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
RTX A2000
CORE STATE
GA106
VRAM
6 GB
CLOCK SPEED
1200 MHz
TDP
70 W
BUS WIDTH
192 bit
ARCHITECTURE
Ampere
PROCESS
8 nm
LAUNCH DATE
2021
PERFORMANCE BENCHMARKS
geekbench_opencl
3dmark_3dmark_steel_nomad_dx12
geekbench_vulkan
DETAILED SPECIFICATIONS
SPECIFICATION
H200 NVL
RTX A2000
Core Specs
Shading Units
16,896
3,328
-80.3%
Shaders
16,896
3,328
-80.3%
TMUs
528
104
-80.3%
ROPs
24
48
+100.0%
SM Count
132
26
-80.3%
Clocks
Base Clock
1365 MHz
562 MHz
Boost Clock
1785 MHz
1200 MHz
Memory Clock
1593 MHz
6.4 Gbps effective
1500 MHz
12 Gbps effective
Memory
Memory Size
141 GB
6 GB
VRAM (MB)
144,384
6,144
-95.7%
Memory Type
HBM3e
GDDR6
Memory Bus
6144 bit
192 bit
Bandwidth
4.89 TB/s
288.0 GB/s
Cache
L1 Cache
256 KB (per SM)
128 KB (per SM)
L2 Cache
50 MB
3 MB
Performance
Pixel Rate
42.84 GPixel/s
57.60 GPixel/s
Texture Rate
942.5 GTexel/s
124.8 GTexel/s
FP32 (TFLOPS)
60.32 TFLOPS
7.987 TFLOPS
FP64 (TFLOPS)
30.16 TFLOPS (1:2)
124.8 GFLOPS (1:64)
FP16 (TFLOPS)
241.3 TFLOPS (4:1)
7.987 TFLOPS (1:1)
AI/RT
RT Cores
—
26
Tensor Cores
528
104
-80.3%
Power
TDP
600 W
70 W
TDP (W)
600
70
-88.3%
Suggested PSU
1000 W
250 W
Power Connectors
8-pin EPS
None
Architecture
Architecture
Hopper
Ampere
GPU Name
GH100
GA106
Generation
Server Hopper
(Hxx)
Workstation Ampere
(Ax000)
Process Size
5 nm
8 nm
Transistors
80,000 million
12,000 million
Die Size
814 mm²
276 mm²
Foundry
TSMC
Samsung
Density
98.3M / mm²
43.5M / mm²
API Support
DirectX
—
12 Ultimate (12_2)
OpenGL
—
4.6
Vulkan
—
1.4
OpenCL
3.0
3.0
CUDA
9.0
8.6
Shader Model
—
6.8
Physical
Slot Width
Dual-slot
Dual-slot
Length
267 mm
10.5 inches
167 mm
6.6 inches
Height
111 mm
4.4 inches
69 mm
2.7 inches
Outputs
No outputs
4x mini-DisplayPort 1.4a
Bus Interface
PCIe 5.0 x16
PCIe 4.0 x16
Other
Launch Price
—
449 USD
Production
Active
End-of-life
Predecessor
Server Ada
Quadro Turing
Successor
Server Blackwell
Workstation Ada