• blah
  • blah

I'm slow
Okay, it took a week, but I've finally included the link for the Bios guide over at 3D Spotlight. I'm officially stupid. -Rat

Blurbage
New board...great work Rat. Speed is noticeably faster and comments have been very favourable, homage due. As for the money ($200) for a registration to allow full implementation of all its bells and whistles several people have pledged funds. Together with zAmboni's recent stats improvements I think this house is on a winner. Roelof has been claiming there will be an update to Setispy to 2.5.1, in a few days, no doubt it will appear and I'll let you know. zAmboni is investigating RAM chip size and effects on times and he'll keep things up to date as he digs deeper into that morass. To think it all started with an innocuous posting in the  alt.sci.seti newsgroup. I've recently updated myself of that resource and after churning through 500 odd emails (10 days or so backlog) became thoroughly depressed. If it wasn't for the several regular contributors I'd have given up, such is the dismal standard of many postings. Now I remember why I stopped looking at it in the first place. Sad and jaded - that's me.
Before launching into the benchmarks, in case any of you need reminding what real power means here's a casual hint from guru with one of the last postings on the old boards...."I just got my phase I server running. I added 220 clients to it. Have fun with the stat's now. 107 of them are dual processor machines." Onto some benching goodies...
Fastest K6-3 title goes to Ralph Morse who's managed a seriously speedy 8:09 with the NonIntel client. Nice to see another Celeron II time from fsgray but it is hard not to feel sorry for a chip running at 850MHz that only manages 7:24 (and with CAS2 RAM). At least you can chop 25 minutes off by using Win98 (see sofaking at 7:03) - not much consolation if Win2K is your OS of choice. Speaking of Win2000 fsgray also ran a bench on a PIII with said OS to return 6:50 which is 2 minutes inside NT4 run on the same system. Finally a reasonably speedy (for an Athlon) time from enabrein of 5:27, fastest Athlon so far but there's not much more to come from these cpu's that definitely don't pull their SETI weight.
Couple of minor points: Xibo those are very sexy numbers you submitted and I'd appreciate if you could get intimate with my email address to give me more info on the hardware. Talaktalan, I have your times and they are included in the results table (update will post soon) but I really need some input before trying to explain/present them!! My AMD shortcomings being exposed here. Right that's all for now, but there are plenty more times to come soon - I know I've just looked in the submission sinbin...the benching wheel of fortune rotates another cycle.
Max out.

Linkage
Guy was nice enough to send in some interesting links for your reading enjoyment. The first is a review of the new AMD Duron, this one running at 700Mhz. It has Seti benchmarks too! While it's not the same work unit as we're using, it does have a database of other systems to compare it to. There's also a review of the new Abit SE6 motherboard, which is based off of the Intel 815E chipset. Kyle 'The Man' Bennet over at the HardOCP gave this one a good once over, and used several interesting words, such as suckage. And lastly, a very handy and informative guide written to help make sense of your Bios settings is up over at 3D Spotlight. It's written in simple English, and it doesn't tend to rattle on and on like I do, which is a huge plus. 
Just on a quick aside here, it looks like I won't be able to setup a Seti forum myself after all. I'm going to be moving in a month and a half, and I found out the only options for connection in my new place are 1)cable- no dedicated IP, or 2)IDSL- slow and expensive. So I'll be going with cable instead of DSL. Damnit.  -Rat

