NVIDIA B100 vs AMD Radeon Instinct MI308X
We compared two Desktop platform GPUs: 96GB VRAM B100 and 192GB VRAM Radeon Instinct MI308X to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA B100 's Advantages
Released 11 months late
AMD Radeon Instinct MI308X 's Advantages
Boost Clock has increased by 14% (2100MHz vs 1837MHz)
More VRAM (192GB vs 96GB)
Larger VRAM bandwidth (10.3TB/s vs 4.10TB/s)
2560 additional rendering cores
Lower TDP (750W vs 1000W)
Score
Benchmark
FP32 (float)
B100
62.08 TFLOPS
Radeon Instinct MI308X
+31%
81.72 TFLOPS
Graphics Card
Nov 2024
Release Date
Dec 2023
Server Blackwell
Generation
Radeon Instinct
Desktop
Type
Desktop
PCIe 5.0 x16
Bus Interface
PCIe 5.0 x16
Clock Speeds
1665 MHz
Base Clock
1000 MHz
1837 MHz
Boost Clock
2100 MHz
2000 MHz
Memory Clock
2525 MHz
Memory
96GB
Memory Size
192GB
HBM3e
Memory Type
HBM3
4096bit
Memory Bus
8192bit
4.10TB/s
Bandwidth
10.3TB/s
Render Config
-
Compute Units
304
132
SM Count
-
16896
Shading Units
19456
528
TMUs
1216
24
ROPs
0
528
Tensor Cores
1216
-
-
-
256 KB (per SM)
L1 Cache
16 KB (per CU)
50 MB
L2 Cache
16 MB
-
L3 Cache
256 MB
Theoretical Performance
44.09 GPixel/s
Pixel Rate
0 MPixel/s
969.9 GTexel/s
Texture Rate
2554 GTexel/s
248.3 TFLOPS
FP16 (half)
653.7 TFLOPS
62.08 TFLOPS
FP32 (float)
81.72 TFLOPS
31.04 TFLOPS
FP64 (double)
81.72 TFLOPS
Board Design
1000W
TDP
750W
1400 W
Suggested PSU
1150 W
No outputs
Outputs
No outputs
-
Power Connectors
None
Graphics Processor
GB102
GPU Name
Aqua Vanjaram
-
-
-
Blackwell
Architecture
CDNA 3.0
TSMC
Foundry
TSMC
5 nm
Process Size
5 nm
104 billion
Transistors
153 billion
Unknown
Die Size
1017 mm²
Graphics Features
N/A
DirectX
N/A
N/A
OpenGL
N/A
3.0
OpenCL
3.0
N/A
Vulkan
N/A
10.1
CUDA
-
N/A
Shader Model
N/A
