Skip to content

Conversation

@tingboliao
Copy link

No description provided.

@martin-frbg
Copy link
Collaborator

I think it should be sufficient to have the declarations of SROTMKERNEL and DROTMKERNEL in the plain "KERNEL" files under each architecture, no need to repeat them in the specific files (except possibly the KERNEL.generic) as long as they are the same. Also, could you please add a line for the "quad precision" version in common_level.1 as
int qrotm_k (BLASLONG, xdouble, BLASLONG, xdouble, BLASLONG, xdouble); and similarly int (*qrotm_k... in common_param.h,
#define ROTM_K QROTM_K in common_macro.h,

ifndef QROTMKERNEL
QROTKERNEL = rotm.S
endif

in kernel/Makefile.L1 and

ifndef QROTMKERNEL
QROTMKERNEL  = ../generic/rotm.c
endif

in kernel/x86_64/KERNEL ? The "quad precision" (EXPRECISION=1) kernels (exclusively for x86_64 are probably not fully functional currently (and may never have been), but that would fix compilation problems arising from your PR.
Also please add srotm_kTS, drotm_kTS or qrotm_kTS as appropriate everywhere you see the corresponding srot_kTS etc in kernel/setparam-ref.c to make the DYNAMIC_ARCH builds work.

@tingboliao tingboliao reopened this Jan 22, 2025
@tingboliao
Copy link
Author

tingboliao commented Jan 22, 2025

I think it should be sufficient to have the declarations of SROTMKERNEL and DROTMKERNEL in the plain "KERNEL" files under each architecture, no need to repeat them in the specific files (except possibly the KERNEL.generic) as long as they are the same. Also, could you please add a line for the "quad precision" version in common_level.1 as int qrotm_k (BLASLONG, xdouble, BLASLONG, xdouble, BLASLONG, xdouble); and similarly int (*qrotm_k... in common_param.h, #define ROTM_K QROTM_K in common_macro.h,

ifndef QROTMKERNEL
QROTKERNEL = rotm.S
endif

in kernel/Makefile.L1 and

ifndef QROTMKERNEL
QROTMKERNEL  = ../generic/rotm.c
endif

in kernel/x86_64/KERNEL ? The "quad precision" (EXPRECISION=1) kernels (exclusively for x86_64 are probably not fully functional currently (and may never have been), but that would fix compilation problems arising from your PR. Also please add srotm_kTS, drotm_kTS or qrotm_kTS as appropriate everywhere you see the corresponding srot_kTS etc in kernel/setparam-ref.c to make the DYNAMIC_ARCH builds work.

Hi, @martin-frbg
I've made the corrections according to the suggestions you provided. However, it seems that there are still some failures for some architectures.

@martin-frbg martin-frbg added this to the 0.3.30 milestone Jan 23, 2025
@martin-frbg martin-frbg merged commit 180ba5e into OpenMathLib:develop Jan 23, 2025
85 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants