adec922361
We would like to start pushing -mcpu=generic towards enabling the set of features that improves performance for some CPUs, without hurting any others. A blend of the performance options hopefully beneficial to all CPUs. The largest part of that is enabling in-order scheduling using the Cortex-A55 schedule model. This is similar to the Arm backend change from eecb353d0e25ba which made -mcpu=generic perform in-order scheduling using the cortex-a8 schedule model. The idea is that in-order cpu's require the most help in instruction scheduling, whereas out-of-order cpus can for the most part out-of-order schedule around different codegen. Our benchmarking suggests that hypothesis holds. When running on an in-order core this improved performance by 3.8% geomean on a set of DSP workloads, 2% geomean on some other embedded benchmark and between 1% and 1.8% on a set of singlecore and multicore workloads, all running on a Cortex-A55 cluster. On an out-of-order cpu the results are a lot more noisy but show flat performance or an improvement. On the set of DSP and embedded benchmarks, run on a Cortex-A78 there was a very noisy 1% speed improvement. Using the most detailed results I could find, SPEC2006 runs on a Neoverse N1 show a small increase in instruction count (+0.127%), but a decrease in cycle counts (-0.155%, on average). The instruction count is very low noise, the cycle count is more noisy with a 0.15% decrease not being significant. SPEC2k17 shows a small decrease (-0.2%) in instruction count leading to a -0.296% decrease in cycle count. These results are within noise margins but tend to show a small improvement in general. When specifying an Apple target, clang will set "-target-cpu apple-a7" on the command line, so should not be affected by this change when running from clang. This also doesn't enable more runtime unrolling like -mcpu=cortex-a55 does, only changing the schedule used. A lot of existing tests have updated. This is a summary of the important differences: - Most changes are the same instructions in a different order. - Sometimes this leads to very minor inefficiencies, such as requiring an extra mov to move variables into r0/v0 for the return value of a test function. - misched-fusion.ll was no longer fusing the pairs of instructions it should, as per D110561. I've changed the schedule used in the test for now. - neon-mla-mls.ll now uses "mul; sub" as opposed to "neg; mla" due to the different latencies. This seems fine to me. - Some SVE tests do not always remove movprfx where they did before due to different register allocation giving different destructive forms. - The tests argument-blocks-array-of-struct.ll and arm64-windows-calls.ll produce two LDR where they previously produced an LDP due to store-pair-suppress kicking in. - arm64-ldp.ll and arm64-neon-copy.ll are missing pre/postinc on LPD. - Some tests such as arm64-neon-mul-div.ll and ragreedy-local-interval-cost.ll have more, less or just different spilling. - In aarch64_generated_funcs.ll.generated.expected one part of the function is no longer outlined. Interestingly if I switch this to use any other scheduled even less is outlined. Some of these are expected to happen, such as differences in outlining or register spilling. There will be places where these result in worse codegen, places where they are better, with the SPEC instruction counts suggesting it is not a decrease overall, on average. Differential Revision: https://reviews.llvm.org/D110830
254 lines
7.4 KiB
LLVM
254 lines
7.4 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=aarch64-unknown-linux-gnu < %s | FileCheck %s
|
|
|
|
;------------------------------------------------------------------------------;
|
|
; Odd divisors
|
|
;------------------------------------------------------------------------------;
|
|
|
|
define i32 @test_urem_odd(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_odd:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #52429
|
|
; CHECK-NEXT: mov w9, #13108
|
|
; CHECK-NEXT: movk w8, #52428, lsl #16
|
|
; CHECK-NEXT: movk w9, #13107, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: cmp w8, w9
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 5
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
define i32 @test_urem_odd_25(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_odd_25:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #23593
|
|
; CHECK-NEXT: mov w9, #28836
|
|
; CHECK-NEXT: movk w8, #49807, lsl #16
|
|
; CHECK-NEXT: movk w9, #2621, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: cmp w8, w9
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 25
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; This is like test_urem_odd, except the divisor has bit 30 set.
|
|
define i32 @test_urem_odd_bit30(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_odd_bit30:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #43691
|
|
; CHECK-NEXT: movk w8, #27306, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: cmp w8, #4
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 1073741827
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; This is like test_urem_odd, except the divisor has bit 31 set.
