This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
Hi, As mentioned by James a few weeks ago, the vqtbl[lx][234] intrinsics are failing on aarch64_be. The attached patch fixes them, and rewrites them using new builtins instead of inline assembly. I wondered about the names of the new builtins, I hope I got them right: qtbl3, qtbl4, qtbx3, qtbx4 with v8qi and v16qi modes. I have modified the existing aarch64_tbl3v8qi and aarch64_tbx4v8qi to use <mode> and share the code with the v16qi variants. In arm_neon.h, I moved the rewritten intrinsics to the bottom of the file, in alphabetical order, although the comment says "Start of optimal implementations in approved order": the previous ones really seem to be in alphabetical order. And I added a new testcase, skipped for arm* targets. This has been tested on aarch64-none-elf and aarch64_be-none-elf targets, using the Foundation model. OK? Christophe.
Attachment:
vqtbx.chlog.txt
Description: Text document
Attachment:
vqtbx.patch.txt
Description: Text document
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |