similar to: [LLVMdev] speed and code size issues

We have some results that are somewhat entertaining and that relate to the size/speed discussion. The basic idea is exhaustive generation of C functions where "exhaustive" is qualified by some structural restrictions (depth of AST, node type, etc.). For one particular set of restrictions we ended up with about 7 million C functions. We then compiled each of these functions with 7

On Fri, Jul 17, 2009 at 4:14 PM, Jonathan Gray<jsg at goblin.cx> wrote: > This seems to go against notes such as > http://clang.llvm.org/features.html#performance > which claim clang is signifcantly faster than gcc. I think the URL you want is actually http://clang.llvm.org/performance.html. The difference isn't as dramatic when you consider code generation, at least for the

On Fri, Jul 17, 2009 at 5:12 PM, Jonathan Gray<jsg at goblin.cx> wrote: > On Fri, Jul 17, 2009 at 04:41:55PM -0700, Eli Friedman wrote: >> On Fri, Jul 17, 2009 at 4:14 PM, Jonathan Gray<jsg at goblin.cx> wrote: >> > This seems to go against notes such as >> > http://clang.llvm.org/features.html#performance >> > which claim clang is signifcantly

I'd look at if_spppsubr.o, which is big in the llvm directories and doesn't appear in the gcc directory listings at all (I assume the listings were fed through tail or something, but there's clearly a sizeable difference there) On Jul 17, 2009, at 5:41 PMPDT, Daniel Dunbar wrote: > On Fri, Jul 17, 2009 at 5:12 PM, Jonathan Gray<jsg at goblin.cx> wrote: >> On

On Fri, Jul 17, 2009 at 04:41:55PM -0700, Eli Friedman wrote: > On Fri, Jul 17, 2009 at 4:14 PM, Jonathan Gray<jsg at goblin.cx> wrote: > > This seems to go against notes such as > > http://clang.llvm.org/features.html#performance > > which claim clang is signifcantly faster than gcc. > > I think the URL you want is actually >

John Regehr wrote: > We have some results that are somewhat entertaining and that relate to the > size/speed discussion. > > The basic idea is exhaustive generation of C functions where "exhaustive" > is qualified by some structural restrictions (depth of AST, node type, > etc.). > > For one particular set of restrictions we ended up with about 7 million C

Very interesting study! However as others have asked, I would like to see some aggregated data. Also, how do you verify that the generated code is correct? How are you systematically generating these tests? In summary, do you have any paper coming along? :) Thanks, Nuno ----- Original Message ----- From: "John Regehr" <regehr at cs.utah.edu> To: "LLVM Developers Mailing

Hi, I'm trying to setup a Linux box with Samba (1.9.18p10) and it isn't having any luck making a home for itself on the network. If I go to a Windows95 machine, I see all (well most) machines on the net there, but not the Linux box. I can also not see any other machines from the Linux box. Here is log.nmbd output that repeats every 5 seconds or so: find_workgroup_on_subnet: workgroup

On Jul 17, 2009, at 4:41 PM, Eli Friedman wrote: > On Fri, Jul 17, 2009 at 4:14 PM, Jonathan Gray<jsg at goblin.cx> wrote: >> This seems to go against notes such as >> http://clang.llvm.org/features.html#performance >> which claim clang is signifcantly faster than gcc. > > I think the URL you want is actually > http://clang.llvm.org/performance.html. The

https://bugs.freedesktop.org/show_bug.cgi?id=76376 Priority: medium Bug ID: 76376 Assignee: nouveau at lists.freedesktop.org Summary: mesa does not build after nouveau loader changes Severity: normal Classification: Unclassified OS: OpenBSD Reporter: jsg at openbsd.org Hardware: Other Status:

All, I have been terribly pleased with openbsd 3.2 I was using it as a mail/samba server with postifx on a pentium pro 200 with 1 gig of memory. Well I wanted to add a larger harddrive than what I have in it now only 6.4 gig. Well the machine would not see it and for some reason after hours and hours of trying I can't get a tekram 395UW and a 18gig scsi to be the boot drive and having failed

Railsoids: FormTestHelper rules. It tests your response.body form at the same time as it triggers your action, with all the parameters in that form. This obviously improves quality. But it took issue with this: submit_form ''validate_stuff'' do |form| form[''IBhidden''] = ''validate'' end The field IBhidden is a hidden field. To be

I just recently purchased an E400P card and am attempting to get it working. My requirements are for 2 ports to run Q.931 and the other 2 ports to run R2 signalling with DTMF inband. The card will operate in a drop and insert fashion with calls arriving on the ISDN side and being terminated on the R2 lines. There don't seem to be any examples out there of how to set up this card for R2

Hi, I was wondering if anybody had this problem. If I use nmbclient to resolve a name, I get 'name_query failed to find name xxxx'. No names get resolved, but if I use smbclient with the same name and specify the IP everything is fine. Just as additional info, ping, telnet etc work fine, I have 'name resolve order = lmhosts wins bcast host' in smb.conf with 'wins support =

Hi, I got Mac OS X 15.6 installed, Locomotive 2.0.8 and MYSQL 5.0.67 running. I''m having this problem when I try to install mysql gem: Last login: Mon Feb 9 13:37:37 on ttys001 source /var/folders/be/bexGvm9pHdq0vcEmR8-WWk+++TM/-Tmp-//39E424DB-A260-4E41-AA2E-30B64FB46F3C-7395-00002254065EA712.environment.bash You have mail. MBPEduCirilo:~ eduardocirilo$ source

https://bugzilla.mindrot.org/show_bug.cgi?id=2491 Bug ID: 2491 Summary: x11-ssh-askpass (ssh-add) sets incorrect flags in WM_SIZE_HINTS Product: Portable OpenSSH Version: 7.1p1 Hardware: All OS: Linux Status: NEW Severity: trivial Priority: P5 Component: ssh-add

I just cranked up Samba 2.0.6 on NetBSD-sun3 1.4. The clients are Windows 98 (non-SE). I have had experience with earlier versions of Samba, so I was able to get 2.0.6 up and running without great difficulty. I exported a directory and was able to mount it on my Win98 machines fine. Using Explorer, everything looked like I expected it to: files were listed with their full names and I was able

Hi, This is mostly a note for downstream distributors of OpenSSH. I've just pushed fixes to the V_9_9 stable branch for a bug in the mlkem768x25519-sha256 key exchange algorithm that was added in this release that causes connection failures when connecting between big-endian and little-endian hosts. The problem is on the big-endian side. No change is required for the more common

Hello, I am attempting to get by broadcom wifi card up and running, am sick of trying to get ndis working, and am attempting to use the bwi driver (originating in dragonflyBSD). I'm hoping others here have tried to do the same and have some pointers. I'm using 7.1-RELEASE (system/source are in-sync) and my card is a BCM94306MP. My dmesg is posted below. Bwi(4) is installed and it

HaoFang client (China's largest platform for online play) HomePage: http://www.cga.com.cn Download(ZIP Version 5.5.0) : http://pv.cga.com.cn/counter.asp?id=952 Installation&Start: > WINEPREFIX=~/.wine-hf winetricks ie6 flash dotnet20 msxml6 vcrun6 riched20 riched30 gdiplus wenquanyi > > cd .wine/drive_c/Program\ Files/platform\ 5.0/ > > WINEPREFIX=~/.wine-hf/