Benchmark for the common man...(think PIII not ELP) plus more action.
A quick look at the 'Oh...my...god' headline (Friday, June 23) on Rizenet should give a chuckle and big pause for thought, the thought being what will Seti crunching times be with that memory bandwidth, gulp. If it hasn't been mentioned elsewhere there are two new Seti@Home newsletters dated June 19th. Number 3 deals with spikes, way back (when total numbers of users was still in 6 digits) they used to post updates but stopped because so many turned out to be interference or hoaxing but if you would like to know what they have discovered take a look. Number. 4 concerns the gaussians we've collectively been sending back by the truckload. No aliens so far but the background on the results analysis and interpretation is very interesting. If you don't like reading there are the obligatory rectangular graphics covered in brightly coloured dots - something for everyone.
Okay lets go straight in and talk benchmarks until I run out of submissions to offer and fingers to ram into my allergy stricken eyes, though yesterdays pool session with leaky goggles could have something to do with it. Chlorine, gotta love it, loathe it or swim in it! An initial offering comes from d2s who rather intriguingly is running a K6 233 (claimed) at 133 (young people nowadays - it's all drugs, loud music and underclocking, the planet is falling apart) and perhaps not surprisingly takes 37:35. So now we know how slow it runs how about really giving it the gun at rated speed! A big and predictable leap to Al Bartenstein coming straight in at 12:58 (sounds more and more like a top 40 chart show from the late 70's, pop-pickers) which is middle upper range for K6-2's in general and only 30 mins off the fastest one running a Linux variant. Next up is Jeremy Crawford who fortunately emailed me with a more sensible time than the 681 hours that first appeared in the submission sheet dungeons! His PIII ran a 5:49 on a modest bus of 117MHz. Respectable is the word, some faster OC'd 450's, some slower so there should be some tweaking leeway in there. Stay keen and enthusiastic and don't let the old-timers drag you down.
We've had multiple benches from several people over the months but I think Steve Common has reached a new level of dedication (read sycophancy) by submitting eight at one sitting. I will simply say thanks and hope you have got that lot out of your system for a while to allow me some respite. Seven of the times are running on a 644MHz PIII at a 143 FSB with a BE6 mobo and CAS2 RAM, so at least the hardware is solid while the clients and OS get swapped around. First up, running BeOS and 2.5 Cli produced 5:34 - the second fastest time for this rare OS. Next up comes DemoLinux with the 386 and 686 2.4 clients both yielding 5:02 (the precise timed difference was 1.43 seconds as if it mattered). This area, 4:20 to 5-ish, is where the seriously OC'd non Xeon PIIIs congregate and relative bus speed determines your placing almost exclusively. Coming next is the 686 client again on DemoLinux but going back to the v2.0 yielding a 4:59 to claim the first sub 5 for Linux, yay! Steve's remaining four times are all on Win98, 4:44 for the v2.0 client, 4:42 the 2.4 and 4:38 for the nonIntel 2.4. Finally after deciding that the nonIntel was quickest (think I could have told you that without wading through the benchmarks) he squeezed the bus to 146MHz and still holding onto CAS2 produced a time of 4:24. Excellent time but unfortunately the same as his previous best using a 160 FSB at CAS3. Even more impressed that someone is using Tweakbios which is a little fiddly but allows many untapped bios settings to be altered. Also the documentation gives explanation to a lot of unusual settings that might help anyone using AGPinfo (from IXBTlabs). Usual warning about locking up applies - it can and will! Suitable (?) for lots of chipsets with the last revision I noted being 1.53 way back in March 1999 so it's either stable and mature or old and ignored...This paragraph of benches is due entirely to Steve Common, thanks.
Even after those efforts there are two more times of real interest (i.e. speed freak worthy).  We have our first 1Gig PIII (hooray, if Intel and AMD can make a fuss who am I to buck the big Chipsters) from Marzio Nieddu who has pushed his PIII 700 (155 bus and CAS2) up to 1085Mhz to achieve 4:23, excellent stuff many thanks for that. A couple of things to note here are 1) how clock speed becomes irrelevant once you get to high FSB's, 2) how PIII's are topping out in this region (4:20 or so) and a Gig clock speed confirms this. A sub 4:15 PIII looks chancy now, let alone nearing the sub 4. I'd be interested to see Marzio's benchmark using a quicker OS such as W98 to see if there's much difference (any chance?). On slower clocked systems Win2K can be half an hour in arrears of Win98. Finally over to our recent smug bastard award winner ('my toy is faster than your toy') Action who has yet again defied the Gods. Swapping from the 2.0 to the 2.4 client has chipped another 2 minutes off to stand at 2:09.  Also realise this is for 2 benchmark WU's crunched with that dual Xeon set-up, very impressive indeed. How about the nonIntel client next? I expect to be potty trained for your next submission.
On an end note the current Ars Seti thread was finally curtailed at page 51. Here's a link to the new  thread, Team Lamb Chop (SETI@Home) - Millenium. I personally like lots of pages but it seems I'm in a parasitic minority of about one. I'll go quietly.
Max out.

