Momin Ali
Member-
Posts
14 -
Joined
-
Last visited
Awards
This user doesn't have any awards
Recent Profile Visitors
The recent visitors block is disabled and is not being shown to other users.
Momin Ali's Achievements
-
Is the combination compatible with 4 GV100? this motherboard supports 4x NVIDIA V100s but not mentioned clearly if it also supports 4 Quadro GV100. Any better alternates to use with 4 Quadro GV100?
-
Any idea when AMD EPYC 7nm processors will release and will be available for purchase? Is it really worth the wait or should I go for EPYC 7601 for now?
-
ASUS ROG Strix RX 590 Gaming
-
Quadro GV100 vs 3x Titan V for Deep learning
Momin Ali replied to Momin Ali's topic in Graphics Cards
GV100 comes with 32GB of HBM2 VRAM whereas Titan V is 12 GB HBM2 and a little lower memory bandwidth but its like paying extra 6000$ just for 24Gigs of VRAM. I really need VRAM but I think GV100 is way too overpriced. I wish there was equal support for AMD GPU accelerators, they seem to like giving away HBM2 memory in bulk for lower prices. Anyway the point is can I train a single DNN on multiple GPUs without NV link or SLI?- 2 replies
-
- graphic card
- gpu
- (and 4 more)
-
So I've been searching alot but couldn't find any source which can confirm that Training of a single DNN can be done on multiple Titan Vs as they dont support SLI or NV link. Any better alternative? or should I go for GV100. P.s datasets can be huge!
- 2 replies
-
- graphic card
- gpu
- (and 4 more)
-
I just figured that Titan V nither supports NV-link nor SLI I am not sure if I can train a single DNN on multiple GPUs this way. Probably another NVIDIA's marketing scheme to make us buy their expensive GPUs for the sake of memory requirement?
-
Thanks!
-
Yes, I think you might be right I might be loosing price per performance. I think I should use PCIe storage along with 2 Titan Vs instead of a single GV100.
-
Datasets are huge.
-
Training single network on 3 GPUS with 12 GB VRAM or 1 GPU with 36 GB VRAM. Which will perform faster?
-
So you are saying that I should go for Threadripper if I plan to add more Quadros?
-
GV100 is primarily 32GB HBM2 and has alot better FP64 performance than its competitors.
-
Hi, I am building a PC for Deep Learning purpose and need suggestions for CPU and Motherboard for NVIDIA Quadro GV100. I am confused between Xeon, Threadripper and Epyc. Current spec in mind: Xeon W-3175X with ROG Dominus Extreme P.S should I go for RDIMMs?