AMDAI GPU · OAM
MI300X
AMD Instinct MI300X is a CDNA 3 data-center accelerator with 192 GB HBM3 and 5.3 TB/s of memory bandwidth.
Specifications
Dense throughput, no sparsity| Architecture | CDNA 3 |
|---|---|
| Form factor | OAM |
| Process | TSMC 5 nm + 6 nm |
| FP16 / BF16 tensor (dense) | 1,307.4 TFLOPS |
| FP8 tensor (dense) | 2,614.9 TFLOPS |
| FP4 tensor (dense) | — |
| Memory | 192 GB HBM3 |
| Memory bandwidth | 5.3 TB/s |
| Power | 750 W |
| Launch | Dec 2023 |
AI index
Tensor × bandwidth100 × √((1,307.4 ÷ 989.4) × (5.3 ÷ 3.35)) = 144.6
The AI index is the geometric mean of dense 16-bit tensor throughput and memory bandwidth, each relative to the H100 SXM. It is a spec-sheet comparison, not a measured benchmark.