SecondSourceJudgment rebuilt from primary sources

NVIDIA has added an algorithm to the new version of cuBLAS, its GPU linear-algebra library, that emulates double-precision matrix multiplication on integer tensor cores. NVIDIA's own tests show it up to 5.5 times as fast as native double precision on Vera Rubin NVL72.

Concept illustration: at the lower left a person with dark hair and a black sweater, seen from behind with a pale chair back at their shoulders, faces a large board whose black top bar is lettered "CUBLAS NVIDIA" next to a green square icon. On the board stand two identical dark server racks: the left one, labelled "NATIVE UNITS" below, has a short stack of beige papers beside it, and the right one, labelled "OZAKI-II" below, has a much taller red stack of papers beside it, and each stack is topped by a sheet with grey lines and a turned-up corner. Printed on the image: Up to 5.5x New cuBLAS algorithm's speed vs native double precision on Vera Rubin NVL72, NVIDIA's own test.

NVIDIA says integer tensor cores built for AI can stand in for double-precision units in this library routine. The 5.5 times comes from NVIDIA's own test against Vera Rubin's native double precision. The new algorithm, Ozaki-II, splits each double-precision number into several integer slices, multiplies them on tensor cores and reassembles the result exactly. NVIDIA writes that the older Ozaki-I needs roughly twice as many tensor-core operations to reach the same precision. The library picks parameters automatically so precision stays no lower than standard double precision, and falls back to native double-precision instructions when it can't. The RTX PRO 6000 has very few native double-precision units, and the new algorithm makes it 10 to 23 times as fast. Even so, with the new algorithm the RTX PRO 6000's double-precision peak is only 37 trillion operations per second, against 175 trillion on Vera Rubin NVL72 (NVIDIA Technical Blog, 2026-11-13). How we checked →

Source and checks

This item ran in the brief of Nov 14, 2026, alongside 14 other items. How we checked it, the quote checks and what was still missing that day are on that issue's How we checked page.

Cite this item

SecondSource (2026). "NVIDIA has added an algorithm to the new version of cuBLAS, its GPU linear-algebra library, that emulates double-precision matrix multiplication on integer tensor cores. NVIDIA's own tests show it up to 5.5 times as fast as native double precision on Vera Rubin NVL72.." https://secondsource.io/en/news/2026-11-14-5/. Accessed access date.