Lights, Action, Music...
On the utility front Setihide was reported to be revised up to 1.3 (beta) on the 16th but I haven't checked so you'll have to take pot luck. Hot on the heels of Setihide is Setidriver which promises many things especially for those of you running multi-cpu systems who need a little more thought in stacking and working with those cached WU's. Bet that got your attention. I haven't tried it (big cop out) but the SetiDriverInformation post from author (Michael Ober) certainly suggests a great deal of pleasantness. Thanks to that nice ColinT for alerting me to that. Hopefully he will keep us informed (yet again) of it's pro's and cons.

Several times to report (which covers the fact that I've already exhausted the info type stuff). So straight in with Mournblade who is still persevering with his system and has made it to the lofty heights of 13:41. I know he's chuffed because he told me so and it's all down to cranking that bus just a little to 105. This originally started out as a plain P233MMX on a socket 7 board and over the weeks he's reduced his time, from 14:32, by close to an hour. Though many have cut off bigger chunks it's good to see someone in the 'middle ground' finally get a result. A new name to me is RatPhish whose OC'd 552 Celeron stumps up a solid 7:56 with NT4. Starting out at as a 400MHz (6 multiplier at 66 MHZ) your OCing is at the point where most Celeries hit their cpu clock limit at 550MHz. Unless seriously cooled or you have a really good chip (or both) any higher bus speed is unlikely to be up and running for long, but you know this already don't you. Finally in the mere mortal time range system's system is in at 7:30. You are pretty much at the FSB limit with 112MHz as PIIs do not take kindly to big OC attempts. But if as you say, you really haven't used any tweaks then there is a long way to go with bios and CAS settings. Sub seven heaven coming up soon. Onwards into the gladiatorial arena that is sub 4.
Steve Green has been hard at work on providing several benchmarks on a Compaq 6400R (2 cpu's installed of 4 possible). To say that the results are close is an understatement, only 3 minutes separates slowest to fastest. Running one client on Win2000 Advanced Server and then again with NT4 Server produced 3:49 for both. A little spooky as I would have expected a few minutes in favour of NT at least. Then running two instances of the client produced 3:51 and 3:52 in favour of NT4. Finally running the client verbose (i.e. it talks a lot - it's been that kind of week, apologies) but only with Win2000 server produced a 3:50. There is a gaggle of 1MB cache PIII Xeon (all at 550MHZ) times just hovering above the 4 hour mark so this Compaq setup does appear to have the edge by about 15 minutes. Thanks for the numbers. Up a notch in the time stakes and Angus makes it to 2:29 with his HPPA 8500 RISC quad processor system which if you are interested is number 6 on the speed list. This improvement of 12 minutes is due to only running a single client whereas the previous 2:41 was also chewing away on other 3 random WU's and covering it's regular workload (!). But competition here in the 2 hour slot is really hotting up and who better to apply some heat than Action who has pushed the bus to 112MHz on that new PIII Xeon. 'Goddammit just tell us the time', okay, okay let me put some fresh diapers on and let you know it's 2:11. A 15 minute improvement in time (from 2:26) is 10% which relates nicely to the 12% increase in bus speed - not that the working set spends much time on it as it all sits in the 1MB cache anyway. Remember we are talking expensive but not exclusive hardware here, you can buy this sucker and mobo for a months salary (better be in a good job, though). That sub 2 looks tantalisingly close. Stay tuned.
Finally, many, many people have been expressing their delight on the thread that zAmboni's stats now go through to 2000th position. This is great news as not knowing your position relative to other LambChoppers was a major downer to many outside the golden glow of the top 200. Appreciation and congrats from this end of the world also.
Max out.

Quick Update
Angus let me know about a small time mix-up and that has been amended in the June 9 news. Also Joey pointed out that his 4:45 with PIII and Rambus was more relevant for its relatively low FSB than for having for a large clock (913MHz). His bus of 114 is way below that of comparable times on PIII's where 133 and up are the norm. So perhaps Rambus is not so lousy after all, merely under-optimised, over-hyped and overpriced. I know things will change when Intel either sorts out its implementation problems (as the 815 mobo purports) or quietly and expensively drops it (not very likely). On the subject of memory there is a long review of the history of PC memory at firing squad. At 15 pages it's fairly long, but if you find this subject as confusing as I do then this might help bring you up to speed in an intelligible way. If you want to skip ahead (5 pages) and miss the cutesy intro stuff then start at main system memory . One time to mention is gkoum with a 6:19 time on a PIII running Redhat which is relatively slow but due mainly to that low bus of 105. Usual tips apply - bios tweaks, then FSB and CAS2 assuming you are headed that way anyway. Look forward to your progress.
Max out.

