Page MenuHomePhabricator

TomHender (Tom Hender)
User

Projects

User does not belong to any projects.

User Details

User Since
Sep 7 2020, 6:07 AM (6 w, 5 d)

Recent Activity

Mon, Oct 19

TomHender added a comment to D89697: * [x86] Implement smarter instruction lowering for FP_TO_UINT from vXf32 to vXi32 for SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction..

@RKSimon I ran it through llvm-mca now. It gives me a reciprocal throughput of 3.5 for Silvermont and 3 for Haswell for the new instruction sequence.

Mon, Oct 19, 9:08 AM · Restricted Project
TomHender updated the diff for D89697: * [x86] Implement smarter instruction lowering for FP_TO_UINT from vXf32 to vXi32 for SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction..

@RKSimon Thank you for your very speedy response.

Mon, Oct 19, 7:45 AM · Restricted Project
TomHender retitled D89697: * [x86] Implement smarter instruction lowering for FP_TO_UINT from vXf32 to vXi32 for SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction. from * [x86] Implement smarter instruction lowering for FP_TO_UINT for vXf32 to vXi32 from SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction. to * [x86] Implement smarter instruction lowering for FP_TO_UINT from vXf32 to vXi32 for SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction..
Mon, Oct 19, 6:39 AM · Restricted Project
TomHender requested review of D89697: * [x86] Implement smarter instruction lowering for FP_TO_UINT from vXf32 to vXi32 for SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction..
Mon, Oct 19, 6:36 AM · Restricted Project

Sat, Oct 10

TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

Can you commit the changes for me?

Sat, Oct 10, 8:53 AM · Restricted Project

Tue, Oct 6

TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

Yes.

Tue, Oct 6, 6:05 AM · Restricted Project

Mon, Oct 5

TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

Unfortunately I couldn't get the "update_analyze_test_checks.py" script to work no matter what arguments I tried (which included --force-update, --opt-binary=T:\llvm-project\llvm\NATIVE\Debug\bin\opt.exe, different quotation mark placement and so on). Most of the time it did just nothing without any error or it would sometimes error with "ERROR: Unexpected opt name: opt.exe".

Mon, Oct 5, 4:10 AM · Restricted Project
TomHender updated the diff for D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..
Mon, Oct 5, 4:10 AM · Restricted Project

Sun, Oct 4

TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

This patch is now just about v8i16 and I updated the array in "getTypeBasedIntrinsicInstrCost".

Sun, Oct 4, 8:32 AM · Restricted Project
TomHender updated the diff for D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..
Sun, Oct 4, 7:29 AM · Restricted Project
TomHender updated the diff for D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..
Sun, Oct 4, 7:24 AM · Restricted Project

Tue, Sep 29

TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

Thank you for your comments.

Tue, Sep 29, 11:31 AM · Restricted Project
TomHender updated the diff for D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..
Tue, Sep 29, 10:50 AM · Restricted Project
TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

Ping.

Tue, Sep 29, 5:23 AM · Restricted Project

Sep 21 2020

TomHender added a comment to D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..

I published a followup patch D88026 that fixes the horizontal reduction regression in the way suggested by craig.topper.

Sep 21 2020, 8:37 AM · Restricted Project
TomHender requested review of D88026: [X86][SSE2] Follow up patch for D87236 to avoid the regression regarding horizontal unsigned 16 bit minimums and maximums..
Sep 21 2020, 8:34 AM · Restricted Project

Sep 7 2020

TomHender updated the summary of D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..
Sep 7 2020, 7:25 AM · Restricted Project
TomHender requested review of D87236: [X86][SSE2] Use smarter instruction patterns for lowering UMIN/UMAX with v8i16..
Sep 7 2020, 7:24 AM · Restricted Project