I'd agree with Miguel. In general, I can't tell the difference in most applications. I did some tests on OpenOffice a few years ago and found the 8 hour compile time wasn't worth it in terms of performance. Where Gentoo shines is in giving you the increased ability to shrink your installed size. I have nothing on this machine that I don't specifically want there and I work overly hard to keep my compile flags exactly right.
<br><br>That said, I'd be interested in seeing your results. It would be nice to know that all of this work is beneficial for more than just keeping me happy.<br><br>Mike<br><br><div><span class="gmail_quote">On 4/30/07,
<b class="gmail_sendername">Miguel Sousa Filipe</b> <<a href="mailto:miguel.filipe@...">miguel.filipe@...</a>> wrote:</span><blockquote class="gmail_quote" style="margin-top: 0; margin-right: 0; margin-bottom: 0; margin-left: 0; margin-left: 0.80ex; border-left-color: #cccccc; border-left-width: 1px; border-left-style: solid; padding-left: 1ex">
Hi there,<br><br>On 4/29/07, Peter A. H. Peterson <<a href="mailto:pedro@...">pedro@...</a>> wrote:<br>> Hi Everyone,<br>><br>> My name is Peter Peterson and I represent a group of a grad students
<br>> at UCLA. We're in a computer systems performance analysis course and we<br>> were hoping to do a general performance comparison of gentoo vs. a<br>> popular binary i386-compatible distribution (probably ubuntu) in some
<br>> "real-world" server tests to try and meaningfully calculate the<br>> performance gains that local compilation provides. (For example,<br>> apache2 requests processed per second on the same hardware.)
<br>><br>> I've subscribed to this list because we want the gentoo community to<br>> be involved in helping us design the tests so that we can hopefully<br>> all feel good about what and how we are testing the systems.
<br>><br>> We have no particular outcome in mind; our group represents a wide<br>> range of computer users, from Mac, Linux, and Windows enthusiasts, and<br>> we have all used a wide variety of Linux distributions. We have simply
<br>> noticed that much of the discussion of gentoo's performance advantage<br>> is anecdotal and we're genuinely hoping to provide some meaningful<br>> experimental data for discussion. Also, if anyone knows of any
<br>> available benchmark data or papers on this subject, we'd love to hear<br>> about them. There was apparently a paper on slashdot a couple of years<br>> ago, but the host it was on appears to now be squatted. For that
<br>> matter, if this is a well understood or closed issue (for example, if<br>> the statistics that people quote are actually from good experimental<br>> data) please let us know.<br>><br>> Is anyone here interested in discussing this project? We are
<br>> specifically interested in discussing methodology, testing suits,<br>> CFLAGS and other options. Our desire is not to "trick out" gentoo or<br>> ubuntu, but rather quantify the performance benefit that gentoo has
<br>> over binary distributions with "normal" compile flags (whatever normal<br>> is).<br>><br><br>A good CFLAGS would be something not very agressive, something like:<br>-march=<cpuType> -O3 or -O2 and at most -fomit-frame-pointer.
<br>(Scientific workloads can speedup considerably with: -ffast-math)<br><br>Having experienced and done some benchmarks with gentoo and other<br>distros on servers and on scientific workstations.<br> What I found is that sometimes gentoo lacks critical performance
<br>patches in glibc that are applied to mainstream distros (redhat,<br>suse..etc) that provide boosts in memcpy, memset, etc..(I remmember a<br>discussion about that some years ago).<br> What I also found out is that the compiler flags only affect
<br>workloads that are very compute intensive. not something that depends<br>almost completely on FSB load or IO load.. like most server<br>workloads... -O3 doesn't do much to a working set full of<br>unpredictable branches (like server workloads usually are) and low IPC
<br>rate.<br><br>I really do believe performance boost from gentoo to be practically<br>negligible. The difference will only be apreciable in very few corner<br>cases. Most distros also optimize critical aplications such has:
<br>openssl, mplayer.. reducing the possible corner cases.<br><br>Anyways, doing a "academic" benchmark would be a good idea.<br><br>Something like:<br>micro-benchmarks:<br>- stream (mem bandwith benchmark)<br>- ??
<br><br>macro-benchmarks:<br>- apache2 + gzip + php(make it cpu intensive, not IO intensive)<br>- xmlmark ?<br>- kernelbench<br>- pybench ?<br>- openssl bench<br><br><br>about methodology:<br>- same system, same bios version, same disks.
<br>- All OSes must be installed in the same disk partitions.<br>- the will be trouble about the kernel config:<br> - for mainstream distros you should use the kernel that is provided.<br> - for gentoo, gentoo-sources configured by someone which is
<br>experienced, and informed about configuration impacts on performance<br>(ideally a kernel hacker?).<br><br>- should use the stable versions in gentoo portage?<br>- or should use the same application versions used on mainstream distro?
<br><br><br><br>--<br>Miguel Sousa Filipe<br>--<br><a href="mailto:gentoo-performance@g.o">gentoo-performance@g.o</a> mailing list<br><br></blockquote></div><br>
|
