4 ms·
The NVIDIA website has a whole section on Data Center graphics cards: https://www.nvidia.com/en-us/data-center/ https://www.nvidia.com/en-us/data-center/ Won't
by Nition 9y ago
The NVIDIA website has a whole section on Data Center graphics cards: https://www.nvidia.com/en-us/data-center/ https://www.nvidia.com/en-us/data-center/
Won't this just be hurting their own sales possibilities? I can only guess they'll be announcing some datacenter-specific version of the software soon, which is essentially the same but mysteriously more expensive.
- Scaevolus 9y agoThose aren't GeForce cards. They don't want their consumer line (GeForce) competing with their datacenter line (Tesla). They previously reduced the FP64/FP16 performance on GeForce cards (making them much worse for neural networks), but it's easier to limit the losses legally than by reducing hardware. Modern games don't want artificially limited ALUs!
- Nition 9y agoAh sorry, I see, so there's already that split in consumer vs. datacenter and they're now forcing the line to be drawn.
- shaklee3 9y agoThis is not technically correct. I think you're referring to cases of Pascal cards where the GeForce version had poor fp64 performance. This is because it's a different asic Rev. Same reason the more expensive p100 Tesla didn't have int8 support -- certainly not because they did it in software. There's no reason a GeForce card for gaming needs high fp64 support. The reason you're seeing it on the Titan v is because it's literally the same asic, but now you're also paying a premium and it's not made for gamers.
- Scaevolus 9y agoI'm referring to the GTX 1080 running FP16 at 1/64th the rate of FP32: https://www.anandtech.com/show/10325/the-nvidia-geforce-gtx-1080-and-1070-founders-edition-review/5 https://www.anandtech.com/show/10325/the-nvidia-geforce-gtx-... "GeForce GTX 1080, on the other hand, is not faster at FP16. In fact it’s downright slow. For their consumer cards, NVIDIA has severely limited FP16 CUDA performance. GTX 1080’s FP16 instruction rate is 1/128th its FP32 instruction rate, or after you factor in vec2 packing, the resulting theoretical performance (in FLOPs) is 1/64th the FP32 rate, or about 138 GFLOPs."
- shaklee3 9y agoRight, and I'm saying it uses the GP104 asic, which has that disabled so that other features can be enabled (higher clocks, for example). I don't think you can find and example where the Tesla shares the same chip as the GeForce and something like fp16 is different. I'm sure they do a lot of shady things, but I believe part of this is just chip design tradeoffs.
- shaklee3 9y agoJust to prove the point, the Tesla p40 has the same chip as the GeForce you site, and it too has slower fp16. This is for a $5k card .
- pepsi 9y agoThis is about the GeForce products, which are consumer/gaming targeted GPUs. Pro-style Quadro, Tesla, etc. cost quite a bit more for similar hardware due to the firmware, drivers and licensing.
- foobarbecue 9y agoi.e. they cost quite a bit more because NVidia likes it that way -- it's tiered pricing.
- LeifCarrotson 9y ago> quite a bit more for similar hardware due to the firmware, drivers and licensing The first two criteria in your list are irrelevant, it's purely due to licensing.
- ixtli 9y agoI searched the site for high end Quadro drivers and they all link to the same "March2009" license. EDIT: I was wrong. Apparently they dont include this "datacenter" provision http://www.nvidia.com/content/DriverDownload-March2009/licence.php?lang=us http://www.nvidia.com/content/DriverDownload-March2009/licen...