Re: [gentoo-user] Changing compilers
Quoting Matthew Finkel matthew.fin...@gmail.com: On Mon, Mar 19, 2012 at 1:26 AM, Andrew Lowe a...@wht.com.au wrote: Hi all, [snip] Any thoughts, greatly appreciated, Andrew Lowe With regard to speed, are you looking for a faster compile time or higher optimization of the compiled code such that the run time is faster? -- Matthew Finkel I'm looking for the faster code, the run time to be faster - I compile the FEA CFD code once but will be running many jobs. Andrew
[gentoo-user] Re: Changing compilers
On 19/03/12 07:26, Andrew Lowe wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. You don't need to change compilers. You can use whatever one you like to build your program. The compiler portage uses to build its packages does not affect your own usage of the others. As for the fastest one, I can only speak for Intel CPUs where Intel C++ gives me the fastest binaries.
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On 03/17/2012 05:11:23 AM, Bruce Hill, Jr. wrote: This item just appeared after eix-sync: HTPC ~ # eselect news read 2012-03-16-udev-181-unmasking Title udev-181 unmasking AuthorWilliam Hubbs willi...@gentoo.org Posted2012-03-16 Revision 1 udev-181 is being unmasked on 2012-03-19. This news item is to inform you that once you upgrade to a version of udev =181, if you have /usr on a separate partition, you must boot your system with an initramfs which pre-mounts /usr. An initramfs which does this is created by =sys-kernel/genkernel-3.4.25.1 or =sys-kernel/dracut-017-r1. If you do not want to use these tools, be sure any initramfs you create pre-mounts /usr. Also, if you are using OpenRC, you must upgrade to = openrc-0.9.9. For more information on why this has been done, see the following URL: http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken Hi, has anybody followed the route shown in http://wiki.gentoo.org/wiki/Basic_initramfs_used_to_check_and_mount_/usr Thanks for sharing your experience, Helmut.
Re: [gentoo-user] Goodbye to gentoo?
Hello! On Sun, 18 Mar 2012 18:44:20 -0600 Canek Peláez Valdés can...@gmail.com wrote: You need to edit the file to remove duplicates and redundant mirrors (I can usually do it inside emacs in two minutes or less), and then check what files you already have in /usr/portage/distfiles (with a tiny bash script). You get the list of files you need, and only select those from the list of URLs, and then you have the files you need to download. You go back to the wi-fi cafe, download the files on a USB drive, and return home to put them on /usr/portage/distfiles. And then you can upgrade world. I just wanted to note that it may be unnecessary to edit a list of files in editor to remove the extra mirrors. I usually do: cat files-to-fetch.lst | cut -d -f1 | wget -c -i - and it downloads only using the first mirror for every file. It doesn't check for the existing files though. So it depends on the situation. I use my broadband Internet connection at work to fetch any number of files for my home PC. So it doesn't matter for me whether I download some files that I already have, but it may be important in your case. Perhaps, it would be useful if emerge had such a feature - to generate a list of files that do really need to be downloaded. Regards, Vladimir - v...@ukr.net
Re: [gentoo-user] Changing compilers
FEA jobs can be parallelized, right? Take a hard look at CUDA and OpenCL. ZZ On Mar 19, 2012 1:29 AM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe
Re: [gentoo-user] Changing compilers
On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Mark
Re: [gentoo-user] Changing compilers
On Mon, Mar 19, 2012 at 5:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Mark Sorry. Meant to include this reference: $15 on Kindle. Reads great on Kindle for PC. http://www.amazon.com/CUDA-Example-Introduction-General-Purpose-ebook/dp/B003VYBOSE/ref=sr_1_4?ie=UTF8qid=1332160431sr=8-4
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Sat, 17 Mar 2012 22:48:54 -0400 (EDT), Bruce Hill, Jr. wrote: And for the Lennart fanboi, his coding is so questionable that Lennartware has become derogatory slang. (Of course, you already know that.) And this is such a common term nowadays that when Googling for Lennartware only one reference to it turn up on the first page, and that is your post! I suppose by quoting your post I have doubled the popularity of this commonplace slang :-O This whole systemd for and against thread has turned up some interesting points - interspersed with vague handwaving from you. -- Neil Bothwick When there's a will, I want to be in it. signature.asc Description: PGP signature
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Sat, 17 Mar 2012 09:00:37 +0100, Andrea Conti wrote: Personally I stopped bothering with a separate /usr ages ago, so I don't really care. Having given this some thought recently, I am coming round to the view that the problem is /usr itself. It may have had a place when boot disks were limited in size, but I really don't see the point it in at all nowadays. This whole question of which bin directory does code belong in should be why do we need so many bin directories? -- Neil Bothwick This is a test of the emergency tagline stealing system. signature.asc Description: PGP signature
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
i want to try this systemd thingy, where do is start at? i have a vm for testing and i might will adopt it on the real one. Thanks, Eliezer On 18/03/2012 09:28, Canek Peláez Valdés wrote: it or at least try it, and given the level of discussion you present, I'm starting to think you don't actually have the capacity to study it. So, in that sense, the one spreading the FUD is you. -- Eliezer Croitoru https://www1.ngtech.co.il IT consulting for Nonprofit organizations elilezer at ngtech.co.il
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Sat, 17 Mar 2012 21:09:14 +0100, Volker Armin Hemmann wrote: Your sarcasm fails because you think that there is an intrinsic reason to keep / seperate. Well, with / filled with usefull binaries to bring a hosed system back from the garbage pile that was true for some peole. But with the current movement there isn't anything there at all. Doesn't everyone use a live distro for that? -- Neil Bothwick WindowError:01B Illegal error. Do NOT get this error. signature.asc Description: PGP signature
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Sat, 17 Mar 2012 08:50:28 -0400, Tanstaafl wrote: How do I find out if I am actually *using* an initramfs right now (I know it is built into the kernel), and If CONFIG_INITRAMFS_SOURCE is not set in your kernel config and you do not have an initrd line in GRUB, you are not using an initramfs. If I am not, how do I do this without using genkernel? Is dracut t he only other option? Is it easy/trivial to set one up manually? There was a post on the dev list explaining how to set up a minimal initramfs, I'll see if I can dig it out if no one beats me to it. -- Neil Bothwick Know thyself. If you need help, call the C.I.A. signature.asc Description: PGP signature
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Mon, Mar 19, 2012 at 6:17 AM, Neil Bothwick n...@digimed.co.uk wrote: On Sat, 17 Mar 2012 09:00:37 +0100, Andrea Conti wrote: Personally I stopped bothering with a separate /usr ages ago, so I don't really care. Having given this some thought recently, I am coming round to the view that the problem is /usr itself. It may have had a place when boot disks were limited in size, but I really don't see the point it in at all nowadays. This whole question of which bin directory does code belong in should be why do we need so many bin directories? -- Neil Bothwick This is a test of the emergency tagline stealing system. +1 http://fedoraproject.org/wiki/Features/UsrMove - Mark
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Sat, 17 Mar 2012 19:45:06 -0600, Canek Peláez Valdés wrote: * Really good in-site customization: The service unit files are trivially overrided with custom ones for specific installations, without needing to touch the ones installed by systemd or a program. With OpenRC, if I modify a /etc/init.d file, chances are I need to check out my next installation so I can see how the new file differs from the old one, and adapt the changes to my customized version. This I like the sound of. -- Neil Bothwick Walking on water and writing software to specification is easy if they're frozen. signature.asc Description: PGP signature
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Sun, 18 Mar 2012 02:49:56 -0600, Canek Peláez Valdés wrote: They ensure that there is an sshd configuration file and give a meaningful message (including where to find the sample) if it is not present, and check for the presence of the hostkeys (again which are needed) and create them if they are not present. Your 9 lines of sshd.service do none of this. That is completely true. I also think that those checks does not belong into the init script: I think the configuration file presence should be guarantee by the package manager at install time, and so the creation of the hostkeys. sshd is a bit of a special case. Think like CDs, like SystemRescueCD. If the keys were created at installation time, every CD would have the same keys, which is not particularly desirable. -- Neil Bothwick I heard someone tried the monkeys-on-typewriters bit trying for the plays of W. Shakespeare but all they got was the collected works of Francis Bacon signature.asc Description: PGP signature
