search for: toddi

Yes that is what he meant. "-dce, -adce, etc" are command line options consumed by tools/opt/opt.cpp to give to the PassManagerBuilder that it creates. The parsing of those options doesn't exist in any of the llvm library code that is linked into clang. Clang has its own code for populating a PassManagerBuilder in tools/clang/lib/CodeGen/BackendUtil.cpp ~Craig On Mon, Jan 8, 2018

//mllvm Options can be dumped by clang -v -help -mllvm and clang -v --help-hidden --> //mllvm Options can be dumped by clang -v -help -mllvm and clang -v --help-hidden -mllvm On Tue, Jan 9, 2018 at 3:09 AM, toddy wang <wenwangtoddy at gmail.com> wrote: > Thanks, Craig. > > So, clang -Xclang -disable-llvm-passes actually disables all the LLVM > passed populated by clang so

Thanks, Craig. So, clang -Xclang -disable-llvm-passes actually disables all the LLVM passed populated by clang so that there is no middle-end optimization on bc files. clang -O2 LULESH.c //clang is the driver, invoking cc1, cc1as, ld //options can be passed through to cc1 directly. //maybe have different names, e.g.

2018-01-08 8:59 GMT-08:00 toddy wang <wenwangtoddy at gmail.com>: > > > On Mon, Jan 8, 2018 at 11:53 AM, Mehdi AMINI <joker.eph at gmail.com> wrote: > >> >> >> 2018-01-08 8:41 GMT-08:00 toddy wang <wenwangtoddy at gmail.com>: >> >>> Hi Medhi, >>> >>> It seems -mllvm does not work as expected. Anything wrong?

2018-01-08 8:41 GMT-08:00 toddy wang <wenwangtoddy at gmail.com>: > Hi Medhi, > > It seems -mllvm does not work as expected. Anything wrong? > > [twang15 at c92 temp]$ clang++ -O3 -mllvm *-deadargelim* LULESH.cc > clang (LLVM option parsing): Unknown command line argument > '-deadargelim'. Try: 'clang (LLVM option parsing) -help' > clang (LLVM option

Mehdi, I found -unroll-max-count can be passed w/ -mllvm. -dce, -adce, etc, are also dumped by 'opt --help-hidden'. However, they cannot be passed w/ -mllvm. Is this what "You can't schedule passes this way, only set parameters like -unroll-threshold=<uint> etc." means? [twang15 at c89 temp]$ clang++ -mllvm -unroll-max-count=4 -mllvm -dce -save-temps LULESH.cc clang

On Mon, Jan 8, 2018 at 11:53 AM, Mehdi AMINI <joker.eph at gmail.com> wrote: > > > 2018-01-08 8:41 GMT-08:00 toddy wang <wenwangtoddy at gmail.com>: > >> Hi Medhi, >> >> It seems -mllvm does not work as expected. Anything wrong? >> >> [twang15 at c92 temp]$ clang++ -O3 -mllvm *-deadargelim* LULESH.cc >> clang (LLVM option parsing):

2018-01-07 23:16 GMT-08:00 toddy wang <wenwangtoddy at gmail.com>: > -mllvm <value> Additional arguments to forward to LLVM's option > processing > > This is dumped by clang. I am not sure what I am supposed to put as value > in order to tune unrolling/inlining threshold. > As the help says, this is used to pass argument to LLVM itself. If you remember

Hi Medhi, It seems -mllvm does not work as expected. Anything wrong? [twang15 at c92 temp]$ clang++ -O3 -mllvm *-deadargelim* LULESH.cc clang (LLVM option parsing): Unknown command line argument '-deadargelim'. Try: 'clang (LLVM option parsing) -help' clang (LLVM option parsing): Did you mean '-regalloc'? [twang15 at c92 temp]$ clang++ -O3 -mllvm *deadargelim* LULESH.cc

For the types of things that you are looking for, you may just want to try a bunch of -mllvm options. You can tune inlining and unrolling threshold like that, for example. On Jan 7, 2018 10:33 PM, "toddy wang via llvm-dev" <llvm-dev at lists.llvm.org> wrote: > Hi Mehdi, > > Now we have 5 pipelines. (In addition to the first 3, which I have > described in detail above,

-mllvm <value> Additional arguments to forward to LLVM's option processing This is dumped by clang. I am not sure what I am supposed to put as value in order to tune unrolling/inlining threshold. On Mon, Jan 8, 2018 at 2:02 AM, Sean Silva <chisophugis at gmail.com> wrote: > For the types of things that you are looking for, you may just want to try > a bunch of

Thanks a lot, it is clear to me now. BTW, for Clang's slowdown, I submit an issue here: https://github.com/flang-compiler/flang/issues/356 I have no idea about the root cause. Maybe due to debug symbols. But, I already use -DCMAKE_BUILD_TYPE=Release. Anyway, I believe there is a bug somewhere. On Sat, Jan 6, 2018 at 3:43 PM, Craig Topper <craig.topper at gmail.com> wrote: >

Why are you using build directions from "flang" which is a fortran compiler and maintained by different people than the LLVM/clang community? But then compiling C/C++ code? Their bug database should be used for filing bugs against the fortran compiler not a C/C++ compiler issue. ~Craig On Sat, Jan 6, 2018 at 1:04 PM, toddy wang <wenwangtoddy at gmail.com> wrote: > Thanks a

Hi Toddy, You can achieve what you're looking for with a pipeline based on `clang -Ox` + `opt -Ox` + `llc -Ox` (or lld instead of llc), but this won't be guarantee'd to be well supported across releases of the compiler. Otherwise, if there are some performance-releated (or not...) command line options you think clang is missing / would benefit, I invite you to propose adding them to

Thanks for pointing this out. The following is a more extensive test. //succeed means dumping help info without Error message. clang -help #succeed 'clang -help-hidden' #Fail 'clang --help-hidden' #Succeed clang -mllvm -v -help #Succeed clang -mllvm -v -help-hidden #Fail clang -mllvm -v --help-hidden #Succeed 'clang -mllvm -help -c LULESH.cc' #Succeed 'clang -mllvm

On Jan 5, 2018 11:30 PM, "toddy wang via llvm-dev" <llvm-dev at lists.llvm.org> wrote: What I am trying is to compile a program with different sets of optimization flags. If there is no fine-grained control over clang optimization flags, it would be impossible to achieve what I intend. LLD has -lto-newpm-passes (and the corresponding -lto-newpm-aa-pipeline) which allows you to

Hi Mehdi, Now we have 5 pipelines. (In addition to the first 3, which I have described in detail above, please refer my latest reply for details) 1. clang + opt + gold 2. clang + opt + lld 3. clang + GNU ld/ gold /lld 4. clang + opt + llc + clang clang -emit-llvm -O1 -Xclang -disable-llvm-passes for c/c++ to .bc generation and minimal front-end optimization opt for single bc file optimization

@Craig and @Michael After installing clang-5.0 (download from http://releases.llvm.org, does not have Flang build's slowdown mention above), 1. clang++ -O0 -Xclang -disable-O0-optnone -Xclang -disable-llvm-passes -c -emit-llvm -o a.bc LULESH.cc; opt -O3 a.bc -o b.bc; llc -O3 -filetype=obj b.bc -o b.o ; clang++ b.o -o b.out; ./b.out 20 runtime: 2.354069e+01 2. clang++ -O1 -Xclang

No, I meant LLD, the LLVM linker. This option for LLD is relevant for exploring different pass pipelines for link time optimization. It is essentially equivalent to the -passes flag for 'opt'. Such a flag doesn't make much sense for 'llc' because llc mostly runs backend passes, which are much more difficult to construct custom pipelines for (backend passes are often required

-disable-O0-optnone has no effect with anything other than -O0. -O0 being passed to clang also causes all functions to be marked noinline. I don't know if there is a command line option to turn that off. I recommend passing "-O1 -Xclang -disable-llvm-passes" to clang. Passing -O0 very specifically means disable optimizations. ~Craig On Sat, Jan 6, 2018 at 12:25 PM, toddy wang