775
93%

Ran 27 Aug 2024 07:43PM UTC

Jobs 1

Files 165

Run time 1min

Badge

Embed ▾

Committed 27 Aug 2024 06:29PM UTC coverage: 93.386% (-0.02%) from 93.406%

Build # 775

Build Type

push

jenkins

Committed by

web-flow

Commit Message

Use cublas<t>matinvBatched() for N <= 32 (#739)

Use the cublas<t>matinvBatched() family of functions to invert linear systems
of size N <= 32. This has two advantages over the more general pair of
getrfBatched() and getriBatched() functions:

1. Higher performance with the single kernel than with split kernels.
2. The matinv functions support in-place transforms and do not modify the
input in the case of out-of-place transforms, so we do not need a temporary
input work buffer if the input is a tensor view.

Run Details

6622 of 7091 relevant lines covered (93.39%)

20869732.79 hits per line

Jobs

ID	Job ID	Ran	Files	Coverage
1	775.1	27 Aug 2024 07:43PM UTC	0	93.39

Source Files on build 775

Detailed source file information is not available for this build.

NVIDIA / MatX / 775
93%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 775

NVIDIA / MatX / 775 93%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 775

NVIDIA / MatX / 775
93%

README BADGES
x