brintos

brintos / llvm-project-archived public Read only

0
0
Text · 2.7 KiB · 79deedc Raw
61 lines · plain
1; RUN: opt %loadNPMPolly '-passes=polly-custom<opt-isl>' -polly-pattern-matching-based-opts=true -polly-target-throughput-vector-fma=1 -polly-target-latency-vector-fma=8 -polly-target-1st-cache-level-associativity=8 -polly-target-2nd-cache-level-associativity=8 -polly-target-1st-cache-level-size=32768 -polly-target-2nd-cache-level-size=262144 -polly-optimized-scops -polly-target-vector-register-bitwidth=256 -disable-output < %s2;3;    /* C := alpha*A*B + beta*C */4;    for (i = 0; i < _PB_NI; i++)5;      for (j = 0; j < _PB_NJ; j++)6;        {7;	   C[i][j] *= beta;8;	   for (k = 0; k < _PB_NK; ++k)9;	     C[i][j] += alpha * A[i][k] * B[k][j];10;        }11;12target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"13target triple = "x86_64-unknown-unknown"14 15define internal void @kernel_gemm(i32 %arg, i32 %arg1, i32 %arg2, double %arg3, double %arg4, ptr %arg5, ptr %arg6, ptr %arg7) #0 {16bb:17  br label %bb818 19bb8:                                              ; preds = %bb29, %bb20  %tmp = phi i64 [ 0, %bb ], [ %tmp30, %bb29 ]21  br label %bb922 23bb9:                                              ; preds = %bb26, %bb824  %tmp10 = phi i64 [ 0, %bb8 ], [ %tmp27, %bb26 ]25  %tmp11 = getelementptr inbounds [1056 x double], ptr %arg5, i64 %tmp, i64 %tmp1026  %tmp12 = load double, ptr %tmp11, align 827  %tmp13 = fmul double %tmp12, %arg428  store double %tmp13, ptr %tmp11, align 829  br label %Copy_030 31Copy_0:                                             ; preds = %Copy_0, %bb932  %tmp15 = phi i64 [ 0, %bb9 ], [ %tmp24, %Copy_0 ]33  %tmp16 = getelementptr inbounds [1024 x double], ptr %arg6, i64 %tmp, i64 %tmp1534  %tmp17 = load double, ptr %tmp16, align 835  %tmp18 = fmul double %tmp17, %arg336  %tmp19 = getelementptr inbounds [1056 x double], ptr %arg7, i64 %tmp15, i64 %tmp1037  %tmp20 = load double, ptr %tmp19, align 838  %tmp21 = fmul double %tmp18, %tmp2039  %tmp22 = load double, ptr %tmp11, align 840  %tmp23 = fadd double %tmp22, %tmp2141  store double %tmp23, ptr %tmp11, align 842  %tmp24 = add nuw nsw i64 %tmp15, 143  %tmp25 = icmp ne i64 %tmp24, 102444  br i1 %tmp25, label %Copy_0, label %bb2645 46bb26:                                             ; preds = %Copy_047  %tmp27 = add nuw nsw i64 %tmp10, 148  %tmp28 = icmp ne i64 %tmp27, 105649  br i1 %tmp28, label %bb9, label %bb2950 51bb29:                                             ; preds = %bb2652  %tmp30 = add nuw nsw i64 %tmp, 153  %tmp31 = icmp ne i64 %tmp30, 105654  br i1 %tmp31, label %bb8, label %bb3255 56bb32:                                             ; preds = %bb2957  ret void58}59 60attributes #0 = { nounwind uwtable "target-cpu"="x86-64" "target-features"="+aes,+avx,+cmov,+cx16,+fxsr,+mmx,+pclmul,+popcnt,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87,+xsave,+xsaveopt" }61