similar to: AVX2 codegen - question reg. FMA generation

On Mon, 2 Sep 2019 at 16:59, Roman Lebedev <lebedev.ri at gmail.com> wrote: > > It appears you need 'reassoc' on fmul/fadd: > https://godbolt.org/z/nuTzx2 Thanks very much, that was it. Either that or providing -enable-unsafe-fp-math to llc yielded FMAs. I didn't expect this since using FMAs here instead of mul/add appears to be safer (the reverse is unsafe). ~ Uday

It probably does not pick the right processor architecture. You could try “clang -mavx” or “clang -march=corei7-avx” for ivy-bridge and “clang -march=core-avx2” or “clang -mavx2" for haswell. $ clang -march=core-avx2 -O3 -S -o - test.c .section __TEXT,__text,regular,pure_instructions .globl _f .align 4, 0x90 _f: ## @f

Hello - I found this post on the llvm blog: http://blog.llvm.org/2012/12/new-loop-vectorizer.html which makes me think that clang / llvm are capable of generating AVX with packed instructions as well as utilizing the full width of the YMM registers… I have an environment where icc generates these instructions (vmulps %ymm1, %ymm3, %ymm2 for example) but I can not get clang/llvm to generate such

Hi, Please consider the following loop: using v4f32 = float __attribute__((__vector_size__(16))); void fct6(v4f32 *x) { #pragma clang loop vectorize(enable) for (int i = 0; i < 256; ++i) x[i] = 7 * x[i]; } After compiling it with: clang++ -O3 -march=native -mtune=native \ -Rpass=loop-vectorize,slp-vectorize -Rpass-missed=loop-vectorize,slp-vectorize

I wonder why AVX is not used automatically if available at the host machine. In contrast to that, SSE41 instructions (like pmulld) are automatically used if the host machine supports SSE41. E.g. $ cat avx.ll define void @_fun1(<8 x float>*, <8 x float>*) { _L1: %x = load <8 x float>* %0 %y = load <8 x float>* %1 %z = fadd <8 x float> %x, %y store

On Thu, 24 May 2012, Pan, Wei wrote: > Very likely AVX is not enabled in your llc. This feature was enabled > just recently (late of April). I forgot to mention that I am using recent LLVM-3.1 and in principle my llc knows about avx as I have shown in the second example. But avx does not seem to be used by default. On Thu, 24 May 2012, Henning Thielemann wrote: > $ llc -o - -mattr

This is the wrong code: declare <16 x float> @foo(<16 x float>) define <16 x float> @test(<16 x float> %x, <16 x float> %y) nounwind { entry: %x1 = fadd <16 x float> %x, %y %call = call <16 x float> @foo(<16 x float> %x1) nounwind %y1 = fsub <16 x float> %call, %y ret <16 x float> %y1 } ./llc -mattr=+avx

On Jan 9, 2012, at 10:00 AM, Jakob Stoklund Olesen wrote: > > On Jan 8, 2012, at 11:18 PM, Demikhovsky, Elena wrote: > >> I'll explain what we see in the code. >> 1. The caller saves XMM registers across the call if needed (according to DEFS definition). >> YMMs are not in the set, so caller does not take care. > > This is not how the register allocator

Hi Elena, You're correct. LLVM does not align the stack to 32-bytes for AVX and unaligned moves should be used for YMM spills. I wrote some code to align the stack to 32-bytes when AVX spills are present; it does break the x86-64 ABI though. If upstream would be interested in this code, I can arrange with my employer to send a patch to the mailing list. -Cameron On Mar 1, 2012, at 4:09 PM,

Hi all, The 213 variant of the FMA3 instructions is currently marked commutable (see X86InstrFMA.td). Is that safe? According to the ISA the FMA3 instructions aren't commutable for non-numeric results, so I'd have thought commuting this would only be valid in fast-math mode? For the curious, the reason that I'm asking is that we currently always select the 213 variant, but this

I am tracking down an x86-64 code generation problem that has to do with AVX instructions. The symptom is: a function is called, and the upper half of the function argument (which is short16) is zero. This happens only when I compile code with pocl, but not when I use clang and/or llc manually. I tracked this down to the following. The call site looks like vmovdqa 24064(%rsp), %ymm0 vmovdqa

Hi Kay, My patch will partially address your bug. For now I'm just looking to switch the default FMA from vfmadd213xx to vfmadd231xx. That will cause the code in PR17229 to compile as desired, but would regress code like: double foo(double a, double b, double c) { return a * b + c; } Which will now require a vmovaps + vfmadd231. If this impacts real benchmarks we could add an

Thanks Sanjay. Interestingly for me, disable-llvm-optmzns did not make a difference in the way the shift was handled. Does the initial IR generated for you show this difference when the option is passed? Best regards Saurabh On 17 February 2017 at 19:03, Sanjay Patel <spatel at rotateright.com> wrote: > I think this is caused by a front-end change (cc'ing clang-dev) because >

Illustrative Example: clang -fveclib=SVML -O3 svml.c -mavx #include <math.h> void foo(double *a, int N){ int i; #pragma clang loop vectorize_width(8) for (i=0;i<N;i++){ a[i] = sin(i); } } Currently, this results in a call to <8 x double> __svml_sin8(<8 x double>) after the vectorizer. This is 8-element SVML sin() called with 8-element argument. On the surface,

The regression for the reported case should be avoided after: https://reviews.llvm.org/rL297232 https://reviews.llvm.org/rL297242 https://reviews.llvm.org/rL297280 It would still be good to understand if the clang change was intentional or if that was a side effect that can be limited. On Sat, Feb 18, 2017 at 9:11 AM, Sanjay Patel <spatel at rotateright.com> wrote: > Yes, there is an

Hi! When compiling the attached module with the JIT engine on an Intel KNL I see wrong code getting emitted. I attach a complete exploit program which shows the bug in LLVM 3.8. It loads and JIT compiles the module and prints the assembler. I stumbled on this since the result of an actual calculation was wrong. So, it's not only the text version of the assembler also the machine

Hi Elena, Thank you very much for looking in to that. I'll go ahead and remove the isCommutable flag from those instructions, since it sounds like that's the right thing to do. I would still like to change the default from the 231 variant to 213 too, as this will reduce code-size for accumulator-style loops. I have at least one benchmark that shows significant speedups when this change

Hi Lang, Unfortunately, I don't have an answer on the commutability question, but I wanted to let you know that I filed a bug on this: http://llvm.org/bugs/show_bug.cgi?id=17229 This also shows a memory operand variant of the fma that you may want to consider in your patch and testcases. Thanks! On Thu, Dec 19, 2013 at 10:45 PM, Lang Hames <lhames at gmail.com> wrote: > Hi all,

Ashutosh, Thanks for the repy. Related earlier topic on this appears in the review of the SVML patch (@mmasten). Adding few names from there. https://reviews.llvm.org/D19544 There, I see Hal's review comment "let's start only with the directly-legal calls". Apparently, what we have right now in the trunk is "not legal enough". I'll work on the patch to stop

I'm seeing a difference in how LLVM 3.3 and 3.2 emit unaligned vector loads on AVX. 3.3 is splitting up an unaligned vector load but in 3.2, it was emitted as a single instruction (details below). In a matrix-matrix inner-kernel, I see a ~25% decrease in performance, which seems to be due to this. Any ideas why this changed? Thanks! Zach LLVM Code: define <4 x double> @vstore(<4 x