NVIDIA H200 NVL vs NVIDIA H100 SXM5 94 GB
We compared two Desktop platform GPUs: 141GB VRAM H200 NVL and 94GB VRAM H100 SXM5 94 GB to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA H200 NVL 's Advantages
Released 1 years and 8 months late
More VRAM (141GB vs 94GB)
Larger VRAM bandwidth (4.89TB/s vs 3.36TB/s)
Lower TDP (600W vs 700W)
NVIDIA H100 SXM5 94 GB 's Advantages
Boost Clock has increased by 11% (1980MHz vs 1785MHz)
Score
Benchmark
FP32 (float)
H200 NVL
60.32 TFLOPS
H100 SXM5 94 GB
+10%
66.91 TFLOPS
Graphics Card
Nov 2024
Release Date
Mar 2023
Server Hopper
Generation
Server Hopper
Desktop
Type
Desktop
PCIe 5.0 x16
Bus Interface
PCIe 5.0 x16
Clock Speeds
1365 MHz
Base Clock
1350 MHz
1785 MHz
Boost Clock
1980 MHz
1593 MHz
Memory Clock
1313 MHz
Memory
141GB
Memory Size
94GB
HBM3e
Memory Type
HBM3
6144bit
Memory Bus
5120bit
4.89TB/s
Bandwidth
3.36TB/s
Render Config
-
-
-
132
SM Count
132
16896
Shading Units
16896
528
TMUs
528
24
ROPs
24
528
Tensor Cores
528
-
-
-
256 KB (per SM)
L1 Cache
256 KB (per SM)
50 MB
L2 Cache
50 MB
-
-
-
Theoretical Performance
42.84 GPixel/s
Pixel Rate
47.52 GPixel/s
942.5 GTexel/s
Texture Rate
1045 GTexel/s
241.3 TFLOPS
FP16 (half)
267.6 TFLOPS
60.32 TFLOPS
FP32 (float)
66.91 TFLOPS
30.16 TFLOPS
FP64 (double)
33.45 TFLOPS
Board Design
600W
TDP
700W
1000 W
Suggested PSU
1100 W
No outputs
Outputs
No outputs
8-pin EPS
Power Connectors
8-pin EPS
Graphics Processor
GH100
GPU Name
GH100
-
-
-
Hopper
Architecture
Hopper
TSMC
Foundry
TSMC
5 nm
Process Size
5 nm
80 billion
Transistors
80 billion
814 mm²
Die Size
814 mm²
Graphics Features
N/A
DirectX
N/A
N/A
OpenGL
N/A
3.0
OpenCL
3.0
N/A
Vulkan
N/A
9.0
CUDA
9.0
N/A
Shader Model
N/A