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On 19/03/2012 04:03, William Kenworthy wrote: On Sun, 2012-03-18 at 18:30 -0600, Canek Peláez Valdés wrote: On Sun, Mar 18, 2012 at 4:28 PM, Alan McKinnon alan.mckin...@gmail.com wrote: On Sun, 18 Mar 2012 15:39:59 -0500 Dale rdalek1...@gmail.com wrote: Eliezer Croitoru wrote: On 18/03/2012 11:52, Dale wrote: Peter Humphrey wrote: On Saturday 17 March 2012 12:54:53 Eliezer Croitoru wrote: genkernel is pretty simple to use if you ask me. just emerege genkernel and then use genkerenl --menuconfig all it will do everything for you the same as in a regular kernel compiling. you have instructions on how to use genkernel on the handbook. What's more, you don't have to keep going through menuconfig if you already have a running self-compiled kernel. Just copy the .config file to somewhere safe (I use, e.g. /boot/config-3.2) and call genkernel with the option to specify the config file it's to use. Sorry but I can't tell you exactly what the parameter is as I don't have genkernel on this box. Someone will be along in a moment though. I used genkernel when I was first installing Gentoo. I let that thing build half a dozen kernels, chroot in between too. You know what, not one of them worked. That was a long time ago but let me check something here. spit spit spit I had to get the bad taste out of my mouth. lol I might also add, I started using a init thingy a few weeks ago, dracut tool. For some crazy reason, when I boot with the init thingy, my system doesn't work right. When I boot without the init thingy, it works fine. Still trying to figure out that one. It's in another thread. I don't see myself using genkernel any time soon. Right now, I'm having flashbacks to hal with regard to dracut and the whole init thingy /usr mess. i have used genkernel for a long time and all of my genkernel compilation works really good. i have (counting, 1 very very big production server, 2 small production server, 3 home server, 4 +5 +6 + 7 +8 of vms runing genkernel with several services such as mail mail filtering web server and monitoring) so what can i say? all these machines will say other then you. Regards, Eliezer Dale :-) :-) Odd, it can work on all those yet fail on a relatively simple system. Makes one wonder. Maybe it is to complicated? Sort of starting to sound like udev isn't it? lol I didn't say it would fail for the OP. I just said it never worked for me. Compiling my own has worked for me. I have only had one failure with that. I might also add, I have read where others have nightmares about genkernel. I'm not the only one. And using genkernel is pretty fucking pointless while it doesn't support suspend/resume right. Anyone feel free to correct me if I'm wrong, I don;t use genkernel so don;t know the truth from experience. I only read what others say, claiming that genkernel doesn't support suspend/resume. Resume/suspend or hibernate/whatever-the-inverse-of-hibernate-is-called? Because resume/suspend has nothing to do with an initramfs (being genkernel or dracut or whatever), since it doesn't "boot" the machine again (contrary to hibernate/whatever-etc.) My laptop has used dracut since months ago, and suspends/resumes just fine, as it does my media center. Regards. Genkernel doesnt, bugs and work arounds on gentoo bugzilla, with angry comments from a dev that it wont be supported and to not file bugs for it - now that dev has moved on I dont know if enough has changed to test the waters and file a bug again. Its missing a hook in the initrd to call the binary that starts the resume process. I was reading where dracut needs a lot of work still, so despite my previous bad experiences with genkernel in the past I went that way as the suspend fix is available. People generally just call it suspend/resume but technically, suspend/resume is often used to refer to suspend to ram, and hibernate is for suspend to disk - I use suspend to disk but generally just call it suspend/resume as (non-tech) people I talk to know what I mean. Calling it hibernate usually has them asking questions. It does work, as I said in a previous post, but the whole initrd thing is a disaster waiting to happen - and dont say to me it works for Red Hat as proof that it must be good because thats the distro where my most major initrd embarrassment occurred (update getting missmatched versions and fail to reboot.) Your experience may be different to mine, but I am of the once bitten, twice shy persuasion. Whatever happened to Linux/Unix and its focus on KISS
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
Eliezer Croitoru writes: i want to try this systemd thingy, where do is start at? http://wiki.gentoo.org/wiki/Systemd Wonko
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
William Kenworthy writes: On Sun, 2012-03-18 at 18:30 -0600, Canek Peláez Valdés wrote: My laptop has used dracut since months ago, and suspends/resumes just fine, as it does my media center. Genkernel doesnt, bugs and work arounds on gentoo bugzilla, with angry comments from a dev that it wont be supported and to not file bugs for it - now that dev has moved on I dont know if enough has changed to test the waters and file a bug again. Its missing a hook in the initrd to call the binary that starts the resume process. Huh? I don't use this at the moment, because suspend-to-ram is enough for me, but it (that is, the initramfs part) used to work just fine out of the box for me, also opening my LUKS-encrypted root volume being on LVM. It also seemed to work on another Gentoo PC I installed recently, although TuxOnIce itself does not work so the resume fails. Argh, this suspend to disk stuff NEVER really worked for me, and I tried for years on different systems. Wonko
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Mon, Mar 19, 2012 at 9:33 AM, Neil Bothwick n...@digimed.co.uk wrote: On Sun, 18 Mar 2012 02:49:56 -0600, Canek Peláez Valdés wrote: They ensure that there is an sshd configuration file and give a meaningful message (including where to find the sample) if it is not present, and check for the presence of the hostkeys (again which are needed) and create them if they are not present. Your 9 lines of sshd.service do none of this. That is completely true. I also think that those checks does not belong into the init script: I think the configuration file presence should be guarantee by the package manager at install time, and so the creation of the hostkeys. sshd is a bit of a special case. Think like CDs, like SystemRescueCD. If the keys were created at installation time, every CD would have the same keys, which is not particularly desirable. I prefer counterexample to special case ... I don't like calling things special cases because it suggests that they're somehow more privileged than anything else, and unnecessarily weighs against software which hasn't been written yet. A similar case which falls into the same kind of circumstance: per-host IDs in mass-deployment scenarios. You see this in large arrays of similar systems; 'sbc-a3d6' 'sbc-a3d9' 'sbc-7721' ... Heck, applying something like that to live installation media would be nice; not having every new install called simply 'gentoo' by default would be very helpful in installfest scenarios. Identical hostnames screw with DHCP-driven DDNS updates. I ran into that on my home network. -- :wq
Re: [gentoo-user] Changing compilers
On Mon, Mar 19, 2012 at 8:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Yes. And as a convenient side-effect, it offers a great excuse to upgrade your video card with some regularity. The performance of mid-grade and high-grade video cards continues to improve rapidly. -- :wq
Re: [gentoo-user] Changing compilers
On 03/19/12 20:34, Mark Knecht wrote: On Mon, Mar 19, 2012 at 5:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Mark Sorry. Meant to include this reference: $15 on Kindle. Reads great on Kindle for PC. http://www.amazon.com/CUDA-Example-Introduction-General-Purpose-ebook/dp/B003VYBOSE/ref=sr_1_4?ie=UTF8qid=1332160431sr=8-4 I'm sorry but I'm doing a PhD, not creating a career in Academia. The concept of writing an FEA or CFD from scratch, with CUDA is laughable, I just don't have the time to learn CUDA, research the field, small displacement, large displacement, dynamics, material nonlinearities, write the code, and then most importantly benchmark it to make sure it's actually correct. This is all bearing in mind that I have 20+ years experience as a C/C++ technical software developer, including FEA and CFD. I'll actually be using Code Aster, an open source FEA code that runs under Linux. Sorry if I sound narky, but compilers is the subject at hand, not how to write FEA code. Anyway, thanks for answering, Andrew
Re: [gentoo-user] Initramfs or move /usr to /, oh my...
On Sun, 18 Mar 2012 21:57:39 -0400 (EDT), Bruce Hill, Jr. wrote: How do I find out if I am actually *using* an initramfs right now (I know it is built into the kernel), and ls -l /boot/ will tell you. No it won't, as the initramfs can be built into the kernel image. There is a difference between an initrd (initial RAM disk) image (simple) and an initramfs (initial RAM filesystem) (complicated). I think you have that the wrong way round. An initramfs is simpler in that you don't have to worry about creating a filesystem or even creating the image file, just give the kernel compilation process a file containing a list of what should be in it. -- Neil Bothwick Middle-age - because your age starts to show at your middle. signature.asc Description: PGP signature
Re: [gentoo-user] Re: Changing compilers
On 03/19/12 17:39, Nikos Chantziaras wrote: On 19/03/12 07:26, Andrew Lowe wrote: Hi all, Has anyone played around with the various better known compilers on [snip] ... ... [snip] You don't need to change compilers. You can use whatever one you like to build your program. The compiler portage uses to build its packages does not affect your own usage of the others. As for the fastest one, I can only speak for Intel CPUs where Intel C++ gives me the fastest binaries. Nikos, Your experience with Intel is what I'm after. Aster, the FEA code I'm going to use is not in Portage hence I will be using it's own build system. When you've used Intel, have you just exported CC=icc or something similar, as make.conf won't be used? Also, I've read somewhere that there are libraries that you have to link against that are specific to the Intel compiler as it does not create libraries that are comparable with the gcc produced ones - is this true or does the compiler now play well with the gcc world? Any thoughts, Andrew
Re: [gentoo-user] Goodbye to gentoo?
