• blah
  • blah

Airborne Duron Gauntlet attempts to defy gravity while P4 whimpers quietly in corner!
New SETI Driver out, v1.6.2 - do I need to say more. No, but I will anyway. Author Mike Ober has a very understated sense of humour. The Driver site has undergone serious cosmetic surgery but the only mention in the 'changes history' is a spelling correction!
It is getting harder to make interesting copy out of benchmarks these days. The surprises have just about been weeded out and most systems/processors fall into line with just a few basic rules. The most important with v3 being clock speed, even the quirky and intriguing OS variations have faded into the background, L2 cache size break point is far lower and the SETI client is heavily Intel optimised (or rather it runs more efficiently on their kit, don't want the conspiracy theorists warmed up). So it was rather fun to read an early bench on a P4 in the forum from Krell: "I finally got my hands on a spanking new Dell Dimension 8100 with a 1.5GHz Pentium IV and 256MB of dual channel PC800 RDRAM @ 400MHz. Just started the benchmark unit on the 3.0 CLI under Windows 2000..." Eventual outcome being 4:26! Now the technically astute (awake in other words) of you might be thinking that with FSB/clock being king then we should have been seeing sub two hours. Vaguely knowledgeable authority figure Roelof came up with a figure of 2:15 in the P4 SETI benchmarks thread. I was just wading into technical notes and P4 comparisons to come up with a vaguely plausible rationale for the 4:26 time when a submission from GMU appeared. His 2:30 seemed to be pleasantly at odds with Krells bench and it spares me trundling out some ignorant fumblings of a techie nature. I will merely state that the P4 could still go either way. But as with so many things, until the SETI coding is optimised for a 64-bit machine (v4 very vague possibility, v5 more likely if ever - think about P4 uptake rate) the architectural advantages of the P4 are all but wasted and the Duron (more later) is still the 'accountants choice' for SETI crunchage. 
On the Duron front a no-holds barred outbreak of overclockage followed by mellowed disbelief have produced some fast times and the chase for fastest TLC Duron is maturing nicely. Falloutboy threw down the mailed fist of competition and several hormonally challenged benchmeisters (well ColinT, subzero and Elgar) lowered themselves to smell the glove. Previous leader was MuelTe at 3:26 which was pushed gently to one side by officeboy's 3:24 which in turn has been ravaged by ColinT's 3:12 (not in table yet). Heatsinks at dawn gentlemen?
Being a determined and obstinate type Betsy managed a 28:10 from an Athlon 600 using the GUI (screensaver enabled), why (?), god knows, but such are folk. A little more respectably jmrojo squeezed a 22:09 from a P200MMX (on NT4) and then a big jump to Stefan at 13:20 with K6-2 (405MHz). While in the area I better mention fsgray (12:35, P250MMX - very respectable overclock), Sith lord's 10:14 (Linux, K6-3 450) and golden_go at 9:45 (PII 266 ramdrive, misguided but laudable). Now a quick dive into Celeron heaven: Mex at 7:17 with a Celery II at 713MHz needing some serious help! Business as usual for Soulegal who improved from 6:31 to 6:21 with a 533 up to 600 MHz boost using RedHat 6.2. Changing to W98, the 533/600 Celeron comparison was 6:23 to 5:39. A bit more like it! JimW managed 6:08 on a 500MHz Celery and tomwyatt posted two completions in 6:01 with a dual Win2000 server and 550 Celerons. Very impressive and the 110 FSB against the previous 83 is the reason why two came out as fast as one. Also at 83MHz FSB for 458 is Tazz slumbering in the doldrums at 6:25. Yet more Celerage with Thunderbox, 6:00 and Smokva, a pleasing 5:45 (high bus & low multiplier to the fore). Our first PIII drops in at 5:32 from baygents followed by our first non-ripped Duron at 5:03 from Moller, who is another ramdisk aficionado, poor deluded fool. Not sure why JokerG is at 5:01 on a 900MHz Thunderbird, at least an hour to skim off there. PIII, 650, W2k and 4:54 for jhoefer is also slightly on the high side. High threes beckon for Idone if he'd ditch the GUI for CLI that's in at 4:19 presently on a 812 PIII. I like happy people and KWSN Sir Loin fits the bill nicely also at 4:19 for a 650MHz PIII. JimN has produced a very quick Athlon (720/120) time of 4:10 but just heading him is Sir Rich's Athlon (850MHz) at 4:03. Last time in the 4's is William M whose 4:01 on an unclocked Thunderbird 850 holds hope of fairly impressive things to come.
