's standpoint:
-> people whines endlessly about lack of features and don't care about performance/watt
-> releases card that sucks at performance BUT has encoding features the rival doesn't
-> ...
-> profit
Regards xD -
King_V Except that Intel A380 has not quite as sucky performance, and MORE encoding options...Reply -
InvalidError
And 2GB more VRAM, and one extra video output, and ultra-basic RT, and a possibility of getting better than it currently is with driver maturity.King_V said:Except that Intel A380 has not quite as sucky performance, and MORE encoding options...
If I had to spend ~$150 on a slot filler because my GTX1050 decided to die before anything decent entered my palatable price-to-performance range under $200, the A380 looks like it may possibly be the least sucky of current options. -
InvalidError
Did they have 4GB of GDDR6? That alone costs $35-40 in bulk.Alvar Miles Udell said:Remember when cards like this cost under $50?
To hit $50 and still make some sort of profit, you would have to drop to 2GB of DDR4, aim under 20W to keep HSF cost at the absolute bare minimum and possibly need a GPU die even smaller than GP108's already tiny 74sqmm. If Nvidia put such a thing together, it may not even be worthy of getting called a GT1610.
Basically, we'd be back to the good ole' days of ultra-basic GPUs getting laughed at as "graphics decelerators" for being generally worse than most contemporary IGPs. -
JWNoctis
Remember when GTX was something like second-to-best?Alvar Miles Udell said:Remember when cards like this cost under $50?
Granted, letters can be rather more confusing than numbers for market segmentation.
Many current and most next-generation processors already have some graphics capability anyway. I'd say that pricing and market segment is no more.InvalidError said:Did they have 4GB of GDDR6? That alone costs $35-40 in bulk.
To hit $50 and still make some sort of profit, you would have to drop to 2GB of DDR4, aim under 20W to keep HSF cost at the absolute bare minimum and possibly need a GPU die even smaller than GP108's already tiny 74sqmm. If Nvidia put such a thing together, it may not even be worthy of getting called a GT1610.
Basically, we'd be back to the good ole' days of ultra-basic GPUs getting laughed at as "graphics decelerators" for being generally worse than most contemporary IGPs. -
Alvar "Miles" Udell InvalidError said:Did they have 4GB of GDDR6? That alone costs $35-40 in bulk.
To hit $50 and still make some sort of profit, you would have to drop to 2GB of DDR4, aim under 20W to keep HSF cost at the absolute bare minimum and possibly need a GPU die even smaller than GP108's already tiny 74sqmm. If Nvidia put such a thing together, it may not even be worthy of getting called a GT1610.
Basically, we'd be back to the good ole' days of ultra-basic GPUs getting laughed at as "graphics decelerators" for being generally worse than most contemporary IGPs.
I was talking about entry level cards such as the HD 5450 which were not gaming cards, but were great in HTPCs due to being inexpensive, passive, low profile, better than integrated, and supported advanced Dolby and DTS audio, not to mention it supported three simultaneous displays. They were also great cards to keep for a spare in the event your primary card died and could take two weeks, or longer, for RMA.
And cards like the 4550, 5550, and 6450 were praised by TomsHardware during their reviews for their HTPC capabilities, passive nature, and low prices. -
InvalidError
There will always be an ultra-low-end market for people who simply want more display outputs than whatever graphics are already present can provide.JWNoctis said:I'd say that pricing and market segment is no more.
The biggest problem with the $150-200 entry-level price range is manufacturers are being greedy, wanting to push their gross profit margins on entry-level GPUs as close to enthusiast margins as possible while providing as little value as possible to force people to take a step up on the price ladder, so we get generation upon generation of either underwhelming entry-level offerings or nothing whatsoever, especially during a component shortage and crypto boom.
The component shortages and crypto boom are mostly over. This should have an interesting effect on next-gen GPU sales and with some luck, AMD and Nvidia will have a renewed interest in competing at the largely untapped and heavily neglected lower-end for sales when their new stuff at likely further inflated MSRPs fails to meet inflated sales expectations from covid and crypto boom.
I wouldn't use a passively cooled GPU, own one, wouldn't buy another. Kind of pointless when you have to make sure there is a case fan pointed at it to prevent it from running obscenely hot since natural convection keeps hot air trapped around the source in a conventional tower setup.Alvar Miles Udell said:And cards like the 4550, 5550, and 6450 were praised by TomsHardware during their reviews for their HTPC capabilities, passive nature, and low prices.
Also, if you aren't going to use your HTPC for 3D, you can just use the IGP until new CODECs not supported by the IGP and too spicy for software-decode on the CPU come along. That is much cheaper, power-efficient and easier to cool than the cheapest GPU.