diff options
Diffstat (limited to 'llvm/lib/Target/AMDGPU/CaymanInstructions.td')
| -rw-r--r-- | llvm/lib/Target/AMDGPU/CaymanInstructions.td | 9 |
1 files changed, 5 insertions, 4 deletions
diff --git a/llvm/lib/Target/AMDGPU/CaymanInstructions.td b/llvm/lib/Target/AMDGPU/CaymanInstructions.td index 1a526675164a0..f4ddbf1131c34 100644 --- a/llvm/lib/Target/AMDGPU/CaymanInstructions.td +++ b/llvm/lib/Target/AMDGPU/CaymanInstructions.td @@ -50,16 +50,19 @@ def COS_cm : COS_Common<0x8E>; def : RsqPat<RECIPSQRT_IEEE_cm, f32>; +def : SqrtPat<RECIPSQRT_IEEE_cm, RECIP_IEEE_cm>; + def : POW_Common <LOG_IEEE_cm, EXP_IEEE_cm, MUL>; defm DIV_cm : DIV_Common<RECIP_IEEE_cm>; // RECIP_UINT emulation for Cayman -// The multiplication scales from [0,1] to the unsigned integer range +// The multiplication scales from [0,1) to the unsigned integer range, +// rounding down a bit to avoid unwanted overflow. def : R600Pat < (AMDGPUurecip i32:$src0), (FLT_TO_UINT_eg (MUL_IEEE (RECIP_IEEE_cm (UINT_TO_FLT_eg $src0)), - (MOV_IMM_I32 CONST.FP_UINT_MAX_PLUS_1))) + (MOV_IMM_I32 CONST.FP_4294966784))) >; def CF_END_CM : CF_CLAUSE_EG<32, (ins), "CF_END"> { @@ -70,8 +73,6 @@ def CF_END_CM : CF_CLAUSE_EG<32, (ins), "CF_END"> { -def : R600Pat<(fsqrt f32:$src), (MUL R600_Reg32:$src, (RECIPSQRT_CLAMPED_cm $src))>; - class RAT_STORE_DWORD <RegisterClass rc, ValueType vt, bits<4> mask> : CF_MEM_RAT_CACHELESS <0x14, 0, mask, (ins rc:$rw_gpr, R600_TReg32_X:$index_gpr), |