Quick Update II
Thanks for all the well wishes from everyone regarding my current illness - I'm starting to feel better now, except for a sore throat from Hell. Also, for those who sent in employment info, thank you very much! I'll be ending my 2 week vacation tomorrow, as I start at American Express as a Help Desk Analyst. Not precisely what I wanted, but it'll pay the bills. As an added perk, I should have a decent view of the Minneapolis skyline too, since I'll be in it. Of course, parking is going to suck hardcore, but that's life for you. Perhaps I'll get lucky and find a spot where I can park for less than $13 dollars a day, that would be a nice bonus. Umm...that's all, I'll stop rambling. I have some things to work on that haven't gotten done due to my illness and I have enough energy to get them done now. So I better get busy. -Rat

Submissions from the dawn of time (roughly 2 weeks ago)... part 1.
A whole shedful of times coming up very shortly after mentioning that Setihide has progressed to beta 1.2, which as mournblade pointed out would make more sense as 0.2 and ColinT who is giving Setihide a good testing suggested to the creator (Mel Brooks and Young Frankenstein come to mind, an errant thought) that a logging feature would be good. Personally I'd rather Roelof who's been quiet lately, adding wu caching to Setispy! But then I want everything. Guru has been in front of the law recently in a number of ways (see thread -'they just do') and my sympathy goes out to a fellow biker, mine's a YZF1000, so yes another person who overclocks cpu's loves to ride around, square off footpegs and waste rubber. Symbiosis on two wheels.

