• blah
  • blah

Exciting?
No, I'm not really. However, I do have a few things to say. First of all, you'll notice the return of the properly working counters. No longer do you have to wait five minutes for the darn counter to load, which results in java script errors when you try and click on something on the page. I'd also like to thank Action for pointing out that my link to the April Archives section was broken. It's fixed now, so feel free to visit and look at the news of yore (the link is at the bottom of the page here). 

I've updated the info in the results table, and yes, I'm aware of the script error there. I'll try and work on that later. It's actually an easy fix, but I'll need Guru to put everything in place for me. So, since I wasn't able to fix that, I finished my update on the speed tips page. I've included more up to date information, such as chipsets, new CPU info, and updated OS info. I removed some old stuff that wasn't relevant anymore, etc.. If you see something that I forgot, misspelled or just plain broke, do feel free to let me know. -Rat

Playing with numbers...

Bench till you drop
Times galore and though the results table might not reflect those I mention, it will catch up eventually. Anyway recognition to those who have submitted is due. Thanks for your efforts. Anyone I've missed apologies in advance. Since several are from 'jonluck' I'll start there. Ranging from 7:40 to 14:03 on a multitude of OS's and Celeron speeds shows devotion to duty. You've been a busy bunny, must be spring. It would be fun to think this was one system with a variety of  mobo's and os's being swapped in and out for the thrill of benching knowledge. Seems unlikely. Anyway jon your times are now peppering the results. In the rare sighting department we have gpriatko with a solid 5:23 using W2K and Williams Mac G3 copping a 6:23, which seems a little slow for such a beast. But I guess some people use their systems in the real world too.
Two times from Guru: Celeron at 583 giving 6:53, good but nothing special and a much improved 6:26 by upping the fsb to 110. Any 366 Celery at 600MHz plus is impressive. Serious cooling or just a really good batch? There is probably still a little more to come though. Bo used CAS2 settings to drop his time to an excellent, for a PII, 7:18. Memory tweaks are the easiest and most stable, assuming you are not at the top end of overclockdom.
In the very hotly contested 4 hour bracket Turbo has upped the bus again to post 4:36. You guys just keep pushing it don't you. And just at the point where I say a 155 fsb is getting silly along comes Tomslik's 4:49 using a 166 bus! Who cares about UT/Quake performance when you can brag a 166 fsb. Action has shaved 4mins off to improve to 4:07, how you ask, gripped? By running one client instead of 2 is the answer. Yet again showing how a 1MB Xeon cache is big enough to give almost no appreciable hit when using 2 clients. Anyone know a cheap source of Xeon's? Great Seti crunchers. Action, the other PIII Xeon time you posted was bit mysterious could you elaborate or was it a bad night?
Honorable mention goes to the new leader in the 'slowest' race. Chris Marble hammered a 105:39 out of his 33MHz box. We are not talking semiconductor dragstrip performance here. Congrats in showing that anything no matter how slow is worthy to crunch for Seti. Without starting an artificially slowed parade of submissions can anyone beat that? Running this slow you could count time by watching the day/night cycle on the Seticam pointed at Arecibo! It's off line at present but you can view your favourite dish (on cam B) in the fall at http://www.seti-inst.edu/seticam/Welcome.html. Beats paint drying, grass growing and snail racing - just. Max out.

Happy Easter!
I'll say this today so that I don't miss everyone tomorrow- Happy Easter. 
Okay, back to business. I'm currently playing with the navbar on the site, so don't worry if something doesn't work- I'll get it fixed in short order. If it's still broken after a couple of days, do let me know. I'm going to link the navbar over to Guru's table, which is a heck of a lot nicer than mine. If I have time later (I dunno, but I'll try), I'm going to update the speed tips, which are sorely in need of updating. Enjoy your holiday, and remember to drive safe. -Rat

Competition Time Folks
Rat here, just posting some news from Max that I had accidentally misplaced in my mailbox. Less than a week old now, it's still interesting!

All of sudden the Register http://www.theregister.co.uk/000413-000003.html has got the Seti bug. This august organ has issued a challenge to find the fastest home owned Seti crunching system. Music to your ears I think. But then those of you submitting benchmarks have been doing just that for quite some time now for merely the respect, admiration and envy of slower rivals. For many it is the satisfaction of wringing a little extra performance and sliding a few minutes up the results table. So an 'Intel bunnyperson' (suit, not contents I assume) as the prize incentive should get you drooling or gagging depending on the sadness in your life. Maybe you always fancied something kinky in a fab plant. I digress. Unfortunately the 'competition' is so full of holes that a large BEM could slime through it. In case you are wondering I have already emailed Mr. Thomas that an obvious candidate as a benchmark wu already exists. Check it out and decide whether you fit the criteria. Now what colour do I want mine in. Max out.

