5 Commits

Author	SHA1	Message	Date
Jay Foad	0d40831765	[AMDGPU] Allow folding to FMAAK with SGPR and immediate operand on GFX10+ (#72266) ... Allow foldImmediate to create instructions like: v_fmaak_f32 v0, s0, v0, 0x42000000 This instruction has two "scalar values": s0 and 0x42000000. On GFX10+ this is allowed. This fold was originally implemented before the compiler supported GFX10, when all ASICs were limited to one scalar value.	2023-11-28 14:36:37 +00:00
Mirko Brkusanin	a657deb42e	[AMDGPU] Update RUN line in test (NFC)	2023-09-22 12:41:54 +02:00
Mirko Brkušanin	ecfdc23dd2	[AMDGPU] Select gfx1150 SALU Float instructions (#66885)	2023-09-21 12:22:55 +02:00
Mirko Brkusanin	1e5359c6ba	[AMDGPU] Treat KIMM32 and KIMM16 operand types as noninlinable ... While they are represent 32/16 bit immediate values they are already included in encoding of the instructions that use them and are not true literals. FMAMK and FMAAK instructions that use them are marked with fixed size so getInstSizeInBytes will not increase the size for these operands. We also add tests whose logic relies on KIMM16 and KIMM32 being considered not inlinable. Differential Revision: https://reviews.llvm.org/D157624	2023-08-11 18:46:39 +02:00
Matt Arsenault	4b1702e87a	AMDGPU: Fix counting source modifiers as literal constants ... This fixes over estimating code size. This was broken by 79f52af4cd9a76485dd50bcdbb5d393eb7a70103. https://reviews.llvm.org/D157103	2023-08-07 18:40:16 -04:00