GPU Comparison
GEFORCE
NVIDIA H200 NVL
CORE STATE
GH100
VRAM
141 GB
CLOCK SPEED
1785 MHz
TDP
600 W
BUS WIDTH
6144 bit
ARCHITECTURE
Hopper
PROCESS
5 nm
LAUNCH DATE
2024
VS
GEFORCE
T600
CORE STATE
TU117
VRAM
4 GB
CLOCK SPEED
1335 MHz
TDP
40 W
BUS WIDTH
128 bit
ARCHITECTURE
Turing
PROCESS
12 nm
LAUNCH DATE
2021
PERFORMANCE BENCHMARKS
geekbench_opencl
geekbench_vulkan
passmark_directx_10
passmark_directx_11
passmark_directx_12
passmark_directx_9
passmark_g2d
passmark_g3d
passmark_gpu_compute
DETAILED SPECIFICATIONS
SPECIFICATION
H200 NVL
T600
Core Specs
Shading Units
16,896
640
-96.2%
Shaders
16,896
640
-96.2%
TMUs
528
40
-92.4%
ROPs
24
32
+33.3%
SM Count
132
10
-92.4%
Clocks
Base Clock
1365 MHz
735 MHz
Boost Clock
1785 MHz
1335 MHz
Memory Clock
1593 MHz
6.4 Gbps effective
1250 MHz
10 Gbps effective
Memory
Memory Size
141 GB
4 GB
VRAM (MB)
144,384
4,096
-97.2%
Memory Type
HBM3e
GDDR6
Memory Bus
6144 bit
128 bit
Bandwidth
4.89 TB/s
160.0 GB/s
Cache
L1 Cache
256 KB (per SM)
64 KB (per SM)
L2 Cache
50 MB
1024 KB
Performance
Pixel Rate
42.84 GPixel/s
42.72 GPixel/s
Texture Rate
942.5 GTexel/s
53.40 GTexel/s
FP32 (TFLOPS)
60.32 TFLOPS
1.709 TFLOPS
FP64 (TFLOPS)
30.16 TFLOPS (1:2)
53.40 GFLOPS (1:32)
FP16 (TFLOPS)
241.3 TFLOPS (4:1)
3.418 TFLOPS (2:1)
AI/RT
Tensor Cores
528
—
Power
TDP
600 W
40 W
TDP (W)
600
40
-93.3%
Suggested PSU
1000 W
200 W
Power Connectors
8-pin EPS
None
Architecture
Architecture
Hopper
Turing
GPU Name
GH100
TU117
Generation
Server Hopper
(Hxx)
Quadro Turing
(Tx000)
Process Size
5 nm
12 nm
Transistors
80,000 million
4,700 million
Die Size
814 mm²
200 mm²
Foundry
TSMC
TSMC
Density
98.3M / mm²
23.5M / mm²
API Support
DirectX
—
12 (12_1)
OpenGL
—
4.6
Vulkan
—
1.4
OpenCL
3.0
3.0
CUDA
9.0
7.5
Shader Model
—
6.8
Physical
Slot Width
Dual-slot
Single-slot
Length
267 mm
10.5 inches
—
Height
111 mm
4.4 inches
—
Outputs
No outputs
4x mini-DisplayPort 1.4a
Bus Interface
PCIe 5.0 x16
PCIe 3.0 x16
Other
Production
Active
End-of-life
Predecessor
Server Ada
Quadro Volta
Successor
Server Blackwell
Workstation Ampere