Working behind the scenes
Rat here, and no, I'm not dead. I feel like it sometimes though. This weekend is a perfect example of what can go wrong will go wrong. As the weather is starting to head to happy spring time cheer, I decided it was time to pull my car out of storage. I had Friday off, and Baby and Mommy were out of town for the weekend. So it's the perfect time to get the car out, and to work on the web page. 
I blew a tire right before I parked it, so I had to replace it. On ThursdayI called Tires Plus, the local tire chain, to find out if they had one. Nope, but they could have one by Friday. Excellent. So I ordered the tire, and went to my Dad's house on Friday morning (that's where I parked the car for the winter). After doing fixing a few things on his computer, I took the rim to town to have the tire mounted. After about 30 minutes, I pay for the tire, and they roll it out. Hmm...that looks like a Buick tire, not a Corvette tire. Wrong one. I told the salesman this. He told me that this one was 'an even better tire'. It is for snow, but not for me. Not to mention, I'm changing one tire. That clue anyone else off? Like I have 3 others that DON'T match? Moron. So, he called around and found one in stock in the Plymouth store, which is where I just came from. About 40 miles away. Damn. I went back to my Dad's, and finished a few other things on the car. After my friend got home from work, we took the rim to the Plymouth store. Only to find them mounting the last one on someone else's car. I was told they had one in Apple Valley though. So helpful..
Feeling rather downtrodden for the day (They closed before I could get to the next store), I went back to my Dad's to drop off my friend. I spent the night there, and was too sick on Saturday to do anything. Yech. Come Sunday, I went to the Apple Valley store, which is about 50 miles away. This is getting old. I dropped off the tire and was told it would be 'a while'. Hmm..I have to go to the airport before too long, but I did allow some extra time. An hour and a half later, it was finally one (it takes 10 minutes to mount and balance a tire). Someone remind me not to do business with them again!

Anyways, enough of my complaining. I could go and and talk about my DOA motherboard that showed up Monday. But I won't. Instead, I'm going to get back to work and try and update some more site info. I've started archiving the news info, which you can find right here. I'll be adding a permanent link too, so don't fret about it. I just want to make sure that all of the interesting information that comes up in the news doesn't get lost like it did with the old page, which wasn't archived at all. I'm not sure how much I'll be able to get done, but I can do more if I don't talk. So I'll shutup now and get back to work. -Rat

Last minute addition- Guru has finished his script for the results table. It's awesome! While we don't have the final version up quite yet (that DOA motherboard is the server's), feel free to take a nice look at the prototype!  You'll like it- I promise. Oh..see they pretty orange headers on the table? The ones that say CPU type, Speed, etc? Try clicking on them...  -Rat

Everything faster than everything else
After posting Dale Hollenbaugh's seriously swift time at the top of the tree others have come out of the rackwork and made the top end a bit more respectable and competitive. Up till now the real excitement has centred on gentler (!) paced systems improving a few % through setup and running tweaks, to gradually sneak up the listing. But some recent submissions have focussed firmly at the sharp end. So we have Angus and Risc 8500 sliding comfortably under the 3 with a 2:40 but to really show the way things are going try imagining something that runs 8 benchmark wu's side by side in 45 mins. About time an UtraSparc showed its grizzly hide. Theoretical output of around 230 wu's per 24hrs - bearing in mind the vagaries of completion times. If there are anymore wild children who fancy putting their kit to the sword be our guest in the spirit of Ars  Team Lamb Chop. Just remember if you don't have privileges/admin rights or pay the wages be careful. I seem to remember Aaron Blosser having a run in with his employer and the law a while back (arrested Sept '98) because of a distributed GIMPS program. It ended in tears all round. His tale is weblore now but it could be you next. Warning over. Also in yesterdays goodie bag was fsgray cementing his (?, let's not be sexist here) fastest NT4 PIII@500 position by another 2 mins to 6:48.

The Ars Seti forum has commenced yet another thread, so update and enjoy http://forum.arstechnica.com/forum/ubb/Forum12/HTML/000948.html. The forum now boasts a mail list so if you want 'thread to door' delivery sign up at http://mail.seaplace.org/mailman/listinfo/teamlambchop . The accent at the moment is very firmly on L2 cache size effects on a slew of processors plus of course some wishful thinking about future developments. If you want hardcore Seti that's the place to go. Need tips on implementation across various platforms and networks then the Seti thread archives contain a wealth of detail. But take your waders! Great to see the discussion moving fast and in depth after a slight lull, good stuff people.

In regards submissions: I don't want to post my guesses and though Rat knows more than he can remember there are times when empathic telepathy just doesn't fill in the blanks...so as a general rule if we can't work out the details we'll leave a ? and you can update us as necessary.

Thanks. Max out. 

The tweak and the thrash (aka tortoise and hare)
Remember Guy Olinger (March 25 update) who gave us his quest for a quick Seti speed predictor and found Sandra 2000 Fpu memory bandwidth as a candidate. Well he's been ferreting more normal routes recently by reducing his ram CAS settings, one of the recognised speed tips. His Cu 781Mhz (142 bus) brings in a slightly under par 5:45 at CAS 3. Remember there are Celerons in this neck of the woods! The system is unstable using CAS 2 so to get a bench he reduced the bus to 130 for reliability. So now he has a Cu 715Mhz (130 bus) with a much more respectable 5:10. Plus now there's scope for improvement if he finds some quality Cas 2 ram to run at 142, sub 5 easy and into the big boys arena.  Moral: less speed can be more...

