From 12f4e2b5a06a37d70ac43a2a9894bf3af2d28a9b Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 14:02:20 +0200 Subject: [PATCH 01/23] try 32bit manylinux build --- azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 7941bf4633..1eb1a9b074 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -22,11 +22,11 @@ jobs: vmImage: 'ubuntu-latest' steps: - script: | - echo "FROM quay.io/pypa/manylinux1_x86_64 + echo "FROM quay.io/pypa/manylinux1_686 COPY . /tmp/openblas RUN cd /tmp/openblas && \ COMMON_FLAGS='DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32' && \ - BTYPE='BINARY=64' CC=gcc && \ + BTYPE='BINARY=32' CC=gcc && \ make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE && \ make -C test $COMMON_FLAGS $BTYPE && \ make -C ctest $COMMON_FLAGS $BTYPE && \ From 779c9b27af7914c43b843b4a7418aa24ba553360 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 14:47:30 +0200 Subject: [PATCH 02/23] Update azure-pipelines.yml --- azure-pipelines.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 1eb1a9b074..dbfddb72d4 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -22,7 +22,7 @@ jobs: vmImage: 'ubuntu-latest' steps: - script: | - echo "FROM quay.io/pypa/manylinux1_686 + echo "FROM quay.io/pypa/manylinux2014_i686 COPY . /tmp/openblas RUN cd /tmp/openblas && \ COMMON_FLAGS='DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32' && \ From 1df0b9237992c131b203b5198b9bc44b42b24dc4 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 15:50:46 +0200 Subject: [PATCH 03/23] Update azure-pipelines.yml --- azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index dbfddb72d4..42e6fb81f9 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -25,9 +25,9 @@ jobs: echo "FROM quay.io/pypa/manylinux2014_i686 COPY . /tmp/openblas RUN cd /tmp/openblas && \ - COMMON_FLAGS='DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32' && \ + COMMON_FLAGS='TARGET=NEHALEM NUM_THREADS=32' && \ BTYPE='BINARY=32' CC=gcc && \ - make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE && \ + make $COMMON_FLAGS $BTYPE && \ make -C test $COMMON_FLAGS $BTYPE && \ make -C ctest $COMMON_FLAGS $BTYPE && \ make -C utest $COMMON_FLAGS $BTYPE" > Dockerfile From ba138375383ba492551ea058c3afd8fcdbe6acc5 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 17:26:26 +0200 Subject: [PATCH 04/23] Update azure-pipelines.yml --- azure-pipelines.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 42e6fb81f9..e06c21c80f 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -22,7 +22,7 @@ jobs: vmImage: 'ubuntu-latest' steps: - script: | - echo "FROM quay.io/pypa/manylinux2014_i686 + echo "FROM quay.io/pypa/manylinux1_i686 COPY . /tmp/openblas RUN cd /tmp/openblas && \ COMMON_FLAGS='TARGET=NEHALEM NUM_THREADS=32' && \ From ff1c2074f43e373b6e289b328f3cb3ce29bb2938 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 17:55:21 +0200 Subject: [PATCH 05/23] Update azure-pipelines.yml --- azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index e06c21c80f..9f18a5dd51 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -22,10 +22,10 @@ jobs: vmImage: 'ubuntu-latest' steps: - script: | - echo "FROM quay.io/pypa/manylinux1_i686 + echo "FROM quay.io/pypa/manylinux2014_i686 COPY . /tmp/openblas RUN cd /tmp/openblas && \ - COMMON_FLAGS='TARGET=NEHALEM NUM_THREADS=32' && \ + COMMON_FLAGS='TARGET=PENRYN NUM_THREADS=32' && \ BTYPE='BINARY=32' CC=gcc && \ make $COMMON_FLAGS $BTYPE && \ make -C test $COMMON_FLAGS $BTYPE && \ From ade9023cd23cc1e6af0b596456a13cd53eea32df Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 18:14:06 +0200 Subject: [PATCH 06/23] Update azure-pipelines.yml --- azure-pipelines.yml | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 9f18a5dd51..2d2eef981b 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -27,10 +27,8 @@ jobs: RUN cd /tmp/openblas && \ COMMON_FLAGS='TARGET=PENRYN NUM_THREADS=32' && \ BTYPE='BINARY=32' CC=gcc && \ - make $COMMON_FLAGS $BTYPE && \ - make -C test $COMMON_FLAGS $BTYPE && \ - make -C ctest $COMMON_FLAGS $BTYPE && \ - make -C