GPUUnlock

The dp4a integer dot product

Four INT8 multiply-accumulates per lane. An integer path, so the SM rate fuse leaves it untouched at 1.00×.

dp4a · TF/s · never rate-limited

1rows
0changed
1.00×peak gain

How this mode was measured

Register-resident dp4a.s32.s32 issue rate, 8 independent accumulator chains per warp, best of three timed runs

Every card measured on this mode

ModeStockUnlockedGainCard
dp4a164.50 TOP/s164.10 TOP/s1.00×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