The figures on this entry are not sourced to a document published by the chip vendor. The reference recorded against it is third-party, so the entry is marked vendor claimed rather than official until a vendor specification can be attached.

NVIDIA B100 SXM 192GB

Blackwell SXM 2024 4nm Spec confidence: Vendor claimed
FP8 (dense)
3,500
TFLOPS
FP32
60
TFLOPS
VRAM
192 GB
Bandwidth
8.0 TB/s
memory
TDP
700 W
85.7 TFLOPS/kW
Download datasheet

Performance

Peak theoretical throughput by precision type

INT8
8-bit integer
3,500 TOPS
sparse not published
FP8
8-bit floating point
3,500 TFLOPS
sparse not published
FP4
4-bit floating point
7,000 TFLOPS
sparse not published
FP16
16-bit floating point
1,750 TFLOPS
sparse not published
BF16
Brain Float 16
1,750 TFLOPS
sparse not published
TF32
TensorFloat-32
875 TFLOPS
sparse not published
FP32
32-bit floating point
60 TFLOPS
FP64
64-bit floating point
30 TFLOPS
5.000 TFLOPS/W FP8 at 700 W
2.500 TFLOPS/W FP16 at 700 W
0.086 TFLOPS/W FP32 at 700 W

The B100 SXM 192GB in the GPU landscape

Peak FP16 TFLOPS (dense) against TDP, single-GPU parts tracked by Flopper

06001,2001,8002,4003,0000 W250 W500 W750 W1000 W1250 W1500 WInstinct MI355XB100 SXM 192GB

Higher and further left is better: more half-precision throughput for less power.

Specifications

Architecture

Blackwell

Form Factor

SXM

Launch Year

2024

Process Node

4nm

Memory

192 GB

Bandwidth

8,000 GB/s

TDP

700 W

Max Power

~805 W

Spec Confidence

Vendor claimed

Full Specifications

Memory
VRAM 192 GB
Bandwidth 8.0 TB/s
Power & Thermal
TDP 700 W
Enterprise Features
Sparsity Yes
General
Form Factor SXM
Architecture Blackwell
Launch Year 2024

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
DataCrunch
Published
2024-04-10

Data Quality

Spec confidence
Vendor claimed
Clock basis
Boost
Precisions tracked
8
Normalization
All values in TFLOPS

Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.

Browse GPU Cloud Providers

Similar GPUs

Systems Using This GPU

Pre-configured systems featuring the NVIDIA B100

SystemGPU CountPeak PerformanceTotal Power
NVIDIA HGX B100 8-GPU
HGX · baseboard · 2024
8x B100 56.00 PFLOPS 5.6 kW View System

Frequently Asked Questions

How many TFLOPS does the NVIDIA B100 have?

The NVIDIA B100 delivers 60 TFLOPS for FP32 operations, 1750 TFLOPS for FP16, and 3500 TFLOPS for FP8 precision.

What is the power consumption of the NVIDIA B100?

The NVIDIA B100 has a TDP (Thermal Design Power) rating of 700 watts.

How much memory does the NVIDIA B100 have?

The NVIDIA B100 is equipped with 192 GB of VRAM with 8000 GB/s memory bandwidth.

What architecture is the NVIDIA B100 based on?

The NVIDIA B100 is based on the Blackwell architecture, launched in 2024.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs