GPU Comparison
GEFORCE
NVIDIA A2
CORE STATE
GA107
VRAM
16 GB
CLOCK SPEED
1770 MHz
TDP
60 W
BUS WIDTH
128 bit
ARCHITECTURE
Ampere
PROCESS
8 nm
LAUNCH DATE
2021
VS
GEFORCE
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
PERFORMANCE BENCHMARKS
geekbench_opencl
DETAILED SPECIFICATIONS
SPECIFICATION
A2
H200 NVL
Core Specs
Shading Units
1,280
16,896
+1220.0%
Shaders
1,280
16,896
+1220.0%
TMUs
40
528
+1220.0%
ROPs
32
24
-25.0%
SM Count
10
132
+1220.0%
Clocks
Base Clock
1440 MHz
1365 MHz
Boost Clock
1770 MHz
1785 MHz
Memory Clock
1563 MHz
12.5 Gbps effective
1593 MHz
6.4 Gbps effective
Memory
Memory Size
16 GB
141 GB
VRAM (MB)
16,384
144,384
+781.3%
Memory Type
GDDR6
HBM3e
Memory Bus
128 bit
6144 bit
Bandwidth
200.1 GB/s
4.89 TB/s
Cache
L1 Cache
128 KB (per SM)
256 KB (per SM)
L2 Cache
2 MB
50 MB
Performance
Pixel Rate
56.64 GPixel/s
42.84 GPixel/s
Texture Rate
70.80 GTexel/s
942.5 GTexel/s
FP32 (TFLOPS)
4.531 TFLOPS
60.32 TFLOPS
FP64 (TFLOPS)
70.80 GFLOPS (1:64)
30.16 TFLOPS (1:2)
FP16 (TFLOPS)
4.531 TFLOPS (1:1)
241.3 TFLOPS (4:1)
AI/RT
RT Cores
10
—
Tensor Cores
40
528
+1220.0%
Power
TDP
60 W
600 W
TDP (W)
60
600
+900.0%
Suggested PSU
250 W
1000 W
Power Connectors
None
8-pin EPS
Architecture
Architecture
Ampere
Hopper
GPU Name
GA107
GH100
Generation
Workstation Ampere
(Ax000)
Server Hopper
(Hxx)
Process Size
8 nm
5 nm
Transistors
8,700 million
80,000 million
Die Size
200 mm²
814 mm²
Foundry
Samsung
TSMC
Density
43.5M / mm²
98.3M / mm²
API Support
DirectX
12 Ultimate (12_2)
—
OpenGL
4.6
—
Vulkan
1.4
—
OpenCL
3.0
3.0
CUDA
8.6
9.0
Shader Model
6.8
—
Physical
Slot Width
Single-slot
Dual-slot
Length
—
267 mm
10.5 inches
Height
—
111 mm
4.4 inches
Outputs
No outputs
No outputs
Bus Interface
PCIe 4.0 x8
PCIe 5.0 x16
Other
Production
End-of-life
Active
Predecessor
Quadro Turing
Server Ada
Successor
Workstation Ada
Server Blackwell