similar to: [LLVMdev] Basic AliasAnalysis: Can GEPs with the same base but different constant indices into a struct alias?

On Sat, Jan 24, 2015 at 5:44 PM, Chandler Carruth <chandlerc at google.com> wrote: > > On Tue, Jan 20, 2015 at 12:27 PM, Ahmed Bougacha <ahmed.bougacha at gmail.com > > wrote: > >> Hi all, >> >> This is covered by (struct-path aware) TBAA, but BasicAA disagrees. >> See the attached testcase, where it prevents us from removing the >> redundant

On Mon, Feb 2, 2015 at 10:59 AM, Chandler Carruth <chandlerc at google.com> wrote: > > On Mon, Feb 2, 2015 at 10:55 AM, Ahmed Bougacha <ahmed.bougacha at gmail.com> > wrote: > >> Ah yes, the structs are what make it messy. >> >> How about the more useful constraint: >> - the (identical) base must point to a (possibly multidimensional) array >>

I've noticed that LLVM does a bad job of optimizing array indexing code for FORTRAN arrays declared using the ALLOCATABLE keyword. For example if you have something like the following: DOUBLE PRECISION,ALLOCATABLE,DIMENSION(:,:,:,:) :: QAV ... ALLOCATE( QAV( -2:IMAX+2,-2:JMAX+2,-2:KMAX+2,ND) ) ... DO L = 1, 5 DO K = K1, K2 DO J = J1, J2 DO I = I1, I2 II = I +

This buildbot seems to have been failing continuously for a couple of weeks now ( http://lab.llvm.org:8011/builders/llvm-mips-linux/builds/14658 ) - is anyone watching it/caring about it? ---------- Forwarded message ---------- From: <llvm.buildmaster at lab.llvm.org> Date: Wed, Sep 30, 2015 at 11:34 PM Subject: buildbot failure in LLVM on llvm-mips-linux To: Ahmed Bougacha

The failure is a bit odd. LLVM is ignoring $PWD because it doesn't have the same inode as '.'. This causes the test failure because DW_AT_comp_dir and $PWD differ. However, $PWD and '.' should be the same inode since $PWD is a symlink to the current directory and stat() looks through symlinks. > Since this latest board only has two cores, it will run slower and it will need

Hi Wonsun, can you please provide a testcase. Best wishes, Duncan. > I've noticed that LLVM does a bad job of optimizing array indexing > code for FORTRAN arrays declared using the ALLOCATABLE keyword. > > For example if you have something like the following: > > DOUBLE PRECISION,ALLOCATABLE,DIMENSION(:,:,:,:) :: QAV > ... > ALLOCATE( QAV(

I want to check if the values a and b in the program alias. int main() { int *a,*b; a=(int *)malloc(sizeof(int)); b=(int *)malloc(sizeof(int)); *a=10; *b=8; return 0; } I use the below code for this (getAnalysisUsage method has been defined) AliasAnalysis::Location loc1=AliasAnalysis::Location(k1); //a AliasAnalysis::Location loc2=AliasAnalysis::Location(k2); //b AliasAnalysis::AliasResult

[+llvmdev] Hi George, You're right, these tests are broken, and have been for a long time. As it turns out, at least in terms of the 2003-12-11-ConstExprGEP.ll test, this is related to a case we've been discussing in another thread ("Basic AliasAnalysis: Can GEPs with the same base but different constant indices into a struct alias?"). It seems like, to some extent, we used to

Thank You, It means vmovdqa64 zmm22, zmmword ptr [rip + .LCPI0_0] # zmm22 = [8,9,10,11,12,13,14,15] zmm22 will contain 64 bit constant values which are indexes here zmm22=8, 9, 10, 11, 12,13,14,15. not the values loaded from these locations. and zmm2 contains constant 4000. so, vpmuludq zmm14, zmm10, zmm2 ; will multiply the indexes values with 4000, as for array b the stride is 4000. zmm14=

