e3cf80c5c1
BlockFrequencyInfo calculates block frequencies as Scaled64 numbers but as a last step converts them to unsigned 64bit integers (`BlockFrequency`). This improves the factors picked for this conversion so that: * Avoid big numbers close to UINT64_MAX to avoid users overflowing/saturating when adding multiply frequencies together or when multiplying with integers. This leaves the topmost 10 bits unused to allow for some room. * Spread the difference between hottest/coldest block as much as possible to increase precision. * If the hot/cold spread cannot be represented loose precision at the lower end, but keep the frequencies at the upper end for hot blocks differentiable.
178 lines
5.7 KiB
LLVM
178 lines
5.7 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc %s -mtriple=thumbv6m-arm-none-eabi -o - -asm-verbose=false | FileCheck %s --check-prefix=CHECK-V6M
|
|
; RUN: llc %s -mtriple=thumbv7m-arm-none-eabi -o - -asm-verbose=false | FileCheck %s --check-prefix=CHECK-V7M
|
|
|
|
define i32 @test_values(i32 %a, i32 %b) minsize optsize {
|
|
; CHECK-V6M-LABEL: test_values:
|
|
; CHECK-V6M: mov r2, r0
|
|
; CHECK-V6M-NEXT: ldr r0, .LCPI0_0
|
|
; CHECK-V6M-NEXT: cmp r2, #50
|
|
; CHECK-V6M-NEXT: beq .LBB0_7
|
|
; CHECK-V6M-NEXT: cmp r2, #1
|
|
; CHECK-V6M-NEXT: beq .LBB0_5
|
|
; CHECK-V6M-NEXT: cmp r2, #30
|
|
; CHECK-V6M-NEXT: beq .LBB0_6
|
|
; CHECK-V6M-NEXT: cmp r2, #0
|
|
; CHECK-V6M-NEXT: bne .LBB0_8
|
|
; CHECK-V6M-NEXT: adds r0, r1, r0
|
|
; CHECK-V6M-NEXT: bx lr
|
|
; CHECK-V6M-NEXT: .LBB0_5:
|
|
; CHECK-V6M-NEXT: adds r0, r0, r1
|
|
; CHECK-V6M-NEXT: adds r0, r0, #1
|
|
; CHECK-V6M-NEXT: bx lr
|
|
; CHECK-V6M-NEXT: .LBB0_6:
|
|
; CHECK-V6M-NEXT: adds r0, r0, r1
|
|
; CHECK-V6M-NEXT: adds r0, r0, #2
|
|
; CHECK-V6M-NEXT: bx lr
|
|
; CHECK-V6M-NEXT: .LBB0_7:
|
|
; CHECK-V6M-NEXT: adds r0, r0, r1
|
|
; CHECK-V6M-NEXT: adds r0, r0, #4
|
|
; CHECK-V6M-NEXT: .LBB0_8:
|
|
; CHECK-V6M-NEXT: bx lr
|
|
; CHECK-V6M-NEXT: .p2align 2
|
|
; CHECK-V6M-NEXT: .LCPI0_0:
|
|
; CHECK-V6M-NEXT: .long 537923600
|
|
;
|
|
; CHECK-V7M-LABEL: test_values:
|
|
; CHECK-V7M: mov r2, r0
|
|
; CHECK-V7M-NEXT: ldr r0, .LCPI0_0
|
|
; CHECK-V7M-NEXT: cmp r2, #50
|
|
; CHECK-V7M-NEXT: beq .LBB0_5
|
|
; CHECK-V7M-NEXT: cmp r2, #1
|
|
; CHECK-V7M-NEXT: ittt eq
|
|
; CHECK-V7M-NEXT: addeq r0, r1
|
|
; CHECK-V7M-NEXT: addeq r0, #1
|
|
; CHECK-V7M-NEXT: bxeq lr
|
|
; CHECK-V7M-NEXT: .LBB0_2:
|
|
; CHECK-V7M-NEXT: cmp r2, #30
|
|
; CHECK-V7M-NEXT: ittt eq
|
|
; CHECK-V7M-NEXT: addeq r0, r1
|
|
; CHECK-V7M-NEXT: addeq r0, #2
|
|
; CHECK-V7M-NEXT: bxeq lr
|
|
; CHECK-V7M-NEXT: .LBB0_3:
|
|
; CHECK-V7M-NEXT: cbnz r2, .LBB0_6
|
|
; CHECK-V7M-NEXT: add r0, r1
|
|
; CHECK-V7M-NEXT: bx lr
|
|
; CHECK-V7M-NEXT: .LBB0_5:
|
|
; CHECK-V7M-NEXT: add r0, r1
|
|
; CHECK-V7M-NEXT: adds r0, #4
|
|
; CHECK-V7M-NEXT: .LBB0_6:
|
|
; CHECK-V7M-NEXT: bx lr
|
|
; CHECK-V7M-NEXT: .p2align 2
|
|
; CHECK-V7M-NEXT: .LCPI0_0:
|
|
; CHECK-V7M-NEXT: .long 537923600
|
|
entry:
|
|
switch i32 %a, label %return [
|
|
i32 0, label %sw.bb
|
|
i32 1, label %sw.bb1
|
|
i32 30, label %sw.bb3
|
|
i32 50, label %sw.bb5
|
|
]
|
|
|
|
sw.bb: ; preds = %entry
|
|
%add = add nsw i32 %b, 537923600
|
|
br label %return
|
|
|
|