On Sun, 18 Mar 2012 22:44:33 -0600, Maxim Wexler wrote: route -n shows nothing except ppp0 (this is from ubuntu, but it was the same for gentoo when it was working) root@gnubu:~# route -n Kernel IP routing table Destination Gateway Genmask Flags Metric RefUse Iface 0.0.0.0 0.0.0.0 0.0.0.0 U 0 00 ppp0 161.184.0.199 0.0.0.0 255.255.255.255 UH 0 00 ppp0 Why show us what it was when it worked? What does route -n show from your broken Gentoo now? -- Neil Bothwick Shotgun wedding: A case of wife or death. signature.asc Description: PGP signature
Re: [gentoo-user] Changing compilers
On 03/19/12 22:02, Michael Mol wrote: On Mon, Mar 19, 2012 at 8:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Yes. And as a convenient side-effect, it offers a great excuse to upgrade your video card with some regularity. The performance of mid-grade and high-grade video cards continues to improve rapidly. Sorry, can't do that, I'm using epic, http://tinyurl.com/83l5o3z which currently ranks at 151 in the top 500 list :) It's amazing how fast this list changes, 6 months ago, this machine was at 107 and 6 months before that 87. Andrew
Re: [gentoo-user] Changing compilers
On Mon, Mar 19, 2012 at 10:05 AM, Andrew Lowe a...@wht.com.au wrote: On 03/19/12 20:34, Mark Knecht wrote: On Mon, Mar 19, 2012 at 5:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Mark Sorry. Meant to include this reference: $15 on Kindle. Reads great on Kindle for PC. http://www.amazon.com/CUDA-Example-Introduction-General-Purpose-ebook/dp/B003VYBOSE/ref=sr_1_4?ie=UTF8qid=1332160431sr=8-4 I'm sorry but I'm doing a PhD, not creating a career in Academia. The concept of writing an FEA or CFD from scratch, with CUDA is laughable, I just don't have the time to learn CUDA, research the field, small displacement, large displacement, dynamics, material nonlinearities, write the code, and then most importantly benchmark it to make sure it's actually correct. This is all bearing in mind that I have 20+ years experience as a C/C++ technical software developer, including FEA and CFD. I'll actually be using Code Aster, an open source FEA code that runs under Linux. Sorry if I sound narky, but compilers is the subject at hand, not how to write FEA code. If you really care about a 1-2% difference, you should not be dismissing GPGPU-accelerated code so easily! If the tools you seem to have already settled on don't support it, you should either use different tools, or correct the ones you're working with. The lead Python guy had an astute observation (which I'll generalize) the other day; for 99% of your program, it doesn't matter what programming language you use. For the 1% where you need speed, you should call out into the faster language. -- :wq
Re: [gentoo-user] Goodbye to gentoo?
cat files-to-fetch.lst | cut -d -f1 | wget -c -i - I know how to generate a fetch list and wget the files. The problem is syncing portage on another machine, my netbook, ubuntu-based. Or maybe, I could boot sysresc on the netbook. It's gentoo based IIRC and do a re-sync back home as per C P Valdes.
Re: [gentoo-user] Goodbye to gentoo?
Why show us what it was when it worked? What does route -n show from your broken Gentoo now? -- Neil Bothwick Good question. Heading over there now. So if I can't get it working there's gonna have to be some awkward shuttling back and forth.
Re: [gentoo-user] Changing compilers
On Mon, Mar 19, 2012 at 10:18 AM, Andrew Lowe a...@wht.com.au wrote: On 03/19/12 22:02, Michael Mol wrote: On Mon, Mar 19, 2012 at 8:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Yes. And as a convenient side-effect, it offers a great excuse to upgrade your video card with some regularity. The performance of mid-grade and high-grade video cards continues to improve rapidly. Sorry, can't do that, I'm using epic, http://tinyurl.com/83l5o3z which currently ranks at 151 in the top 500 list :) It's amazing how fast this list changes, 6 months ago, this machine was at 107 and 6 months before that 87. That does change things a bit. I don't know Epic's structure or their upgrade plans, but if you're confident it's not going to have GPGPU capabilities, then CUDA and OpenCL are less useful for you. OpenCL, at least, still handles per-CPU and per-node job dispatching, though. And that's still likely to be useful for performing on huge matrices. To answer your original question: No, I haven't done much with anything other than gcc on Gentoo. What you *should* do is grab each compiler (trial versions, if necessary) and test them to find which gives you the best results. It's my understanding PhD programs involve getting things done right, not so much quickly or easily. Best to be methodical about it. -- :wq
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On March 19, 2012 at 9:13 AM Neil Bothwick n...@digimed.co.uk wrote: On Sat, 17 Mar 2012 22:48:54 -0400 (EDT), Bruce Hill, Jr. wrote: And for the Lennart fanboi, his coding is so questionable that Lennartware has become derogatory slang. (Of course, you already know that.) And this is such a common term nowadays that when Googling for Lennartware only one reference to it turn up on the first page, and that is your post! I suppose by quoting your post I have doubled the popularity of this commonplace slang :-O This whole systemd for and against thread has turned up some interesting points - interspersed with vague handwaving from you. -- Neil Bothwick mingdao@workstation ~ $ grep Lennartware irclogs/* irclogs/#gentoo-dev.log:09:01 @bonsaikitten Caster: do you see now why I don't appreciate Lennartware? irclogs/#gentoo.log:10:56 @bonsaikitten Zaba: Lennartware. Linux needs to be more like MacOS https://s6-us2.startpage.com/do/search?cmd=process_searchpid=04014d679c59b80b606405a6fe33495a --- 4 references Various other mentions of systemd being nefarious software are mostly amongst kernel devs and might not use the word Lennartware, but the logical reasons why systemd is a _bad_ idea are the same. -- Happy Penguin Computers`) 126 Fenco Drive( \ Tupelo, MS 38801^^ 662-269-2706; 662-491-8613 support at happypenguincomputers dot com http://www.happypenguincomputers.com
Re: [gentoo-user] Re: Changing compilers
Am 19.03.2012 15:11, schrieb Andrew Lowe: On 03/19/12 17:39, Nikos Chantziaras wrote: On 19/03/12 07:26, Andrew Lowe wrote: Hi all, Has anyone played around with the various better known compilers on [snip] ... ... [snip] You don't need to change compilers. You can use whatever one you like to build your program. The compiler portage uses to build its packages does not affect your own usage of the others. As for the fastest one, I can only speak for Intel CPUs where Intel C++ gives me the fastest binaries. Nikos, Your experience with Intel is what I'm after. Aster, the FEA code I'm going to use is not in Portage hence I will be using it's own build system. When you've used Intel, have you just exported CC=icc or something similar, as make.conf won't be used? Also, I've read somewhere that there are libraries that you have to link against that are specific to the Intel compiler as it does not create libraries that are comparable with the gcc produced ones - is this true or does the compiler now play well with the gcc world? Any thoughts, Andrew Look here: http://en.gentoo-wiki.com/wiki/Intel_C%2B%2B_Compiler Regards, Florian Philipp signature.asc Description: OpenPGP digital signature
[gentoo-user] *** urgent *** root needs non existing service 'fsck'
Hi, suddenly might Gentoo system doesn't boot anymore, not normally, at least. I get 'root' needs non existing service 'fsck' but I've no idea why. rc-update -u tells me about this, as well. reemerging linux-utils and openrc didn't help. /etc/init.d/fsck is a normal file which contents looks normally. I can invoke /sbin/fsck. I'm very grateful for any hints to get out of this, Helmut.
