NVIDIA CMP 40HX vs NVIDIA H200 NVL
GPU · Specs Compare
Graphics Card
Release Date
Feb 2021
Nov 2024
Generation
Mining GPUs
Server Hopper
Type
Desktop
AI GPU
Bus Interface
PCIe 1.0 x4
PCIe 5.0 x16
Clock Speeds
Base Clock
1470 MHz
1365 MHz
Boost Clock
1650 MHz
1785 MHz
Memory Clock
1750 MHz
1593 MHz
Memory
Memory Size
8GB
141GB
Memory Type
GDDR6
HBM3e
Memory Bus
256bit
6144bit
Bandwidth
448.0GB/s
4.89TB/s
Render Config
SM Count
36
132
Shading Units
2304
16896
TMUs
144
528
ROPs
64
24
Tensor Cores
288
528
RT Cores
36
—
L1 Cache
64 KB (per SM)
256 KB (per SM)
L2 Cache
4 MB
50 MB
Theoretical Performance
Pixel Rate
105.6 GPixel/s
42.84 GPixel/s
Texture Rate
237.6 GTexel/s
942.5 GTexel/s
FP16 (half)
15.21 TFLOPS
241.3 TFLOPS
FP32 (float)
7.603 TFLOPS
60.32 TFLOPS
FP64 (double)
237.6 GFLOPS
30.16 TFLOPS
Graphics Processor
GPU Name
TU106
GH100
GPU Variant
TU106-100-A1
—
Architecture
Turing
Hopper
Foundry
TSMC
TSMC
Process Size
12 nm
5 nm
Transistors
10.8 billion
80 billion
Die Size
445 mm²
814 mm²
Board Design
TDP
185W
600W
Suggested PSU
450 W
1000 W
Outputs
No outputs
No outputs
Power Connectors
1x 8-pin
8-pin EPS
Graphics Features
DirectX
12 Ultimate (12_2)
N/A
OpenGL
4.6
N/A
OpenCL
3.0
3.0
Vulkan
1.3
N/A
CUDA
7.5
9.0
Shader Model
6.8
N/A
Benchmark
Blender
1,174
—
▲ Specs Compare
Data sourced from official specifications and benchmark tests.