GPU Comparison
GEFORCE
NVIDIA A10G
CORE STATE
GA102
VRAM
24 GB
CLOCK SPEED
1710 MHz
TDP
150 W
BUS WIDTH
384 bit
ARCHITECTURE
Ampere
PROCESS
8 nm
LAUNCH DATE
2021
VS
GEFORCE
B200
CORE STATE
GB100
VRAM
90 GB
CLOCK SPEED
1965 MHz
TDP
1000 W
BUS WIDTH
4096 bit
ARCHITECTURE
Blackwell
PROCESS
5 nm
LAUNCH DATE
—
PERFORMANCE BENCHMARKS
geekbench_opencl
geekbench_vulkan
DETAILED SPECIFICATIONS
SPECIFICATION
A10G
B200
Core Specs
Shading Units
9,216
18,944
+105.6%
Shaders
9,216
18,944
+105.6%
TMUs
288
592
+105.6%
ROPs
96
24
-75.0%
SM Count
72
148
+105.6%
Clocks
Base Clock
1320 MHz
700 MHz
Boost Clock
1710 MHz
1965 MHz
Memory Clock
1563 MHz
12.5 Gbps effective
2000 MHz
8 Gbps effective
Memory
Memory Size
24 GB
90 GB
VRAM (MB)
24,576
92,160
+275.0%
Memory Type
GDDR6
HBM3e
Memory Bus
384 bit
4096 bit
Bandwidth
600.2 GB/s
4.10 TB/s
Cache
L1 Cache
128 KB (per SM)
256 KB (per SM)
L2 Cache
6 MB
50 MB
Performance
Pixel Rate
164.2 GPixel/s
47.16 GPixel/s
Texture Rate
492.5 GTexel/s
1,163.3 GTexel/s
FP32 (TFLOPS)
31.52 TFLOPS
74.45 TFLOPS
FP64 (TFLOPS)
985.0 GFLOPS (1:32)
37.22 TFLOPS (1:2)
FP16 (TFLOPS)
31.52 TFLOPS (1:1)
1,191.2 TFLOPS (16:1)
AI/RT
RT Cores
72
—
Tensor Cores
288
592
+105.6%
Power
TDP
150 W
1000 W
TDP (W)
150
1,000
+566.7%
Suggested PSU
450 W
1400 W
Power Connectors
8-pin EPS
—
Architecture
Architecture
Ampere
Blackwell
GPU Name
GA102
GB100
Generation
Server Ampere
(Axx)
Server Blackwell
(Bxx)
Process Size
8 nm
5 nm
Transistors
28,300 million
104,000 million
Die Size
628 mm²
—
Foundry
Samsung
TSMC
Density
45.1M / mm²
—
API Support
DirectX
12 Ultimate (12_2)
—
OpenGL
4.6
—
Vulkan
1.4
—
OpenCL
3.0
3.0
CUDA
8.6
10.0
Shader Model
6.8
—
Physical
Slot Width
Single-slot
SXM Module
Length
267 mm
10.5 inches
—
Height
112 mm
4.4 inches
—
Outputs
No outputs
No outputs
Bus Interface
PCIe 4.0 x16
PCIe 5.0 x16
Other
Production
End-of-life
Active
Predecessor
Tesla Turing
Server Hopper
Successor
Server Ada
Server Rubin