On with the times and first up is Urugami with a rather lengthy 80:46 on a K5 at 133MHz which makes any Cyrix look good! I would imagine you were hoping for something slower to be able to take over the bottom spot though if squeamish or RAOF (see June 5) ran a benchmark you'd have no chance, if you two are out there let me know what your using. Talaktalan has been very busy (again) with old Intel chipsets (where old means anything pre 'son of Rat'). A 120Mhz Pentium on an FX and FPM DRAM produced a 34:31 and the 166 on a VX with SDRAM 20:59, just goes to show that old and slow  will still push out units if you have no better use for them. Another pair of WU's ran on a 366 Celeron, the 2.4 686 linux client giving 12:31 and the 386 an 11:20. After a previous comparison (18:52/20:29) using the linux 386/686 clients I think talaktalan has given in to the obvious conclusion that the 686 is just slower. Finally back with the linux 386 client he upped the Celery bus from 66 to 75 but only improved his time 13 minutes to 11:07. Which is a little odd and is probably something to do with the chipset (Intel EX) performance limits. The only direct Linux comparison I can find is Steve Common's old time of 9:19 for a 400 Celeron though using a 686 client to confuse things. Feel free to let me know if I'm missing something deeper.
Still in the slow but sure lane a rare appearance of Win95 coughed up a 25:32 for wehringer and as he comments it's no match for SuSE running on the same setup (18:41). After a long absence Redbeard has reappeared with a salvaged Pentium saved from dusty oblivion - a small overclock (naturally) to end up with 17:43 and he seems pleased. Okay that will do for times on the geological scale, lets get into the middle ground.
In with a whimper is Mirage with a 10:51 from a PenPro sporting a ramdrive.  Don't let the ramdrive fool you this very impressive time is due to the 1MB cache L2. Though PenPro's at 166/180/200 were ubiquitous workhorses only a couple of years ago, I have the vague feeling that the 1MB cache is rare. Unfortunately not very OC-able either. Steve Mac G submitted a K6-3 time of 10:34 which has got to be a slight disappointment, running W98 and the 2.4 client you should be looking to knock about an hour off that.
I find great comfort in the slower times but right now I need single digits. A second time for Urugami using a 'new' (why, do they wear out...) 500 Athlon giving 7:56 is okay but could do better (school reports haunting me there). Don't hold out too much hope for overclocking as a scan of the table only shows a couple of Athlons over 110 FSB, find a cool, dry cellar, big fans, rip the case off (heresy) and go from there. Another fast Athlon (600MHZ) but slow time for shirty at 7:54. There is some slack to picked up on, but you'll struggle to compete even with 500 Athlons (see Urugami) while using NT4. At present this is the only Athlon in the table not using Win98, could this be related to a lack of gaming pretensions dare I wonder, not that any of you would forego a couple of hours crunching alien signals as opposed to crunching aliens. Lastly in the Athlon benefit arena we have Joe Bucker whose 600 Athlon has produced a very respectable 6:32 - bit more like it. Next up is Angus, a Dell laptop and a battery bursting 7:37, don't leave home without your mains adaptor. It's a bit behind Actions' Inspiron7500 at 6:56 but is a reminder not to leave valuable Seti resources untouched. Suggest you go get a Coffee/Benzedrine/favourite Shari Lewis poster/whatever as there's more to come...
Part 2
Slight discourse: Ram latency reduction is a very worthwhile improvement but L2 cache latency is not something for most of us to be worried about, (unless of course you are a Rambus owner) as you might get a % or two if you're lucky. Speeding up movement of Seti data on the bus is going to hugely outweigh any L2 latency reduction for almost all cpu's. At 1MB cache, the Seti working set fits in the cache and L2 latency theoretically becomes important. We are now in Xeon territory where the cache is heavily optimised and L2 latency reductions are going to produce instability rather than speed. One school of thought is that increasing L2 latency can improve stability and allow a greater overclock, this of course is only true if it is the L2 that is the major factor in limiting your overclocking. One way to find out is by disabling L2 in bios, does it boot, yes, then probably L2 cannot handle the CPU speed. The Celeron is a good example where a higher bus is possible if L2 disabled but of course without the 128kb cache, performance drops dramatically. Increasing L2 latency might allow you to POST or even partly boot at a higher FSB but to actually achieve OS stability and then stress it with Seti (or RC5 or Prime95) is a whole different game. Your max. FSB is most likely going to be the highest FSB you can run reliably at default L2 latency, 15 minutes and blue screens and/or locks do not count as stable! It would be lovely to report that latency reductions were going to help but overall it's a bit of a sideshow. So after all that thanks to fsgray who dropped his L2 and recouped 79 seconds to go to 6:47, was it worth it. A  Speed Demonz article (8 pages by Adrian Wong, if the link holds up) from way back (Dec '99) attempted to show what changed L2 settings could do, the outcome being 'very little and not a lot'. The article has questionable methodology but at least it's something. If you still feel like a little tweaking then WCPUL2 is at H.Oda.
A Celeron II time now from Steve A with the words left to him. "I got my first Celemine 533 the other day and set out to SETI it right off. I obtained a FSB of 110MHz for 880 on chip with 1.75V. This produced a time of 6:36 on the same platform that yields 4:18 with a 550E at 825. Yawn. My 533a failed at 112FSB (896 on chip), even at 1.8V...Since we can't raise the FSB on the Celemine over 110-115MHz, even good 300a times with a 124FSB are better. I have speculated that this is not absolute (wu's with guassians profit from the extra on chip clock), but even then it's within minutes of the truth." Because of that unpleasantly large multiplier (8) on even this the lowest clocked of the new Celeron II's the opportunities to OC are very limited. 880MHz sounds great but for Seti remember, as always, the bus comes first and the clock speed follows way behind in importance.
Bylee's 400MHz PII Xeon managed two WU's in 5:39 with W2K, that 1MB cache working rather well. A rather delicious time but now get rid of the crud you say you have running in the background and get a sub 5. Continuing into the mid 5's is nethead with 5:25 on a PIII/133/800. A little more bus might see you to a sub 5, then give us that BeOS (see his/her comments) time, think I smell rubber burning. Another Xeon (PIII 400MHz, NT4) time now from Hance_wu with 4:48 which is just about where it should be. Next is a dual PII Xeon from Angus at 4:46 also running NT4 tied with a much faster dual PIII (133/800MHz) from Feisal also at 4:46 which just happens to be the fastest Win2K box so far. These high 4's seem to be the place where the really tight competition hangs out. Anything to gain a minute even perhaps dodgy L2 latency settings! A short shuffle along is Joey at 4:45 who has the honour of submitting the fastest cpu to date (as if it mattered) at 913MHz (PIII with PC800) and another confirmation of how far Rambus RDRAM has to go before it's worth shelling out the big money for such poor memory. It's a useful improvement on Charitys 5:12 RDRAM equipped box but is still way short of exciting. A staggering 166 bus for Tomslik on his PIII is yummy but running the RAM at CAS3 naturally keeps the time to a sedate 4:43. There are several of you trying to get really high FSB's (150+) at CAS2 to run but I'm beginning to wonder whether it will really happen. Is this another barrier? A big leap now to a real goody. A monster time from Actions new baby, a 700MHz PIII Xeon, coming in first time at 2:26 for 2 bench units! Think I just wet myself.
I'm glad to read in the thread (zAmboni June 5, 22:37 ) that I'm not the only person who mailed the Register and didn't get a reply about Seti matters. Since I wrote seriously and on topic I consider that rather unprofessional. Shame. There have been a bargeload of Duron/Thunderbird/Timna offerings the last few days and several interesting links from posts. It seemed better to get this out and done rather than chase after stuff that will have lots of linkable reviews in the next few weeks. I'm just three people after all. Finally as this is a big update the usual proviso about errors/corrections holds, you tell me and I'll sort it. 
Max out.