Into the three's goes manny (3:58) and some minor exotica from kforce in the way of an Alpha EV6 which for a 466MHz clock kicks out a solid 3:45. A minute better (3:44) is Kont3's  PII 770 (140FSB) and Tazz again with a PII at 720 (144FSB) for a 3:32. The Crow didn't know that feathers were flying for the fastest Duron title, which is good as he was losing from the start with 3:31. A superb time but still shy of the professional ranks though not many can claim to have extracted 146MHz from PC100 RAM, tee hee. Still in the 3's with mmtt (I should have been a poet) at 3:25 and Joey at 3:18 (PIII 912/112, very tasty for such a low FSB) brings me nicely to Andbad who has the dubious distinction of benching the first gigahertz Celeron! Ticking along at 1062MHz it managed to finish in a glorious 3:18. In at 3:15 is Chboss (1050MHz Tbird) chased down by MattC's remarkable Athlon (1008MHz) which is going to cause a few raised eyebrows and envious queries with 3:10. PIII time of 3:08 for peroni using a 140 bus is indicative of the way things are going and we are still the wrong side of three hours. Spare a thought for the humbled PIII Xeons of Action that are slapping them down at 3:08 a pair from this dual 700MHz 1MB L2 slab of greyness. Excellent performance...but not for the money! When you call yourself jobbigvolvo you deserve a good time and I guess 3:07 on a PIII at 938 (125) just about qualifies. One minute to the good via a big bus of 155MHz is Amy's PIII (980) newly in at 3:06 and last of tonight's plus 3 times. So sneaking under the wire for 2:59 is Schnider with Gary D in close attendance at 2:58. If neck-wringing was a spectator sport I guess I'd park my chair in Roelof's shed and listen to that Thunderbird squawk! Firstly at 1177MHz from a 1Gig Tbird, remember these things are being pushed pretty hard at their rated speed let alone above,  extracted a 2:57. But not letting the corpse rest he gave it another twist up to 1189MHz to post 2:52. Delightfully edging Marzio's 2:52 out by 11 seconds. It's a hard world boys. But of course to spoil the party Jay gutted a PIII (1001/154) and leapfrogged the chasing pack to record 2:48. But to round off a very long and somewhat dry overview of the TLC numbers game comes a sweet poke in the eye from Roger W. Alpha 21264 running Tru64 Unix and a single client on this multi-proc beastie bansheed a 2:27 to be top of the sandpile for the present. It's going to take a big wodge of money to get past that one. I remain calm having learnt some semblance of self-control.
Max out

Double whammy at a site near you...
Minor (to some) news  SETI Driver  now at 1.6.1.1 - yet more minor improvements, under the shell fixes and tweaks to this excellent caching software. Now with W95 specific support. Also SETI Spy goes to 3.0.2, smooth utility that just keeps on getting better. Among the fixes is proper checkbox support allowing closure of both Driver and Spy when exiting Spy, a very pleasing enhancement. More results up soon!
Housekeeping: If you are out there, playa, could you get in touch, thanks.

Benchmark mayhem - PIII/Thunderbird slugfest...the plot begins to boil.
First off an update to SETI Driver today, which scales the giddy heights of 1.6.0.0. As you might expect there are fixes and a few enhancements. Main addition is the processor affinity enabling which will be of value to those TLC'ers who were Mikes beta crew for this specific add-on. A TLC critical update I'd say. 
