Scalar FP32 fused multiply-add

The CUDA-core path, not the tensor cores. Included as a control: it holds 1.00× on every card, showing the unlock is scoped to tensor paths.

ffma · TF/s · never rate-limited

1rows

0changed

1.00×peak gain

How this mode was measured

Every card measured on this mode

Mode	Stock	Unlocked	Gain	Card
`ffma`	79.10 TFLOP/s	79.10 TFLOP/s	1.00×	GeForce RTX 4090 · 128 SMs

Same card, same driver, same binary. The unlocked figure is the only variable.

Which cards this covers

Only cards with a same-card measurement on this mode appear in the table. A card missing from it has not been measured on this path, which is not the same as failing.

Other measured modes

FP16 with an FP16 accumulator · 1
FP16 with an FP32 accumulator · 2
BF16 with an FP32 accumulator · 1
TF32 with an FP32 accumulator · 1
INT8 with an INT32 accumulator · 1
INT4 with an INT32 accumulator · 1
FP8 E4M3 through CUTLASS · 1
The dp4a integer dot product · 1
FP16/FP32 as a cuBLAS GEMM · 4