similar to: [LLVMdev] lli and runtime error

Hello Pushkar, >I get the error Could not resolve external global address: __dso_handle Based on what I've read on the mailing list in the past, that's because the interpreter doesn't support non-bitcode external addresses such as the function calls used to do input and output of data. As such the interpreter is quite useless in its current state. If you use JIT compilation it

Hello, I am trying to use Clang to generate AST for C code in a different format. This format is specific to the term rewriting system that we use for source-to-source transformations. Studying the ast generated by Clang using the "-ast-print" option to the clang-cc, I see that the pre-processor expands the code(expected). Is there a way to get around and not print all the information

Hi Nick, > Fundamentally the issue is that the system linker is supposed to define > __dso_handle when linking, but since there is no system linker between > your build of the .bc files and running lli, nobody has defined it. > > It seems reasonable to me that lli should define __dso_handle if it's > declared in the module. > > You could module tools/lli/lli.cpp to add

Ashok Nalkund wrote: > > Resending :(. Any pointers? Fundamentally the issue is that the system linker is supposed to define __dso_handle when linking, but since there is no system linker between your build of the .bc files and running lli, nobody has defined it. It seems reasonable to me that lli should define __dso_handle if it's declared in the module. You could module

Hi, I'm trying to follow the R-admin instructions for using MKL10 as the external BLAS compiling R-2.6.2 under Linux on a RH EL head node of a cluster. The configure process seems to have problems when it checks for dgemm in the BLAS. I'm using configure as: ./configure CC=icc F77=ifort --with-lapack="$MKL" --with-blas="$MKL" where $MKL is defined as in R-admin

> Resending :(. Any pointers? Are you seeing missing __dso_handle with the interpreter or with the JIT? Ciao, Duncan. > > tia > > On 5/21/2012 2:46 PM, Ashok Nalkund wrote: >> On 5/21/2012 11:15 AM, Nick Lewycky wrote: >>> Ashok Nalkund wrote: >>>> Resending, any pointers? I demangled the symbol and it turns out to be: >>>>

Hi, I am trying to compile a single module (dgemm.c) using clang and generate dgemm.ll. Command: clang -emit-llvm dgemm.c -S -o dgemm.ll Some of the functions are declared as static and clang ignores these functions: One of the functions is: static void innerloop(double a, const double* b, double* c, double beta) { *b = a*beta; return; } Is there any way to make clang generate llvm code

>>>>> Cohn, Robert S <robert.s.cohn at intel.com> >>>>> on Sat, 7 Jan 2017 16:41:42 +0000 writes: > I am using R to multiply some large (30k x 30k double) > matrices on a 64 core machine (xeon phi). I added some timers > to src/main/array.c to see where the time is going. All of the > time is being spent in the matprod function, most of that

Hi Arvind, you need to use it effectively : static int add(int a, int b) { return a+b; } int use() { return add(4,5); } ==> ; ModuleID = '<stdin>' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i386-pc-linux-gnu" define i32 @use()

I am using R to multiply some large (30k x 30k double) matrices on a 64 core machine (xeon phi). I added some timers to src/main/array.c to see where the time is going. All of the time is being spent in the matprod function, most of that time is spent in dgemm. 15 seconds is in matprod in some code that is checking if there are NaNs. > system.time (C <- B %*% A) nancheck: wall time

Ah ok. Sorry. :) I think that argument "-femit-all-decls" will help you. Olivier. On Mon, Dec 14, 2009 at 9:18 PM, Arvind Sudarsanam < arvind.sudarsanam at aggiemail.usu.edu> wrote: > Hi, > > Sorry for not being specific. I just wanted to know if there is any > way at all to force clang to generate intermediate code for static > functions when they are not being

Hi I am new to writing C code and am trying to write an R extension in C. I have hit a wall with F77_CALL(dgemm) in that it produces wrong results. The code below is a simplified example that multiplies the matrices Ab and Bm to give Cm. The results below show clearly that Cm is wrong. Am= 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Bm= 1 1 1 1 1

Dear R-devel, I've written a numerical solver for SOCPs (second order cone programs) in R, and now I want to move most of the solver code into C for speed. I've written combined R/C packages before, but in this case I need to do matrix operations in my C code. As I have never done that before, I'm trying to write some simple examples to make sure I understand the basics. I am stuck

Hello again, I'm exploring llvm's intermediary representation with this code: //////////////////////////////////// %.LC0 = internal constant [13 x sbyte] c"hello world\0A\00" declare int %puts(sbyte*) implementation int %main() { %cast210 = getelementptr [13 x sbyte]* %.LC0, long 0, long 0; call int %puts(sbyte* %cast210) ret int 0 }

Hi, Sorry for not being specific. I just wanted to know if there is any way at all to force clang to generate intermediate code for static functions when they are not being called anywhere inside the current module. Other compilers seem to generate intermediate code (lcc, for instance). Thanks for your reply..Olivier. Sincerely Arvind On Mon, Dec 14, 2009 at 1:10 PM, Olivier Meurant

Hi Robert, thanks for the report and your suggestions how to make the NaN checks faster. Based on my experiments it seems that the "break" in the loop actually can have positive impact on performance even in the common case when we don't have NaNs. With gcc on linux (corei7), where isnan is inlined, the "break" version uses a conditional jump while the

On 5/21/2012 11:15 AM, Nick Lewycky wrote: > Ashok Nalkund wrote: >> Resending, any pointers? I demangled the symbol and it turns out to be: >> std::__1::locale::use_facet(std::__1::locale::id&) const > > My guess is that you've got a .bc file produced on a mac using libc++ > (hence the ::_1 part) and you're trying to run it on linux with > libstdc++ (which

I am trying to figure out how theora timestamps map to absolute time values. This information would be needed to seek by 1 sec, 5 secs etc. I understand the granulepos is interpreted as Absolute number of key frames so far | number of frames since last key frame. E.g. 1000|1, 1000|2, ... How does this map to absolute time in millisecs since the start, I can think of the following formula: Total

Initially, tt was not a problem with the verifier, we were just inserting a wrong instruction in our pass. Also we were able to figure out the error you pointed out and our pass is running as intended!!! Thanks a lot. On Fri, May 7, 2010 at 12:02 AM, Nick Lewycky <nicholas at mxc.ca> wrote: > Adarsh Yoga wrote: > >> I was able to solve that >> > > What was wrong? I

> Do you have R code (including set.seed(.) if relevant) to show on how to generate > the large square matrices you've mentioned in the beginning? So we get to some > reproducible benchmarks? Hi Martin, Here is the program I used. I only generate 2 random numbers and reuse them to make the benchmark run faster. Let me know if there is something I can do to help--alternate