Linux and the Desktop Pentium M: Uncommon Performance
by Kristopher Kubicki on December 24, 2004 12:00 PM EST- Posted in
- Linux
Summing it all up
Overall, Dothan provided us with some sporadic, but interesting, performance gains and losses. Unfortunately, Pentium M just doesn't scale similarly to Pentium 4 or Athlon 64 in any application, although it does seem to mimic the performance of one or the other occasionally. On our OpenSSL tests, Dothan continually out-nudged even our mid-range Athlons, but then fell far behind in compilation and some content creation tests.There are, however, bottlenecks in the performance. High speed memory is something that our Dothan severely lacked on Linux, and we would certainly appreciate the next generation Alviso chipset to support something a little faster than DDR400. However, as Pentium M is a notebook chipset first and a blade/desktop chip second, the demands of low power notebook memory certainly take priority over a niche SFF/HTPC crowd.
The first surprise in our analysis came with the SQL database tests. Our windows benchmarks have shown in the past that the additional L3 cache can be quite helpful for database applications, and the 2MB L2 cache found on the Dothan plays a huge part in boosting performance. On the other hand, the additional cache might have been the same reason why GCC performed so poorly - although we hope that the Linux compile test was just a fluke (Update: Please see the note on the Compiling page. We believe we had an isolated fluke with the PATA driver that limited our performance). Other benchmarks put Dothan right in the upper middle of the pack, usually beating out the Pentium 4 offerings, but occasionally beating out the best that our Athlon 64s could produce as well.
Dothan isn't the miracle chip that we would have liked it to be. For starters, it is horribly expensive still. The 2.1GHz Dothan that we previewed today runs at around $500, and the motherboard costs another $270. For just a barebones configuration, our Pentium M desktop runs at around $1000. Granted, the overclockability on Pentium M seems outstanding, but finding slower, cheaper Dothans in socket 479 pin configurations may be a problem.
Unfortunately, we are only getting a small glimpse of the story here today. Our preliminary benchmarks on Windows show that Dothan does some awesome things on Windows; the compilers and operating system get a little more help from Intel in the design phase. Unfortunately, the extremely powerful and free Linux compiler remains dully unaware of many of the benefits that Pentium M has to offer, and as a result, it gets hurt painfully under the default or wrong compile flags.
All in all, Dothan does some very exciting things. The promise of cool, efficient powerhouses - from Intel, nonetheless - certainly has our attention. We will be keeping a very close eye on Pentium M over the next few months, particularly with the upcoming Alviso launch. If Dothan's Linux performance keeps up this well on the 855 chipset, we can't wait to see what it does on faster memory and the 915 Northbridge.
Facebook
Twitter
RSS
47 Comments
View All Comments
Lynx516 - Monday, December 27, 2004 - link
What yozza says is true gcc 3.4.1 is old especially a pre release. As yozza said it has some bugs in it with respect to the pentiu_m march flag.I have been running gcc 3.4.3 for atleast 2 months and i definatly was not one of the first to use itKristopherKubicki - Monday, December 27, 2004 - link
Yozza: I made some corrections. Saying GCC 3.4.1 is "pretty old" seems pretty hard for me to swallow, but the rest of what you say seems correct. I would be interested in seeing your P-M compile time benchmarks.Kristopher
Yozza - Sunday, December 26, 2004 - link
The march=pentium-m flag was pretty broken on earlier gcc 3.4 releases, and it seems that you're using quite an early "3.4.1 (prerelease)" version, which could explain a few things, especially your TSCP benchmarks, where the Pentium-M is the only one to have its performance _decrease_ with optimised (march=pentium-m) compiler flags. This clearly indicates some issues with pentium-m optimisations in your gcc revision (3.4.1 is pretty old these days).The extremely slow kernel compile time is especially surprising though. I did some test just now on my 1.7GHz Banias P-M, and the kernel compile times do NOT appear to correspond with your results. So I guess something seems amiss with your system configuration.
There was certainly some pretty impressive performance in the integer-dependent tests such as the database one -- we already know that the P-M's fp performance isn't that great, which explains some of the more fp-dependent benchmark results. I for one was pretty impressed by its performance in the majority of the benchmarks and by its scaling possibilities, both wrt FSB/memory and core freqs. If only Intel would upgrade the platform to 800MHz FSB with dual channel DDR400; such a configuration would be appear to hold a lot of promise.
The argument that "Dothan is adherently a linear processor" doesn't hold water either (it should be "inherently" too), since the kernel compile uses one thread by default. Regardless, it should have been possible to test different CPU schedulers to determine how well Dothan deals with multi-tasking loads, particularly wrt compile times by comparing different "MAKEOPTS=-jX" settings. Behaviour under such loads is as dependent on the CPU scheduler as it is on the CPU itself anyway.
Hence, clearly, the comment "When we stack multiple jobs on the processing queue, Dothan spends a huge majority of its time swapping around" is flawed and incorrect. The implication that the CPU 'swaps around' somewhat like memory paging to disk is rather inaccurate to say the least.
larson0699 - Sunday, December 26, 2004 - link
"...The only additional offering that the 855GME feature provides is a 64-bit PCI-X (not to be confused with PCI-Express) bus..."PCI-X is a "feature" of the 6300ESB ("Hance Rapids") southbridge, NOT the 855GME northbridge.
Other makers used the standard ICH5 southbridge, hence no PCI-X slots.
Such a system would be awesome for a mini RAID server. An adapter from 3Ware would be the perfect utilization of the PCI-X slot provided. Ahh...
vaystrem - Sunday, December 26, 2004 - link
Why not Gentoo?You could do two boxes - compile the entire OS optomized for the Pentium M with latest GCC revision and do the exact same for the Athlon 64 Box.
That would be a lot more interesting comparison and useful than doing these benchmarks on top of Suse whose default optomizations are certainly hurting the PentiumM.
bhtooefr - Saturday, December 25, 2004 - link
BTW, I mean that maybe Intel shut them down...bhtooefr - Saturday, December 25, 2004 - link
Hmm... maybe you're right... After all, it seemed that PowerLeap was dodging my questions about their P-M adaptor (and then tried to pimp the PL-AXP (basically a golden fingers card for Socket A) - if I wanted to unlock an AXP, I'd get a pencil ;-))...Here's the chat: http://cpu-museum.de/forum/download.php?id=334
sprockkets - Saturday, December 25, 2004 - link
Oh well, I just looked at SuSE 9.1 again and 9.2 has the ability with kpowersave to select easily the profiles it will use to run, and will say what processor speed it currently is throttling to.ElFenix - Saturday, December 25, 2004 - link
i'd like to point out that 'heat sink' is actually two words. thanks.KristopherKubicki - Saturday, December 25, 2004 - link
bhtooefr: I am pretty sure intel wont let them do that.Kristopher