5 Commits

Author SHA1 Message Date
Pierre van Houtryve
9ab601fa53 Reland "[NFC][AMDGPU] Do not flush after printing every instruction (#95237)"
It's very expensive and doesn't achieve anything.

I one test I did, it saves almost 10s on a 2m23s build, bringing it down
to 2m15s using a downstream branch.
2024-06-13 08:30:06 +02:00
pvanhout
04c4cf45fa Revert "[NFC][AMDGPU] Do not flush after printing every instruction (#95237)"
This reverts commit ad9fe3b2a949fb3379e0a1bafbcd2ca81f5fa414.
2024-06-12 15:24:21 +02:00
Pierre van Houtryve
ad9fe3b2a9
[NFC][AMDGPU] Do not flush after printing every instruction (#95237)
It's very expensive and doesn't achieve anything.

I one test I did, it saves almost 10s on a 2m23s build, bringing it down
to 2m15s using a downstream branch.
2024-06-12 14:57:41 +02:00
Kazu Hirata
2bc095037e [AMDGPU] Use llvm::bit_cast (NFC)
AFAICT, this is the only place where we pass a 64-bit value into
BitsToFloat.
2023-02-12 21:59:14 -08:00
Daniil Fukalov
48958d02d2 [NFC][AMDGPU] Reduce includes dependencies.
1. Splitted out some parts of R600 target to separate modules/headers.
2. Reduced some include lists in headers.
3. Found and fixed issue with override `GCNTargetMachine::getSubtargetImpl()`
   and `R600TargetMachine::getSubtargetImpl()` had different return value type
   than base class.
4. Minor forward declarations cleanup.

Reviewed By: foad

Differential Revision: https://reviews.llvm.org/D108596
2021-08-25 12:01:55 +03:00