https://www.techpowerup.com/267079/nvidia-ampere-a100-has-54-billion-transistors-worlds-largest-7nm-chip
 
So this is also the world's largest 7nm chip

GTXJackBauer
For only $200K a pop. 

Let’s see... a Porsche 911 Turbo or a graphics card...

🤔🤔🤔

this is what I found as far as something that has some specs :https://www.nvidia.com/en-us/data-center/a100/
 
no mention of any core counts, but 40GB of ram (compared to 32 on the volta counterpart) and 1,555 GB memory bandwidth compared 1134GB memory Bandwidth of the volta version.
 
Still hard to pull speculation from that.
 

“The more you buy, the more you save.”
Jensen Huang

Yes.... he went there...

z999z3mystorys
this is what I found as far as something that has some specs :https://www.nvidia.com/en-us/data-center/a100/
 
no mention of any core counts, but 40GB of ram (compared to 32 on the volta counterpart) and 1,555 GB memory bandwidth compared 1134GB memory Bandwidth of the volta version.
 
Still hard to pull speculation from that.
 

 
Here's a pretty good comparison, I think...?
 
https://pcper.com/2020/05/nvidia-introduces-7nm-ampere-a100-gpu/
 
It's the follow on to the V100, datacenter card.  Not really even a graphics GPU. There is no TU100 chip that would allow a direct comparison to Turing chips.

Good discussion on..........Reddit.
 
https://www.reddit.com/r/...ans_for_rtx_3000_gpus/

The full implementation of the NVIDIA Ampere GA100 GPU includes the following units:

• 8 GPCs, 8 TPCs/GPC, 2 SMs/TPC, 16 SMs/GPC, 128 SMs per full GPU
• 64 FP32 CUDA Cores/SM, 8192 FP32 CUDA Cores per full GPU
• 4 third-generation Tensor Cores/SM, 512 third-generation Tensor Cores per full GPU
• 6 HBM2 stacks, 12 512-bit memory controllers

The A100 Tensor Core GPU implementation of the NVIDIA Ampere GA100 GPU includes the following units:

• 7 GPCs, 7 or 8 TPCs/GPC, 2 SMs/TPC, up to 16 SMs/GPC, 108 SMs
• 64 FP32 CUDA Cores/SM, 6912 FP32 CUDA Cores per GPU
• 4 third-generation Tensor Cores/SM, 432 third-generation Tensor Cores per GPU
• 5 HBM2 stacks, 10 512-bit memory controllers

 
 
https://wccftech.com/nvidia-ampere-ga100-gpu-7nm-architecture-specifications-performance-deep-dive/
 
 

 
 

 
https://wccftech.com/nvidia-dgx-a100-next-gen-ampere-ga100-gpu-based-supercomputing-system-spotted/
 
this would be tesla Professional cards .....not Consumer based cards ....  still have to wait til later this year like AUG  like the hotchips show in AUG
 
Intel's Xe will make a debut there as well
 
 
 
May i also speculate that 5 stack of HBM on teslas mean 40GB vs the full 48GB
 
so any Consumer card following the logic above means 20GB GDDR6 on the consumer based cards - since we only get 1/2 that of tesla cards
 
PS: mighty sexy NVLink connector there
 
 
Next-Gen Hopper GPUs Based on Samsung’s 5nm EUV Process, Alleges Rumor – Mass Production Begins This Quarter
 
..... There also exists an opportunity for NVIDIA to go with Samsung's 5nm for its consumer-based graphics card lineup since mass production is indeed happening this quarter and GeForce cards replacing the Turing lineup could arrive by the end of 2020.
 
 
https://wccftech.com/nvidias-ampere-gpus-tsmcs-7nm-and-next-gen-hopper-gpus-samsungs-5nm-euv-rumor/
 
 
 
What I figure is we won't see Ampere for Consumer at all....Volta, Ampere, Watt and Ohm will be used only for Professional cards - physicist names
 
and Consumer base cards will now be based on Programmer names....like turing and hopper
 
this leave s us with two different paths - and the reason for two different naming paths- one AI and server based and on CoWoS chips - specifically focused  there
 
and consumer with GDDR6 and with different needs
 
5nm for consumer cards???? - probably trying to get power consumption down.....I hear Ampere eats a lot
 
if you are doing 5nm now in 6 months you could be ready for launch  for hopper - limited supply until TSMC is up - but seeing that not many people are up to buying high $700 or more expensive cards then early low availability won't be an issue 
 
 
yeah they locked down tsmc's 5nm production as well
 
The GPU giant has also locked down TSMC’s 5-nm capacity for 2021, which will most likely go toward the next generation of cards, reportedly code-named Hopper.

Nice industrial strength computing power

kevinc313
 
 
why go to all the trouble developing these for professional graphics but not release consumer versions?  Nvidia wants to sell every chip it can, where are the poorly binned chips going to go?  

That's where they make the big dough.  Us consumer folks are the small market for them just like the CPU mfgers.  Once they've set which silicons will enter that space, the disabled leftovers is for us.