search for: proving

...uire non-wrapping. Looks like we only consider an inbounds gep non-wrapping if it uses a unit stride. I am not sure why… Adam > > Part of the problem is that in this case ScalarEvolution does not try > to prove that {0,+,4} is <nsw> when the expression is constructed > (since proving that has non-trivial cost) [1]. To get ScalarEvolution > to try to prove that {0,+,4} has no-wrap properties, the client needs > to construct a sign-extend expression of {0,+,4}. SCEV will try to > change a sext of an add-rec to an add-rec of sexts and try to prove > no-wrap in the pr...

I am testing vectorization on the following test case: float x[1024], y[1024]; void myloop1() { for (long int k = 0; k < 512; k++) { x[2*k] = x[2*k]+y[k]; } } Vectorization failed due to "unsafe dependent memory operation". I traced the LoopAccessAnalysis.cpp and found the reason is the NoWrapFlag for SCEVAddRecExpr is not set and consequently the

Dear R-helper, Please suggest some methods for my question below. We measured the amount of protein A in patient blood in pre-treatment and post-treatment condition from 32 patients. -------------------------------------------- Pre-treatment Post-treatment Pat1 25 28 Pat2 19 15 Pat3 94

...trap values. Right; this is the second of the two approaches Eli originally suggested. For this approach, the IR is already sufficient to allow the optimizer to prove that i >=0 and to subsequently prove that the relevant pointers don't alias. BasicAA doesn't have any logic related to proving either i >= 0, or that _length and _data don't alias even if it somehow knew i >= 0. It could be taught both of those things, if someone were interested. SCEV-AA does have logic related to proving that i >= 0, but it happens to fail in this testcase; I haven't investigated it in d...

On 10 August 2013 15:32, Nuno Lopes <nunoplopes at sapo.pt> wrote: > Hi Jakob, > > Thanks for creating this interesting proposal. > Let me just comment on this part: > > >>> What might be better is to put some abstract interface between >>> instcombine and the IR, so that instcombine can be run on these pseudo-MIs >>> as well as on IR. >>

Dear List, I'm now working on MLE and OSL estimators.I just noticed that the textbook argues they are joint normal distributed.But how to prove the conclusion? Thanks for your time in advance! Best, Ning

...gep non-wrapping if it uses a unit stride. I am not sure why… >> >> Adam >> >>> >>> Part of the problem is that in this case ScalarEvolution does not try >>> to prove that {0,+,4} is <nsw> when the expression is constructed >>> (since proving that has non-trivial cost) [1]. To get ScalarEvolution >>> to try to prove that {0,+,4} has no-wrap properties, the client needs >>> to construct a sign-extend expression of {0,+,4}. SCEV will try to >>> change a sext of an add-rec to an add-rec of sexts and try to prov...

...nite > loops. I believe the advantage of writing down the optimizations in a DSL is that the representation is way more succinct. The code in instcombine is a bit boring and repetitive (and huge!). We could, however, try to automatically convert from C++ to the DSL. Not impossible, AFAICT. Proving absence of loops is a bit more complicated than traditional static analysis. It often requires things like synthesis of ranking functions (to prove that the transition relation is well-founded), which is not trivial. Nuni

On Tue, Jan 20, 2015 at 3:27 PM, Michael Hennebry <hennebry at web.cs.ndsu.nodak.edu> wrote: > Unix and ntp handle leap seconds a bit differently. > Unix time increases during the leap second and drops back a second after. > Ntp freezes time during the leap second. > OS kernels may do either or neither. Does anyone have a succinct summary of how to prove to management-types that

Colleagues, I am trying to learn to use uniroot to solve a simple linear equation. I define the function, prove the function and a call to the function works. When I try to use uniroot to solve the equation I get an error message, Error in yfu n(x,10,20) : object 'x' not found. I hope someone can tell we how I can fix the problem

Skipped content of type multipart/alternative-------------- next part -------------- A non-text attachment was scrubbed... Name: PGP.sig Type: application/pgp-signature Size: 194 bytes Desc: This is a digitally signed message part Url : http://wrath.rubyonrails.org/pipermail/rails/attachments/20060206/cf123b71/PGP.bin

...be able to (in reasonable time) prove that UB /does/ happen (in many cases this would require complex interprocedural analysis - the array is defined in one function, maybe with a complex dynamic bound, then passed to another function and indexed using a non-trivial dynamic expression... statically proving that to be true or false is complex/expensive and so basically not done by any compiler - so any cases that are caught by the compiler are relatively trivial ("oh, you declared a const null pointer value, then dereferenced it within the same function", etc) & so don't happen very...

> only if the compiler can prove that the called function has > no side effects (such as modifying some global variables or > causing the program to exit). can it prove if the function resides in a shared library? -- View this message in context: http://old.nabble.com/DCE-and-external-function-tp29932485p29942236.html Sent from the LLVM - Dev mailing list archive at Nabble.com.

...SCEV was written) that SCEV is tuned for C/C++ and for C/C++ add recurrences are almost always nsw on construction so you don't have these, let's say "phase ordering", issues. However, for Java, I was under the impression that despite this caching issue we are fairly effective at proving ranges. Of course "fairly effective" does not help if the one thing we don't optimize is 90% of an Important Benchmark(TM). I see a couple of ways out of this: 1. Eagerly compute nsw/nuw on AddRec construction. I'd be happy with this if you can show that we don't regress...

Thanks very much for the quick response. I have read the text many times, but it was not very clear to me why checking the two conditions involving dominance frontiers is equivalent to proving the pair {entry, exit} defines a refined region. I was asking for an mathematical proof really. It sounds to me like there should be a theorem or two in the graph theory endorsing it. Or do you mean, the implementation is solely based on empirical observation instead of theory. The reason I am inte...

Hello Everyone, I want to raise a discussion about reasonability of late setting of nsw/nuw/nw flags to SCEV AddRecs through setNoWrapFlags method. A discussion about this have already happened in August last year, there was a concern about different no-wrap flags that come from different sequences of SCEV flags invocations. It was mentioned there that late setting of flags is actually a hack to

Hi Jakob, Thanks for creating this interesting proposal. Let me just comment on this part: >> What might be better is to put some abstract interface between >> instcombine and the IR, so that instcombine can be run on these >> pseudo-MIs as well as on IR. > > I like the idea of sharing code between IR and MI passes through an > abstract interface. I think that later

...SCEV was written) that SCEV is tuned for C/C++ and for C/C++ add recurrences are almost always nsw on construction so you don't have these, let's say "phase ordering", issues. However, for Java, I was under the impression that despite this caching issue we are fairly effective at proving ranges. Of course "fairly effective" does not help if the one thing we don't optimize is 90% of an Important Benchmark(TM). I see a couple of ways out of this: 1. Eagerly compute nsw/nuw on AddRec construction. I'd be happy with this if you can show that we don't regress...

Hi Nadav, > I almost missed your email because you replied only to the list. I understand your argument. I think that my fix addresses a part of it. It says that instructions that do not return should not be removed. > The current implementation of mayReturn is to check the 'noreturn' function attribute. Are you suggesting to add a new attribute that is called 'mustreturn'

Remember James McGovern and his articles about Rails and enterprise? Well... Check this one out where he "puts his money where his mouth is" http://duckdown.blogspot.com/2006/04/ruby-community-proved-mcgovern-wrong.html -- -------------- Jon Gretar Borgthorsson http://www.jongretar.net/