[gentoo-user] [HEADS UP] udev-181
I just had a bit of a scare after updating to udev-181, but all is well now, finally. (I hope :) In addition to the separate /usr problem that has already been discussed at length here, there are other important changes in udev-181 to be aware of. First, I had to add two new items to my kernel config: CONFIG_DEV_TMPFS (which I thought I'd had for years but didn't) and CONFIG_TMPFS_POSIX_ACL. I also elected to make the devfs automounting, but I don't think that was really necessary. Second, don't forget like I did to update the udev initscripts with etc-update or your machine won't be able to find the udev files in their new locations (just like mine didn't) and none of your kernel modules will auto-load, either. Oh, and of course you need to pre-mount /usr before udev starts if you have a separate /usr partition -- but you already knew that ;)
Re: [gentoo-user] [HEADS UP] udev-181
On Mon, Mar 19, 2012 at 10:56 AM, walt w41...@gmail.com wrote: I just had a bit of a scare after updating to udev-181, but all is well now, finally. (I hope :) In addition to the separate /usr problem that has already been discussed at length here, there are other important changes in udev-181 to be aware of. First, I had to add two new items to my kernel config: CONFIG_DEV_TMPFS (which I thought I'd had for years but didn't) and CONFIG_TMPFS_POSIX_ACL. I also elected to make the devfs automounting, but I don't think that was really necessary. Second, don't forget like I did to update the udev initscripts with etc-update or your machine won't be able to find the udev files in their new locations (just like mine didn't) and none of your kernel modules will auto-load, either. Oh, and of course you need to pre-mount /usr before udev starts if you have a separate /usr partition -- but you already knew that ;) Is there an ENOTICE warning in the ebuild to hit people over the head with these? Also, how trivial would it be to have the ebuild check the running kernel config (if available under /proc or wherever) for the necessary config options? -- :wq
[gentoo-user] Re: Changing compilers
Andrew Lowe agl at wht.com.au writes: Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. Octave is in portage, as it is a matlab sorta package for fluids. google for Octave : Computational Fluid Dynamics : Finite Element Analysis for starters. I understand you are smart, but why push the issues on compiler efficiency, if you are not willing to hack the compiler code? Why not push the mathematical envelop and use: Sparse Matrix Techniques ?? http://www.cise.ufl.edu/research/sparse/matrices/ http://blogs.mathworks.com/loren/2007/03/01/ creating-sparse-finite-element-matrices-in-matlab/ hth, James
[gentoo-user] udev-181 and kmod vs module-init-tools
udev-181 merge is blocked because : [blocks B ] sys-apps/module-init-tools (sys-apps/module-init-tools is blocking sys-apps/kmod-7) [blocks B ] sys-apps/kmod (sys-apps/kmod is blocking sys-apps/module-init-tools-3.16-r1) * Error: The above package list contains packages which cannot be * installed at the same time on the same system. (sys-apps/kmod-7::gentoo, ebuild scheduled for merge) pulled in by =sys-apps/kmod-5 required by (sys-fs/udev-181::gentoo, ebuild scheduled for merge) (sys-apps/module-init-tools-3.16-r1::gentoo, installed) pulled in by =sys-apps/module-init-tools-3.2 required by (virtual/modutils-0::gentoo, installed) Googling, it looks like kmod is the replacement for module-init-tools, so it should just work, unless of course any packages explicitly depend on module-init-tools instead of the virtual... However, module-init-tools is part of system profile and I get the scary red text and 10-second countdown when I try to unmerge, so I chickened out and did a ctrl-c. :) Have any of you already changed from module-init-tools to kmod? Anything to be afraid of? It's a remote machine and I can boot rescue CD if it breaks, but I'd rather not have to do that.
Re: [gentoo-user] Goodbye to gentoo?
Hello, On Sun, 18 Mar 2012, Maxim Wexler wrote: root@gnubu:~# route -n Kernel IP routing table Destination Gateway Genmask Flags Metric RefUse Iface 0.0.0.0 0.0.0.0 0.0.0.0 U 0 00 ppp0 161.184.0.199 0.0.0.0 255.255.255.255 UH0 00 ppp0 I'm on a Suse ATM[1], but as I do my networking config by hand that should not matter. Above looks wrong. It should look like: # route -n 161.184.0.199 0.0.0.0 255.255.255.255 UH0 00 ppp0 [or:161.184.0.0 0.0.0.0 255.255.255.0 U 0 00 ppp0] 0.0.0.0 161.184.0.199 0.0.0.0 UG0 00 ppp0 ^ ^ your ppp assigned P-t-P gateway flag So, have a look at how the route gets set. Manually, a # route del default # route add default gw 161.184.0.199 should suffice. Don't know the 'ip' syntax, but luckily that doesn't matter, as both commands just push stuff to the kernel ;) Looking at my much outdated gentoo's /etc/conf.d/net.eth0, that'd have to be in your /etc/conf.d/ppp0 (or current equivalent) routes_ppp0=( default via 161.184.0.199 ) mtu_ppp0=1492 No idea at all how this is handled with dialup on gentoo. When I still used dialup (10 years ago, IIRC?), I used wvdial, since then, I used my own shell-scripts for DSL calling ifconfig/pppd (with pppoe at first). The scripts/config-files should still be somewhere on disk ;) If in doubt, do as I do, write your own script containing the relevant commands to get your internet up and running. Even if it'll just be for future reference in case of trouble. (also from the ubuntu side) root@gnubu:~# ifconfig ppp0 ppp0 Link encap:Point-to-Point Protocol inet addr:161.184.44.73 P-t-P:161.184.0.199 Mask:255.255.255.255 UP POINTOPOINT RUNNING NOARP MULTICAST MTU:1500 Metric:1 That MTU is IIRC too big for PPP. Change that to 1492 or less. RX packets:5867 errors:0 dropped:0 overruns:0 frame:0 TX packets:6439 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:3 RX bytes:1694892 (1.6 MB) TX bytes:746705 (746.7 KB) That looks like an established connection (despite the MTU) to the gateway, as well as you said there a dns got assigned. -dnh [1] not enough time to tweak two systems on my main box -- Me? No, why me? She's much more interesting. An enigma wrapped up in a riddle with a tail in the middle.-- Harper about Trance Gemini -- Andromeda 1x14 - Harper 2.0
[gentoo-user] Re: [HEADS UP] udev-181
On 03/19/2012 08:15 AM, Michael Mol wrote: On Mon, Mar 19, 2012 at 10:56 AM, walt w41...@gmail.com wrote: I just had a bit of a scare after updating to udev-181, but all is well now, finally. (I hope :) In addition to the separate /usr problem that has already been discussed at length here, there are other important changes in udev-181 to be aware of. First, I had to add two new items to my kernel config: CONFIG_DEV_TMPFS (which I thought I'd had for years but didn't) and CONFIG_TMPFS_POSIX_ACL. I also elected to make the devfs automounting, but I don't think that was really necessary. Second, don't forget like I did to update the udev initscripts with etc-update or your machine won't be able to find the udev files in their new locations (just like mine didn't) and none of your kernel modules will auto-load, either. Oh, and of course you need to pre-mount /usr before udev starts if you have a separate /usr partition -- but you already knew that ;) Is there an ENOTICE warning in the ebuild to hit people over the head with these? Also, how trivial would it be to have the ebuild check the running kernel config (if available under /proc or wherever) for the necessary config options? Yes, I found out about the added kernel config items from the ENOTICE, but I wanted to point out that it needs to be done *before* you reboot or you'll be in trouble. Even better if you do it before you update udevd, I'd say. If the warning to run etc-update appeared, I missed it :( which I usually don't.
