@@ -54,10 +54,15 @@ Building OpenBLAS requires the following to be installed:
54
54
55
55
Simply invoking ` make ` (or ` gmake ` on BSD) will detect the CPU automatically.
56
56
To set a specific target CPU, use ` make TARGET=xxx ` , e.g. ` make TARGET=NEHALEM ` .
57
- The full target list is in the file ` TargetList.txt ` . For building with ` cmake ` , the
58
- usual conventions apply, i.e. create a build directory either underneath the toplevel
59
- OpenBLAS source directory or separate from it, and invoke ` cmake ` there with the path
60
- to the source tree and any build options you plan to set.
57
+ The full target list is in the file ` TargetList.txt ` , other build optionss are documented in Makefile.rule and
58
+ can either be set there (typically by removing the comment character from the respective line), or used on the
59
+ ` make ` command line.
60
+ Note that when you run ` make install ` after building, you need to repeat all command line options you provided to ` make `
61
+ in the build step, as some settings like the supported maximum number of threads are automatically derived from the
62
+ build host by default, which might not be what you want.
63
+ For building with ` cmake ` , the usual conventions apply, i.e. create a build directory either underneath the toplevel
64
+ OpenBLAS source directory or separate from it, and invoke ` cmake ` there with the path to the source tree and any
65
+ build options you plan to set.
61
66
62
67
### Cross compile
63
68
@@ -117,7 +122,7 @@ Use `PREFIX=` when invoking `make`, for example
117
122
``` sh
118
123
make install PREFIX=your_installation_directory
119
124
```
120
-
125
+ (along with all options you added on the ` make ` command line in the preceding build step)
121
126
The default installation directory is ` /opt/OpenBLAS ` .
122
127
123
128
## Supported CPUs and Operating Systems
@@ -137,7 +142,7 @@ Please read `GotoBLAS_01Readme.txt` for older CPU models already supported by th
137
142
- ** AMD Bulldozer** : x86-64 ?GEMM FMA4 kernels. (Thanks to Werner Saar)
138
143
- ** AMD PILEDRIVER** : Uses Bulldozer codes with some optimizations.
139
144
- ** AMD STEAMROLLER** : Uses Bulldozer codes with some optimizations.
140
- - ** AMD ZEN** : Uses Haswell codes with some optimizations.
145
+ - ** AMD ZEN** : Uses Haswell codes with some optimizations for Zen 2/3 (use SkylakeX for Zen4)
141
146
142
147
#### MIPS32
143
148
@@ -169,13 +174,16 @@ Please read `GotoBLAS_01Readme.txt` for older CPU models already supported by th
169
174
- ** TSV110** : Optimized some Level-3 helper functions
170
175
- ** EMAG 8180** : preliminary support based on A57
171
176
- ** Neoverse N1** : (AWS Graviton2) preliminary support
172
- - ** Apple Vortex** : preliminary support based on ARMV8
177
+ - ** Neoverse V1** : (AWS Graviton3) optimized Level-3 BLAS
178
+ - ** Apple Vortex** : preliminary support based on ThunderX2/3
179
+ - ** A64FX** : preliminary support, optimized Level-3 BLAS
180
+ - ** ARMV8SVE** : any ARMV8 cpu with SVE extensions
173
181
174
182
#### PPC/PPC64
175
183
176
184
- ** POWER8** : Optimized BLAS, only for PPC64LE (Little Endian), only with ` USE_OPENMP=1 `
177
185
- ** POWER9** : Optimized Level-3 BLAS (real) and some Level-1,2. PPC64LE with OpenMP only.
178
- - ** POWER10** :
186
+ - ** POWER10** : Optimized Level-3 BLAS including SBGEMM and some Level-1,2.
179
187
180
188
#### IBM zEnterprise System
181
189
0 commit comments