Re: [gmx-users] intel compiler + amd64: help needed.

Jones de Andrade wrote: Ok, I think I got it. Compiled. ;) About a thousand tricks. I'll publish a tutorial later. But, basically: 1 - Change all invsrt calls to gmx_inv-sqrt (57 entries, 12 files); 2 - Edit the Makefile.in in the nb_kernels directories, changing the line "

Re: [gmx-users] intel compiler + amd64: help needed.

Ok, I think I got it. Compiled. ;) About a thousand tricks. I'll publish a tutorial later. But, basically: 1 - Change all invsrt calls to gmx_inv-sqrt (57 entries, 12 files); 2 - Edit the Makefile.in in the nb_kernels directories, changing the line "LTCCASCOMPILE = $(LIBTOOL) --mode=compile $(CCAS

Re: [gmx-users] intel compiler + amd64: help needed.

> > If it is, just let me know, cause I'm tracking those down (finding a script > way to back up original files before edition.) and at the same time trying > to find some sort of "magical flag" to convince icc to *not* use the > "invsqrt" function or the "math.h" file at all (second choice seems

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all. Of course I want SSE enabled. There is no sense in benching it against different compilers if in one you enable the speed and on the other, and in the other you remove almost every optimization. That's why in the last message I've sent the eror I'm getting in the compilation: ***

Re: [gmx-users] intel compiler + amd64: help needed.

Hi, Erik: I'm sure about the flags. Not using the ones for SSE3, despite my processor having those. I'll just try to activate those as soon as I get a compiling procedure. By now, I even using the -- disable-x86-64-sse flag for configure, just to be sure. ;) Also, I tried the "CC_FOR_BUI

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all. :) Last message today. It seems that I succeded in switching all "invsqrt" functins to a new "gmx_invsqrt". Just 57 in the 12 files below: *** include/vec.h src/mdlib/csettle.c src/mdlib/vsite.c src/mdlib/clincs.c src

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all. On 6/1/06, David van der Spoel <[EMAIL PROTECTED]> wrote: > Trying withou the "-ipo" flag now. It still yelds the same weird warning > (warning #147: declaration is incompatible with "double invsqrt(double)"> (declared at line 565 of "/usr/local/intel/include/math.h")rename all gromacs inv

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all!. First, sorry for the delay in getting in touch again. Bureaucracy. :P Second, thanks a lot for everybody for advising on this subject. It's helping me a lot to solve this! :) Anyway, here is where I'm now: Erik: I'm sure about the flags. Not using the ones for SSE3, despite my

Re: [gmx-users] intel compiler + amd64: help needed.

Hi Jones,There are two things I can think of immediately:1. Make sure to set optimization and architecture options for portable code. By default, the intel compiler might generate binaries that use SSE3-specific stuff, although it is unlikely in this part of the code.2. Try setting "CC_FOR_BUILD=gc

Re: [gmx-users] intel compiler + amd64: help needed.

make[3]: ** [grompp] Erro 1 make[3]: Leaving directory `/home/johannes/src/gromacs/gromacs-3.3/src/kernel' make[2]: ** [all-recursive] Erro 1 make[2]: Leaving directory `/home/johannes/src/gromacs/gromacs-3.3/src' make[1]: ** [all] Erro 2 make[1]: Leaving directory `/home/johannes/src/gromacs/

Re: [gmx-users] intel compiler + amd64: help needed.

Are you using the ICC with EM64T extension? I would like you to look at the nature of the program: GROMACS code is unique as comparing to other computational software: its core code is written in assembly code which simply can not be optimized further by the compiler. What can be optimized is

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all. First of all, thanks everybody for the help. ;) Yang, I know, I know. I should keep to the GNU. But lets say that I get addicted to the intel compiler on a cpmd winter school, when it appeared in front of me as the only freely available for both amd and intel cpus fortran90 compiler. Sinc

Re: [gmx-users] intel compiler + amd64: help needed.

Can you try to generate a backtrace for the mknb crash using gdb, in case this offers some insight?$ gdb mknbgdb> r -software_invsqrt...gdb> bt(backtrace output appears here)--TomOn May 31, 2006, at 11:33 AM, Jones de Andrade wrote:Hi David. First, thanks for your time. ;) Unfortunatelly, it seem

Re: [gmx-users] intel compiler + amd64: help needed.

Hi, Jones I can understand your sentiment after seeing these texts from Mars. I shared the same when I was trying aggressive optimization (with Gentoo Linux, FYI) and later with whatever software. But don't keep trying such as it is not worthy the efforts you put (knowing the intrinsics of sy

Re: [gmx-users] intel compiler + amd64: help needed.