To hide or not to hide that is the software question...
A few bug fixes for Setihide and it becomes version 1.1, still considered  beta though. Folks on the thread seem enthusiastic which means that someone else is trying it out and reporting back, good. ColinT has some useful impressions and a thoughtful review on page 40 (June 05, 2000 16:49). Remember how in the mists of time you ditched the GUI for the CLI and received a pleasant time reduction, so why go back a step? Who needed visuals anyway? But If you hanker for something on your screen that shows more than numbers then this is a good compromise. With the WU caching working properly it could be very tempting to use. The peaks and troughs of the main window's graph show many of us what we already knew that the client work rate varies considerably depending on the FFT set size, the frequency resolution and gaussian considerations (where present). Yes, you can assign an overall efficiency to a processor but at any particular point it could be well above or below this. So it visually answers some peoples questions about instantaneous processing rates that Setispy shows up. Personally, I still have the fundamental reservation that it is taking too much cpu time, as you might expect with anything that provides some mild graphics output. Which means you times will increase and production down - nothing like stating the blindingly obvious. I guess that means I'm in the speed freak camp in spite of the odd sympathies I have for tortoises. If you can come up with some real benefits of Setihide let me know and I'll mention them. I don't want to appear to negative about someone else's hard work. As before try it and decide.
Guy Olinger pointed me to an article at tweakmax about memory timings so if you like your comparisons simple, clear and with Sandra FPU/CPU numbers over a range of different FSB settings then CAS2 vs CAS3 (3 pages) is for you.
First there was the competition then the results and now the Rambus rumble which is not as exciting as I've made it out to be! Considering our views on the sad implementation of Rambus on Intel  820 & 840 chipsets and it's disappointing Seti crunching times (the oft quoted Charity and 5:12 with PC800 springs to mind) the Register is still looking for a champion of sorts. But it is good that a more mainstream site occasionally covers Seti stuff even if the pace is a little on the slow side.
Those of you who have submitted benchmarks or pointed out minor errors (thanks - honest) are being patient and sometime soon I will get down to putting them in the results table and making unreasoned and vaguely caustic comments about your efforts, that's why I get so much fan mail. At present the slight hitch is that they are all located on Rat Bastard's old server (mediaone.net) and though not lost are presently unavailable. He is trying to sort this out along with looking after a baby, finding a job, flying radio controlled planes, writing here, setting up a forum and sleeping - really not busy at all.
Finally, since helping (?) out here I have subscribed to, signed up and chased down so many links and mailings that my inbox is a little crowded most days, very ego boosting. Shame that so little is worth passing on. A while ago I came across some freeware called mailman that allows me to see what's there and delete the dross before downloading from my mail server. It's on beta v5 but seems stable and very clean. As mentioned previously I'm on a modem connection and anything that cuts out the crap is useful. Also on a non-Seti note if anyone has a serious link to the 'Eureka, scientists break speed of light' mailing I just got let me know. Update - Ars front page has provided one to the NY Times (not a scientific organ the last time I read it), thanks people. Here's a quote:
"The work was carried out by Dr Lijun Wang, of the NEC research institute in Princeton, who transmitted a pulse of light towards a chamber filled with specially treated caesium gas. Before the pulse had fully entered the chamber it had gone right through it and travelled a further 60ft across the laboratory." (Feel free to giggle). The team claim 300x c which sounds like a windup but you just never know, but then I am writing this from the far future...  I'm sure a Danish scientist was mentioned on the BBC science site a while back about using lithium ions to slow c to walking pace, the punchline being that she was able to go get a cup of coffee and come back before the beam crossed the room. Enough already.
Max out.

