In D94155#2511894, @xbolva00 wrote:

http://llvm-compile-time-tracker.com/compare.php?from=51f4958057d6c246e85c3fbc65353bc0d7c1049b&to=20013d02f3352a88d0838eed349abc9a2b0e9cc0&stat=instructions

Compile time regression is too high IMHO.

cc @nikic

In D94155#2511894, @xbolva00 wrote:

http://llvm-compile-time-tracker.com/compare.php?from=51f4958057d6c246e85c3fbc65353bc0d7c1049b&to=20013d02f3352a88d0838eed349abc9a2b0e9cc0&stat=instructions

Compile time regression is too high IMHO.

cc @nikic

In D94155#2508556, @pengfei wrote:

LGTM, but let's wait for one day or two to see others' opinions.

Rebase.

Address Pengfei's comments.

Address Pengfei's comments.

Address Pengfei's comments.

Address Xiang's comments.

Make the logic more clean.

Get tile register number from backend.

We also need check tile config register interference. Since we
don't model the config register we should check interference from the
ldtilecfg to each tile data register def.

      ldtilecfg
      /       \
    BB1      BB2
     /         \
    call       BB3
    /           \
%1=tileload   %2=tilezero

We can start from the instruction of each tile def, and backward to
ldtilecfg. If there is any call instruction, and tile data register is
not preserved, we should insert ldtilecfg after the call instruction.

Address Pengfei's comments.

Support tile_zero and fix bugs for tile_load and tile_store.

In D94372#2491970, @pengfei wrote:

LGTM, but I'd like to see if @lebedev.ri has any objections.

Address Pengfei's comments.

Thank David!

In D94372#2489048, @lebedev.ri wrote:

What specifically is having problems after that transformation?

Address Pengfei's comments.

Fix typo in the comments.

In D93898#2483571, @wxiao3 wrote:

LGTM. But I suggest you waiting for one or two days to see if @qcolombet or others object.

Add test case for IPRA.

Remove dead code.

Address Wei's comments.

Rebase.

Rebase.

Add avx512f in test case.

Thank @pengfei and @MaskRay.

Rebase.

Remove datalayout from test case.

Rebase

In D91927#2471140, @pengfei wrote:

LGTM. Thanks for the refactors. Maybe better to wait for a few days to see if others have objections.

Refine comments.

Address Pengfei's comments.

Address Pengfei's comments.

Improve the comments.

Address Pengfei's comments.

In D91927#2469977, @pengfei wrote:

In my test case, it is transformed after Combine redundant instructions.

Can we disable it for AMX type? The pointer to AMX type is meaningless and may result in bad perfomance.

Scalarize tilestore.

Move comments.

Address Pengfei's comments.

Rebase and fix lit test case failure.