|
1 | 1 | OpenBLAS ChangeLog
|
| 2 | +==================================================================== |
| 3 | +Version 0.3.19 |
| 4 | + 19-Dec-2021 |
| 5 | + |
| 6 | + general: |
| 7 | + - reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 |
| 8 | + - fixed a potential thread race in the thread buffer reallocation routines |
| 9 | + that were introduced in 0.3.18 |
| 10 | + - fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE |
| 11 | + - fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG |
| 12 | + - made automatic library suffix for CMAKE builds with INTERFACE64 available |
| 13 | + to CBLAS-only builds |
| 14 | + |
| 15 | +x86_64: |
| 16 | + - DYNAMIC_ARCH builds now fall back to the cpu with most similar capabilities |
| 17 | + when an unknown CPUID is encountered, instead of defaulting to Prescott |
| 18 | + - added cpu detection for Intel Alder Lake |
| 19 | + - added cpu detection for Intel Sapphire Rapids |
| 20 | + - added an optimized SBGEMM kernel for Sapphire Rapids |
| 21 | + - fixed DYNAMIC_ARCH builds on OSX with CMAKE |
| 22 | + - worked around DYNAMIC_ARCH builds made on Sandybridge failing on SkylakeX |
| 23 | + - fixed missing thread initialization for static builds on Windows/MSVC |
| 24 | + - fixed an excessive read in ZSYMV |
| 25 | + |
| 26 | +POWER: |
| 27 | + - added support for POWER10 in big-endian mode |
| 28 | + - added support for building with CMAKE |
| 29 | + - added optimized SGEMM and DGEMM kernels for small matrix sizes |
| 30 | + |
| 31 | +ARMV8: |
| 32 | + - added basic support and cputype detection for Fujitsu A64FX |
| 33 | + - added a generic ARMV8SVE target |
| 34 | + - added SVE-enabled SGEMM and DGEMM kernels for ARMV8SVE and A64FX |
| 35 | + - added optimized CGEMM and ZGEMM kernels for Cortex A53 and A55 cpus |
| 36 | + - fixed cpuid detection for Apple M1 and improved performance |
| 37 | + - improved compiler flag setting in CMAKE builds |
| 38 | + |
| 39 | +RISCV64: |
| 40 | + - fixed improper initialization in CSCAL/ZSCAL for strided access patterns |
| 41 | + |
| 42 | +MIPS: |
| 43 | + - added a GENERIC target for MIPS32 |
| 44 | + - added support for cross-compiling to MIPS32 on x86_64 using CMAKE |
| 45 | + |
| 46 | +MIPS64: |
| 47 | + - fixed misdetection of MSA capability |
| 48 | + |
2 | 49 | ====================================================================
|
3 | 50 | Version 0.3.18
|
4 | 51 | 02-Oct-2021
|
|
0 commit comments