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
Quadro K620
CORE STATE
GM107
VRAM
2 GB
CLOCK SPEED
1124 MHz
TDP
45 W
BUS WIDTH
128 bit
ARCHITECTURE
Maxwell
PROCESS
28 nm
LAUNCH DATE
2014
PERFORMANCE BENCHMARKS
geekbench_opencl
geekbench_vulkan
DETAILED SPECIFICATIONS
SPECIFICATION
H200 NVL
Quadro K620
Core Specs
Shading Units
16,896
384
-97.7%
Shaders
16,896
384
-97.7%
TMUs
528
24
-95.5%
ROPs
24
16
-33.3%
SM Count
132
—
Clocks
Base Clock
1365 MHz
1058 MHz
Boost Clock
1785 MHz
1124 MHz
Memory Clock
1593 MHz
6.4 Gbps effective
900 MHz
1800 Mbps effective
Memory
Memory Size
141 GB
2 GB
VRAM (MB)
144,384
2,048
-98.6%
Memory Type
HBM3e
DDR3
Memory Bus
6144 bit
128 bit
Bandwidth
4.89 TB/s
28.80 GB/s
Cache
L1 Cache
256 KB (per SM)
64 KB (per SMM)
L2 Cache
50 MB
2 MB
Performance
Pixel Rate
42.84 GPixel/s
17.98 GPixel/s
Texture Rate
942.5 GTexel/s
26.98 GTexel/s
FP32 (TFLOPS)
60.32 TFLOPS
863.2 GFLOPS
FP64 (TFLOPS)
30.16 TFLOPS (1:2)
26.98 GFLOPS (1:32)
FP16 (TFLOPS)
241.3 TFLOPS (4:1)
—
AI/RT
Tensor Cores
528
—
Power
TDP
600 W
45 W
TDP (W)
600
45
-92.5%
Suggested PSU
1000 W
200 W
Power Connectors
8-pin EPS
None
Architecture
Architecture
Hopper
Maxwell
GPU Name
GH100
GM107
Generation
Server Hopper
(Hxx)
Quadro Kepler
(Kx200)
Process Size
5 nm
28 nm
Transistors
80,000 million
1,870 million
Die Size
814 mm²
148 mm²
Foundry
TSMC
TSMC
Density
98.3M / mm²
12.6M / mm²
API Support
DirectX
—
12 (11_0)
OpenGL
—
4.6
Vulkan
—
1.4
OpenCL
3.0
3.0
CUDA
9.0
5.0
Shader Model
—
6.7 (5.1)
Physical
Slot Width
Dual-slot
Single-slot
Length
267 mm
10.5 inches
160 mm
6.3 inches
Height
111 mm
4.4 inches
69 mm
2.7 inches
Outputs
No outputs
1x DVI1x DisplayPort 1.2
Bus Interface
PCIe 5.0 x16
PCIe 2.0 x16
Other
Production
Active
End-of-life
Predecessor
Server Ada
Quadro Fermi
Successor
Server Blackwell
Quadro Maxwell