summaryrefslogtreecommitdiff
path: root/param.h
AgeCommit message (Expand)AuthorFilesLines
2021-01-27Merge branch 'develop' into msvcxoviat1-133/+511
2021-01-11Added definitions for GEMM_PREFERED_SIZE and SWITCH_RATIO to the POWER9 and P...Gordon Fossum1-0/+6
2020-12-10Merge pull request #3026 from martin-frbg/revert747Martin Kroeker1-2/+4
2020-12-09Add msa support for loongsongxw1-17/+31
2020-12-08Remove GEMM_DEFAULT_UNROLL_MN parameters for Haswell and ZEN (introduced in P...Martin Kroeker1-2/+4
2020-12-06Change comments to C style for compatibilityMartin Kroeker1-9/+9
2020-12-03POWER10: Update param.hRajalakshmi Srinivasaraghavan1-5/+34
2020-11-10Refs #2899Xianyi Zhang1-0/+39
2020-11-10Merge branch 'develop' into risc-vXianyi Zhang1-0/+4
2020-10-31POWER10: Change dgemm unroll factorsRajalakshmi Srinivasaraghavan1-0/+4
2020-10-16Merge branch 'develop' into risc-vZhang Xianyi1-27/+277
2020-10-15Add the support for RISC-V Vector.damonyu1-0/+78
2020-10-11Rename "HALF" and "sh" to "BFLOAT16" and "sb"Martin Kroeker1-16/+16
2020-08-13Enable COOPERLAKE build targetChen, Guobing1-0/+118
2020-08-11s390x/SGEMM: adjust default P and Q to multiples of MMarius Hillenbrand1-2/+2
2020-07-26ARM64: Add THUNDERX3T110 TargetAshwin Sekhar T K1-0/+29
2020-07-14Use POWER6 GEMM parameters on 32bit POWER8Martin Kroeker1-2/+12
2020-06-25powerpc: Optimized SHGEMM kernel for POWER10Rajalakshmi Srinivasaraghavan1-0/+13
2020-06-11powerpc: Add support for future processorRajalakshmi Srinivasaraghavan1-1/+1
2020-06-03Change PPCG4 CGEMM_M to match kernel changeMartin Kroeker1-1/+1
2020-05-20split cortex-a53 param to match 8x8 kernel张丹枫1-1/+30
2020-05-12s390x/Z14: Change register blocking for SGEMM to 16x4Marius Hillenbrand1-1/+1
2020-04-24Increase POWER8 ZGEMM_R and use same R values for POWER9Martin Kroeker1-1/+6
2020-04-18Typo fix in MIPS24K additionMartin Kroeker1-1/+1
2020-04-18Handle MIPS24K like P5600Martin Kroeker1-1/+7
2020-04-12Increase default BUFFER_SIZE on ARM, ZARCH and newer x86_64, add GEMM_R for P...Martin Kroeker1-9/+20
2020-03-30Merge pull request #2520 from wjc404/developMartin Kroeker1-2/+2
2020-03-20Update param.hwjc4041-2/+2
2020-02-29Merge pull request #2422 from wjc404/developMartin Kroeker1-8/+8
2020-02-29Add Neoverse-N1 coreAli Saidi1-0/+29
2020-02-27Change default RISC-V 64-bit corename to RISCV64_GENERICXianyi Zhang1-1/+1
2020-02-27Merge branch 'develop' into risc-vXianyi Zhang1-97/+296
2020-02-26Always assume server-class cpu count for TSV110 and EMAG8180Martin Kroeker1-1/+1
2020-02-19Add preliminary support for EMAG8180 ARMV8 processorMartin Kroeker1-1/+1
2020-02-16Update param.hwjc4041-8/+8
2020-02-04Update param.hwjc4041-1/+1
2020-02-03Update param.hwjc4041-4/+4
2020-01-22fix a few performance drop in some matrix size per data typeWang,Long1-4/+14
2020-01-13improve skylakex paralleled sgemm performancewjc4041-7/+2
2020-01-06optimize AVX2 SGEMMwjc4041-6/+6
2019-12-30Update param.hwjc4041-8/+8
2019-12-27Update param.hwjc4041-8/+8
2019-12-23Update param.hwjc4041-6/+6
2019-12-21Adjust Haswell ZGEMM blocking parameterswjc4041-2/+2
2019-11-28Update param.hwjc4041-3/+3
2019-11-17Use "generic" S/CGEMM unroll M on big-endian PPC970Martin Kroeker1-0/+8
2019-11-06Merge pull request #2300 from wjc404/developMartin Kroeker1-2/+2
2019-11-02Add files via uploadwjc4041-1/+1
2019-11-01update sgemm_q on skylakex cpuswjc4041-1/+1
2019-10-25Remove special parameter set for obsolete IOS/ARMV8 workaroundMartin Kroeker1-34/+0
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-29 05:17:01 +0000