GPU Comparison
GEFORCE
NVIDIA CMP 70HX
CORE STATE
GA104
VRAM
8 GB
CLOCK SPEED
1395 MHz
TDP
—
BUS WIDTH
256 bit
ARCHITECTURE
Ampere
PROCESS
8 nm
LAUNCH DATE
—
VS
GEFORCE
GB10
CORE STATE
GB20B
VRAM
128 GB
CLOCK SPEED
2418 MHz
TDP
140 W
BUS WIDTH
256 bit
ARCHITECTURE
Blackwell 2.0
PROCESS
5 nm
LAUNCH DATE
2025
PERFORMANCE BENCHMARKS
geekbench_opencl
geekbench_vulkan
DETAILED SPECIFICATIONS
SPECIFICATION
CMP 70HX
GB10
Core Specs
Shading Units
3,840
6,144
+60.0%
Shaders
3,840
6,144
+60.0%
TMUs
120
384
+220.0%
ROPs
64
48
-25.0%
SM Count
30
48
+60.0%
Clocks
Base Clock
1365 MHz
1665 MHz
Boost Clock
1395 MHz
2418 MHz
Memory Clock
1188 MHz
19 Gbps effective
1067 MHz
8.5 Gbps effective
Memory
Memory Size
8 GB
128 GB
VRAM (MB)
8,192
131,072
+1500.0%
Memory Type
GDDR6X
LPDDR5X
Memory Bus
256 bit
256 bit
Bandwidth
608.3 GB/s
273.2 GB/s
Cache
L1 Cache
128 KB (per SM)
256 KB (per SM)
L2 Cache
4 MB
50 MB
Performance
Pixel Rate
89.28 GPixel/s
116.1 GPixel/s
Texture Rate
167.4 GTexel/s
928.5 GTexel/s
FP32 (TFLOPS)
10.71 TFLOPS
29.71 TFLOPS
FP64 (TFLOPS)
167.4 GFLOPS (1:64)
464.3 GFLOPS (1:64)
FP16 (TFLOPS)
10.71 TFLOPS (1:1)
29.71 TFLOPS (1:1)
AI/RT
RT Cores
30
48
+60.0%
Tensor Cores
120
384
+220.0%
Power
TDP
—
140 W
TDP (W)
—
140
Suggested PSU
200 W
300 W
Power Connectors
1x 12-pin
None
Architecture
Architecture
Ampere
Blackwell 2.0
GPU Name
GA104
GB20B
Generation
Mining GPUs
Server Blackwell
(Bxx)
Process Size
8 nm
5 nm
Transistors
17,400 million
—
Die Size
392 mm²
—
Foundry
Samsung
TSMC
Density
44.4M / mm²
—
API Support
DirectX
12 Ultimate (12_2)
—
OpenGL
4.6
—
Vulkan
1.4
—
OpenCL
3.0
3.0
CUDA
8.6
12.1
Shader Model
6.8
—
Physical
Slot Width
Dual-slot
IGP
Length
267 mm
10.5 inches
150 mm
5.9 inches
Height
112 mm
4.4 inches
51 mm
2 inches
Outputs
No outputs
1x HDMI
Bus Interface
PCIe 1.0 x4
PCIe 5.0 x16
Other
Launch Price
—
3,999 USD
Production
End-of-life
Active
Predecessor
—
Server Hopper
Successor
—
Server Rubin