B300
NVIDIA B300 Full Specs
18,944 Shaders 1.98GHz |
288GB HBM3e7.67TB/s |
·· 75 TFLOPS |
Form Factor SXM6 |
TDP 1100W |
B30018 PFLOPSMXFP4 Tensor SparseClock Speed ·· |
Peak OPS 18 PFLOPSMXFP4 Tensor Sparse |
Tensor FP4 13.5 PFLOPSFP4 Tensor Sparse 18 PFLOPSMXFP4 Tensor 13.5 PFLOPSMXFP4 Tensor Sparse 18 PFLOPSNVFP4 Tensor 13.5 PFLOPSNVFP4 Tensor Sparse 18 PFLOPS |
FP6 Tensor 4.5 PFLOPSFP6 Tensor Sparse 9 PFLOPSMXFP6 Tensor 4.5 PFLOPSMXFP6 Tensor Sparse 9 PFLOPS |
Tensor FP8-16 4.5 PFLOPSFP8-16 Tensor Sparse 9 PFLOPSTensor FP8-32 4.5 PFLOPSFP8-32 Tensor Sparse 9 PFLOPSMXFP8 Tensor 4.5 PFLOPSMXFP8 Tensor Sparse 9 PFLOPS |
FP16 75 TFLOPSTensor FP16-16 2.25 PFLOPSFP16-16 Tensor Sparse 4.5 PFLOPSTensor FP16-32 2.25 PFLOPSFP16-32 Tensor Sparse 4.5 PFLOPS |
Tensor BF16 2.25 PFLOPSBF16 Tensor Sparse 4.5 PFLOPS |
Tensor TF32 1.13 PFLOPS |
FP32 75 TFLOPS |
FP64 1.17 TFLOPS |
Tensor INT8 140.6 TOPS |
Pixel Rate 63.3 GPixel/s |
Texture Rate 1.17 TTexel/s |
Shaders 18,944 Shaders |
TMUs 592 TMUs |
ROPs 32 ROPs |
Tensor Cores 592 T-Cores |
SMs 148 SMs |
Boost Clock 1.98GHz |
Tensor Clock 1.86GHz |
L2 Cache 126MB shared |
288GB HBM3e |
Memory Bus 7680-bit |
Memory Speed 8GT/s |
Memory Bandwidth 7.67TB/s |
ECC Yes |
TDP 1100W |
CUDA 10.3 |
Decoder 7x NVDEC 6 |
Codec MPEG-1 MPEG-2 MPEG-4 VC-1 VP8 VP9 AVC (H.264) HEVC (H.265) AV1 |
Form Factor SXM6 |
Cooling Passive |
Multi-GPU Multi-GPU Type NVLink |
Manufacturer |
Chip Designer |
Architecture |
Family |
Branding  |
Market Segment Server |
Release Date Sep 11, 2025 |
Foundry TSMC |
Fabrication Node 4NP |
Die Size 1620mm² |
Transistor Count 208B |
Transistor Density 128.4 MTr/mm² |