Re: [gentoo-user] Changing compilers
On Mon, Mar 19, 2012 at 7:05 AM, Andrew Lowe a...@wht.com.au wrote: On 03/19/12 20:34, Mark Knecht wrote: On Mon, Mar 19, 2012 at 5:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Mark Sorry. Meant to include this reference: $15 on Kindle. Reads great on Kindle for PC. http://www.amazon.com/CUDA-Example-Introduction-General-Purpose-ebook/dp/B003VYBOSE/ref=sr_1_4?ie=UTF8qid=1332160431sr=8-4 I'm sorry but I'm doing a PhD, not creating a career in Academia. The concept of writing an FEA or CFD from scratch, with CUDA is laughable, I just don't have the time to learn CUDA, research the field, small displacement, large displacement, dynamics, material nonlinearities, write the code, and then most importantly benchmark it to make sure it's actually correct. This is all bearing in mind that I have 20+ years experience as a C/C++ technical software developer, including FEA and CFD. I'll actually be using Code Aster, an open source FEA code that runs under Linux. Sorry if I sound narky, but compilers is the subject at hand, not how to write FEA code. Anyway, thanks for answering, Andrew Nahh, be as snarky as you like as long as you don't really mean it personally. My experience with CUDA, and I'm not a programmer, is that there is a fairly steep learning curve. However changing C compilers will get you maybe 5%. Changing to CUDA will get you 30,000%, assuming a mid-high range CUDA card and that you can parallel-ize FEA. I did a little Googling and it seems that FEA is a pretty common CUDA topic so I don't think at the outset that you'd find yourself all alone. Good luck whatever you do and know that I didn't mind the response at all! :-) Cheers, Mark
[gentoo-user] ppp-gentoo woes cont'd
Just got back from gentoo land. Arrrgh, gmail won't let me attach files, just sits there spinning. So I'll have to make do with pastebin. http://paste.ubuntu.com/890854/ Is a chronicle of the commands entered. First having booted and not changing anything, I do #ifconifg, then I do #route -n, then #pon isp to connect. Then there is the running tail of the messages log, ifconfig, route -n Next, rmmod the drivers. I do #poff isp to bring down ppp0, tail the messages, And so on... NB: the crash of the time daemon doesn't matter. Like everything else I need to work around the problem; in this case via a script in /etc/ppp/ip-up.d and /ip-down.d. The problem persists if I don't start the daemon at all. At the bottom of the file I've included the /etc/ppp/ip-up script. The scripts it refers to are here: 30-wins.sh http://paste.ubuntu.com/890854/ 40-dns.sh http://paste.ubuntu.com/890857/ 50-initd.sh http://paste.ubuntu.com/890857/ 90-ntpd.sh http://paste.ubuntu.com/890857/ The first two don't apply. 50-initd.sh, I don't quite grok. Hope somebody has the patience to go through this ;) MW
Re: [gentoo-user] udev-181 and kmod vs module-init-tools
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi, it worked flawless for me. Nothing bad happened. If it's a server you could also build a monolithic Kernel and remove the dependency completly since most servers don't need loadable modules. It even adds a little security-wise... with kind regards, Hinnerk On 19.03.2012 16:47, Paul Hartman wrote: udev-181 merge is blocked because : [blocks B ] sys-apps/module-init-tools (sys-apps/module-init-tools is blocking sys-apps/kmod-7) [blocks B ] sys-apps/kmod (sys-apps/kmod is blocking sys-apps/module-init-tools-3.16-r1) * Error: The above package list contains packages which cannot be * installed at the same time on the same system. (sys-apps/kmod-7::gentoo, ebuild scheduled for merge) pulled in by =sys-apps/kmod-5 required by (sys-fs/udev-181::gentoo, ebuild scheduled for merge) (sys-apps/module-init-tools-3.16-r1::gentoo, installed) pulled in by =sys-apps/module-init-tools-3.2 required by (virtual/modutils-0::gentoo, installed) Googling, it looks like kmod is the replacement for module-init-tools, so it should just work, unless of course any packages explicitly depend on module-init-tools instead of the virtual... However, module-init-tools is part of system profile and I get the scary red text and 10-second countdown when I try to unmerge, so I chickened out and did a ctrl-c. :) Have any of you already changed from module-init-tools to kmod? Anything to be afraid of? It's a remote machine and I can boot rescue CD if it breaks, but I'd rather not have to do that. -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.18 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJPZ2LNAAoJEJwwOFaNFkYc04UH/0FoJECyQ1FeguUT4IYyksv3 ddXb/00Pb0soMLmYYymJxz4Wpvp6N9SeQV1EKQOJPwTxaSdRE0RK/IeeA/2goJrm utkMfK9rbpxIdv5gqbOjcmsm5mj/8DK4o4WwUFuLf1rN+rFpDouMWblbpF7maH89 w2SkGyR7rsKquQ/iK2BJSC5fbtTkfWkQz96XxNRkJAKsS3n9RlMHI5C8onLExBG2 WsXmV8kWUcDuJqcLYtkjilM1/J6Cmp5yME2VK3oj3Z5gaOg06GoF3PCr6y52ujDR 3Bqu+3EpC/MDQSSHhDNlikJrLYpnSOni7D4Uwz7MRBn+XwIAKQfaHQ1Fkvf2Vt0= =fO+O -END PGP SIGNATURE-
Re: [gentoo-user] Changing compilers
On 03/20/12 00:03, Mark Knecht wrote: On Mon, Mar 19, 2012 at 7:05 AM, Andrew Lowe a...@wht.com.au wrote: On 03/19/12 20:34, Mark Knecht wrote: On Mon, Mar 19, 2012 at 5:32 AM, Mark Knecht markkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowe a...@wht.com.au wrote: Hi all, Has anyone played around with the various better known compilers on Gentoo? By better known, I'm referring to gcc, Intel, llvm, pathscale. My situation is that I've just started my PhD which requires me to do Finite Element Analysis, FEA, and Computational Fluid Dynamics, CFD, and I want to find the best compiler for the job. Before anyone says Why bother, XXX compiler is only 1 - 2% faster than gcc, in the context of the work I'm doing this 1 - 2% IS important. What I'm looking for is any feedback people may have on ability to compile the Gentoo environment, the ability to change compilers easily, gcc-config or flags in make.conf, as to whether the compiler/linker can use the libraries as compiled by gcc on a standard gentoo install and so on. Obviously there is much web trawling to be done to find what other people are saying as well. Any thoughts, greatly appreciated, Andrew Lowe Think CUDA Mark Sorry. Meant to include this reference: $15 on Kindle. Reads great on Kindle for PC. http://www.amazon.com/CUDA-Example-Introduction-General-Purpose-ebook/dp/B003VYBOSE/ref=sr_1_4?ie=UTF8qid=1332160431sr=8-4 I'm sorry but I'm doing a PhD, not creating a career in Academia. The concept of writing an FEA or CFD from scratch, with CUDA is laughable, I just don't have the time to learn CUDA, research the field, small displacement, large displacement, dynamics, material nonlinearities, write the code, and then most importantly benchmark it to make sure it's actually correct. This is all bearing in mind that I have 20+ years experience as a C/C++ technical software developer, including FEA and CFD. I'll actually be using Code Aster, an open source FEA code that runs under Linux. Sorry if I sound narky, but compilers is the subject at hand, not how to write FEA code. Anyway, thanks for answering, Andrew Nahh, be as snarky as you like as long as you don't really mean it personally. My experience with CUDA, and I'm not a programmer, is that there is a fairly steep learning curve. However changing C compilers will get you maybe 5%. Changing to CUDA will get you 30,000%, assuming a mid-high range CUDA card and that you can parallel-ize FEA. I did a little Googling and it seems that FEA is a pretty common CUDA topic so I don't think at the outset that you'd find yourself all alone. Good luck whatever you do and know that I didn't mind the response at all! :-) Cheers, Mark The thing is that I agree that CUDA is the way to go for things like FEA CFD, in fact the mob that runs the super computer I'm using is installing another one that is top heavy in CUDA cards. But the thing is as I'm using the FEA as a tool, rather than playing around with the innards of the code, I need an established code, one that has verification behind it. My topic looks at the way that steel connections behave so I need an established FEA code that is verified to provide the correct answers, I don't get that if I write my own code. Most likely I'll write a short paper covering a comparison of existing C/C++ compilers and their relative speeds, spend the next 18 months - 2 years doing my research and then to close things off, I'll probably be able to write another short paper concerning CUDA speedups as the FEA code bases will have caught up and been verified. Thanks for peoples replies, Andrew p.s. Writing this I just had a sudden horrific though and checked the FEA code I'm using, Aster. It's written mostly in FORTRAN - fat chance I'm going to be hacking that code
[gentoo-user] [SOLVED BUT] *** urgent *** root needs non existing service 'fsck'
On 03/19/2012 03:54:35 PM, Helmut Jarausch wrote: Hi, suddenly might Gentoo system doesn't boot anymore, not normally, at least. I get 'root' needs non existing service 'fsck' but I've no idea why. rc-update -u tells me about this, as well. reemerging linux-utils and openrc didn't help. /etc/init.d/fsck is a normal file which contents looks normally. I can invoke /sbin/fsck. I'm very grateful for any hints to get out of this, By comparing my broken system to a recent backup it showed up there was a new file in /etc/ concerning an initramfs. qfile showed that this file has been installed by genkernel. It's unbelievable but just emerging genkernel broke my system though I've never used it, yet. Just emerge -C sys-kernel/genkernel and my system was sane again. Helmut.
[gentoo-user] Re: Changing compilers
On 19/03/12 16:11, Andrew Lowe wrote: On 03/19/12 17:39, Nikos Chantziaras wrote: On 19/03/12 07:26, Andrew Lowe wrote: Hi all, Has anyone played around with the various better known compilers on [snip] ... ... [snip] You don't need to change compilers. You can use whatever one you like to build your program. The compiler portage uses to build its packages does not affect your own usage of the others. As for the fastest one, I can only speak for Intel CPUs where Intel C++ gives me the fastest binaries. Nikos, Your experience with Intel is what I'm after. Aster, the FEA code I'm going to use is not in Portage hence I will be using it's own build system. When you've used Intel, have you just exported CC=icc or something similar, as make.conf won't be used? Also, I've read somewhere that there are libraries that you have to link against that are specific to the Intel compiler as it does not create libraries that are comparable with the gcc produced ones - is this true or does the compiler now play well with the gcc world? No special libs required. The binaries I get (both C and C++) don't use anything extra. I checked both with ldd as well as with lsof at runtime (in case it dlopens anything). For building, you use CC=icc and CXX=icpc for regular makefiles or autoconf scripts. I mostly use qmake though (I use Qt for my GUIs). In that case, you call qmake like this: qmake -spec linux-icc and it creates a makefile that will use ICC. This is also an example of ICC using C and C++ libs (Qt is C++) that were built by GCC without issues; its ABI is fully GCC compatible. There are way to use ICC for portage too. I tried that once. It worked quite well. But I didn't went with it since too much of a bother. Note that the link Florian posted is a bit outdated. For example the sections that tells you that binaries compiled with icc won't work after icc is uninstalled is not true. They will work just fine. The exception of course if when you specifically use an ICC library, like the Intel math kernel library.
