haraldwolff/CLBlast
1
0
Fork 0
mirror of https://github.com/CNugteren/CLBlast.git synced 2024-07-07 12:23:46 +02:00
Code Issues Packages Projects Releases Wiki Activity

Fixed an issue where the xNRM2 and xASUM testers would incorrectly report failures for complex inputs

Browse source
This commit is contained in:
cnugteren 2016-05-08 18:07:55 +02:00
parent 25a25dbd6f
commit eaf1de5745
2 changed files with 16 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
scripts/generator/generator.py
View file

@ -266,6 +266,7 @@ def wrapper_cblas(routines):
# Special case for scalar outputs
assignment = ""
postfix = ""
endofline = ""
extra_argument = ""
for output_buffer in routine.outputs:
if output_buffer in routine.ScalarBuffersFirst():
@ -274,12 +275,17 @@ def wrapper_cblas(routines):
indent += " "
extra_argument += ",\n"+indent+"reinterpret_cast<return_pointer_"+flavour.buffertype[:-1]+">(&"+output_buffer+"_buffer["+output_buffer+"_offset])"
else:
assignment = output_buffer+"_buffer["+output_buffer+"_offset] = "
assignment = output_buffer+"_buffer["+output_buffer+"_offset]"
if (flavour.name in ["Sc","Dz"]):
assignment = assignment+".real("
endofline += ")"
else:
assignment = assignment+" = "
indent += " "*len(assignment)
result += " "+assignment+"cblas_"+flavour.name.lower()+routine.name+postfix+"("
result += (",\n"+indent).join([a for a in arguments])
result += extra_argument+");"
result += extra_argument+endofline+");"
result += "\n}\n"
return result

16
test/wrapper_cblas.h
View file

@ -338,14 +338,14 @@ void cblasXnrm2(const size_t n,
void cblasXnrm2(const size_t n,
std::vector<float2>& nrm2_buffer, const size_t nrm2_offset,
const std::vector<float2>& x_buffer, const size_t x_offset, const size_t x_inc) {
nrm2_buffer[nrm2_offset] = cblas_scnrm2(n,
reinterpret_cast<const float*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
nrm2_buffer[nrm2_offset].real(cblas_scnrm2(n,
reinterpret_cast<const float*>(&x_buffer[x_offset]), static_cast<int>(x_inc)));
}
void cblasXnrm2(const size_t n,
std::vector<double2>& nrm2_buffer, const size_t nrm2_offset,
const std::vector<double2>& x_buffer, const size_t x_offset, const size_t x_inc) {
nrm2_buffer[nrm2_offset] = cblas_dznrm2(n,
reinterpret_cast<const double*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
nrm2_buffer[nrm2_offset].real(cblas_dznrm2(n,
reinterpret_cast<const double*>(&x_buffer[x_offset]), static_cast<int>(x_inc)));
}
// Forwards the Netlib BLAS calls for SASUM/DASUM/ScASUM/DzASUM
@ -364,14 +364,14 @@ void cblasXasum(const size_t n,
void cblasXasum(const size_t n,
std::vector<float2>& asum_buffer, const size_t asum_offset,
const std::vector<float2>& x_buffer, const size_t x_offset, const size_t x_inc) {
asum_buffer[asum_offset] = cblas_scasum(n,
reinterpret_cast<const float*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
asum_buffer[asum_offset].real(cblas_scasum(n,
reinterpret_cast<const float*>(&x_buffer[x_offset]), static_cast<int>(x_inc)));
}
void cblasXasum(const size_t n,
std::vector<double2>& asum_buffer, const size_t asum_offset,
const std::vector<double2>& x_buffer, const size_t x_offset, const size_t x_inc) {
asum_buffer[asum_offset] = cblas_dzasum(n,
reinterpret_cast<const double*>(&x_buffer[x_offset]), static_cast<int>(x_inc));
asum_buffer[asum_offset].real(cblas_dzasum(n,
reinterpret_cast<const double*>(&x_buffer[x_offset]), static_cast<int>(x_inc)));
}
// Forwards the Netlib BLAS calls for iSAMAX/iDAMAX/iCAMAX/iZAMAX