NVIDIA launches A100 PCIe accelerator with 80GB HBM2e memory

Published: 28th Jun 2021, 07:20 GMT   Comments

NVIDIA A100 PCIe 80GB

NVIDIA formally introduces its A100 variant with a PCI Express interface and 80GB of memory.

Back in November NVIDIA introduced an 80GB variant of the A100 accelerator based on the SXM4 form factor. This variant offered twice the memory capacity of the original Ampere GA100 based model as well as higher bandwidth. Seven months later, NVIDIA also introduces a PCIe-based model featuring the exact same features, except on a standard interface and with lower TDP.

NVIDIA A100 PCIe 80GB is based on 7nm Ampere GA100 GPU featuring 6192 CUDA cores. The bandwidth on this variant increases to 2039 GB/s (over 484 GB/s more than A100 40GB). This is achieved using faster memory with an effective speed of 3186 Gbps.

This GPU is computer-oriented which means it has no gaming purposes, at least not in this form. This product is strictly for high-performance computing to accelerate training using deeplearning algorithms.

Furthermore, NVIDIA announced its GPUDirect Storage feature, which is similar to consumer Microsoft DirectStorage technology. In consumer space, it gives access to fast NVMe storage which can boost loading times in certain workloads. NVIDIA’s technology appears to focus on a similar type of access except to the large memory pool on the GPU, in this case, 80GB of faster HBM2e memory.

NVIDIA Compute Accelerator Series (Formely Tesla)
VideoCardz.comA100 PCIeA100 SXMTesla V100sTesla V100Tesla P100
Picture
GPU7nm GA1007nm GA10012nm GV10012nm GV10016nm GP100
Die Size
 
826 mm^2
 
826 mm^2
 
815 mm^2
 
815 mm^2
 
610 mm^2
Transistors
 
54 billion
 
54 billion
 
21.1 billion
 
21.1 billion
 
15.3 billion
SMs
 
108
 
108
 
80
 
80
 
56
CUDA Cores
 
6912
 
6912
 
5120
 
5120
 
3840
Tensor Cores
 
432
 
432
 
640
 
640
NA
FP16 Compute
 
78 TFLOPS
 
78 TFLOPS
 
32.8 TFLOPS
 
31.4 TFLOPS
 
21.2 TFLOPS
FP32 Compute
 
19.5 TFLOPS
 
19.5 TFLOPS
 
16.4 TFLOPS
 
15.7 TFLOPS
 
10.6 TFLOPS
FP64 Compute
 
9.7 TFLOPS
 
9.7 TFLOPS
 
8.2 TFLOPS
 
7.8 TFLOPS
 
5.3 TFLOPS
Boost Clock
 
~1410MHz
 
~1410MHz
 
~1601 MHz
 
~1533 MHz
 
~1480MHz
Bandwidth
 
2039/1555 GB/s
 
2039/1555 GB/s
 
1134 GB/s
 
900 GB/s
 
721 GB/s
Eff. Memory Clock
 
3186/2430 Gbps
 
3186/2430 Gbps
 
2214 Gbps
 
1760 Gbps
 
1408 Gbps
Memory Config.
 
80/40GB HBM2e
 
80/40GB HBM2e
 
32GB HBM2
 
16/32GB HBM2
 
16GB HBM2
Memory Bus
 
5120-bit
 
5120-bit
 
4096-bit
 
4096-bit
 
4096-bit
TDP
 
250
 
400
 
250W
 
300W
 
300W
Form FactorPCIe 4.0SXM4PCIe 3.0SXM2 / PCIe 3.0SXM

Source: HardwareLuxx




Comment Policy
  • Comments must be written in English.
  • 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.
  • Comments and usernames containing language or concepts that could be deemed offensive will be deleted. Note this may include abusive, threatening, pornographic, offensive, misleading, or libelous language.
  • Comments complaining about the article subject or its source will be removed.
  • A failure to comply with these rules will result in a warning and, in extreme cases, a ban.
  • Please note that comments that attack or harass an individual directly will be deleted and such comments will result in a ban.
  • VideoCardz Moderating Team reserves the right to edit or delete any comments submitted to the site without notice.
  • If you would like to appeal for a comment section ban to be removed, please use this page.
  • If you have any questions about the commenting policy, please let us know through the Contact Page.
  • NEW: Due to the recent increased amount of spam all links will require moderator approval. 
Hide Comment Policy
Comments