In praise of the Tortoise...
News and hot topics in the Seti world seem to come in bursts and right now things are a little on the cool side. So I think I will go for the soft option and provide a few links, thoughts and minor ramblings. The thread as always is full of interest and thanks to ColinT for giving me an Athlon overclocking link that suggests going lower on the multiplier and thereby being able to up the bus. Note, the article has faq links about using 'gold fingers' to change the multiplier or try this little explanation  and the benchmarks are game oriented but the idea is sound.  We know higher fsb's are better for Seti but not everyone can change the multiplier. ColinT took this to heart and tried it out going from 8.5x104 (884MHz)to 8x111 (888MHz). Sorry no results yet, I will be as interested as you when they appear. Anything that can give the hamstrung Athlons a bit more help will be good.
ColinT didn't end there and reported on a new utility for Seti monitoring called Setihide. Two programs are in the 305kb zip, one for managing/displaying Seti info and one for hiding the client.  It is definitely beta and a shade flaky but always good to see new ideas. I'm sure Roelof will have a few comments as it is stepping into his territory but as yet nowhere near Setispy in quality or info. It's main visual attribute is a graph of wu processing % per hour, interesting but not much more illuminating than the GUI display. Unfortunately even minimised it is consuming about 5-6% of cpu time (Setispy takes around 0.15% when monitoring) so has quite an impact on Seti crunching time. Claims to be able to cache 50 units but I haven't tried so unable to comment. Download it, have a look and then decide if it's any good for you. I'm sure that it will improve in efficiency and presentation with updates but at present is not recommended.
Also from the thread was Mr Petey's observation in praise of team members with high average times citing the 53 wu production from Xyberknight averaging 117hrs. Extremely impressive, I always maintain that the slow stuff merits more attention. There are several members averaging over 100hrs with special mention going to Immortal1 who's produced 10 at 130hrs each to be at 1713th place. Spare a thought for squeamish (1886th place)  who's 4 units average 167 hours. Finally, to do one unit slowly is impressive but then to knowingly do another is real class so I think RAOF (1990th) averaging 242 hours for 2 units must take pride of place. A special breed of Seti cruncher indeed.
In the quest for faster times keeping your system cool is paramount. Before persuing peltiers/fridges/water cooling/liquid nitrogen a simpler first step is a new case. More fans equals better (and noisier) so a box with 3 extracting and one blowing in from the bottom as standard has got to be worth considering...but I was more interested in the testing, stuff the box into a green bin liner for 30mins and see what happens - answer, it gets hot. Note: If you are on a modem (me for instance) the pictures take a while to download.
If you can't get enough comment on Rambus then check out the Registers SiSoft Sandra numbers for a 933MHz PIII on an Intel i840 equipped (dual channel memory chipset compared to the i820) board. Yes the mem bandwidth numbers are heading in the right direction but the comparison with an 800 MHz PIII BX chipset with a 100FSB is a little misleading considering we know that the BX is good for far more. Rambus is not for Seti yet, spend your money more efficiently elsewhere.
Finally here's a reason for those of you who don't post to the TeamLambChop thread to continue not to, courtesy of Mr Petey...you have been warned.
"Hello, my name is Mr Petey, and I am a Arsaholic. I have been reading Ars for a year and a half now. Then about nine months ago I started hanging around in the forum reading the odd thread here and there, what harm could it do? Within days I was posting. Soon, all I could think of was posting. 
Then my girlfriend deserted me. "Good riddance to the Cold-Hearted, Blood-Sucking Bitch anyway!" Said I, for she would never understand the Joy of Posting. Who needed her cold comfort anyway when I could Post?
With nothing to hold me back I posted like a man possessed. Eventually my fingers screamed stop, but I posted any way. I posted like no man has ever posted before...I was the GOD of the newsgroups. Nerds and web surfers from across the globe came to bask in the wisdom of my posts. Then, my fingers gave up. Forced to walk among the Netless I knew I had truly reached my darkest hour. As if from nowhere, a ray of light appeared. Voice recognition software! Joy of joys! I could post again! ...Here I sit in a broken down shell of a house. Guarding the last functional telephone line in town with my twelve gauge and ring of proximity mines, with the only the glow of the monitor for heat and light. Alone in the knowledge that I am responsible for the social and economic collapse of middle England. All I know is that I WILL Post. Maybe not while installing the server, maybe not till after lunch, but soon and for the rest of the afternoon...."
I know I'm into computers too much when mention of the Rascasse corner at the the Monaco Grand Prix only makes me think of memory timings - sad.
Max out.