utest $COMMON_FLAGS $BTYPE" > Dockerfile + make TARGET=NEHALEM BINARY=32 && \ + make -C utest TARGET=NEHALEM BINARY=32" > Dockerfile docker build . displayName: Run manylinux1 docker build - job: Intel_SDE_skx From c5e331c109cdb3b2f8744306cdf39eedfab8b7ee Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 18:34:08 +0200 Subject: [PATCH 07/23] Update azure-pipelines.yml --- azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 2d2eef981b..7d48200280 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -27,8 +27,8 @@ jobs: RUN cd /tmp/openblas && \ COMMON_FLAGS='TARGET=PENRYN NUM_THREADS=32' && \ BTYPE='BINARY=32' CC=gcc && \ - make TARGET=NEHALEM BINARY=32 && \ - make -C utest TARGET=NEHALEM BINARY=32" > Dockerfile + make TARGET=NEHALEM BINARY=32 DEBUG=1&& \ + make -C utest TARGET=NEHALEM BINARY=32 DEBUG=1" > Dockerfile docker build . displayName: Run manylinux1 docker build - job: Intel_SDE_skx From f1ed097ae1aa8e8d670cc65a41b3b04a32da26c4 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 19:24:39 +0200 Subject: [PATCH 08/23] Update azure-pipelines.yml --- azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 7d48200280..c0a5283852 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -27,8 +27,8 @@ jobs: RUN cd /tmp/openblas && \ COMMON_FLAGS='TARGET=PENRYN NUM_THREADS=32' && \ BTYPE='BINARY=32' CC=gcc && \ - make TARGET=NEHALEM BINARY=32 DEBUG=1&& \ - make -C utest TARGET=NEHALEM BINARY=32 DEBUG=1" > Dockerfile + make TARGET=NEHALEM BINARY=32 && \ + make -C utest TARGET=NEHALEM BINARY=32 " > Dockerfile docker build . displayName: Run manylinux1 docker build - job: Intel_SDE_skx From 8bf3c49baf2c6c5d43d5c9422200586a02ca5f23 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 19:25:48 +0200 Subject: [PATCH 09/23] Update test_potrs.c --- utest/test_potrs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/utest/test_potrs.c b/utest/test_potrs.c index bcb1f753b0..a9f450e981 100644 --- a/utest/test_potrs.c +++ b/utest/test_potrs.c @@ -32,7 +32,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. **********************************************************************************/ #include "openblas_utest.h" -#if defined(ARCH_LOONGARCH64) +#if defined(ARCH_LOONGARCH64) || defined(BINARY32) #pragma GCC optimize("no-gcse") #endif /* From 1c54b8cce94d8e255de6c9ca49b4a2edf7287c0a Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 19:42:05 +0200 Subject: [PATCH 10/23] Update test_potrs.c --- utest/test_potrs.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/utest/test_potrs.c b/utest/test_potrs.c index a9f450e981..bcdbbab107 100644 --- a/utest/test_potrs.c +++ b/utest/test_potrs.c @@ -32,9 +32,10 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. **********************************************************************************/ #include "openblas_utest.h" -#if defined(ARCH_LOONGARCH64) || defined(BINARY32) +#if defined(ARCH_LOONGARCH64) #pragma GCC optimize("no-gcse") #endif +#pragma GCC optimize("O0") /* void BLASFUNC(cpotrf)(char*, BLASINT*, complex float*, BLASINT*, BLASINT*); void BLASFUNC(zpotrs_(char*, BLASINT*, BLASINT*, complex double*, From 4f7ae3865a9f7cfe16202cb692fdcc16000650f5 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 19:56:59 +0200 Subject: [PATCH 11/23] Update zscal.c --- kernel/arm/zscal.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/arm/zscal.c b/kernel/arm/zscal.c index b210f9af30..d00e748f13 100644 --- a/kernel/arm/zscal.c +++ b/kernel/arm/zscal.c @@ -33,7 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * TEST : OK * **************************************************************************************/ - +#pragma GCC optimize("O0" #include "common.h" // The c/zscal_k function is called not only by cblas_c/zscal but also by other upper-level interfaces. From 5e9ecb4b8618b7b2852f8b2349e62653aa110068 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 20:30:33 +0200 Subject: [PATCH 12/23] Update azure-pipelines.yml --- azure-pipelines.