If you go with an RX 6000 GPU, you will get a 2:1 ratio for FP16 and a 1:16 for FP64. Here's where things get a bit interesting today: That's actually where Titan cards became a problem, because they had much better FP64 than the other GeForce cards but a small fraction of the cost compared to Quadros. In some cases, a mid-tier workstation GPU could run laps around a flagship desktop GPU in workloads heavy with FP64, while in gaming it would be the complete opposite. The half-precision and double-precision floats can vary across different performance tiers, or in some cases, a GPU could be could be entirely incompatible with them (even for workstation models). I'm sure you're thinking "yeah obviously a lower-end GPU will perform worse" but perhaps not in the way you're thinking. Again, even the same program with different workloads will yield different winners, but also, performance can vary substantially between performance tiers or pro-grade models. A single benchmark, including a real-world use case, is inconclusive. That's why I was reminded of having seen such indications in the past, that AMD cards don't pull their weight in these applications, due to poor driver support.ĪMD has some catching up to do with Vulkan but like I said, performance varies drastically with OpenCL. They're now in the same camp as AMD, and it seems like Intel is making an effort to improve non-CUDA compute.Ĭan 3060 Ti beat 6800 in raw power in Vulkan and barely lose in OpenCl? That seems unlikely. Their OpenCL performance was fine for what the chips were, but still not noteworthy. Intel never really cared that much about compute because their GPUs were too weak to be worthwhile. Even if their performance is worse, being compatible is more important. That's the only way they can convince people to switch. Key word is "can" though - if they truly want to succeed, they will need to adopt CUDA. If AMD plays their cards right, I believe they can compete with Nvidia faster than Intel can compete in the gaming market. The good news is, they actually have a chance to catch up - AMD's hardware is fine, they already have their toes in the GPU server market, and compute is a lot easier to optimize for than games. They're basically playing catch up with Nvidia at this point. As a result, they've been working on ROCm and HIP. In the past 3 years or so, AMD has realized how much money they've been losing out on in the server GPU market. CUDA is so much better that many open-source developers use it despite the fact it requires closed-source binaries to work. CUDA is also significantly easier to implement. They wrote most of the libraries themselves. Nvidia really optimized CUDA for their platform. The reason why they're never chosen is because they lack CUDA, which not only limits what they can do but CUDA is also just so much better. AMD has actually been very competitive in OpenCL pretty much since TeraScale2 and they've done practically nothing to optimize until maybe a year ago. Overall it seems that Intel's best and mot fast GPU A770 seems to switch places with the RTX 3060 in both tests at best (table courtesy videocardz).Īren't AMD's drivers always considered the reason for AMD underperfoming in these kind of workloads? I haven't yet heard anyone praising the Intel graphics card drivers, so it ought to be unknown how optimised they are throughout the whole spectrum of stuff a GPU can be put through. In the OpenCL test, the difference between the A770 and the A750 is again around 12%, which is in favor of the full ACM-G10 model. This is a difference of 10.4% between the two cards. The A76 points on the Vulkan test, while the A750 only got 66609 points. Both cards have been seen on the Geekbench website in an ASUS ROG Z690 Apex & Core i9-12900KS system with both cards installed. It was tested with the A750, which has 448 Vector Engines and 8 GB of VRAM (3584 FP32 cores). The A770 tested is the Limited Edition with all 512 Xe Vector Engines and 16GB of VRAM (or 4096 FP32 cores). Only the first GPU could hint at its actual performance, which may or may not translate to raw gaming power, videocardz noticed. Two graphics APIs, Vulkan and OpenCL, were used to test both cards. The first benchmark results for desktop cards with Intel's Limited Edition of Arc have been spotted.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |