Your GeForce runs AI at half speed. This fixes it for 20/30/40 series cards. Up to 2× FP16 / BF16 / TF32 tensor throughput
C
docs: add RTX 3090 reports #148, #152, #155, #158 to the tested cards table
viwe-monai2e4af94437db
diff --git a/README.md b/README.mdindex 53c8f77..ff7175b 100644--- a/README.md+++ b/README.md@@ -28,7 +28,7 @@ whether it worked or not.| Chip | Card | PCI ID | Result | BF16 GEMM gain | Reports || ----- | --------------------------------- | ----------- | ----------------------------------- | -------------- | ------- || GA102 | RTX 3090 Ti | `10de:2203` | Unlocked | +68–80% | [#1](https://github.com/Zagitta/gpu-unlock/issues/1) |-| GA102 | RTX 3090 | `10de:2204` | Unlocked | +25–67% | [#5](https://github.com/Zagitta/gpu-unlock/issues/5), [#9](https://github.com/Zagitta/gpu-unlock/issues/9), [#10](https://github.com/Zagitta/gpu-unlock/issues/10), [#17](https://github.com/Zagitta/gpu-unlock/issues/17), [#25](https://github.com/Zagitta/gpu-unlock/issues/25), [#27](https://github.com/Zagitta/gpu-unlock/issues/27), [#33](https://github.com/Zagitta/gpu-unlock/issues/33) |+| GA102 | RTX 3090 | `10de:2204` | Unlocked | +25–67% | [#5](https://github.com/Zagitta/gpu-unlock/issues/5), [#9](https://github.com/Zagitta/gpu-unlock/issues/9), [#10](https://github.com/Zagitta/gpu-unlock/issues/10), [#17](https://github.com/Zagitta/gpu-unlock/issues/17), [#25](https://github.com/Zagitta/gpu-unlock/issues/25), [#27](https://github.com/Zagitta/gpu-unlock/issues/27), [#33](https://github.com/Zagitta/gpu-unlock/issues/33), [#148](https://github.com/Zagitta/gpu-unlock/issues/148), [#152](https://github.com/Zagitta/gpu-unlock/issues/152), [#155](https://github.com/Zagitta/gpu-unlock/issues/155), [#158](https://github.com/Zagitta/gpu-unlock/issues/158) || GA102 | RTX 3080 Ti 20 GB (OEM, GA102-250) | `10de:2205` | Unlock completes, BF16 stays at half rate | — | [#28](https://github.com/Zagitta/gpu-unlock/issues/28) || GA102 | RTX 3080 10 GB | `10de:2206` | Unlocked | +70–73% | [#14](https://github.com/Zagitta/gpu-unlock/issues/14) || GA102 | RTX 3080 20 GB (memory upgrade) | `10de:2206` | Unlocked | +56–59% | [#10](https://github.com/Zagitta/gpu-unlock/issues/10), [#15](https://github.com/Zagitta/gpu-unlock/issues/15) |