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/azure-pipelines.yml b/azure-pipelines.yml index c0a5283852..55a4c6e29f 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -27,8 +27,8 @@ jobs: RUN cd /tmp/openblas && \ COMMON_FLAGS='TARGET=PENRYN NUM_THREADS=32' && \ BTYPE='BINARY=32' CC=gcc && \ - make TARGET=NEHALEM BINARY=32 && \ - make -C utest TARGET=NEHALEM BINARY=32 " > Dockerfile + make TARGET=NEHALEM BINARY=32 USE_THREAD=0 && \ + make -C utest TARGET=NEHALEM BINARY=32 USE_THREAD=0" > Dockerfile docker build . displayName: Run manylinux1 docker build - job: Intel_SDE_skx From 1290c317b672ea279738f884c1a4dbc4b1607077 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 21:28:04 +0200 Subject: [PATCH 13/23] Update potrf_L_single.c --- lapack/potrf/potrf_L_single.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lapack/potrf/potrf_L_single.c b/lapack/potrf/potrf_L_single.c index 0edadf3213..116721e146 100644 --- a/lapack/potrf/potrf_L_single.c +++ b/lapack/potrf/potrf_L_single.c @@ -35,7 +35,7 @@ /* interpreted as representing official policies, either expressed */ /* or implied, of The University of Texas at Austin. */ /*********************************************************************/ - +#pragma GCC optimize("O0") #include #include "common.h" From 4964105ebecaa6c6fe007e57598d35fbf8300137 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 21:29:04 +0200 Subject: [PATCH 14/23] Update zpotf2_L.c --- lapack/potf2/zpotf2_L.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lapack/potf2/zpotf2_L.c b/lapack/potf2/zpotf2_L.c index d265c5d884..aa4786f50d 100644 --- a/lapack/potf2/zpotf2_L.c +++ b/lapack/potf2/zpotf2_L.c @@ -35,7 +35,7 @@ /* interpreted as representing official policies, either expressed */ /* or implied, of The University of Texas at Austin. */ /*********************************************************************/ - +äpragma GCC optimize("O0") #include #include #include "common.h" From 70c430c4b5ddce53c3f0dce506bbeb5e686c55b4 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 21:29:47 +0200 Subject: [PATCH 15/23] Update zpotf2_U.c --- lapack/potf2/zpotf2_U.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lapack/potf2/zpotf2_U.c b/lapack/potf2/zpotf2_U.c index 594a90ad3f..a88a6d2053 100644 --- a/lapack/potf2/zpotf2_U.c +++ b/lapack/potf2/zpotf2_U.c @@ -35,7 +35,7 @@ /* interpreted as representing official policies, either expressed */ /* or implied, of The University of Texas at Austin. */ /*********************************************************************/ - +#pragma GCC optimize("O0") #include #include #include "common.h" From d33a9c5395c9bcc7e2805a34d7bb8a6bbb9ca491 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Mon, 16 Jun 2025 22:37:05 +0200 Subject: [PATCH 16/23] Update zpotf2_L.c --- lapack/potf2/zpotf2_L.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lapack/potf2/zpotf2_L.c b/lapack/potf2/zpotf2_L.c index aa4786f50d..11167031a4 100644 --- a/lapack/potf2/zpotf2_L.c +++ b/lapack/potf2/zpotf2_L.c @@ -35,7 +35,7 @@ /* interpreted as representing official policies, either expressed */ /* or implied, of The University of Texas at Austin. */ /*********************************************************************/ -äpragma GCC optimize("O0") +#pragma GCC optimize("O0") #include #include #include "common.h" From f4f0f44703585c0b8fad7cb506a49a168dc0e2c1 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Tue, 17 Jun 2025 08:36:50 +0200 Subject: [PATCH 17/23] Update zscal.c --- kernel/arm/zscal.c | 43 ++++++++++++++++++++++++++++++------------- 1 file changed, 30 insertions(+), 13 deletions(-) diff --git a/kernel/arm/zscal.c b/kernel/arm/zscal.c index d00e748f13..51509ecdae 100644 --- a/kernel/arm/zscal.c +++ b/kernel/arm/zscal.c @@ -51,22 +51,39 @@ int CNAME(BLASLONG n, BLASLONG dummy0, BLASLONG dummy1, FLOAT da_r,FLOAT da_i, F inc_x2 = 2 * inc_x; if (dummy2 == 0) { - for (i = 0; i < n; i++) + for ( i=0; i Date: Tue, 17 Jun 2025 11:22:06 +0200 Subject: [PATCH 18/23] Update zscal.c --- kernel/arm/zscal.