Branch:
release-0.3.0
Travisppc
arm_soft_fp_abi
armv7
bulldozer
ck860v
cmake
develop
fix3126
gh-pages
haswell
integer_datatype
issue2419
issue3321
jenkins
loongson3a
loongson3b
master
optimized_for_deeplearning
piledriver
power8
release-0.2.21
release-0.3.0
revert-1233-cpuid-fix
revert-1239-cgroups
revert-1254-xbmv_range
revert-1333-haswell32
revert-1464-issue1461
revert-1510-aa/travis-macos
revert-1600-noyield
revert-1642-develop
revert-1798-cmake-avx512
revert-2051-issue2048
revert-2142-issue1912-2
revert-2272-thread-sqrt-of-negative
revert-2325-travisosx
revert-2566-azurewin
revert-3118-issue3018-2
revert-3518-elbrus
revert-4103-windows_perf
revert-4153-dgemv
revert-4628-CI-c910v-mips64-loongarch64
revert-4920-issue4917
revert-5180-openmp_use_cmake
revert-5251-issue5250
risc-v
sandybridge
small_matrices
timeout-riscv-ci
utest-riscv
z13
zhbmv_smp
v0.1.0
v0.1.1
v0.1alpha1
v0.1alpha2
v0.1alpha2.1
v0.1alpha2.2
v0.1alpha2.3
v0.1alpha2.4
v0.1alpha2.5
v0.2.0
v0.2.1
v0.2.10
v0.2.10.rc1
v0.2.10.rc2
v0.2.11
v0.2.12
v0.2.13
v0.2.14
v0.2.15
v0.2.16
v0.2.16.rc1
v0.2.17
v0.2.18
v0.2.19
v0.2.2
v0.2.20
v0.2.3
v0.2.4
v0.2.5
v0.2.6
v0.2.7
v0.2.8
v0.2.9
v0.2.9.rc1
v0.2.9.rc2
v0.3.0
v0.3.1
v0.3.10
v0.3.11
v0.3.12
v0.3.13
v0.3.14
v0.3.15
v0.3.16
v0.3.17
v0.3.18
v0.3.19
v0.3.2
v0.3.20
v0.3.21
v0.3.22
v0.3.23
v0.3.24
v0.3.25
v0.3.26
v0.3.27
v0.3.28
v0.3.29
v0.3.3
v0.3.30
v0.3.4
v0.3.5
v0.3.6
v0.3.7
v0.3.8
v0.3.9
1 Commits (release-0.3.0)
1 Commits (release-0.3.0)
| Author | SHA1 | Message | Date |
|---|---|---|---|
 Chen, Guobing
|
deaeb6c5b8 |
Add bfloat16 based dot and conversion with single/double
1. Added bfloat16 based dot as new API: shdot
2. Implemented generic kernel and cooperlake-specific (AVX512-BF16) kernel for shdot
3. Added 4 conversion APIs for bfloat16 data type <=> single/double: shstobf16 shdtobf16 sbf16tos dbf16tod
shstobf16 -- convert single float array to bfloat16 array
shdtobf16 -- convert double float array to bfloat16 array
sbf16tos -- convert bfloat16 array to single float array
dbf16tod -- convert bfloat16 array to double float array
4. Implemented generic kernels for all 4 conversion APIs, and cooperlake-specific kernel for shstobf16 and shdtobf16
5. Update level1 thread facilitate functions and macros to support multi-threading for these new APIs
6. Fix Cooperlake platform detection/specify issue when under dynamic-arch building
7. Change the typedef of bfloat16 from unsigned short to more strict uint16_t
Signed-off-by: Chen, Guobing <guobing.chen@intel.com>
|
5 years ago |