similar to: [LLVMdev] IR optimization pass ideas for backend porting before ISel

The scoreboard hazard detector that I've added for the PPC 440 is not detecting hazards as it should (which certainly could be my fault somehow, but...). For example, it will produce a schedule that looks like... SU(28): 0x127969b0: f64,ch = LFD 0x12793aa0, 0x1277b4f0, 0x127965b0<Mem:LD8[%scevgep100](tbaa=!"double")> [ORD=41] [ID=28] SU(46): 0x12796ab0: f64 = FADD 0x127969b0,

In the case when coming from C it was probably the loop unroller and SLP vectorizer which vectorized the code. Potentially I could do the same in the IR. However, the loop body that is generated in the IR can get very large. Thus, the loop unroller will refuse to unroll the loop in a large number of (important) cases. Isn't there a way to convince the loop vectorizer that it should

I am trying a setup where the one loop is rewritten as two loops. This avoids the 'rem' and 'div' instructions in the index calculation (which give the loop vectorizer a hard time). However, with this setup the loop vectorizer complains about a too small loop. LV: Checking a loop in "main" LV: Found a loop: L3 LV: Found a loop with a very small trip count. This loop

Hello, I was trying to trace the Loop vectorizer of the LLVM, I wrote a simple loop with a clear dependency. But found that the debug shows that 'we can vectorize this loop' Here you are my loop with dependency: for(int k=20;k<50;k++) dataY[k] = dataY[k-1]; And the debug prints: LV: Checking a loop in "main" LV: Found a loop: for.body4 LV: Found an

Notice that the code you provided, for globals and stack allocations, at least, is semantically equivalent to: int a = d[19]; for(int k = 20; k < 50; k++) dataY[k] = a; Like so, the load you see missing was redundant, probably hoisted by GVN/PRE and replaced with "%.pre". H. On Sun, Nov 3, 2013 at 11:26 AM, Sara Elshobaky <sara.elshobaky at gmail.com>wrote: >

Hi Sarah, the loop vectorizer runs not on the C code but on LLVM IR this c code was lowered to. Before the loop vectorizer runs many other optimization change the shape of this IR. You can see in the LLVM IR you referenced below, a preceding LLVM IR transformation has change your loop from: > for(int k=20;k<50;k++) > dataY[k] = dataY[k-1]; to > int a = d[19]; >

Actually what I meant in my original loop, that there is a dependency between every two consecutive iterations. So, how the loop vectorizer says 'we can vectorize this loop'? for(int k=20;k<50;k++) dataY[k] = dataY[k-1]; From: Henrique Santos [mailto:henrique.nazare.santos at gmail.com] Sent: Sunday, November 03, 2013 4:28 PM To: Sara Elshobaky Cc: <llvmdev at

Codeprepare and independent blocks are introducing these loads and stores. These are prepasses that polly runs prior to building the dependence graph to transform scalar dependences into data dependences. Ether was working on eliminating the rewrite of scalar dependences. On Thu, Aug 15, 2013 at 5:32 AM, Star Tan <tanmx_star at yeah.net> wrote: > Hi all, > > I have investigated the

Hi Nick, Thanks for looking into it. I have tried that as well but it didn't worked. "AddExpr->getOperand(0))" node is: " (4 * (sext i32 {2,+,2}<%for.body4> to i64))<nsw>" When I cast this to "SCEVAddRecExpr" it returns NULL. Regards, Ashutosh -----Original Message----- From: Nick Lewycky [mailto:nicholas at mxc.ca] Sent: Thursday, March 19,

I do not think that running SROA before polly is a good idea: it would defeat the purpose of the code preparation passes that polly intentionally schedules for the data dependence analysis. If you remove the data references before polly runs, you would miss them in the dependence graph: that could lead to incorrect transforms. On Thu, Aug 15, 2013 at 7:28 PM, Star Tan <tanmx_star at

Yes, I can get "SCEVAddRecExpr" from operands of "(sext i32 {2,+,2}<%for.body4> to i64)". So whenever SCEV cast to "SCEVAddRecExpr" fails, we have drill down for such patterns ? Is that the right way ? Regards, Ashutosh -----Original Message----- From: Nick Lewycky [mailto:nicholas at mxc.ca] Sent: Thursday, March 19, 2015 1:02 PM To: Nema, Ashutosh Cc:

Hi Sebpop, Thanks for your explanation. I noticed that Polly would finally run the SROA pass to transform these load/store instructions into scalar operations. Is it possible to run such a pass before polly-dependence analysis? Star Tan At 2013-08-15 21:12:53,"Sebastian Pop" <sebpop at gmail.com> wrote: >Codeprepare and independent blocks are introducing these loads and

Sorry typo in test case, Please ignore previous mail. Consider below case: for (j=1; j < itr; j++) { - - - - for (i=1; i < itr; i++) { { temp= var[i << 1]; - - - - - } } In the above example, we are unable to get "SCEVAddRecExpr" for "var[i << 1]" Its "SCEVAddRecExpr" is computable in *Outer Loop* I

Thanks Sanjoy. > To be pedantic, "var[i<<1]" is not an add recurrence, but "&var[i << > 1]" is an add recurrence. I'll assume that's that you meant. Yes, I meant the same. > I think that is because in C, multiplication is nsw but left shift is > not and so "i << 1" can legitimately sign-overflow but i * 2 cannot >

Hi all, I have investigated the 6X extra compile-time overhead when Polly compiles the simple nestedloop benchmark in LLVM-testsuite. (http://188.40.87.11:8000/db_default/v4/nts/31?compare_to=28&baseline=28). Preliminary results show that such compile-time overhead is resulted by the complicated polly-dependence analysis. However, the key seems to be the polly-prepare pass, which introduces

Hi, I'm trying to cast one of the SCEV node to "SCEVAddRecExpr". Every time cast return NULL, and I'm unable to do this. SCEV Node: ((4 * (sext i32 {2,+,2}<%for.body4> to i64))<nsw> + %var)<nsw> Casting: const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(SCEVNode); 'var' is of type float pointer (float*). Without 'sext' it works, but

Hello Yes, I agree that the SCEV cannot be simplified. Is my understanding correct that it is passed to a function like "isKnownNegative"? Which could still be able to prove is always true. The delinearisation may be valid, depending on exactly how you define delinearisation (under what conditions it should be giving results). It would be invalid for DA to return a dependency of [0

On 08/15/2013 03:32 AM, Star Tan wrote: > Hi all, Hi, I tried to reproduce your findings, but could not do so. > I have investigated the 6X extra compile-time overhead when Polly compiles the simple nestedloop benchmark in LLVM-testsuite. (http://188.40.87.11:8000/db_default/v4/nts/31?compare_to=28&baseline=28). Preliminary results show that such compile-time overhead is resulted by

Hi there, I've installed the whole cygwin package on an NT4 machine. I've been trying to rsync from the NT box to a box running openbsd (an account on an isp). I'm using a test directory to see if things are working. At the end of transmission - I get an error - does anyone have any ideas as to what it means? (i.e should I talk to the ISP because its something that's a

In https://bugs.llvm.org/show_bug.cgi?id=38011 (see also https://reviews.llvm.org/D48721) a problem was revealed related to llvm.loop metadata. The fault was that clang added the !llvm.loop metadata to branches outside of the loop (not only the loop latch). That was not handled properly by some opt passes (simplifying cfg) since it ended up merging branch instructions with different !llvm.loop