Seti and the search for intelligent conversation
Well, I've received a lot of mail concerning the possibility of our own dedicated forum for Seti discussions. And with the exception of one person, it was agreed that the dedicated forum would be a good idea. Of course, I'll have to post the question in the current forum too, I'd like to get everyone's opinion.  However, I think we'll probably end up with a dedicated forum. So I went ahead and tripled the CPU capacity of my server today, and quadrupled the memory. So now it's no longer a POS, and should be able to hand the UBB just fine. So, where is it? Umm..hold on there, I'm not done yet. Okay, I'm not close to done yet. I'm still waiting for some parts to arrive so I can get my SCSI drives online, and then I have to configure everything and stuff. In short, it'll be awhile. But I'm working on it  :)  And with that, I'm going to wrap this up and get back to work.  -Rat

The hunt for meaning
I finally went ahead and did the one thing that I've been meaning to do for a while, I quit my previous employer. I think it was a good move overall, as I had no potential for advancement with the company. After all, they want techs to have 4 year degrees- and make $15 dollars an hour. I don't think so. So I'm currently looking elsewhere to see what I can find. It's tough because there isn't a huge market for hardware support people, and I'm hampered by the fact that I don't have a degree. So I'll find whatever I can, and I think I'll be spending a few years at a local University too. About 4 of them. I had too much spare time anyways  :)
If anyone in the Minneapolis MN area needs a fool to come in and overclock all of their workstations (and put hidden Seti installs on!  :P ), email me! I doubt I'll hear anything, but it never hurts. After all, the last time I mentioned something like this, I was looking for a bed. A gentleman with Intel forwarded my request to a cousin who sells mattresses wholesale. It's amazing how things work sometimes. 

Continuing on my thread of ranting, I'd like to congratulate Geordie, who scored an excellent deal on a Quad Xeon board this weekend. Sounds like he'll populate it with 400 MHz Xeons, meaning he can crank out 20 odd units per day from that one machine. Nice!  I'd also like to welcome Knight to the top 200, 'bout time he got here, the lazy bastard. Knight is a good friend of mine, we go out and crash R/C planes together when we have the time. It's a blast. 

On to some Seti related content. I'm currently rebuild my webserver, and I've been toying with some ideas. I've noticed that the thread over at Ars is a little disjointed due to several different discussions being carried out at once. And it's hard to find info sometimes, because it's somewhere between page 1 and page 38. If we started separate threads for each item, they'd quickly become lost in the mass threading that exists in the Ars forums. Therefore, I've been kicking around the idea of hosting a UBB service just like the OpenForum at Ars Technica. However, this one would be dedicated solely to us.  We can ramble on about the project, hardware considerations, and stats in separate threads. I'm curious what everyone thinks about this. Should I? I'd really love to get some feedback on this one, so feel free to flood my mailbox with your opinion. Be short and state yes or no, or spend time telling me why you think that I should or should not setup another forum. I honestly want to hear what you have to say. Thanks! -Rat