sw.bb1: ; preds = %entry
|
|
%add2 = add nsw i32 %b, 537923601
|
|
br label %return
|
|
|
|
sw.bb3: ; preds = %entry
|
|
%add4 = add nsw i32 %b, 537923602
|
|
br label %return
|
|
|
|
sw.bb5: ; preds = %entry
|
|
%add6 = add nsw i32 %b, 537923604
|
|
br label %return
|
|
|
|
return: ; preds = %entry, %sw.bb5, %sw.bb3, %sw.bb1, %sw.bb
|
|
%retval.0 = phi i32 [ %add6, %sw.bb5 ], [ %add4, %sw.bb3 ], [ %add2, %sw.bb1 ], [ %add, %sw.bb ], [ 537923600, %entry ]
|
|
ret i32 %retval.0
|
|
}
|
|
|
|
define i32 @test_addr(i32 %a, ptr nocapture readonly %b) {
|
|
; CHECK-V6M-LABEL: test_addr:
|
|
; CHECK-V6M: mov r2, r0
|
|
; CHECK-V6M-NEXT: movs r0, #19
|
|
; CHECK-V6M-NEXT: lsls r3, r0, #4
|
|
; CHECK-V6M-NEXT: movs r0, #0
|
|
; CHECK-V6M-NEXT: cmp r2, #29
|
|
; CHECK-V6M-NEXT: bgt .LBB1_4
|
|
; CHECK-V6M-NEXT: cmp r2, #0
|
|
; CHECK-V6M-NEXT: beq .LBB1_8
|
|
; CHECK-V6M-NEXT: cmp r2, #1
|
|
; CHECK-V6M-NEXT: bne .LBB1_9
|
|
; CHECK-V6M-NEXT: adds r3, r3, #1
|
|
; CHECK-V6M-NEXT: b .LBB1_8
|
|
; CHECK-V6M-NEXT: .LBB1_4:
|
|
; CHECK-V6M-NEXT: cmp r2, #30
|
|
; CHECK-V6M-NEXT: beq .LBB1_7
|
|
; CHECK-V6M-NEXT: cmp r2, #50
|
|
; CHECK-V6M-NEXT: bne .LBB1_9
|
|
; CHECK-V6M-NEXT: adds r3, r3, #3
|
|
; CHECK-V6M-NEXT: b .LBB1_8
|
|
; CHECK-V6M-NEXT: .LBB1_7:
|
|
; CHECK-V6M-NEXT: adds r3, r3, #2
|
|
; CHECK-V6M-NEXT: .LBB1_8:
|
|
; CHECK-V6M-NEXT: ldrb r0, [r1, r3]
|
|
; CHECK-V6M-NEXT: .LBB1_9:
|
|
; CHECK-V6M-NEXT: bx lr
|
|
;
|
|
; CHECK-V7M-LABEL: test_addr:
|
|
; CHECK-V7M: mov r2, r0
|
|
; CHECK-V7M-NEXT: movs r0, #0
|
|
; CHECK-V7M-NEXT: cmp r2, #29
|
|
; CHECK-V7M-NEXT: bgt .LBB1_4
|
|
; CHECK-V7M-NEXT: cbz r2, .LBB1_7
|
|
; CHECK-V7M-NEXT: cmp r2, #1
|
|
; CHECK-V7M-NEXT: it ne
|
|
; CHECK-V7M-NEXT: bxne lr
|
|
; CHECK-V7M-NEXT: .LBB1_3:
|
|
; CHECK-V7M-NEXT: movw r0, #305
|
|
; CHECK-V7M-NEXT: b .LBB1_9
|
|
; CHECK-V7M-NEXT: .LBB1_4:
|
|
; CHECK-V7M-NEXT: cmp r2, #30
|
|
; CHECK-V7M-NEXT: beq .LBB1_8
|
|
; CHECK-V7M-NEXT: cmp r2, #50
|
|
; CHECK-V7M-NEXT: bne .LBB1_10
|
|
; CHECK-V7M-NEXT: movw r0, #307
|
|
; CHECK-V7M-NEXT: b .LBB1_9
|
|
; CHECK-V7M-NEXT: .LBB1_7:
|
|
; CHECK-V7M-NEXT: mov.w r0, #304
|
|
; CHECK-V7M-NEXT: b .LBB1_9
|
|
; CHECK-V7M-NEXT: .LBB1_8:
|
|
; CHECK-V7M-NEXT: mov.w r0, #306
|
|
; CHECK-V7M-NEXT: .LBB1_9:
|
|
; CHECK-V7M-NEXT: ldrb r0, [r1, r0]
|
|
; CHECK-V7M-NEXT: .LBB1_10:
|
|
; CHECK-V7M-NEXT: bx lr
|
|
entry:
|
|
switch i32 %a, label %return [
|
|
i32 0, label %return.sink.split
|
|
i32 1, label %sw.bb1
|
|
i32 30, label %sw.bb4
|
|
i32 50, label %sw.bb7
|
|
]
|
|
|
|
sw.bb1: ; preds = %entry
|
|
br label %return.sink.split
|
|
|
|
sw.bb4: ; preds = %entry
|
|
br label %return.sink.split
|
|
|
|
sw.bb7: ; preds = %entry
|
|
br label %return.sink.split
|
|
|
|
return.sink.split: ; preds = %entry, %sw.bb1, %sw.bb4, %sw.bb7
|
|
%.sink = phi i32 [ 307, %sw.bb7 ], [ 306, %sw.bb4 ], [ 305, %sw.bb1 ], [ 304, %entry ]
|
|
%arrayidx8 = getelementptr inbounds i8, ptr %b, i32 %.sink
|
|
%0 = load i8, ptr %arrayidx8, align 1
|
|
%phitmp = zext i8 %0 to i32
|
|
br label %return
|
|
|
|
return: ; preds = %return.sink.split, %entry
|
|
%retval.0.shrunk = phi i32 [ 0, %entry ], [ %phitmp, %return.sink.split ]
|
|
ret i32 %retval.0.shrunk
|
|
}
|
|
|