AMD announced Radeon Vega earlier this year and followed up with news of its Vega Frontier Edition at its 2017 Financial Analyst Day. AMD claims the Frontier Edition is the "world's first" GPU geared for AI (Artificial Intelligence), creatives, and science pioneers.
Nvidia just announced its new Tesla V100 based on the Volta architecture, so it's not surprising that AMD is firing back with details of its new Frontier Edition. Vega marks AMD's re-entrance into the high-performance desktop GPU arena, but it also serves as a platform for penetrating into the data center market.
GPUs offer tremendous advantages over most other forms of compute for machine learning training workloads, and the high-margin data center segment offers a rapidly expanding market with lucrative profit margins. The Frontier Edition sports 64 Next-Generation Compute Units, which AMD has bestowed with the "nCU" moniker due to support new data types, that feature 4,096 stream processors. AMD has optimized the nCUs for higher clocks and has also integrated larger instruction buffers to feed them.
AMD provided estimated performance specifications of 12.5 TFLOPS in peak FP32 single precision compute and 25 TFLOPS of peak FP16 compute. The card features 16GB of HBC (High Bandwidth Cache), which is AMD's new term for HBM (High Bandwidth Memory). The new card also features support for 8K displays, and it can access up to 256TB of virtual memory. This voluminous capacity is useful for crunching, video processing, and deep learning algorithms.
AMD noted that the gold shroud is hand-made for each GPU, and there are both liquid- and air-cooled models available.
AMD isn't providing a lot of new information outside of the previously-announced Vega highlights (EDIT - AMD shared some basic performance data on its blog), but the company did provide a rough overview of the new Frontier Edition:
Machine learning. Together with AMD’s ROCm open software platform, Radeon Vega Frontier Edition enables developers to tap into the power of Vega for machine learning algorithm development. Frontier Edition delivers more than 50 percent more performance than today’s most powerful machine learning GPUs.Advanced visualization. Radon Vega Frontier Edition provides the performance required to drive increasingly large and complex models for real-time visualization, physically-based rendering and virtual reality through the design phase as well as rendering phase of product development.VR workloads. Radeon Vega Frontier Edition is ideal for VR content creation supporting AMD’s LiquidVR technology to deliver the gripping content, advanced visual comfort and compatibility needed for next-generation VR experiences.Revolutionized game design workflows. Radeon Vega Frontier Edition simplifies and accelerates game creation by providing a single GPU optimized for every stage of a game developer’s workflow, from asset production to playtesting and performance optimization.
The Vega Frontier Edition is slated for release in June 2017, but the company has not provided any pricing information.
Stay on the Cutting Edge
Join the experts who read Tom's Hardware for the inside track on enthusiast PC tech news — and have for over 25 years. We'll send breaking news and in-depth reviews of CPUs, GPUs, AI, maker hardware and more straight to your inbox.
Paul Alcorn is the Managing Editor: News and Emerging Tech for Tom's Hardware US. He also writes news and reviews on CPUs, storage, and enterprise hardware.
What a joke. So when are they planning to announce a consumer Vega card? Probably the day after 1180Ti drops.Reply
They're going to announce it at E3Reply
19699347 said:What a joke. So when are they planning to announce a consumer Vega card? Probably the day after 1180Ti drops.
They haven't "announced" it yet, but you'll see Vega Nova, Eclipse, and Core releasing June 5th after Computex reveal on the 31st.
AMD noted that the gold shroud is hand-made for each GPUNo thanks. Save yourselves the trouble and save us some $$$. The only place I want gold, in my PC, is plating the various pins & connectors (and there are even better options for that). Of course, I'm not exactly planning on buying a Frontier Edition anyways, but the professionals they listed as the primary market probably don't care either. Few will even have windowed cases.
the company has not provided any pricing informationWell, if they want performance pricing parity with Nvidia's GV100 (which I think won't be shipping to Q3, FWIW), they could charge as much as ~$3750. That's using a back-of-the-envelope figure of $18k for GV100, dividing by 120 fp16 TFLOPS, then scaling by Vega's fp16 performance of 25 TFLOPS. I don't know how big their die is, but I'd be a surprised if they weren't making decent margins at that price point.
Uh, why is "hand-made" a plus for a graphics card shroud?Reply
Annnd I just got my EVGA GTX 1080 Ti FTW3 card yesterday. Yep, consumer Vega is officially late to the party.Reply
Is this the card that has been tossed around in rumors and leaks for a year now? Looks like the answer is yes. And it is a disappointment. AMD might have a market for 2-3 months until Nvidia releases Volta. Anyone who is interested in AI deep learning already has a Pascal card, and will likely just wait a couple months until this card is destroyed outright.Reply
If AMD does not pull a fast one, and release a monster consumer GPU at a good price point, I will be giving up and moving on. So long, and thanks for wasting my time AMD
Oh, and AMD literally has to the end of the month for me. I have been waiting for over a year at this point for Vega. They have officially run out of time. Any delays will no longer be tolerated. Thin ice does not begin to describe what AMD is skating on with meReply
Very nice. AMD haters rule here as always.Reply
I really don't get all the hostility. You'd think that AMD had promised a cure for cancer and then only announced a new treatment.Reply
These "haters" should really get a life.
AMD and nVidia are battling it out for the GPU and AI markets and the winner has to be the consumer.