NVIDIA Blackwell architecture
E804860
NVIDIA Blackwell architecture is a next-generation GPU microarchitecture designed to power advanced AI, data center, and high-performance computing workloads, succeeding the Ada Lovelace generation.
All labels observed (1)
| Label | Occurrences |
|---|---|
| NVIDIA Blackwell architecture canonical | 1 |
Statements (46)
| Predicate | Object |
|---|---|
| instanceOf | GPU microarchitecture ⓘ |
| announcedBy | NVIDIA Corporation NERFINISHED ⓘ |
| architectureType | parallel processing architecture ⓘ |
| belongsToSeries | NVIDIA Blackwell GPU series NERFINISHED ⓘ |
| codeNameOrigin | Blackwell NERFINISHED ⓘ |
| compatibleWith |
CUDA software ecosystem
ⓘ
NVIDIA AI software stack NERFINISHED ⓘ |
| componentOf | NVIDIA accelerated computing platform NERFINISHED ⓘ |
| designedFor |
AI supercomputers
ⓘ
GPU-accelerated data centers ⓘ |
| developer | NVIDIA NERFINISHED ⓘ |
| enables |
real-time AI inference at scale
ⓘ
training trillion-parameter models ⓘ |
| family | NVIDIA GPU architectures NERFINISHED ⓘ |
| focus |
energy efficiency for data centers
ⓘ
high memory bandwidth for AI ⓘ scalability across multi-GPU systems ⓘ |
| intendedUse |
artificial intelligence workloads
ⓘ
data center workloads ⓘ high-performance computing workloads ⓘ |
| marketSegment |
AI accelerators
ⓘ
HPC accelerators ⓘ data center GPUs ⓘ |
| optimizationTarget |
AI performance per watt
ⓘ
inference latency ⓘ training throughput ⓘ |
| predecessor | NVIDIA Ada Lovelace architecture NERFINISHED ⓘ |
| successorOf | NVIDIA Ada Lovelace architecture NERFINISHED ⓘ |
| supports |
deep learning frameworks
ⓘ
generative AI workloads ⓘ large language model inference ⓘ large language model training ⓘ matrix multiplication acceleration ⓘ mixed-precision computation ⓘ tensor operations ⓘ |
| targetEnvironment |
cloud GPU instances
ⓘ
on-premises data centers ⓘ server GPUs ⓘ |
| technologyDomain |
accelerated computing
ⓘ
graphics processing units ⓘ high-performance computing ⓘ |
| useCase |
cloud computing
ⓘ
data analytics ⓘ enterprise AI ⓘ machine learning at scale ⓘ scientific computing ⓘ |
Referenced by (1)
Full triples — surface form annotated when it differs from this entity's canonical label.