|
|
define i32 @test_urem_odd_bit31(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_odd_bit31:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #43691
|
|
; CHECK-NEXT: movk w8, #10922, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: cmp w8, #2
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 2147483651
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
;------------------------------------------------------------------------------;
|
|
; Even divisors
|
|
;------------------------------------------------------------------------------;
|
|
|
|
define i16 @test_urem_even(i16 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_even:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #28087
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: and w9, w8, #0xfffc
|
|
; CHECK-NEXT: lsr w9, w9, #1
|
|
; CHECK-NEXT: bfi w9, w8, #15, #17
|
|
; CHECK-NEXT: ubfx w8, w9, #1, #15
|
|
; CHECK-NEXT: cmp w8, #2340
|
|
; CHECK-NEXT: cset w0, hi
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i16 %X, 14
|
|
%cmp = icmp ne i16 %urem, 0
|
|
%ret = zext i1 %cmp to i16
|
|
ret i16 %ret
|
|
}
|
|
|
|
define i32 @test_urem_even_100(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_even_100:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #23593
|
|
; CHECK-NEXT: mov w9, #23593
|
|
; CHECK-NEXT: movk w8, #49807, lsl #16
|
|
; CHECK-NEXT: movk w9, #655, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: ror w8, w8, #2
|
|
; CHECK-NEXT: cmp w8, w9
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 100
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; This is like test_urem_even, except the divisor has bit 30 set.
|
|
define i32 @test_urem_even_bit30(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_even_bit30:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #20165
|
|
; CHECK-NEXT: movk w8, #64748, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: ror w8, w8, #3
|
|
; CHECK-NEXT: cmp w8, #4
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 1073741928
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; This is like test_urem_odd, except the divisor has bit 31 set.
|
|
define i32 @test_urem_even_bit31(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_even_bit31:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #64251
|
|
; CHECK-NEXT: movk w8, #47866, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: ror w8, w8, #1
|
|
; CHECK-NEXT: cmp w8, #2
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 2147483750
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
;------------------------------------------------------------------------------;
|
|
; Special case
|
|
;------------------------------------------------------------------------------;
|
|
|
|
; 'NE' predicate is fine too.
|
|
define i32 @test_urem_odd_setne(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_odd_setne:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #52429
|
|
; CHECK-NEXT: mov w9, #858993459
|
|
; CHECK-NEXT: movk w8, #52428, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: cmp w8, w9
|
|
; CHECK-NEXT: cset w0, hi
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 5
|
|
%cmp = icmp ne i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; The fold is only valid for positive divisors, negative-ones should be negated.
|
|
define i32 @test_urem_negative_odd(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_negative_odd:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #858993459
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: cmp w8, #1
|
|
; CHECK-NEXT: cset w0, hi
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, -5
|
|
%cmp = icmp ne i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
define i32 @test_urem_negative_even(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_negative_even:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #37449
|
|
; CHECK-NEXT: movk w8, #51492, lsl #16
|
|
; CHECK-NEXT: mul w8, w0, w8
|
|
; CHECK-NEXT: ror w8, w8, #1
|
|
; CHECK-NEXT: cmp w8, #1
|
|
; CHECK-NEXT: cset w0, hi
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, -14
|
|
%cmp = icmp ne i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
;------------------------------------------------------------------------------;
|
|
; Negative tests
|
|
;------------------------------------------------------------------------------;
|
|
|
|
; We can lower remainder of division by one much better elsewhere.
|
|
define i32 @test_urem_one(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_one:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w0, #1
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 1
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; We can lower remainder of division by powers of two much better elsewhere.
|
|
define i32 @test_urem_pow2(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_pow2:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: tst w0, #0xf
|
|
; CHECK-NEXT: cset w0, eq
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 16
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; The fold is only valid for positive divisors, and we can't negate INT_MIN.
|
|
define i32 @test_urem_int_min(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_int_min:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: tst w0, #0x7fffffff
|
|
; CHECK-NEXT: cset w0, eq
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 2147483648
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|
|
|
|
; We can lower remainder of division by all-ones much better elsewhere.
|
|
define i32 @test_urem_allones(i32 %X) nounwind {
|
|
; CHECK-LABEL: test_urem_allones:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: neg w8, w0
|
|
; CHECK-NEXT: cmp w8, #2
|
|
; CHECK-NEXT: cset w0, lo
|
|
; CHECK-NEXT: ret
|
|
%urem = urem i32 %X, 4294967295
|
|
%cmp = icmp eq i32 %urem, 0
|
|
%ret = zext i1 %cmp to i32
|
|
ret i32 %ret
|
|
}
|