search for: tidx

hi, list, I found AMDGPUPromoteAlloca calculates newly ptr as follows: std::tie(TCntY, TCntZ) = getLocalSizeYZ(Builder); Value *TIdX = getWorkitemID(Builder, 0); Value *TIdY = getWorkitemID(Builder, 1); Value *TIdZ = getWorkitemID(Builder, 2); Value *Tmp0 = Builder.CreateMul(TCntY, TCntZ, "", true, true); Tmp0 = Builder.CreateMul(Tmp0, TIdX); Value *Tmp1 = Builder.CreateMul(TIdY, TCntZ, "", tr...

...e pointer refers to an array of values, one for each thread that will execute the region. After the region completes, a loop must be created to actually perform the requested reduction. Inside the parallel region, each thread accesses its value using its thread id as the index. See the nthreads and tidx intrinsics below. -- Special handling regions -- - Serial Regions - Serial regions within parallel blocks (called 'single' in OpenMP) are executed only by one thread. As with parallel regions themselves, they are lowered as functions; the call site(s) of these functions are tagged with &...

...PRS.TCGP..PPOR.T 1980: 43 47 50 0a 05 54 50 4d 56 8c 54 43 47 50 0a 06 CGP..TPMV.TCGP.. 1990: 4d 4f 52 5f 8c 54 43 47 50 0a 07 54 56 45 4e 5b MOR_.TCGP..TVEN[ 19a0: 80 54 43 47 43 01 0a 72 0a 02 5b 81 10 54 43 47 .TCGC..r..[..TCG 19b0: 43 11 54 49 44 58 08 54 50 44 41 08 5b 86 1c 54 C.TIDX.TPDA.[..T 19c0: 49 44 58 54 50 44 41 11 00 40 20 54 50 50 31 08 IDXTPDA..@ TPP1. 19d0: 50 50 4c 4f 08 54 50 50 33 08 5b 80 53 4d 49 50 PPLO.TPP3.[.SMIP 19e0: 01 0a b2 0a 02 5b 81 10 53 4d 49 50 02 53 4d 49 .....[..SMIP.SMI 19f0: 54 08 53 4d 49 44 08 5b 80 2f 03 50 43 49 30 4c T.SMID.[./...