Nvidia’s first Ampere GPU is designed for data centers and AI, not your PC

[Windows9]

 

Quote

Nvidia is unveiling its next-generation Ampere GPU architecture today. The first GPU to use Ampere will be Nvidia’s new A100, built for scientific computing, cloud graphics, and data analytics. While there have been plenty of rumors around Nvidia’s Ampere plans for GeForce “RTX 3080” cards, the A100 will primarily be used in data centers.

Nvidia’s latest data center push comes amid a pandemic and a huge increase in demand for cloud computing. Describing the coronavirus situation as “terribly tragic,” Nvidia CEO Jensen Huang noted that “cloud usage of services are going to see a surge,” in a press briefing attended by The Verge. “Those dynamics are really quite good for our data center business ... My expectation is that Ampere is going to do remarkably well. It’s our best data center GPU ever made and it capitalizes on nearly a decade of our data center experience.”

The A100 sports more than 54 billion transistors, making it the world’s largest 7nm processor. “That is basically at nearly the theoretical limits of what’s possible in semiconductor manufacturing today,” explains Huang. “The largest die the world’s ever made, and the largest number of transistors in a compute engine the world’s ever made.”

Nvidia is boosting its Tensor cores to make them easier to use for developers, and the A100 will also include 19.5 teraflops of FP32 performance, 6,912 CUDA cores, 40GB of memory, and 1.6TB/s of memory bandwidth.

 

https://www.theverge.com/2020/5/14/21258419/nvidia-ampere-gpu-ai-data-centers-specs-a100-dgx-supercomputer

 

[Guest]

I feel like this happens every launch. No one should expect consumer GPUs, but they do anyway.

[RollinLower]

every time a new datacenter GPU comes out the specs just absolutely blow me away.

what a freakin beast! i bet it would cost an arm and a leg tho.

[AbydosOne]

10 minutes ago, Windows9 said:

It’s our best data center GPU ever made

I never understand why tech companies use this line of reasoning, ever.

 

Like, does the general population not completely expect technology to move forward?

[Guest]

14 minutes ago, AbydosOne said:

I never understand why tech companies use this line of reasoning, ever.

 

Like, does the general population not completely expect technology to move forward?

Intel 10nm has entered the chat...

[cajjpants]

45 minutes ago, Den-Fi said:

Intel 10nm has entered the chat...

sideways and more power is progress, right?

[Chimi]

4 hours ago, cajjpants said:

sideways and more power is progress, right?

made me think of this

[Results45]

From the other threads:

 

https://www.anandtech.com/show/15801/nvidia-announces-ampere-architecture-and-a100-products

https://devblogs.nvidia.com/nvidia-ampere-architecture-in-depth/

 

Here's additional specs from TechPowerUp: https://www.techpowerup.com/gpu-specs/tesla-a100-sxm4.c3506

 

So the A100 spits out 19.49 raw TFLOFS at 1410Mhz, but since Tech Jesus comfortably overclocked the Titan V to more than 1600Mhz and the Titan RTX to more than 1900Mhz (before switching to water or LN2) I think it's safe to say that an OC'd "Titan A" card will easily hit 22.12 TFLOPS if based on GA100.

[igormp]

13 minutes ago, Results45 said:

So the A100 spits out 19.49 raw TFLOFS at 1410Mhz, but since Tech Jesus comfortably overclocked the Titan V to more than 1600Mhz and the Titan RTX to more than 1900Mhz (before switching to water or LN2) I think it's safe to say that an OC'd "Titan A" card will easily hit 22.12 TFLOPS if based on GA100.

The next titan probably won't be based on the GA100, but on the GA102.

[Dash Lambda]

39 minutes ago, igormp said:

The next titan probably won't be based on the GA100, but on the GA102.

Honestly, that bugged me back when they started it, but seeing how marvelously absurd the Gx100 chips have gotten I totally understand it now.

[RejZoR]

I always chuckle when companies brag how this is their biggest and most complex chip ever done by them. Yeah, no shit, when you're progressing further, it's kind expected that today's latest and greatest chip will be the biggest and most complex one compared to everything you've done in the past. Apple sort of started doing this nonsense and everyone is doing it now. I think even AMD did it with Lisa Su on stage. It's just such a cringy statement. I'd only give it a pass if somehow NVIDIA, AMD, Apple or anyone else managed to make smaller, less complex chip that actually obliterates older generations in performance that were big and complex. But last time I checked transistors don't work that way unless you deploy some sort of dramatic architectural change.

[NumLock21]

They announced their gaming card during their GeForce gaming event, not in GTC.