GPU FUNCTIONAL ARCHITECTURE SERIES — ENGLISH
Conceptual functional diagrams, not official physical teardown drawings.
The separated layers show roles; they do not represent a literal mechanical stack.
Only explicitly labeled component counts represent specifications.
GB200 NVL72 uses rack-wide figures; H100, H200 and B200 use the named SXM configurations.
AMD specifications use individual OAM accelerators.
Peak figures are not workload benchmark results.
System cooling, networking and software depend on the deployment.

NVIDIA GB200 NVL72
https://www.nvidia.com/en-us/data-center/gb200-nvl72/
https://www.nvidia.com/en-us/data-center/technologies/blackwell-architecture/

NVIDIA B200
https://docs.nvidia.com/enterprise-reference-architectures/hgx-ai-factory-h100-h200-b200/latest/components.html
https://www.nvidia.com/en-us/data-center/technologies/blackwell-architecture/

NVIDIA H200
https://www.nvidia.com/en-us/data-center/h200/
https://docs.nvidia.com/enterprise-reference-architectures/hgx-ai-factory-h100-h200-b200/latest/components.html

NVIDIA H100
https://www.nvidia.com/en-us/data-center/h100/
https://docs.nvidia.com/enterprise-reference-architectures/hgx-ai-factory-h100-h200-b200/latest/components.html

AMD Instinct MI300X
https://www.amd.com/en/products/accelerators/instinct/mi300/mi300x.html
https://rocm.docs.amd.com/en/latest/reference/gpu-arch/mi300.html

AMD Instinct MI355X
https://www.amd.com/en/products/accelerators/instinct/mi350/mi355x.html
https://rocm.docs.amd.com/en/latest/reference/gpu-arch/mi350.html