search for: floord

...m/gemm.c -I utilities/ utilities/polybench.c -mllvm -polly-ignore-aliasing -DPOLYBENCH_USE_SCALAR_LB :: init_array : entry.split => for.end56 if ((nj >= 1) && (nk >= 1) && (p_1 >= 1) && (p_4 >= 1)) { for (c2=0;c2<=p_4-1;c2+=32) { for (c3=max(-32*floord(p_1-12*p_4+10,32)-32*p_4,-32*c2-32*floord(-12*c2+p_1+10,32)-640);c3<=-20*c2;c3+=32) { for (c4=max(ceild(-c3-p_1-30,20),c2);c4<=min(min(floord(-c3,20),c2+31),p_4-1);c4++) { for (c5=max(c3,-20*c4-p_1+1);c5<=min(-20*c4,c3+31);c5++) { Stmt_for_body41(c4,-20*c4-c5)...

Dear Tobias, Thank you very much for your very helpful advice. Yes, -debug-pass and -time-passes are two very useful and powerful options when evaluating the compile-time of each compiler pass. They are exactly what I need! With these options, I can step into details of the compile-time overhead of each pass. I have finished some preliminary testing based on two randomly selected files from

On 04/30/2013 04:13 PM, Star Tan wrote: > Hi all, [...] > How could I find out where the time is spent on between two adjacent Polly passes? Can anyone give me some advice? Hi Star Tan, I propose to do the performance analysis using the 'opt' tool and optimizing LLVM-IR, instead of running it from within clang. For the 'opt' tool there are two commands that should help

...utilities/polybench.c -mllvm -polly-ignore-aliasing >-DPOLYBENCH_USE_SCALAR_LB >:: init_array : entry.split => for.end56 >if ((nj >= 1) && (nk >= 1) && (p_1 >= 1) && (p_4 >= 1)) { > for (c2=0;c2<=p_4-1;c2+=32) { > for >(c3=max(-32*floord(p_1-12*p_4+10,32)-32*p_4,-32*c2-32*floord(-12*c2+p_1+10,32)-640);c3<=-20*c2;c3+=32) >{ > for >(c4=max(ceild(-c3-p_1-30,20),c2);c4<=min(min(floord(-c3,20),c2+31),p_4-1);c4++) >{ > for (c5=max(c3,-20*c4-p_1+1);c5<=min(-20*c4,c3+31);c5++) { > Stmt_...

...+ 5*new_iv_2 + 6*new_iv_3 Passing a symbolic expression, as you propose, could allow further simplifictions, however it also requires s to translate an isl_ast_expr to some kind of ScevExpr. This is non-trivial as we would need to teach SCEV about the different operands isl codegen could produce (floord, ceild, min, max, %, ....). I would suggest to leave this optimization for later and to first remove the dependence on the ivdeps pass. We can than evaluate, if this optimization should be done in SCEV or if we rather rely on instcombine or something similar to optimize the generated instructions f...

Hi Tobi, I would like to remove the SCEVRewriter code and replace it with a call to SCEVAddRec::apply (see attached a patch that adds just this function). More precisely I want to add another function called apply_map that applies a map (loop -> expr) on a given scev. This is the apply function on a multi-variate polynomial. So here is an overview of how I would like the scev code generator