GLM 5.3 Flash on 4x 8 GB/Confirmation 3
From CMP Wiki
More actions
| Confirms | GLM 5.3 Flash on 4x 8 GB |
|---|---|
| Result | Reproduced |
| By | @mountains1605 |
| Date | 2026-10-08 |
| Cards | 8 GB |
| Numbers | PP4: 90 to 140 tok/s decode, 1M context window, prefill around 3k tok/s (2026-10-04). 340 tok/s total at 8 concurrent on real agent work (2026-10-08). |
Four 8 GB cards on PCIe Gen2 x4 links (no x16 mod), PP4 layout. These are the field numbers for PP4 on x4 links, which the recipe itself has not measured. Source: @mountains1605, 2026-10-04
Source: @mountains1605, 2026-10-08