NVIDIA Blackwell architecture
GPTKB entity
Statements (30)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:graphics_card
|
| gptkbp:announced |
March 2024
|
| gptkbp:architecture |
Parallel computing
|
| gptkbp:developer |
gptkb:NVIDIA
|
| gptkbp:firstGPU |
NVIDIA GB200
|
| gptkbp:memoryType |
gptkb:HBM3E
|
| gptkbp:namedAfter |
gptkb:David_Blackwell
|
| gptkbp:notableFeature |
gptkb:NVLink_5.0
Multi-die GPU design Blackwell Transformer Engine Enhanced memory bandwidth Second-generation NVSwitch |
| gptkbp:predecessor |
gptkb:NVIDIA_Hopper_architecture
|
| gptkbp:processNode |
gptkb:TSMC_4NP
|
| gptkbp:releaseYear |
2024
|
| gptkbp:successor |
NVIDIA Rubin architecture
|
| gptkbp:supports |
gptkb:TensorFloat-32
BF16 FP16 FP8 INT4 INT8 |
| gptkbp:targetMarket |
gptkb:artificial_intelligence
gptkb:HPC |
| gptkbp:usedIn |
NVIDIA B200 GPU
NVIDIA GB200 Grace Blackwell Superchip |
| gptkbp:bfsParent |
gptkb:NVIDIA_Hopper_architecture
gptkb:NVIDIA_Ada_Lovelace_architecture |
| gptkbp:bfsLayer |
8
|
| https://www.w3.org/2000/01/rdf-schema#label |
NVIDIA Blackwell architecture
|