Dear list, I have a vector of characters and a list of two named elements : i <- c("a","a","b","b","b","c","c","d") j <- list(j1 = c("a","c"), j2 = c("b","d")) I'm looking for a fast way to obtain a vector with names, as follows : [1] "j1" "j1"

Hi all, I am struggling to work out how to use the rownames from an edited dataframe rather than the row names from the original dataframe. In my data set i'm trying to extract several rows of data on particular individuals, i don't doubt i'm using the long way round but what i have in the way of a script is this: ##selecting the IDs from the dataframe individually

Dear R users, I have a few hundred .csv files which I need to put together (full outer joins on a common variable) to do a factor analysis. Each file may contain anywhere from a few hundred to a few thousand rows. What would be the most efficient way to do this in R? Please include some sample code if applicable. Thank you, b.

Hi Adarsh, > I want to check if the values a and b in the program alias. > > int main() { > int *a,*b; > a=(int *)malloc(sizeof(int)); > b=(int *)malloc(sizeof(int)); > *a=10; > *b=8; > return 0; > } > > I use the below code for this (getAnalysisUsage method has been defined) > > AliasAnalysis::Location loc1=AliasAnalysis::Location(k1); //a >

Hi, is there any way to keep a sub matrix dimension? exemple : i1<-1; i2<-1 j1<-2; j2<-3; ret <-matrix(1,4,4)[i1:i2,j1:j2] ; dim(ret) is NULL because the submatrix single col or single row is coerce to a vector automaticaly. How can i bypass this cast : submatrix->vector ??????? Thank you. --------------------------------- [[alternative HTML version

On 27 February 2015 at 21:26, Ahmed Bougacha <ahmed.bougacha at gmail.com> wrote: > Which brings us to my fallback proposal: what about disabling the > pass on darwin only? That's a decision for Jim/Evan. I'm ok if they are. > As for other targets, as a first step, making the pass run under -O3 > rather than -O1 is hopefully agreeable to everyone? Sounds reasonable.

Hi, I was looking into the semantics of GEP inbounds and some BasicAA rules and I'm wondering if it's valid in LLVM IR to allocate more than half of the address space with a global variable or an alloca. If that's a scenario want to consider, then we have problems :) Consider this C code (32 bits): #include <string.h> char obj[0x80000008]; char f() { char *p = obj +

I want to create a new array which selects values from an original array based on a function of the indices. That is: I want to create a new matrix Vnew[i,j,k]=Vold[i,j,ks] where ks is a function of the index elements i,j,k. I want to do this WITHOUT a loop. Call the function "ksfunction", and the array dimensions nis,njs,nks. I can do this using a loop as follows: # Loop version:

I filed a couple more, in case they're actually different issues: - http://llvm.org/bugs/show_bug.cgi?id=22412 - http://llvm.org/bugs/show_bug.cgi?id=22413 And that's pretty much it for internal changes. I'm fine with flipping the switch; Quentin, are you? Also, just to have an idea, do you (or someone else!) plan to tackle these in the near future? -Ahmed On Thu, Jan 29, 2015 at

Dear R People: Here is an interesting question(I think) Suppose I want to calculate Centered Moving Averages; i.e. x[1] <- ( sum(y[1:12]) )/12 x[2] <- ( sum(y[2:13]) )/12 and so on. Of course, this is easily done through loops. However, I have been trying to do this more elegantly, but have failed. I have tried things like j1 <- 1:109 j2 <- 12:120 x[1:109] <- ( sum( y[j1:j2])

On Fri, Feb 27, 2015 at 2:01 PM, Renato Golin <renato.golin at linaro.org> wrote: > On 27 February 2015 at 21:26, Ahmed Bougacha <ahmed.bougacha at gmail.com> wrote: >> Which brings us to my fallback proposal: what about disabling the >> pass on darwin only? > > That's a decision for Jim/Evan. I'm ok if they are. Jim, thoughts? > >> As for other