Tuesday, March 22nd 2022

NVIDIA H100 is a Compute Monster with 80 Billion Transistors, New Compute Units and HBM3 Memory

During the GTC 2022 keynote, NVIDIA announced its newest addition to the accelerator cards family. Called NVIDIA H100 accelerator, it is the company's most powerful creation ever. Utilizing 80 billion of TSMC's 4N 4 nm transistors, H100 can output some insane performance, according to NVIDIA. Featuring a new fourth-generation Tensor Core design, it can deliver a six-fold performance increase compared to A100 Tensor Cores and a two-fold MMA (Matrix Multiply Accumulate) improvement. Additionally, new DPX instructions accelerate Dynamic Programming algorithms up to seven times over the previous A100 accelerator. Thanks to the new Hopper architecture, the Streaming Module structure has been optimized for better transfer of large data blocks.

The full GH100 chip implementation features 144 SMs, and 128 FP32 CUDA cores per SM, resulting in 18,432 CUDA cores at maximum configuration. The NVIDIA H100 GPU with SXM5 board form-factor features 132 SMs, totaling 16,896 CUDA cores, while the PCIe 5.0 add-in card has 114 SMs, totaling 14,592 CUDA cores. As much as 80 GB of HBM3 memory surrounds the GPU at 3 TB/s bandwidth. Interestingly, the SXM5 variant features a very large TDP of 700 Watts, while the PCIe card is limited to 350 Watts. This is the result of better cooling solutions offered for the SXM form-factor. As far as performance figures are concerned, the SXM and PCIe versions provide two distinctive figures for each implementation. You can check out the performance estimates in various precision modes below. You can read more about the Hopper architecture and what makes it special in this whitepaper published by NVIDIA.
NVIDIA H100
Add your own comment

29 Comments on NVIDIA H100 is a Compute Monster with 80 Billion Transistors, New Compute Units and HBM3 Memory

#26
TheoneandonlyMrK
CyberCTThere's GPU availability now. The higher end 3000 series are still inflated, price wise but the 3070 TI can now be had for under $900 and is in stock now. What is the "supposed" launch price? $600 before all the OEMs said prices were going up regardless? We're almost there ...
So at +50% that's near MSRP to you, nah ,no ,I think not.
Posted on Reply
#27
CyberCT
It was @ $1200 a few months ago with much less availability. And you'll never see MSRP on these cards, doubtful even on launch day of the 4000 series. If you're waiting for MSRP you won't be upgrading for a long time. Enjoy your current card.
Posted on Reply
#28
TheoneandonlyMrK
CyberCTIt was @ $1200 a few months ago with much less availability. And you'll never see MSRP on these cards, doubtful even on launch day of the 4000 series. If you're waiting for MSRP you won't be upgrading for a long time. Enjoy your current card.
I am and will do , if you're replying to someone ,do so.

It's likely 2k in Russia, should I care.

MSRP isn't + 50% , I will happily wait until the next generations low end card stomps that 3070ti and buy that then for less or not if the price isn't right.

You think these prices are reasonable, crack on but please don't insinuate these prices are normal and yes I DO expect prices to get back closer to where they were and I am happy to wait or do without.

At 900£ for a 3070 ti or 6800Xt both companies can f#@£ the f££k OFF.
Posted on Reply
#29
lexluthermiester
Tomorrow30 series has been available for months. The problem has been the price.
So what you're saying is that they're still unavailable. Whether or not cards are on shelves is irrelevant if the price is out of range for the general consumer. The cards might as well not be there.
Posted on Reply
Add your own comment
Nov 23rd, 2024 02:27 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts