similar to: [LLVMdev] [Polly] Summary of some expensive compiler passes, especially PollyDependence

On 08/08/2013 01:29 AM, Star Tan wrote: > Hi all, > > > I have summarized the top 10 compiler passes for Polly when compiling LLVM test-ssuite. Results can be viewed on: > https://gist.github.com/tanstar/581bcea1e4e03498f935/raw/f6a4ec4e8565f7a7bbdb924cd59fcf145caac039/Polly-top10 > > > Based on the comparison between "clang -O3" and "polly -O3"

At 2013-08-08 22:28:33,"Tobias Grosser" <tobias at grosser.es> wrote: >On 08/08/2013 01:29 AM, Star Tan wrote: >> Hi all, >> >> >> I have summarized the top 10 compiler passes for Polly when compiling LLVM test-ssuite. Results can be viewed on: >>    

At 2013-08-09 10:20:46,"Tobias Grosser" <tobias at grosser.es> wrote:>On 08/08/2013 06:27 PM, Star Tan wrote: >> At 2013-08-08 22:28:33,"Tobias Grosser" <tobias at grosser.es> wrote: >>> On 08/08/2013 01:29 AM, Star Tan wrote: >>>> Hi all, >>>> >>>>

On 08/08/2013 06:27 PM, Star Tan wrote: > At 2013-08-08 22:28:33,"Tobias Grosser" <tobias at grosser.es> wrote: >> On 08/08/2013 01:29 AM, Star Tan wrote: >>> Hi all, >>> >>> >>> I have summarized the top 10 compiler passes for Polly when compiling LLVM test-ssuite. Results can be viewed on: >>>

On 08/08/2013 07:45 PM, Star Tan wrote: > At 2013-08-09 10:20:46,"Tobias Grosser" <tobias at grosser.es> wrote:>On 08/08/2013 06:27 PM, Star Tan wrote: >>> At 2013-08-08 22:28:33,"Tobias Grosser" <tobias at grosser.es> wrote: >>>> On 08/08/2013 01:29 AM, Star Tan wrote: >>>>> Hi all, >>>>>

Hi Sebastian, Recently, I found the "Polly - Calculate dependences" pass would lead to significant compile-time overhead when compiling some loop-intensive source code. Tobias told me you found similar problem as follows: http://llvm.org/bugs/show_bug.cgi?id=14240 My evaluation shows that "Polly - Calculate dependences" pass consumes 96.4% of total compile-time overhead

On 08/16/2013 02:42 AM, Star Tan wrote: > At 2013-08-16 12:44:02,"Tobias Grosser" <tobias at grosser.es> wrote: >> Hi, >> >> I tried to reproduce your findings, but could not do so. > > > Sorry, I did not put all code in my previous email because the code seems a little too long and complicated. > You can refer to the detailed C code and LLVM IR

At 2013-08-16 12:44:02,"Tobias Grosser" <tobias at grosser.es> wrote: >Hi, > >I tried to reproduce your findings, but could not do so. Sorry, I did not put all code in my previous email because the code seems a little too long and complicated. You can refer to the detailed C code and LLVM IR code on http://llvm.org/bugs/show_bug.cgi?id=16843 There are four attachments

At 2013-08-16 22:32:30,"Tobias Grosser" <tobias at grosser.es> wrote: >> >> Yes, I have changed the original code to the form you suggested: >> for (i >> for (j >> ... >> x=1 > >Sorry, I meant > x[0] += > It is interesting that Polly would run much faster if we change the

On 07/28/2013 06:52 AM, Star Tan wrote: > Hi Tobias, > > I tried to investigated the problem related to ScopInfo, but I need your > help on handling some problems about ISL and SCEV. I copied the list as the discussion may be helpful for others. @Sven, no need to read all. Just search for your name. [..] >>The interesting observation is, that Polly introduces three parameters

Hi, I would like to know if it is possible to extract the source level names of variables that create loop-carried dependencies. For example, for the following code: for (int i = 0; i < A_ROW; i++) { for (int j = 1; j < B_COL; j++) { a_matrix[i][j] = a_matrix[i][j - 1]; } } I get the following AST: #pragma omp parallel for

Hi Duncan & Tobi, Thanks a lot for your interest, and for pointing out differences in GIMPLE I missed. Attached is simplified test case. Is it good? Tobi, regarding runtime alias analysis: in KernelGen we already do it along with runtime values substitution. For example: <------------------ __kernelgen_main_loop_17: compile started ---------------------> Integer args substituted:

On 07/29/2013 09:15 AM, Sven Verdoolaege wrote: > On Mon, Jul 29, 2013 at 07:37:14AM -0700, Tobias Grosser wrote: >> On 07/29/2013 03:18 AM, Sven Verdoolaege wrote: >>> On Sun, Jul 28, 2013 at 04:42:25PM -0700, Tobias Grosser wrote: >>>> Sven: In terms of making the behaviour of isl easier to understand, >>>> it may make sense to fail/assert in case

Hi all, I have evaluated Polly's performance on LLVM test-suite with latest LLVM (r188054) and Polly (r187981).  Results can be viewed on: http://188.40.87.11:8000. There are mainly five new tests and each test is run with 10 samples: clang (run id = 27):  clang -O3 pollyBasic (run id = 28):  clang -O3 -load LLVMPolly.so pollyNoGen (run id = 29):  pollycc -O3 -mllvm -polly-optimizer=none

On 01/01/2013 02:45 PM, Duncan Sands wrote: > Hi Dmitry, > >> >> In our compiler we use a modified version LLVM Polly, which is very >> sensitive to >> proper code generation. Among the number of limitations, the loop region >> (enclosed by phi node on induction variable and branch) is required to >> be free >> of additional memory-dependent

On 08/10/2013 06:59 PM, Star Tan wrote: > Hi all, > > I have evaluated Polly's performance on LLVM test-suite with latest LLVM (r188054) and Polly (r187981). Results can be viewed on: http://188.40.87.11:8000. Hi Star Tan, thanks for the update. > There are mainly five new tests and each test is run with 10 samples: > clang (run id = 27): clang -O3 > pollyBasic (run id =

Hi Michael, Thanks for your answer. Checking the indexes looks as an another straightforward approach to identify the actual loop-carried dependencies. What is not clear to me still is how it looks like the code to extract the ScopArrayInfo pointer from the user part. Miguel On 08/05/2016 02:32 PM, Michael Kruse wrote: > I just realized this is a crossposting llvm-dev/polly-dev. It has

Hi, I want to find simple loops in LLVM bytecode, and extract the basic information of the loop. For example: for (i=0; i<1000; i++) sum += i; I want to extract the bound [0, 1000), the loop variable "i" and the loop body (sum += i). What should I do? I read the LLVM API document, and find some useful classes like "Loop", "LoopInfo". But I do

Hi Tobias and Sven, Thanks for your discussion and suggestion. @Sven: ISL actually allows users to have different identifiers with the same name. The problem that we have discussed is caused by incorrect usage of isl_space in Polly, so please do not worry about ISL library. You can skip the following information related to Polly implementation. @Tobias and Polly developers: I have attached

At 2013-08-12 01:18:30,"Tobias Grosser" <tobias at grosser.es> wrote: >On 08/10/2013 06:59 PM, Star Tan wrote: >> Hi all, >> >> I have evaluated Polly's performance on LLVM test-suite with latest LLVM (r188054) and Polly (r187981).  Results can be viewed on: http://188.40.87.11:8000. > >Hi Star Tan, > >thanks for the update. >