Re: [gentoo-user] Re: Changing compilers
On 03/20/12 01:17, Nikos Chantziaras wrote: On 19/03/12 16:11, Andrew Lowe wrote: On 03/19/12 17:39, Nikos Chantziaras wrote: On 19/03/12 07:26, Andrew Lowe wrote: Hi all, Has anyone played around with the various better known compilers on [snip] ... ... [snip] You don't need to change compilers. You can use whatever one you like to build your program. The compiler portage uses to build its packages does not affect your own usage of the others. As for the fastest one, I can only speak for Intel CPUs where Intel C++ gives me the fastest binaries. Nikos, Your experience with Intel is what I'm after. Aster, the FEA code I'm going to use is not in Portage hence I will be using it's own build system. When you've used Intel, have you just exported CC=icc or something similar, as make.conf won't be used? Also, I've read somewhere that there are libraries that you have to link against that are specific to the Intel compiler as it does not create libraries that are comparable with the gcc produced ones - is this true or does the compiler now play well with the gcc world? No special libs required. The binaries I get (both C and C++) don't use anything extra. I checked both with ldd as well as with lsof at runtime (in case it dlopens anything). For building, you use CC=icc and CXX=icpc for regular makefiles or autoconf scripts. I mostly use qmake though (I use Qt for my GUIs). In that case, you call qmake like this: qmake -spec linux-icc and it creates a makefile that will use ICC. This is also an example of ICC using C and C++ libs (Qt is C++) that were built by GCC without issues; its ABI is fully GCC compatible. There are way to use ICC for portage too. I tried that once. It worked quite well. But I didn't went with it since too much of a bother. Note that the link Florian posted is a bit outdated. For example the sections that tells you that binaries compiled with icc won't work after icc is uninstalled is not true. They will work just fine. The exception of course if when you specifically use an ICC library, like the Intel math kernel library. Thanks for that. The library question was the reason I didn't proceed with playing around with icc ages ago. Your experience tells me it's now rectified. Andrew
[gentoo-user] Re: Changing compilers
On 19/03/12 19:24, Andrew Lowe wrote: On 03/20/12 01:17, Nikos Chantziaras wrote: On 19/03/12 16:11, Andrew Lowe wrote: On 03/19/12 17:39, Nikos Chantziaras wrote: On 19/03/12 07:26, Andrew Lowe wrote: Hi all, Has anyone played around with the various better known compilers on [snip] ... ... [snip] You don't need to change compilers. You can use whatever one you like to build your program. The compiler portage uses to build its packages does not affect your own usage of the others. As for the fastest one, I can only speak for Intel CPUs where Intel C++ gives me the fastest binaries. [...] Also, I've read somewhere that there are libraries that you have to link against that are specific to the Intel compiler as it does not create libraries that are comparable with the gcc produced ones - is this true or does the compiler now play well with the gcc world? No special libs required. The binaries I get (both C and C++) don't use anything extra. I checked both with ldd as well as with lsof at runtime (in case it dlopens anything). [...] Thanks for that. The library question was the reason I didn't proceed with playing around with icc ages ago. Your experience tells me it's now rectified. Just to verify that I'm not mistaken about this, I just compiled a non-trivial project that uses C++ libraries, then uninstalled icc and all its deps (with --depclean), and the binary still ran without issues.
Re: [gentoo-user] [HEADS UP] udev-181
Just a minor correction. It's CONFIG_DEVTMPFS not CONFIG_DEV_TMPFS :) I just had a bit of a scare after updating to udev-181, but all is well now, finally. (I hope :) In addition to the separate /usr problem that has already been discussed at length here, there are other important changes in udev-181 to be aware of. First, I had to add two new items to my kernel config: CONFIG_DEV_TMPFS (which I thought I'd had for years but didn't) and CONFIG_TMPFS_POSIX_ACL. I also elected to make the devfs automounting, but I don't think that was really necessary. Second, don't forget like I did to update the udev initscripts with etc-update or your machine won't be able to find the udev files in their new locations (just like mine didn't) and none of your kernel modules will auto-load, either. Oh, and of course you need to pre-mount /usr before udev starts if you have a separate /usr partition -- but you already knew that ;)
Re: [gentoo-user] [HEADS UP] udev-181
On Mon, Mar 19, 2012 at 8:15 AM, Michael Mol mike...@gmail.com wrote: SNIP Is there an ENOTICE warning in the ebuild to hit people over the head with these? Also, how trivial would it be to have the ebuild check the running kernel config (if available under /proc or wherever) for the necessary config options? Concerning the kernel config stuff file an enhancement request. Dracut checks for the same variable. They should be able to just copy it from there. - Mark
Re: [gentoo-user] [SOLVED BUT] *** urgent *** root needs non existing service 'fsck'
On Mon, Mar 19, 2012 at 12:09 PM, Helmut Jarausch jarau...@igpm.rwth-aachen.de wrote: On 03/19/2012 03:54:35 PM, Helmut Jarausch wrote: Hi, suddenly might Gentoo system doesn't boot anymore, not normally, at least. I get 'root' needs non existing service 'fsck' but I've no idea why. rc-update -u tells me about this, as well. reemerging linux-utils and openrc didn't help. /etc/init.d/fsck is a normal file which contents looks normally. I can invoke /sbin/fsck. I'm very grateful for any hints to get out of this, By comparing my broken system to a recent backup it showed up there was a new file in /etc/ concerning an initramfs. qfile showed that this file has been installed by genkernel. It's unbelievable but just emerging genkernel broke my system though I've never used it, yet. Just emerge -C sys-kernel/genkernel and my system was sane again. Wow, that is strange. I have genkernel installed, never used it, don't use initramfs, and have not hit this problem yet. The /etc/initramfs.mounts file only has one uncommented line: /usr No idea what that's for...
Re: [gentoo-user] udev-181 and kmod vs module-init-tools
On Mon, Mar 19, 2012 at 11:46 AM, Hinnerk van Bruinehsen h.v.bruineh...@fu-berlin.de wrote: it worked flawless for me. Nothing bad happened. If it's a server you could also build a monolithic Kernel and remove the dependency completly since most servers don't need loadable modules. It even adds a little security-wise... Thanks... I tried it, but didn't reboot yet. I was going to upgrade kernel at the same time, it tells me this during make install phase: DEPMOD 3.2.11-gentoo Warning: you may need to install module-init-tools See http://www.codemonkey.org.uk/docs/post-halloween-2.6.txt But looking in the scripts seems the test is not valid for kmod: if ! $DEPMOD -V 2/dev/null | grep -q module-init-tools; then echo Warning: you may need to install module-init-tools 2 echo See http://www.codemonkey.org.uk/docs/post-halloween-2.6.txt; 2 sleep 1 fi so I think it's a false alarm. Now I just found http://comments.gmane.org/gmane.linux.kernel.modules/659 this seems to confirm it. And of course while I was compiling, the new kernel 3.3 has showed up in portage, with this problem fixed. :) Now I'm rebooting, watching ping replied, fingers crossed... it's alive! Success. :) Thanks, Paul
[gentoo-user] Lua libraries
Hello everybody, I need to install luaimap4 library on my computer. I can't find any ebuilds and I was thinking Is there another way to install lua libraries, which are written mostly in LUA? I have found luarocks. Are there any other options? Thanks very much Ignas
[gentoo-user] Attachments with Yahoo Mail in Firefox
I can't sent a reply email with an attachment when using Yahoo mail in Firefox. However, on a Windows box it works just fine with explorer. This seem very strange because I can create a new email and a file attaches without error, but when I reply to an email the attachment fails. Does this happen to anyone else? It seems very strange.Thanks--dhk
Re: [gentoo-user] Goodbye to gentoo?