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/kernel/arm/zscal.c b/kernel/arm/zscal.c index 51509ecdae..ca1696df91 100644 --- a/kernel/arm/zscal.c +++ b/kernel/arm/zscal.c @@ -33,7 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * TEST : OK * **************************************************************************************/ -#pragma GCC optimize("O0" + #include "common.h" // The c/zscal_k function is called not only by cblas_c/zscal but also by other upper-level interfaces. @@ -48,9 +48,10 @@ int CNAME(BLASLONG n, BLASLONG dummy0, BLASLONG dummy1, FLOAT da_r,FLOAT da_i, F if ((n <= 0) || (inc_x <= 0)) return(0); - +printf("arm zscal kernel\n"); inc_x2 = 2 * inc_x; if (dummy2 == 0) { + printf("arm zscal kernel, dummy2 is zero\n"); for ( i=0; i Date: Tue, 17 Jun 2025 06:04:48 -0700 Subject: [PATCH 19/23] Add files via upload --- kernel/arm/zscal.c | 47 ++++++++++++++-------------------------------- 1 file changed, 14 insertions(+), 33 deletions(-) diff --git a/kernel/arm/zscal.c b/kernel/arm/zscal.c index ca1696df91..b210f9af30 100644 --- a/kernel/arm/zscal.c +++ b/kernel/arm/zscal.c @@ -48,46 +48,27 @@ int CNAME(BLASLONG n, BLASLONG dummy0, BLASLONG dummy1, FLOAT da_r,FLOAT da_i, F if ((n <= 0) || (inc_x <= 0)) return(0); -printf("arm zscal kernel\n"); + inc_x2 = 2 * inc_x; if (dummy2 == 0) { - printf("arm zscal kernel, dummy2 is zero\n"); - for ( i=0; i Date: Tue, 17 Jun 2025 15:10:38 +0200 Subject: [PATCH 20/23] Update zscal.c --- kernel/arm/zscal.c | 1 + 1 file changed, 1 insertion(+) diff --git a/kernel/arm/zscal.c b/kernel/arm/zscal.c index b210f9af30..57c095a47c 100644 --- a/kernel/arm/zscal.c +++ b/kernel/arm/zscal.c @@ -51,6 +51,7 @@ int CNAME(BLASLONG n, BLASLONG dummy0, BLASLONG dummy1, FLOAT da_r,FLOAT da_i, F inc_x2 = 2 * inc_x; if (dummy2 == 0) { + printf("arm zscal da_r %f da_i %f\n",da_r,da_i); for (i = 0; i < n; i++) { if (da_r == 0.0 && da_i == 0.0) From 20080ab91a0c3a14152adf2f987c7ab86dde99a0 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Tue, 17 Jun 2025 15:12:13 +0200 Subject: [PATCH 21/23] Update test_potrs.c --- utest/test_potrs.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/utest/test_potrs.c b/utest/test_potrs.c index bcdbbab107..7293b59ce6 100644 --- a/utest/test_potrs.c +++ b/utest/test_potrs.c @@ -395,7 +395,8 @@ CTEST(potrf, bug_695){ }; #ifdef BUILD_COMPLEX BLASFUNC(cpotrf)(&up, &n, (float*)(A3), &n, info); - // printf("%g+%g*I\n", creal(A3[91]), cimag(A3[91])); + printf("%g+%g*I\n", creal(A3[90]), cimag(A3[90])); + printf("%g+%g*I\n", creal(A3[91]), cimag(A3[91])); if(isnan(CREAL(A3[91])) || isnan(CIMAG(A3[91]))) { CTEST_ERR("%s:%d got NaN", __FILE__, __LINE__); } From 413d68a4b428fe84bd36bec5ce5d899b6c2d11b1 Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Tue, 17 Jun 2025 16:09:07 +0200 Subject: [PATCH 22/23] Update zpotf2_U.c --- lapack/potf2/zpotf2_U.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/lapack/potf2/zpotf2_U.c b/lapack/potf2/zpotf2_U.c index a88a6d2053..8efe88a8c1 100644 --- a/lapack/potf2/zpotf2_U.c +++ b/lapack/potf2/zpotf2_U.c @@ -66,7 +66,7 @@ blasint CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, for (j = 0; j < n; j++) { ajj = CREAL(DOTC_K(j, a, 1, a, 1)); - +printf("zpotf2_U ajj from CDOT is %f\n",ajj); ajj = *(a + j * 2) - ajj; if (ajj <= 0){ @@ -86,7 +86,7 @@ blasint CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, a + lda * 2, lda, a, 1, a + (j + lda) * 2, lda, sb); - +printf("zpotf2_U calling scal with 1./%f\n",ajj); SCAL_K(i, 0, 0, ONE / ajj, ZERO, a + (j + lda) * 2, lda, NULL, 0, NULL, 0); } From 4fed23268b7e8cd3d840ab1edfc520a12ba592ef Mon Sep 17 00:00:00 2001 From: Martin Kroeker Date: Tue, 17 Jun 2025 21:30:36 +0200 Subject: [PATCH 23/23] Update KERNEL --- kernel/x86/KERNEL | 2 ++ 1 file changed, 2 insertions(+) diff --git a/kernel/x86/KERNEL b/kernel/x86/KERNEL index 3ae268e6c9..0be5ef5c54 100644 --- a/kernel/x86/KERNEL +++ b/kernel/x86/KERNEL @@ -203,3 +203,5 @@ endif CSCALKERNEL = ../arm/zscal.c ZSCALKERNEL = ../arm/zscal.c +CDOTKERNEL = ../arm/zdot.c +ZDOTKERNEL = ../arm/zdot.c