projects / platform / upstream / openblas.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: cb097be) | patch
dgemv optimization for POWER10
authorRajalakshmi Srinivasaraghavan <rajis@linux.ibm.com>
Wed, 29 Jul 2020 23:59:32 +0000 (18:59 -0500)
committerRajalakshmi Srinivasaraghavan <rajis@linux.ibm.com>
Wed, 29 Jul 2020 23:59:32 +0000 (18:59 -0500)
commitf77b6a83f4c20ca4e4769a999a69b0f47f7f4bb1
tree31f0298873b3455977cb4c7561308e965b836c47tree | snapshot
parentcb097beba27dd2b0c6b3e941ad8f263fcc808015commit | diff
dgemv optimization for POWER10

Making use of new vector pair POWER10 instructions in dgemv_n and dgemv_t.
Also adding a new block 4x128 to make use of Matrix-Multiply Assist (MMA)
feature introduced in POWER ISA v3.1.  Tested on simulator and there
are no new test failures.
kernel/power/KERNEL.POWER10 diff | blob | history
kernel/power/dgemv_n_microk_power10.c [new file with mode: 0644] blob
kernel/power/dgemv_n_power10.c [new file with mode: 0644] blob
kernel/power/dgemv_t_power10.c [new file with mode: 0644] blob
Domain: Machine Learning / ML Framework;