I was going to post my little OpenMP self-tuition examples, but in the process of cleaning them up for public consumption, I discovered that Visual Studio 2010 doesn’t support OpenMP 3.0. Since about half of my examples focus on the use of the task directive for recursion and descent, it’s kinda pointless =/
OpenMP dissapointment from Visual Studio
kfsone · March 28, 2010 · Coding · intel compiler, openmp, visual studio · 7 Comments
Categories
Recent Posts
Recent Comments
Jonathan Witty on Spread wide before installing… | |
Jonathan Witty on Spread wide before installing… | |
kfsone on Vested Interests | |
fsdfsdfsd on C++0x lambdas suck | |
JonKalb on CPPCon 2017 | |
kfsone on Frontal lobe massage | |
Mike Wallis on Wink 2 review |
Top Posts
Social
Archives
Blog Stats
- 861,010 squirrels
1.31
adobe
apple
austin
battleground europe
battlegroundeurope
c++
c++0x
c++11
capture
cats
chrome
Coding
databases
dba
elite
firefox
games
gamevideos
Gaming
gcc
google
htc trophy
intel
kittens
linux
localization
mars
mea culpa
mercurial
microsoft
mmo
mmorpg
mysql
networking
online games
parallelism
parallelization
patches
python
rants
satans sphincter
space
spam
spore
squads
stargate
stargate atlantis
star wars
strat
subversion
svn
ubuntu
version control
video games
videos
virtualization
vista
visual source safe
visual studio
visualsvn
vmware
vss
web
windows
windows 7
windows 7 phone
wwii
WWIIOL
wwiionline
wwii online
x3
x3 terran conflict
x3:tc
zeromq
7 Comments
I just got the Intel compiler for my machine, and I noticed that it came with the Thread Building Blocks. It seems to do largely the same function that OpenMP does, did you see if it would work?
I’m only using evaluation versions of the Intel Compiler for Linux at this point: we’re at the wrong point in the release cycle for me to go changing compiler :)
TBB is tempting … But I’d want to replace all my existing threading stuff. Given the way I’ve done it, that ought not to take too long.
I’ve actually got a CMake project for building the hosts these days, and it took me about 2 hours to get the host projects compiling under ICC — just dumb little differences between GCC and ICC; and actually it caught a few things GCC has been ignoring, which translated to crash-bug fixes :)
I’m not sure if our Windows licenses for ICC translate to Linux licenses. If they do, then I might bump up my usage of it for testing purposes. I’ll have to ask Killer.
The licenses are architecture-specific, I think. At least the ones I got on the academic discount ($160 for compiler, MKL and TBB seems like a decent deal…)
I had the same experience of getting a bunch of warnings on icpc that I never got under gcc, some of which were useful. Less useful are those that are left now that warns that a function that says “if (cond) return a; else return b;” gets warnings that “a non-void function ends without a return”. Clearly they aren’t putting the same amount of work into their warnings as they are into optimizations…
It doesn’t cover the Mac client, I think its Windows only license.
Lutorm: They’re pretty good about fixing stuff; most of the time. But yeah, I found their warnings were a little too tightly bunched together – so that I wind up turning off some that would be useful so as not to get others that are just plain dumb.
Trackbacks and Pingbacks
[…] Categories: Coding Tags: openmp, parallelism, threading, visual studio, vs2010 An update from my earlier post, I got a reply to my OpenMP 3.0 ticket, for those of you who don’t have a Windows Live […]