Draft:Blackwell (microarchitecture)
Draft article not currently submitted for review.
This is a draft Articles for creation (AfC) submission. It is not currently pending review. While there are no deadlines, abandoned drafts may be deleted after six months. To edit the draft click on the "Edit" tab at the top of the window. To be accepted, a draft should:
It is strongly discouraged to write about yourself, your business or employer. If you do so, you must declare it. Where to get help
How to improve a draft
You can also browse Wikipedia:Featured articles and Wikipedia:Good articles to find examples of Wikipedia's best writing on topics similar to your proposed article. Improving your odds of a speedy review To improve your odds of a faster review, tag your draft with relevant WikiProject tags using the button below. This will let reviewers know a new draft has been submitted in their area of interest. For instance, if you wrote about a female astronomer, you would want to add the Biography, Astronomy, and Women scientists tags. Editor resources
Last edited by 2001:E68:5423:95A3:3D8D:A3B:8D1:CABB (talk | contribs) 2 months ago. (Update) |
Launched | March 18, 2024 |
---|---|
Designed by | Nvidia |
Manufactured by | |
Fabrication process | TSMC 4NP |
Specifications | |
L1 cache | KB (per SM) |
L2 cache | MB |
Memory support | HBM3E |
PCIe support | PCIe 6.0 |
Supported Graphics APIs | |
DirectX | DirectX 12 Ultimate (Feature Level 12_2) |
Direct3D | Direct3D 12 |
Shader Model | Shader Model 6.8 |
OpenCL | OpenCL 3.0 |
OpenGL | OpenGL 4.6 |
CUDA | Compute Capability 10.x |
Vulkan | Vulkan 1.3 |
Supported Compute APIs | |
CUDA | CUDA Toolkit |
DirectCompute | Yes |
History | |
Predecessor | Ada Lovelace (consumer) Hopper (datacenter) |
Blackwell is a graphics processing unit (GPU) microarchitecture developed by Nvidia as the successor to Hopper microachitecture and Ada Lovelace microarchitecture.
Named after statistician and mathematician David Blackwell, the Blackwell architecture was leaked in 2022 and B100 & B40 was officially revealed in October 2023 in an official Nvidia roadmap in an official Nvidia investor presentation for investors[1] and was officially announced at Nvidia GTC 2024 keynote on March 18, 2024.[2]
Details[edit]
Architectural improvements of the Blackwell architecture include the following:[3]
- CUDA Compute Capability 10.0 for B100 & B200
- High Bandwidth Memory 3E (HBM3E) for B100 & B200
- Fifth Generation Tensor Cores[4]
- Second Generation Transformer Engine
- NVLink 5.0 for B100 & B200
References[edit]
Citations[edit]
- ^ https://investor.nvidia.com/events-and-presentations/presentations/presentation-details/2023/NVIDIA-Investor-Presentation-October-2023/default.aspx
- ^ https://nvidianews.nvidia.com/news/nvidia-blackwell-platform-arrives-to-power-a-new-era-of-computing
- ^ https://www.nvidia.com/en-us/data-center/technologies/blackwell-architecture/
- ^ https://www.nvidia.com/en-us/data-center/tensor-cores/#blackwell