Right, now for an attempt to make the words match the results table...I'll still be way behind after this update but making headway. Looking at the submissions sheet there are at least another 25 that are not included here. If you haven't already checked the v3 results table there are many additions from my last round up of likely suspects. In at 12:40 is Steve Common, a man more used to low single digit crunching. But that's what you get for K6-2's at 500MHz these days with CAS3 RAM. Slotting nicely ahead of him is Chboss and K6-3 at 420MHz for 10:09, nearly next generation cpu 2 hours to the good! But ultimately both are hampered by having their L2 (512kb) cache on the board - it's never going to compete with full speed but smaller L2 on die. Be good to have some more AMD K times in the near future. Haps pops up with a 9:14 Win2K dual PII, 2 every nine hours is a goodly output. "Really, really stock"...(enter system name) has almost become tantamount to an excuse and Jim W is no stranger to these "hey it's slow but what can I do?" tactics. We understand, we really, really do. But your 8:55 is nothing to be ashamed of on a 300 PII, with 66 bus, half speed cache and all, I'd be fairly happy. Returning to the fray is Chboss (several more times to come) whose Celery is suffering the dreaded Apollo pro chipset fate and tortoising to an 8:29. Crazy to think that his lower clocked 450 Celeron on the venerable Intel LX chipset takes nearly 2 hours less, posting 6:41. A useful 7:33 from fsgray and OmniBook XE2 (PII-366) which rather pleasantly, exactly mirrors Roelofs 7:33 on a Dell Inspiron with the same processor. Ignacio is two minutes the wrong side of seven hours with PII at 434 using Free BSD and respective client for 7:02. But it seems a slow OS as Jeffrey Ottie's Dell Dimension benchmarks are all an hour faster (6:03) for nominally similar hardware. Whether with ramdisk/TweakBios or without made only one minutes difference to his times. In fact the minute improvement to 6:02 was after changing to Win2000 (from 98) and dispensing with ramdisk and TweakBios (only Win 9x compatible). Ramdisks have always caught the tweakers imagination but never lived up to the hype and the benchmarks just seem to reinforce how needless they are. Laptops wanting to save power (by letting the HD spin down) seem the only sensible application of a ramdisk, unless you can come up with something really exotic. I know there are a few diehard proponents out there but so is the truth! If Scully used a ramdisk, The Lone Gunmen would soon shoot her down. Clipping in under six hours is Chboss at 5:51 with a 500MHz Athlon and a few minutes to the good (5:45) is his PII on a 112 bus. Our first PIII this update comes in at 5:04 from fsgray with a 550 Katmai. We are just entering the area where faster clock starts to equate closely to lower time. Brave Betsy set the priority to 'real-time' for a 4:54 with an Athlon 600, which is foolhardy to say the least. I think 'unresponsive' would be a mild term to apply to that particular system, Colin. Using the GUI slowed Sir Rich to a motley 4:36 with an 850 Athlon, down 20 minutes with the CLI perhaps? Kim S and PIII Katmai with a 112 bus slaps down a healthy 4:19 but I would hardly call enabling L2 error correction a tweak, well not in the 'speeds you up bracket' anyway! 
Oh look mother! Rambus (PC800) how quaint. I remember back in the old days when this stuff was going to rip the guts out of SDRAM and makes us all feel very small and envious. So a 'stock' Vectra (fsgray) turning in a 4:13 with 128MB of Rambus just confirms (again) that the technology is more exciting than the results. Though prices have dropped, it's still an expensive way to claim bragging rights.