Hello! On Mon, 19 Mar 2012 08:26:21 -0600 Maxim Wexler maxim.wex...@gmail.com wrote: ... The problem is syncing portage on another machine, my netbook, ubuntu-based. You could just download the portage snapshot and unpack it on your Gentoo machine. Regards, - v...@ukr.net
Re: [gentoo-user] ppp-gentoo woes cont'd
Have you checked it's not DNA related? - used IP numbers rather than urls in pings etc? Try panga/trace route to upstream IPs. BillK On 20/03/2012, at 0:33, Maxim Wexler maxim.wex...@gmail.com wrote: Just got back from gentoo land. Arrrgh, gmail won't let me attach files, just sits there spinning. So I'll have to make do with pastebin. http://paste.ubuntu.com/890854/ Is a chronicle of the commands entered. First having booted and not changing anything, I do #ifconifg, then I do #route -n, then #pon isp to connect. Then there is the running tail of the messages log, ifconfig, route -n Next, rmmod the drivers. I do #poff isp to bring down ppp0, tail the messages, And so on... NB: the crash of the time daemon doesn't matter. Like everything else I need to work around the problem; in this case via a script in /etc/ppp/ip-up.d and /ip-down.d. The problem persists if I don't start the daemon at all. At the bottom of the file I've included the /etc/ppp/ip-up script. The scripts it refers to are here: 30-wins.sh http://paste.ubuntu.com/890854/ 40-dns.sh http://paste.ubuntu.com/890857/ 50-initd.sh http://paste.ubuntu.com/890857/ 90-ntpd.sh http://paste.ubuntu.com/890857/ The first two don't apply. 50-initd.sh, I don't quite grok. Hope somebody has the patience to go through this ;) MW
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Mon, Mar 19, 2012 at 12:35:26AM +0200, Alan McKinnon wrote systemd is like Captain Picard of STTNG (Start Trek The Next Generation) always saying make it so. *HOW DO YOU MAKE IT SO? That intelligence has to be somewhere. So what alternative do you propose? A bash or ash script is more guaranteed to run than a binary. Shoving all that intelligence into the service itself, means that the service has to start up in order to determine whether it's safe for the service to start up. What's wrong with this picture? The intelligence goes in the init system's config file for that service of course. I know I didn't clearly say so, but that's where it goes. The config file can specify upper/lower limits, variables, settings, etc, etc. But in the end, some executable somewhere is going to have to parse the config file, check the actual environment, and decide whether or not to launch the service, and with what parameters. Note also that many open source programs are multiplatform. I.e. they run on FreeDOS with DJGPP, multiple flavours of Windows, multiple BSDs (including Apple), linux, and multiple commercial unix flavours. Do you really want to throw multiple platform-specific IFDEFs into the program code to allow the services to do the appropriate platform-specific initialization? Isn't it be easier to move the service setup out of the main service, and let the maintainers of the specific platforms figure it out? One last question. Let's go back in time 20 years, and assume that you're the maintainer for a program that runs as a service. A small handfull of end-users come along. They're running a hobby OS that fits on a couple floppies. Said hobby OS has been cobbled together by a university student. Would you... * download that university student's hobby OS, and install it * throw in a bunch of additional IFDEF initialization code in your program * test and debug the program to make sure it runs under that OS -- Walter Dnes waltd...@waltdnes.org
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Mon, 19 Mar 2012 13:17:01 + Neil Bothwick n...@digimed.co.uk wrote: On Sat, 17 Mar 2012 09:00:37 +0100, Andrea Conti wrote: Personally I stopped bothering with a separate /usr ages ago, so I don't really care. Having given this some thought recently, I am coming round to the view that the problem is /usr itself. It may have had a place when boot disks were limited in size, but I really don't see the point it in at all nowadays. This whole question of which bin directory does code belong in should be why do we need so many bin directories? There are some separations that do make sense in a Unix context: - */bin vs */sbin is one. Nothing to do with security, but */sbin can go in root's PATH and apps that only makes sense when run as root (eg mkfs) go there. This avoids cluttering the display with useful crap from tab-completion. - / vs /usr/local. I like this one, everything I build and install myself without help from the package manager goes here. On FreeBSD it means I used ports to install the stuff and it's not in world. I do need this distinction in my world. Perl CPAN too for the same reasons. - /opt. Um yeah, OK. So we have these things called proprietary apps where devs just want to make a directory specially for their app and dump everything belong it. OK, as a scheme, it works. I don't like it but I don't have a better idea. / vs /usr is the only one I don't need myself, as /usr is not read-only (a very valid use case) and I don't have thin clients on the network. -- Alan McKinnnon alan.mckin...@gmail.com
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Mon, 19 Mar 2012 10:33:11 -0400 (EDT), Bruce Hill, Jr. wrote: And for the Lennart fanboi, his coding is so questionable that Lennartware has become derogatory slang. (Of course, you already know that.) And this is such a common term nowadays that when Googling for Lennartware only one reference to it turn up on the first page, and that is your post! mingdao@workstation ~ $ grep Lennartware irclogs/* irclogs/#gentoo-dev.log:09:01 @bonsaikitten Caster: do you see now why I don't appreciate Lennartware? irclogs/#gentoo.log:10:56 @bonsaikitten Zaba: Lennartware. Linux needs to be more like MacOS Wow, 2 mentions on IRC - the term really has invaded the English language. --- 4 references Still not enough for Google to see it, barely enough for a contrived allegation. Various other mentions of systemd being nefarious software are mostly amongst kernel devs and might not use the word Lennartware, but the logical reasons why systemd is a _bad_ idea are the same. Where does systemd come into it? Gentoo is following udev's upstream requirements. These may have been triggered by udev's support for systemd but that in no way means that systemd is required. Greg K-H is also in favour of making /usr available to early boot, are you going to accuse him of shoddy coding too? -- Neil Bothwick Of course, I could switch back to Windows. At least there, if I have a problem, I don't suffer under the illusion that I could ever fix it. - Unknown (paraphrased) signature.asc Description: PGP signature
Re: [gentoo-user] Re: systemd? [ Was: The End Is Near ... ]
On Mon, 19 Mar 2012 18:58:22 -0400 Walter Dnes waltd...@waltdnes.org wrote: On Mon, Mar 19, 2012 at 12:35:26AM +0200, Alan McKinnon wrote systemd is like Captain Picard of STTNG (Start Trek The Next Generation) always saying make it so. *HOW DO YOU MAKE IT SO? That intelligence has to be somewhere. So what alternative do you propose? A bash or ash script is more guaranteed to run than a binary. Shoving all that intelligence into the service itself, means that the service has to start up in order to determine whether it's safe for the service to start up. What's wrong with this picture? The intelligence goes in the init system's config file for that service of course. I know I didn't clearly say so, but that's where it goes. The config file can specify upper/lower limits, variables, settings, etc, etc. But in the end, some executable somewhere is going to have to parse the config file, check the actual environment, and decide whether or not to launch the service, and with what parameters. Note also that many open source programs are multiplatform. I.e. they run on FreeDOS with DJGPP, multiple flavours of Windows, multiple BSDs (including Apple), linux, and multiple commercial unix flavours. Do you really want to throw multiple platform-specific IFDEFs into the program code to allow the services to do the appropriate platform-specific initialization? Isn't it be easier to move the service setup out of the main service, and let the maintainers of the specific platforms figure it out? One last question. Let's go back in time 20 years, and assume that you're the maintainer for a program that runs as a service. A small handfull of end-users come along. They're running a hobby OS that fits on a couple floppies. Said hobby OS has been cobbled together by a university student. Would you... * download that university student's hobby OS, and install it * throw in a bunch of additional IFDEF initialization code in your program * test and debug the program to make sure it runs under that OS I'm not sure where you're going with this. We're discussing an init system and good, simple ways to start services. App maintainers are going to continue to do whatever they feel they ought to do, some might write the systemd files, some might not - that is what already happens. Someone has to write it and what goes in it depends on what the app code does, not the other way round. I'm not punting the merits of systemd, I don;t know enough about it. I started off by saying a nice clean easy way to do init would be awesome, as I'm sick and tired of having to deal with sysvinit. That's all, don't read more into it than that. As for the last question, I really have no idea where you're taking this. I don't know the answer, I've never been a maintainer in that position. Being the arrogant shit that I am, I reckon I would probably tell the user to piss off and I don't support hobby crap. But hey, that's just what I think I might say while sitting here on my couch. Any other answer would be equally made up. -- Alan McKinnnon alan.mckin...@gmail.com
Re: [gentoo-user] Re: Heads Up - sys-apps/net-tools-1.60_p20120127084908 might break your system -- openvpn still fails
On Sat, 17 Mar 2012 08:48:08 -0700, fe...@crowfix.com wrote: This appears to help with eth0 etc, but my openvpn fails with Linux ifconfig failed: could not execute external program Leaving the sbin symlink in place fixes this. I didn't see any gentoo bugs for this -- I can create a new one, but is there another one related to the symlink that I should add this note to? There is a bug, the fix is to re-emerge openvpn. Apparently openvpn has hardcoded paths for ifconfig and friends, which are set when it is compiled, based on the locations at the time. Re-emerging builds it with the new locations. -- Neil Bothwick Kludge: (v., adj., or n.) to fix a program in the usual way. signature.asc Description: PGP signature