> But now I have a (big) question: this code > compiles properly with gcc. > Ok, maybe that means difference of tolerance > between different compilers. But the code also compiles with intel on intel > machines, and seems to work properly on older amd machines (not quite sure > if they need to be

Re: [gmx-users] intel compiler + amd64: help needed.

Jones de Andrade wrote: Hi all! Ok, got what should be done. But now I have a (big) question: this code compiles properly with gcc. Ok, maybe that means difference of tolerance between different compilers. But the code also compiles with intel on intel machines, and seems to work properly on

Re: [gmx-users] intel compiler + amd64: help needed.

Hi David. And how could I do it? I mean, isn't mknb a significative piece of the code? What is it for? Would not that it yeld any problem? And how to do it? Am I wrong, or should I first I run configure as it is for gcc, save the mknb msakefile on it's directory, then configure for icc, substitut

Re: [gmx-users] intel compiler + amd64: help needed.

Jones de Andrade wrote: Hi all. Program received signal SIGSEGV, Segmentation fault. 0x2b15a88c in __find_specmb () from /lib64/tls/libc.so.6 (gdb) where #0 0x2b15a88c in __find_specmb () from /lib64/tls/libc.so.6 #1 0x2b140e6f in vfprintf () from /lib64/tls/libc.so.6

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all! Ok, got what should be done. But now I have a (big) question: this code compiles properly with gcc. Ok, maybe that means difference of tolerance between different compilers. But the code also compiles with intel on intel machines, and seems to work properly on older amd machines (not quite

Re: [gmx-users] intel compiler + amd64: help needed.

> > Program received signal SIGSEGV, Segmentation fault. > 0x2b15a88c in __find_specmb () from /lib64/tls/libc.so.6 > (gdb) where > #0 0x2b15a88c in __find_specmb () from /lib64/tls/libc.so.6 > #1 0x2b140e6f in vfprintf () from /lib64/tls/libc.so.6 > #2 0x2b15e2

Re: [gmx-users] intel compiler + amd64: help needed.

Hi all. Sorry for answering twice, but I tried a bit more with the debuggers and, despite not finding how to fix what I'm doing wrong, I could get some extra information: Using "idb": Intel(R) Debugger for Intel(R) EM64T -based Applications, Version 9.0-16, Build 20051121 (idb) load ./mkn

Re: [gmx-users] intel compiler + amd64: help needed.

David van der Spoel wrote: > The mknb_func.coul is of another type (a subset of int) there you can > give its value to an integer, but not necessarily the other way around. > I find it very unlikely that this would crash the program. The enum and int variables should be the same size so the ac

Re: [gmx-users] intel compiler + amd64: help needed.

Jones de Andrade wrote: Hi David. Well, here is what it yelds: GNU gdb 6.3 Copyright 2004 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copyin

Re: [gmx-users] intel compiler + amd64: help needed.

Hi David. Well, here is what it yelds: GNU gdb 6.3 Copyright 2004 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. Th

Re: [gmx-users] intel compiler + amd64: help needed.

Jones de Andrade wrote: Hi David. Thanks a lot. But almost there. first, coulsave is already n "int" type variable: int vdwsave,coulsave,read_from_mem,write_to_mem; (line 294) So, the change from: mknb_func.coul=coulsave; mknb_func.vdw=vdwsave; To: mknb_func.coul=(int)coulsave; mknb_func.

Re: [gmx-users] intel compiler + amd64: help needed.

Hi David. Thanks a lot. But almost there. first, coulsave is already n "int" type variable: int vdwsave,coulsave,read_from_mem,write_to_mem; (line 294) So, the change from: mknb_func.coul=coulsave; mknb_func.vdw=vdwsave; To: mknb_func.coul=(int)coulsave; mknb_func.vdw=(int)vdwsave; Was point

Re: [gmx-users] intel compiler + amd64: help needed.

> mknb_innerloop.c(362): warning #188: enumerated type mixed with another type > mknb_func.coul=coulsave; > ^ I don't have the most recent source code on my machine, but the way around this may be to use an explicit cast. Assuming that

Re: [gmx-users] intel compiler + amd64: help needed.

Hi David. First, thanks for your time. ;) Unfortunatelly, it seems that the simple flag it is not quite enough still... :( If it was not a well known fact that intel compiler yelds you some extra performance for gromacs, and if I would not want to try to make QM/MM of gromacs+cpmd (and, in this

Re: [gmx-users] intel compiler + amd64: help needed.

Jones de Andrade wrote: Hi all. Ok, I admit that I've posted it before. But I still not finding the error that makes the gromacs compilation with intel compilers not work on my AMD64 machine here. The error I get is still the following: **

[gmx-users] intel compiler + amd64: help needed.

Hi all. Ok, I admit that I've posted it before. But I still not finding the error that makes the gromacs compilation with intel compilers not work on my AMD64 machine here. The error I get is still the following: ** ./mk