NVIDIA A100 SXM vs NVIDIA B200 Full Specs

Summary

6,912 Shaders 1.41GHz	18,944 Shaders 1.98GHz
80GB HBM2e2.04TB/s	192GB HBM3e7.7TB/s
Peak·FP32·FP16 19.49 TFLOPS	Peak·FP32·FP16 75.02 TFLOPS
Form Factor SXM4	Form Factor SXM6
TDP 400W	TDP 1000W

Highlights

Peak OPS

Peak OPS FP32 FP16 Tensor FP4 FP4 Tensor Sparse Tensor FP8-16 FP8-16 Tensor Sparse Tensor FP8-32 FP8-32 Tensor Sparse Tensor FP16-16 FP16-16 Tensor Sparse Tensor FP16-32 FP16-32 Tensor Sparse FP64 Tensor FP64 BF16 Tensor BF16 BF16 Tensor Sparse Tensor TF32 Tensor INT8 Tensor INT4 Memory Bandwidth TDP

A100 SXM2.5 POPSINT4 Tensor Sparse

x1

B20018 PFLOPSFP4 Tensor Sparse

x7.22

Show Top Score

Theoretical Performance

Clock Speed1410MHz UC·Base·Boost·OC	Clock Speed1980MHz UC·Boost·OC
Peak OPS 2.5 POPSINT4 Tensor Sparse	Peak OPS 18 PFLOPSFP4 Tensor Sparse
-	Tensor FP4 9 PFLOPS FP4 Tensor Sparse 18 PFLOPS
-	Tensor FP8-16 4.5 PFLOPS FP8-16 Tensor Sparse 9 PFLOPS Tensor FP8-32 4.5 PFLOPS FP8-32 Tensor Sparse 9 PFLOPS
Tensor FP16-16 311.9 TFLOPS FP16-16 Tensor Sparse 623.7 TFLOPS Tensor FP16-32 311.9 TFLOPS FP16-32 Tensor Sparse 623.7 TFLOPS	Tensor FP16-16 2.25 PFLOPS FP16-16 Tensor Sparse 4.5 PFLOPS Tensor FP16-32 2.25 PFLOPS FP16-32 Tensor Sparse 4.5 PFLOPS
BF16 38.98 TFLOPS Tensor BF16 311.9 TFLOPS BF16 Tensor Sparse 623.7 TFLOPS	BF16 150 TFLOPS Tensor BF16 2.25 PFLOPS BF16 Tensor Sparse 4.5 PFLOPS
Tensor TF32 155.9 TFLOPS	Tensor TF32 1.13 PFLOPS
FP32 19.49 TFLOPS	FP32 75.02 TFLOPS
FP64 9.75 TFLOPS Tensor FP64 19.49 TFLOPS	FP64 37.51 TFLOPS Tensor FP64 35.16 TFLOPS
Tensor INT4 1.25 POPS	Tensor INT4 -
Tensor INT8 623.7 TOPS	Tensor INT8 4.5 POPS
Pixel Rate 225.6 GPixel/s	Pixel Rate 63.4 GPixel/s
Texture Rate 609.1 GTexel/s	Texture Rate 1.17 TTexel/s

Core Config

Shaders 6,912 Shaders	Shaders 18,944 Shaders
TMUs 432 TMUs	TMUs 592 TMUs
ROPs 160 ROPs	ROPs 32 ROPs
Tensor Cores 432 T-Cores	Tensor Cores 592 T-Cores
SMs 108 SMs	SMs 148 SMs

Clocks

Base Clock 1.27GHz	Base Clock -
Boost Clock 1.41GHz	Boost Clock 1.98GHz
Tensor Clock -	Tensor Clock 1.86GHz

Cache

L2 Cache 41MB shared

L2 Cache 65.5MB shared

Memory

80GB HBM2e	192GB HBM3e
Memory Bus 5120-bit	Memory Bus 8192-bit
Memory Speed 3.2GT/s	Memory Speed 7.5GT/s
Memory Bandwidth 2.04TB/s	Memory Bandwidth 7.7TB/s

Power & Thermals

TDP 400W

TDP 1000W

Decoder

Decoder Model 5x NVDEC 4

Decoder Model 7x NVDEC 6

Card

Form Factor SXM4	Form Factor SXM6
Cooling Passive	Cooling Passive

Chip

Manufacturer NVIDIA	Manufacturer NVIDIA
Chip Designer NVIDIA	Chip Designer NVIDIA
Architecture Ampere	Architecture Blackwell
Family Server	Family Server
Branding	Branding
Codename NV170	Codename NV190
Chip Variant -	Chip Variant Umbriel
Market Segment Server	Market Segment Server
Release Date Jun 28, 2021	Release Date Mar 18, 2024

Fabrication

Foundry TSMC	Foundry TSMC
Fabrication Node 7N	Fabrication Node 4NP
Die Size 826mm²	Die Size 1620mm²
Transistor Count 54.2 Billion	Transistor Count 208 Billion
Transistor Density 65.62 MTr/mm²	Transistor Density 128.4 MTr/mm²

Images

No images available

No images available

Competitors

NVIDIAA100 SXM vs GPU NVIDIAA100 SXM 40GB NVIDIAA100 SXM vs GPU NVIDIAA100X NVIDIAA100 SXM vs GPU NVIDIAA100 Show 12 more

NVIDIAB200 vs GPU NVIDIAB100 NVIDIAB200 vs GPU NVIDIAGB200 NVIDIAB200 vs GPU AMDInstinct MI350X Show 12 more