openblas/2672.patch

From 478898b37a91836a39d046f8c70e26c6c9fc06c7 Mon Sep 17 00:00:00 2001
From: Marius Hillenbrand <mhillen@linux.ibm.com>
Date: Wed, 17 Jun 2020 16:08:48 +0200
Subject: [PATCH 1/2] cpp_thread_test/dgemv: cap concurrency to number of hw
 threads on small systems

... instead of (number of hw threads - 4) to avoid invalid numbers on
smaller systems. Currently, systems with 4 or fewer CPUs (e.g., small CI
VMs) would fail the test. Fixes one of the issues discussed in #2668

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
---
 cpp_thread_test/dgemv_thread_safety.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/cpp_thread_test/dgemv_thread_safety.cpp b/cpp_thread_test/dgemv_thread_safety.cpp
index 5411fec29..277594ff0 100644
--- a/cpp_thread_test/dgemv_thread_safety.cpp
+++ b/cpp_thread_test/dgemv_thread_safety.cpp
@@ -18,7 +18,7 @@ int main(int argc, char* argv[]){
 	uint32_t maxHwThreads = omp_get_max_threads();
 	
 	if (maxHwThreads < 52)
-		numConcurrentThreads = maxHwThreads -4;
+		numConcurrentThreads = maxHwThreads;
 	
 	if (argc > 4){
 		std::cout<<"ERROR: too many arguments for thread safety tester"<<std::endl;

From de838c38ef9db98a635de1dcedeba8b578ec87b3 Mon Sep 17 00:00:00 2001
From: Marius Hillenbrand <mhillen@linux.ibm.com>
Date: Wed, 17 Jun 2020 16:15:44 +0200
Subject: [PATCH 2/2] cpp_thread_test/dgemv: fail early if concurrency is zero

The two test cases dgemv_tester and dgemm_tester accept the degree of
concurrency as command line argument (amongst others). Fail early if
value 0 has been specified, instead of later with less-clear symptoms.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
---
 cpp_thread_test/cpp_thread_safety_common.h | 8 ++++++++
 cpp_thread_test/dgemm_thread_safety.cpp    | 2 ++
 cpp_thread_test/dgemv_thread_safety.cpp    | 2 ++
 3 files changed, 12 insertions(+)

diff --git a/cpp_thread_test/cpp_thread_safety_common.h b/cpp_thread_test/cpp_thread_safety_common.h
index 60ab5bb2f..8005369a8 100644
--- a/cpp_thread_test/cpp_thread_safety_common.h
+++ b/cpp_thread_test/cpp_thread_safety_common.h
@@ -5,6 +5,14 @@ inline void pauser(){
     std::getline(std::cin, dummy);
 }
 
+void FailIfThreadsAreZero(uint32_t numConcurrentThreads) {
+	if(numConcurrentThreads == 0) {
+		std::cout<<"ERROR: Invalid parameter 0 for number of concurrent calls into OpenBLAS!"<<std::endl;
+		std::cout<<"CBLAS DGEMV thread safety test FAILED!"<<std::endl;
+		exit(-1);
+	}
+}
+
 void FillMatrices(std::vector<std::vector<double>>& matBlock, std::mt19937_64& PRNG, std::uniform_real_distribution<double>& rngdist, const blasint randomMatSize, const uint32_t numConcurrentThreads, const uint32_t numMat){
 	for(uint32_t i=0; i<numMat; i++){
 		for(uint32_t j = 0; j < static_cast<uint32_t>(randomMatSize*randomMatSize); j++){
diff --git a/cpp_thread_test/dgemm_thread_safety.cpp b/cpp_thread_test/dgemm_thread_safety.cpp
index 1c5287524..104c64f2a 100644
--- a/cpp_thread_test/dgemm_thread_safety.cpp
+++ b/cpp_thread_test/dgemm_thread_safety.cpp
@@ -46,6 +46,8 @@ int main(int argc, char* argv[]){
 	std::cout<<"Number of concurrent calls into OpenBLAS : "<<numConcurrentThreads<<'\n';
 	std::cout<<"Number of testing rounds : "<<numTestRounds<<'\n';
 	std::cout<<"This test will need "<<(static_cast<uint64_t>(randomMatSize*randomMatSize)*numConcurrentThreads*3*8)/static_cast<double>(1024*1024)<<" MiB of RAM\n"<<std::endl;
+
+	FailIfThreadsAreZero(numConcurrentThreads);
 	
 	std::cout<<"Initializing random number generator..."<<std::flush;
 	std::mt19937_64 PRNG = InitPRNG();
diff --git a/cpp_thread_test/dgemv_thread_safety.cpp b/cpp_thread_test/dgemv_thread_safety.cpp
index 277594ff0..20ea38138 100644
--- a/cpp_thread_test/dgemv_thread_safety.cpp
+++ b/cpp_thread_test/dgemv_thread_safety.cpp
@@ -47,6 +47,8 @@ int main(int argc, char* argv[]){
 	std::cout<<"Number of concurrent calls into OpenBLAS : "<<numConcurrentThreads<<'\n';
 	std::cout<<"Number of testing rounds : "<<numTestRounds<<'\n';
 	std::cout<<"This test will need "<<((static_cast<uint64_t>(randomMatSize*randomMatSize)*numConcurrentThreads*8)+(static_cast<uint64_t>(randomMatSize)*numConcurrentThreads*8*2))/static_cast<double>(1024*1024)<<" MiB of RAM\n"<<std::endl;
+
+	FailIfThreadsAreZero(numConcurrentThreads);
 	
 	std::cout<<"Initializing random number generator..."<<std::flush;
 	std::mt19937_64 PRNG = InitPRNG();