Re: [gentoo-user] Beta test Gentoo with mdev instead of udev; version 5 - failure :-(
On Sat, 17 Mar 2012 16:11:58 +0200, Alan McKinnon wrote: Why not connect that TV? ;) Because the hardware to do so would cost around £100, USB sticks cost rather less :P The hardware is more shiny than the USB stick. Go on, do it. You know you want to. If I had £100 to spend on shiny, I'd spend it on shiny for me. If I don't have £100 to spare, it's probably because I've already spent it on shiny for me :) -- Neil Bothwick I've got a mind like a... a... what's that thing called? signature.asc Description: PGP signature
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Tue, 20 Mar 2012 01:04:04 +0200, Alan McKinnon wrote: - */bin vs */sbin is one. Nothing to do with security, but */sbin can go in root's PATH and apps that only makes sense when run as root (eg mkfs) go there. This avoids cluttering the display with useful crap from tab-completion. Agreed - / vs /usr/local. I like this one, everything I build and install myself without help from the package manager goes here. On FreeBSD it means I used ports to install the stuff and it's not in world. I do need this distinction in my world. Perl CPAN too for the same reasons. That too, or you can move all system stuff from /usr to / and put user stuff in a directory with an appropriate name, something that reflects its purpose, maybe something like /usr. - /opt. Um yeah, OK. So we have these things called proprietary apps where devs just want to make a directory specially for their app and dump everything belong it. OK, as a scheme, it works. I don't like it but I don't have a better idea. Yep. / vs /usr is the only one I don't need myself, as /usr is not read-only (a very valid use case) and I don't have thin clients on the network. Separating system and user-compiled/installed software makes sense. Separating root and general programs makes sense. Separating system programs and libraries based on fairly arbitrary, and moveable, criteria does not make sense to me. As for making /usr read-only; it is generally only writeable by root and anyone with the root password could remount rw anyway, so there's not much point there. -- Neil Bothwick Ninety-Ninety Rule Of Project Schedules - The first ninety percent of the task takes ninety percent of the time, and the last ten percent takes the other ninety percent of the time. signature.asc Description: PGP signature
Re: [gentoo-user] Beta test Gentoo with mdev instead of udev; version 5 - failure :-(
On Mon, 19 Mar 2012 23:23:51 + Neil Bothwick n...@digimed.co.uk wrote: On Sat, 17 Mar 2012 16:11:58 +0200, Alan McKinnon wrote: Why not connect that TV? ;) Because the hardware to do so would cost around £100, USB sticks cost rather less :P The hardware is more shiny than the USB stick. Go on, do it. You know you want to. If I had £100 to spend on shiny, I'd spend it on shiny for me. If I don't have £100 to spare, it's probably because I've already spent it on shiny for me :) You don't have an 'undred quid to buy shiny for your missus? And you've got 10 years on me by my reckoning, so odds are good your missus has been your missus for 30 odd years How'd you manage to make it to 30 years without knowing the value of an 'undred quid to buy shiny for your missus? -- Alan McKinnnon alan.mckin...@gmail.com
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
On Mon, 19 Mar 2012 23:33:39 + Neil Bothwick n...@digimed.co.uk wrote: As for making /usr read-only; it is generally only writeable by root and anyone with the root password could remount rw anyway, so there's not much point there. I was thinking here more of /usr mounted -t nfs root on nfs client != root on nfs server hence the need for rootsquash. But these days that setup is becoming a niche thing, the last one I saw was in a university lab and I've never actually admined one myself. -- Alan McKinnnon alan.mckin...@gmail.com
Re: [gentoo-user] The End Is Near ... or, get the vaseline, they're on the way!
Have two here - disk less atoms as mythtv front ends - seems a common use case in the mythtv world. And another advantage is they sidestep the whole /user mess :) BillK On 20/03/2012, at 7:49, Alan McKinnon alan.mckin...@gmail.com wrote: On Mon, 19 Mar 2012 23:33:39 + Neil Bothwick n...@digimed.co.uk wrote: As for making /usr read-only; it is generally only writeable by root and anyone with the root password could remount rw anyway, so there's not much point there. I was thinking here more of /usr mounted -t nfs root on nfs client != root on nfs server hence the need for rootsquash. But these days that setup is becoming a niche thing, the last one I saw was in a university lab and I've never actually admined one myself. -- Alan McKinnnon alan.mckin...@gmail.com
Re: [gentoo-user] Initramfs or move /usr to /, oh my...
On 03/18/2012 06:44 PM, Tanstaafl wrote: Ok, I have never used genkernel, and have no desire to... I have no idea what dracut is or how to use it... While genkernel also can generate kernel configs for you, both dracut and genkernel are initramfs creators: they take repeated creation (after each kernel update) of a potentially complex initramfs off your shoulders. I have a remote system that has /usr on a separate partition. So... How do I find out if I am actually *using* an initramfs right now (I know it is built into the kernel) Afaik the in-kernel one is a dummy, only. If you don't remember anything about genkernel or dracut from the past you do not have an initramfs in your Gentoo installation(s). , and If I am not, how do I do this without using genkernel? Is dracut the *only* other option? Is it easy/trivial to set one up manually? A manual initramfs is not that easy, no. And it becomes outdated more easily than an initramfs-creator-based approach. On that note - is it possible, and if so, does anyone have any decent detailed How-to's on how I might be able to convert a separate /user to one on directly on / on a running system? From my current understanding (please double-check, no warrenties): 0. Make backups 1. Boot some sort of live/rescue CD (so you can fiddle with /usr without shooting your foot) (2. Enlarge space on partition/device root (the one holding /)) (3. Enlarge file system sitting on partition/device root) 4. Make a new folder root/usr 5. Copy content from usr/ to root/usr/ - Watch out for use of Xattr (extended file attributes) - Watch out for use of POSIX ACLs - Use something like --archive with cp/rsync to maintain attributes 6. Update root/etc/fstab 7. Reboot 8. Resolve partition usr Good luck. Best, Sebastian
Re: [gentoo-user] Changing compilers
On 19/03/2012 10:18 PM, Andrew Lowe wrote: On 03/19/12 22:02, Michael Mol wrote: On Mon, Mar 19, 2012 at 8:32 AM, Mark Knechtmarkkne...@gmail.com wrote: On Sun, Mar 18, 2012 at 10:26 PM, Andrew Lowea...@wht.com.au wrote: Hi all, [snip] ... ... [snip] which currently ranks at 151 in the top 500 list :) It's amazing how fast this list changes, 6 months ago, this machine was at 107 and 6 months before that 87. Andrew Just in closing on this subject, thanks to those who responded, have a quick look at this page from the llvm/clang project: http://clang.llvm.org/diagnostics.html For my non FEA/CFD programming, I don't care if clang is 5 - 10% slower than gcc, the diagnostic output that clang produces looks to be spectacular in comparison to gcc and will be enough for me to dump gcc and shift to clang. Andrew
Re: [gentoo-user] Changing compilers
On Mon, Mar 19, 2012 at 02:00:18PM +0800, Andrew Lowe wrote
I'm looking for the faster code, the run time to be faster - I compile
the FEA CFD code once but will be running many jobs.
Andrew
Since you're asking on the Gentoo list, can I safely assume you use
Gentoo? Gentoo gives *MUCH MORE* than 1% or 2% improvement, *IF YOU
OPTIMIZE PROPERLY*. This does not mean going into Gentoo-ricer
territory, but it does mean using the features of Gentoo to their
fullest. Gentoo allows you to build binaries with gcc that are tuned to
*YOUR* cpu. The advantage is that it gets the maximum out of your cpu.
The disadvantage is that a binary compiled for a newer cpu will probably
not run on an older cpu on another machine. In your /etc/make/conf, I
recommend...
CFLAGS=-O2 -march=native -mfpmath=sse -fomit-frame-pointer -pipe
CXXFLAGS=${CFLAGS}
MAKEOPTS=-j1
The -march=native setting tells gcc to build binaries to use all the
cpu's available features. The CXXFLAGS variable is specific to C++.
Set it identical to CFLAGS, so that C++ code gets the same settings.
The MAKEOPTS=-j1 setting slows down the build process slightly, but it
does *NOT* affect the final binary. It does avoid some occasional
mysterious hard-to-reproduce errors that stop builds dead in their
tracks. The first time you bang your head against the wall for a couple
of hours trying to figure out the problem, you'll waste more time than
you've saved with a higher numbered j value.
Note also that if you've done a recent fresh install, you should...
* emerge system
* emerge world
* rebuild the kernel entirely and reboot
...in that order. The binaries from the install CD are generic i686 (32
bit) or amd64 (64 bit) with no optional machine instructions selected..
They have to be this way to install properly on 8-year-old machines.
But this doesn't take advantage of the faster instructions on newer
machines. The following is a true story that happened to me, not a
friend-of-a-friend. I have a 4 1/2 year old Dell with an onboard Intel
GPU. Right after the install, it could not keep up with 1080i video
from my TV tuner box, or even teh slowest speed for NHL GameCenter Live.
After emerging system+world and rebuilding, the same machine was able to
view 1080i TV and run the low-bandwidth version of NHL GameCenter Live.
That is a very significat difference.
Note that merely optimizing the program itself isn't enough. The
binary is dynamically linked to various math libraries. The program
reads data from and writes output to disk. And there are always kernel
calls along the way. So optimizing every math library, disk I/O code,
and kernel code contributes to faster execution.
--
Walter Dnes waltd...@waltdnes.org
