similar to: [LLVMdev] LLVM 3.3 JIT code speed

On Thu, Jul 18, 2013 at 9:07 AM, Stéphane Letz <letz at grame.fr> wrote: > Hi, > > Our DSL LLVM IR emitted code (optimized with -O3 kind of IR ==> IR passes) runs slower when executed with the LLVM 3.3 JIT, compared to what we had with LLVM 3.1. What could be the reason? > > I tried to play with TargetOptions without any success… > > Here is the kind of code we use to

Le 18 juil. 2013 à 19:07, Eli Friedman <eli.friedman at gmail.com> a écrit : > On Thu, Jul 18, 2013 at 9:07 AM, Stéphane Letz <letz at grame.fr> wrote: >> Hi, >> >> Our DSL LLVM IR emitted code (optimized with -O3 kind of IR ==> IR passes) runs slower when executed with the LLVM 3.3 JIT, compared to what we had with LLVM 3.1. What could be the reason? >>

I understand you to mean that you have isolated the actual execution time as your point of comparison, as opposed to including runtime loading and so on. Is this correct? One thing that changed between 3.1 and 3.3 is that MCJIT no longer compiles the module during the engine creation process but instead waits until either a function pointer is requested or finalizeObject is called. I would

Since updating to LLVM 3.3, the system is generating attributes such as: attributes #0 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"="true" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "unsafe-fp-math"="false"

On Tue, Jul 16, 2013 at 8:16 AM, Stéphane Letz <letz at grame.fr> wrote: > Hi, > > Our DSL emit sub-optimal LLVM IR that we optimize later on (LLVM IR ==> LLVM IR) before dynamically compiling it with the JIT. We would like to simply follow what clang/clang++ does when compiling with -O1/-O2/-O3 options. Our strategy up to now what to look at the opt.cpp code and take part of it

Hi, Our DSL emit sub-optimal LLVM IR that we optimize later on (LLVM IR ==> LLVM IR) before dynamically compiling it with the JIT. We would like to simply follow what clang/clang++ does when compiling with -O1/-O2/-O3 options. Our strategy up to now what to look at the opt.cpp code and take part of it in order to implement our optimization code. It appears to be rather difficult to follow

Sent from my Verizon Wireless 4G LTE DROID On Nov 17, 2016 5:53 PM, Mehdi Amini <mehdi.amini at apple.com<mailto:mehdi.amini at apple.com>> wrote: > > >> On Nov 17, 2016, at 4:33 PM, Hal Finkel <hfinkel at anl.gov<mailto:hfinkel at anl.gov>> wrote: >> >> >> ________________________________ >>> >>> From: "Warren

----- Original Message ----- > From: "Sanjay Patel" <spatel at rotateright.com> > To: "Hal J. Finkel" <hfinkel at anl.gov> > Cc: "Mehdi Amini" <mehdi.amini at apple.com>, "llvm-dev" > <llvm-dev at lists.llvm.org>, "cfe-dev" <cfe-dev at lists.llvm.org>, > "andrew kaylor" <andrew.kaylor at

Hi Sanjay, Thanks, I saw this flag and it's definitely should be considered, but it appeared to me to be static characteristic of target platform. I'm not sure how appropriate it would be to change its value from a front-end. It says "Has", while optional flag would rather say "Uses" meaning that implementation cares about floating point exceptions. Regards, Sergey

Hello everyone, I'd like to propose the attached patch to form FMA intrinsics aggressively, but in order to do so I need some clarification on the intended semantics for the various FP precision-related TargetOptions. I've summarized the three relevant ones below: UnsafeFPMath - Defaults to off, enables "less precise" results than permitted by IEEE754. Comments specifically

> On Dec 6, 2017, at 02:28, Haidl, Michael <michael.haidl at uni-muenster.de> wrote: > > The IR goes through a backend agnostic preparation phase that brings it into SSA from and changes the AS from 0 to 1. This sounds possibly problematic to me. The IR should be created with the correct address space to begin with. Changing this in the middle sounds suspect. > After this

Hi Owen, Having looked into this due to Clang failing PlumHall with it recently I can give an opinion... I think !NoExcessFPPrecision covers FMA completely. There are indeed some algorithms which give incorrect results when FMA is enabled, examples being those that do floating point comparisons such as: a * b + c - d. If c == d, it is still possible for that result not to equal a*b, as "+c

> > <32 x float> takes up 8 SSE registers; you're likely running into > issues with register pressure. Does it work better if you use > something smaller like <4 x float>? > > Besides that, I don't see any obvious issues. > > -Eli You are right yes. The code works faster with <4 x float> types, with still works a bit slower than the scalar

Le 2 juin 2010 à 12:21, Eli Friedman a écrit : > On Wed, Jun 2, 2010 at 2:59 AM, Stéphane Letz <letz at free.fr> wrote: >> Hi, >> >> We need to generate "Floating point constants" in our code. In http://llvm.org/docs/LangRef.html it is explained that FP has to follow special encoding rules to be handled by LLVM later one (hexadecimal coding...) >>

> ------------------------------ > > Message: 4 > Date: Wed, 2 Jun 2010 11:07:39 -0700 > From: Dale Johannesen <dalej at apple.com> > Subject: Re: [LLVMdev] Generating Floating point constants > To: St?phane Letz <letz at free.fr> > Cc: llvmdev at cs.uiuc.edu > Message-ID: <AEC895CC-E887-4329-8743-FA606BD401F6 at apple.com> > Content-Type:

Hi, We need to generate "Floating point constants" in our code. In http://llvm.org/docs/LangRef.html it is explained that FP has to follow special encoding rules to be handled by LLVM later one (hexadecimal coding...) Is there any code available in LLVM to handle this kind of "standard float to LLVM float" conversion? Thanks. Stéphane Letz

On Sat, May 29, 2010 at 1:23 AM, Stéphane Letz <letz at grame.fr> wrote: >> >> <32 x float> takes up 8 SSE registers; you're likely running into >> issues with register pressure.  Does it work better if you use >> something smaller like <4 x float>? >> >> Besides that, I don't see any obvious issues. >> >> -Eli > >

Le 5 juil. 2013 à 04:11, Tobias Grosser <tobias at grosser.es> a écrit : > On 07/04/2013 01:39 PM, Stéphane Letz wrote: >> Hi, >> >> Our DSL can generate C or directly generate LLVM IR. With LLVM 3.3, we can vectorize the C produced code using clang with -O3, or clang with -O1 then opt -O3 -vectorize-loops. But the same program generating LLVM IR version cannot be

Hi, We are experimenting directly generating vectorized LLVM IR (using <8 x float> kind of types), then compiling the code to SSE on a 64 bits machine. Right now the equivalent code in scalar mode sill outperform the SSE one. What is the quality of the SSE support in X86 LLVL backend? Are they any specific things to be aware of to improve the speed? Thanks Stéphane Letz

Le 5 juil. 2013 à 17:23, Arnold Schwaighofer <aschwaighofer at apple.com> a écrit : > > On Jul 5, 2013, at 9:50 AM, Stéphane Letz <letz at grame.fr> wrote: > >> >> Le 5 juil. 2013 à 04:11, Tobias Grosser <tobias at grosser.es> a écrit : >> >>> On 07/04/2013 01:39 PM, Stéphane Letz wrote: >>>> Hi, >>>>