If you have been keeping up with performance issues you'd know that Thunderbirds are go! Had to get that in somewhere. In at 4:07 is frm220 with an 832MHz Tbird. Just ahead is Steve Common's more normal time-type offering of 4:04. Unclocked at 600MHz this PIIIEB harks back to v2 days where CAS2, 133 bus, low multiplier and a good board (BE6) were in direct competition with big L2 Xeon cache and Athlon/Thunderbirds were sidelined. This is a very impressive time and shows that not all slower systems are going to be humbled by the new King Clock. Naturally having said that it is only right to reaffirm King Clock by banging in a couple of monster Duron times. Both tsubasa and Beyond have warmed the price-conscious SETI-crunchers hearts by overclocking from 600MHz to a Gigahertz, most pleasant. Times are respectively 3:51 and a delicious 3:38. Remeber this is a bargain basement cpu. Splitting them is Hellburner (great name, is there a Hellfreezer and some little HellsBells?) at 3:43 from a PIII at 866. Make that CAS3 a 2 and there will be tears before bedtime. Heading into the 'seriously quick' is Targa and an unclocked 1GHz Thunderbird for a 3:32, not that you'll be able to ramp it up much anyway. The battle rages on as another OC'd PIII (at 866) belonging to Ifkommer flays the bench WU for  3:27. Perhaps it should be renamed the RatMark. That Chboss person, is back again with 3:25 and a Thunderbird at a Gig, quite a range of times from this farm! Be ye humble, for the uncached shall inhabit the stats. Not too surprisingly Hellburner returned after deciding that CAS2 was the next step in wringing that PIII's neck. Tears of joy as it shed 27 minutes to 3:16. Tremendous but I don't think you will ever see the other side of three hours. Prove me wrong, please! Still a little way to go as Chad ups the ante to an inspiring, but also very disappointing 3:01 with a 1.133GHz Thunderbird and ME. How painful to be that close to a 2 hour time. Finally, knocking MadMac off the top spot by a minute is Marzio's 2:52. His PIII 700 OC'd to 1085 wrests the crown back (for now) from the upstart Tbird brigade. I've been weakly twitching and gurgling the last couple of weeks as the number of people pushing their systems under 4 hours keeps increasing. I jealously salute you in this strange game of numbers!
Please don't think that outright speed is everything, fun and informative it is certainly, but the sheer weight of numbers of slower systems make up a large chunk of TLC output. The big-hitters of TLC are great to have onboard but 'the team' is it's members and that means all of you.
Max out.

What all the best crunchers wear...
There's a very recent update to TaskInfo 2000, 575kb zipped, available on Igor Arsenin's  download page.  How many of  you know the name of the author of the software you rely on? I guess his would be would be one of the few. Since many of you use it to see what your system internals are doing while chugging away, it seemed reasonable to post news of this great bit of software. I've been using the 2.2 beta for several months with no problems (on W2K) and it's good to see the new version up at last. Windows 2000 does have Task Manager, but if you want the real dirt on what's running and hogging your cycles, TaskInfo 2000 is the beastie. Also works on 95, 98 & NT. After SETI Spy and SETI Driver this is next in my personal 'must have' list followed closely by Oda bits and pieces. Now for a little bit of miscellaneous update-age...
Also in similar it might be-of-interest-some-of-you type vein if you have a "Windows 2000-based computer that contains the VIA AGP chip set" and it hangs when using 3D apps then the MS knowledge base (MSKB) has an article and patch. Thanks to BetaNews for that one. Got a relatively recent Maxtor drive? Yes. Ever heard of Acoustic Management, nope, me neither. Some newer disks have their seek times under-optimised to keep them quieter. That really got me interested, how dare they slow down my drive, am I affected, can I get more speed, slobber, drool..? So if you want to make it faster (those once a minute sah file writes can really add up, got to be worth a few milliseconds per WU) then have a look at this Maxtor AMSET document which links to the utility you need. The info was spotted at Rojaks pot (Thurs November 2 - "Maxtor replies") but starts in the archives, Sunday 29th Oct "Acoustic Management for Maxtor Drives". My 20Gig DiamondMax Plus 6800 was to old apparently. Oh well, the chase was fun but now I need to go and clean the carpet, again.Yet more hardware updates surfaced from AMDZone  (Nov. 1st item) for AMD chipsets or go straight to the AMD drivers page. New AGP driver v5.21 - not sure how it links in, if at all, with the MS patch. 
Okay, back to the reality of TLC housekeeping and there are more times in the pipes. These include some serious Duron and Thunderbird speed demons. Words and numbers will appear sometime over the next few days (I hope), ready for a Monday morning office onslaught after a hard weekend down on the 'farm'. If Maxx and SETIan are reading this could you get in touch about those slaughterhouse times (no I'm not sure what that means either). Thanks. I'm approximately ten days in arrears as far as submissions go, so if you think you've been ignored or omitted it's more than likely that you are in my backlog. How does zAmboni keep up, is he human? The jury is still out on recently co-moderator enabled RB!
Max out.