GPU Comparison

NVIDIA
GEFORCE

NVIDIA B200

CORE STATE GB100
VRAM 90 GB
CLOCK SPEED 1965 MHz
TDP 1000 W
BUS WIDTH 4096 bit
ARCHITECTURE Blackwell
nm
PROCESS 5 nm
LAUNCH DATE
VS
NVIDIA
GEFORCE

PG506-232

CORE STATE GA100
VRAM 24 GB
CLOCK SPEED 1440 MHz
TDP 165 W
BUS WIDTH 3072 bit
ARCHITECTURE Ampere
nm
PROCESS 7 nm
LAUNCH DATE 2021

PERFORMANCE BENCHMARKS

geekbench_opencl
345,482
225,124

DETAILED SPECIFICATIONS

SPECIFICATION
B200
PG506-232
Core Specs
Shading Units
18,944
3,584 -81.1%
Shaders
18,944
3,584 -81.1%
TMUs
592
224 -62.2%
ROPs
24
96 +300.0%
SM Count
148
56 -62.2%
Clocks
Base Clock
700 MHz
930 MHz
Boost Clock
1965 MHz
1440 MHz
Memory Clock
2000 MHz 8 Gbps effective
1215 MHz 2.4 Gbps effective
Memory
Memory Size
90 GB
24 GB
VRAM (MB)
92,160
24,576 -73.3%
Memory Type
HBM3e
HBM2
Memory Bus
4096 bit
3072 bit
Bandwidth
4.10 TB/s
933.1 GB/s
Cache
L1 Cache
256 KB (per SM)
192 KB (per SM)
L2 Cache
50 MB
24 MB
Performance
Pixel Rate
47.16 GPixel/s
138.2 GPixel/s
Texture Rate
1,163.3 GTexel/s
322.6 GTexel/s
FP32 (TFLOPS)
74.45 TFLOPS
10.32 TFLOPS
FP64 (TFLOPS)
37.22 TFLOPS (1:2)
5.161 TFLOPS (1:2)
FP16 (TFLOPS)
1,191.2 TFLOPS (16:1)
10.32 TFLOPS (1:1)
AI/RT
Tensor Cores
592
224 -62.2%
Power
TDP
1000 W
165 W
TDP (W)
1,000
165 -83.5%
Suggested PSU
1400 W
450 W
Power Connectors
8-pin EPS
Architecture
Architecture
Blackwell
Ampere
GPU Name
GB100
GA100
Generation
Server Blackwell (Bxx)
Server Ampere (Axx)
Process Size
5 nm
7 nm
Transistors
104,000 million
54,200 million
Die Size
826 mm²
Foundry
TSMC
TSMC
Density
65.6M / mm²
API Support
OpenCL
3.0
3.0
CUDA
10.0
8.0
Physical
Slot Width
SXM Module
Dual-slot
Length
267 mm 10.5 inches
Height
112 mm 4.4 inches
Outputs
No outputs
No outputs
Bus Interface
PCIe 5.0 x16
PCIe 4.0 x16
Other
Production
Active
End-of-life
Predecessor
Server Hopper
Tesla Turing
Successor
Server Rubin
Server Ada
View B200 Details View PG506-232 Details