Max out.

Physics defied, explained. Chemistry tomorrow? 
I came across this at The Register a few days ago and grinned knowingly at the 2nd paragraph. No mention is made of using the ars benchmark wu so conclusions were open to interpretation. Anyway here's some content...

Pentium defies law of physics :

07/04/2000 5:00pm by Andrew Thomas  http://www.theregister.co.uk/000407-000011.html

"A third machine has a Coppermine 700MHz, 256MB and a much faster, 20GB hard disk. It takes an average of between 10.5 and 11 hours to complete a SETI@Home work unit, with Windows 2K reporting 100 per cent CPU utilisation. The Lancewood takes between nine and ten hours to complete a SETI unit, Win2K indicating that only one processor is used to handle the task, using 95 per cent CPU time, the second processor merely ticking over at two to three per cent.

So we have a 700Mhz machine with twice the RAM running flat out, being soundly thrashed by just one 500 chip in the dual box ambling along at 95 per cent of full speed."

Didn't think too much of it, when has ambling been 95%? But a few days later an interesting update appeared  and here are a few snippets (note the word 'benchmark' in first para) http://www.theregister.co.uk/000411-000015.html

Guy Morrogh adds: "One processor runs all the system processes and handles all the interrupts, including handlers for the clock, disk, screen (drawing graphics etc). These have to be swapped into the cache, meaning that the benchmark gets pushed out and has to be reloaded into the cache. With two processors, only one processor gets hit when these run".
"SETI likes a bigger cache - you'll probably find it works better with a larger cache, even if it is slower. If it overflows a 256K cache, but not a 512K cache, it will have to make a lot of hits to main memory, which is going to take a lot of clock ticks with a 7x clock ratio."

"I can perhaps help you with the Lancewood mystery," Dean Johnson says. "I have a SGI 1200 with dual 700e's and 1Gb of memory and most of the time it is running SETI@Home, for which I have the source because I am on the porting team.
"Under Linux, the SETI@Home client uses about 13Mb of memory, so the overall memory isn't an important thing. The key, however, is the size of the cache.The data for a Seti work unit is about 350K, so the difference between smaller and larger caces is huge due to the poorly optimised FFT (fast fourier transform) that eats up most of the CPU time. For instance, A PII 350 is twice as fast as a Celeron 466.
So if you're manipulating large wodges of data, you're better off with an old 500MHz Katmai PIII than a shiny new 700+MHz Coppermine..."

Draw your own conclusions: if you intend buying something for seti, better to o/c the fsb on your present crate than fork out dosh for a faster shiny new proc. Well that's my poor mans thought anyway. Night all. Max out.

Just!
Did you spot the deliberate mistake on the results - well I didn't until they were up! Perhaps I should have let the nameless version stay to remove the ego element from the results but where's the fun in that? So accept my apologies for leaving out the one really important column ie your names and rest assured that I have ordered a tall glass of  Hemlock for breakfast. Will explain personnel changes soon but better get this up rather than gas.  Max out.

An Update!
Rat here, just thought I'd introduce Max to you all. Max has been active in the background of the page, sending in all sorts of cool info that he digs up. Max also volunteered his services to help out with the page, as I don't have enough time to do everything. I'm hoping to be able to spend more time than I have, but we'll see what happens. I think that with Max taking some of the load off, I'll be able to expand some, and do some things that I've been meaning to do, like update the speed tips page with all the latest info.  -Rat
Ps- I just noticed the lack of names in the table. Oops.  :)

Life!
My god there is life on this page after all...I'm sure there are some new words to ingest. RB is extremely busy with life in general and babies in particular so things might take a while to get rolling. He is also trying to explain the intricacies of web page management and construction by remote control because I was fool enough to offer a little time to help. there is still plenty of room for someone with serious knowledge to reply to his job advert of a few days back.

An updated table will appear soon. Which means when I have learnt enough html/XL/Fpx, not to mention patience, to do a job worthy of what has gone before. Or perhaps you number-hungry people would rather a disjointed mess but with your times on!! I have tried to reduce the lumpiness of the comments by pruning a little harder while maintaining their accuracy (!) and intent. Errors/corrections/useful criticism always welcome and on slow days all things considered.

There are numerous new entries (20+) and probably a shed full more to come...obviously some people have been either very busy benching for glory or very slothful at work (or both)- stylish. RB has commented on several of these new times already and in the words of the master I urge you to find the new ones for yourself. I am not suitably qualified to compare one blatting overstretched mobo against another at present. But there is one very iteresting item at the top off the tree - say no more! And to those who submit gentler paced times you work is appreciated. Someone, somewhere finds this useful.

On the list of requests comes a call to Schmo to give a little more detail about that Xeon time and setup - thanks. 
Max out.