Linux - ServerThis forum is for the discussion of Linux Software used in a server related context.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
Here's an idea - download an flv movie from somewhere and see how quick your computer can convert it from flv format to avi. Here's the command I used:
$ time ffmpeg -i ./somemovie.flv -ab 128 -ar 44100 test.avi
real 0m12.203s
user 0m12.029s
sys 0m0.152s
Takes 12 seconds on my PC which is a 2.33Ghz E6550 Core 2 Duo - I'm betting my money heavily on your machine though...
Something that might make it work a little harder is MDCrack - I believe it can use multiple cores, but being Windows-only I'm not sure... I've used it successfully to see how well my PC could crunch numbers. You can get MDCrack here: http://c3rb3r.openwall.net/mdcrack/
Generate some ciphertext like this:
$ md5sum -t "test12"
Then try cracking it using MDCrack - sorry, but I can't remember the syntax for MDCrack. It'll be on their website. I think my PC would take about 5 minutes to exhaust all the possible hashes up to 7 alphanumeric characters.
I think the best thing to do is load it up to do whatever it needs to. There is a tool called "stress", but I don't know how useful it would be to you since I have no idea what you'll be testing. Some things can't really be properly tested without a special environment to test in (for example, multiple high-speed network cards + several individual computers to send requests for http, database queries, etc).
The characteristics which are most important to you really depends on your use. For example, when processing satellite data into images it really is the computation speed and memory which is important; even if it took as long as 5s to load the data into memory, the processing time (depending on what you want to do) may be on the order of minutes. If you're running a high-throughput web server, memory, storage retrieval, and network bandwidth are very important, but raw processor speed is not the highest priority.
Since I run compute-intensive tasks, another thing which is very important to me is that the results are correct. Some optimizations, hidden approximations (some people thinking they're 'clever' for using a fast - but wrong - shortcut), and buggy hardware or software may produce fast results - but if they're wrong, what's the point?
As for how well CentOS makes use of the processors, one factor affecting that would be the kernel configuration (and to some degree the software running); specifically, settings affecting the 'CPU affinity' of a process. A complication comes up with "threads" - there is the Linux way (a copy of the parent process with shared resources, similar to a 'thread' in WinDuhs) and the older UNIX way where you essentially have an unrelated process and all passing of information between 'threads' (really parent and children processes) has to be done via Inter Process Communications mechanisms. Otherwise the latest scheduler does a pretty good job of distributing the load and you needn't be too concerned - unless real operating conditions show that your load balancing is weird.
i have got a IBM eServer Xseries 335 and i would like to run somehting on it that is going to make it work fairly hard to see how well it performs?
Thanks
Steve
1) Run one or several folding@home clients.
2) See how quickly it can encode movie into x264 (the only codec I know that can use several processors at once during encoding).
Here's simple x264 2-pass encoding sciprt, which uses all available CPUs:
Honestly, apart from encoding x264 video (or encoding several non-x264 videos at once), compiling stuff with -j switch or making several archives at once I can't think about any good use for multicore/multicpu machine.
Quote:
Originally Posted by halon1301
$ time ffmpeg -i ./somemovie.flv -ab 128 -ar 44100 test.avi
I agree with pinniped. Install what the system is going to be used for and use specific benchmark tests to test performance, etc. Just running a command to encode a movie isn't going to give you accurate results if say, you were going to run a webserver with a database backend on it. Unless you're going to run this server to encode movies that is..
I agree with pinniped. Install what the system is going to be used for and use specific benchmark tests to test performance, etc. Just running a command to encode a movie isn't going to give you accurate results if say, you were going to run a webserver with a database backend on it. Unless you're going to run this server to encode movies that is..
thats fair enough.
it is used to sever an internal website with database and Samaba shares. What can i bench mark this with.
it is used to sever an internal website with database and Samaba shares. What can i bench mark this with.
Steve
Well if you're running apache for the webserver portion of it, you could use ab which comes with apache, it is it's own benchmarking tool.
As for samba, not sure, maybe pound the crap out of the web server and test by copying large files to and from it, etc. I try to avoid samba so never had to benchmark it.
There are plenty of benchmarks for databases as well, depending on which one you're using.
Run VirtualBox with 15 guest computers at the same time Believe me, that will show some damage to your resources.
My desktop has 2 Xeon 2.2 GHz processors, and I can really tell when I'm running a couple virtual boxes.
But that doesn't give him accurate results in what the machine is going to be used for. Why test a machine or benchmark a server with virtual machines when he has no intention of running virtual machines? That's not going to tell him how well a webserver + database and samba shares are going to hold up under heavy load?
Come on people, think before you reply. Try to give real world answers instead of just pointing users to unreliable solutions that have nothing to do with the question or problem on hand.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.