NVIDIA GH100 Hopper GPU comes with 48MB of L2 Cache, only one GPC has graphics enabled

Published: Mar 21st 2022, 09:33 GMT   Comments

NVIDIA GH100 Hopper: only one GPC with graphics enabled

An interesting part of the NVIDIA leak has just resurfaced. 

Twitter user Locuza specializing in creating visual summaries of the existing knowledge on upcoming hardware products, has made a new GPU diagram featuring NVIDIA GH100 GPU, the upcoming data-center accelerator and company’s first 5 nm processor.

His diagram features two details that have been revealed in the past 2 weeks, details that we have missed. The origin of this leak is not official data, but the information that has leaked after hacking group published leaked confidential data from NVIDIA servers.

Although NVIDIA is more than likely to announce its Hopper GPUs tomorrow, some details might not immediately be confirmed. For instance, the fact that only one Graphics Processing Cluster (GPC) has a 3D engine, while 7 do not. This means that not all GPCs in Hopper GPU are identical, and NVIDIA is clearly saving space for a functionality that is simply not as important for a datacenter GPU as it would have been for a consumer product.

NVIDIA GH100 GPC configuration, Source: @xinoassassin1

Furthermore, it is said that GH100 GPU would feature 48 MB of L2 Cache. This is not an upgrade over Ampere GA100 GPU (48MB) though. However, it is three times as much as AMD’s Instinct MI250 “Aldebaran” GPU with 16MB of total L2 cache. Interestingly, NVIDIA’s RTX 40 series codenamed “Ada” are rumored to feature even more L2 cache, up to 96MB for AD102 GPU.

Thanks to the leak, NVIDIA GH100 is confirmed to feature 8 GPCs. Each cluster is to feature 9 TPCs and each TPC comes with two Streaming Multiprocessors. Now assuming that the architecture has not changed in regard to CUDA cores, this would give 144 SMs and 9216 CUDA cores (or 17152 if FP32 cores are doubled).

NVIDIA GH100 GPU Block Diagram, Source: @Locuza

It is worth noting that GH100 is a single die chip (monolithic), whereas the rumored GH202 is a MCM design, possibly featuring two GH100 dies. Naturally, the end-product such as H100 Tensor Core for SXM or PCIe accelerator would not have the full 144 SMs enabled. It is expected that around 15 to 20% of SMs will be disabled.

NVIDIA is to unveil Hopper architecture at GTC tomorrow during CEO Jensen Huang’s keynote at GTC 2022.

RUMORED NVIDIA Data-Center GPUs Specifications
VideoCardz.comNVIDIA H100NVIDIA A100NVIDIA Tesla V100NVIDIA Tesla P100
Die Size814 mm²828 mm²815 mm²610 mm²
Fabrication NodeTSMC N4TSMC N712nm FFN16nm FinFET+
GPU Clusters1321088056
CUDA Cores16896/14592*691251203584
L2 Cache50MB40MB6MB4MB
Tensor Cores528/456*432320
Memory Bus5120-bit5120-bit4096-bit4096-bit
Memory Size80 GB HBM3/HBM2e*40/80GB HBM2e16/32 HBM216GB HBM2
InterfaceSXM5/*PCIe Gen5SXM4/PCIe Gen4SXM2/PCIe Gen3SXM/PCIe Gen3
Launch Year2022202020172016

Source: @Locuza, @xinoassassin1

Comment Policy
  1. Comments must be written in English and should not exceed 1000 characters.
  2. Comments deemed to be spam or solely promotional in nature will be deleted. Including a link to relevant content is permitted, but comments should be relevant to the post topic. Discussions about politics are not allowed on this website.
  3. Comments and usernames containing language or concepts that could be deemed offensive will be deleted.
  4. Comments complaining about the post subject or its source will be removed.
  5. A failure to comply with these rules will result in a warning and, in extreme cases, a ban. In addition, please note that comments that attack or harass an individual directly will result in a ban without warning.
  6. VideoCardz has never been sponsored by AMD, Intel, or NVIDIA. Users claiming otherwise will be banned.
  7. VideoCardz Moderating Team reserves the right to edit or delete any comments submitted to the site without notice.
  8. If you have any questions about the commenting policy, please let us know through the Contact Page.
Hide Comment Policy