• blah
  • blah

Oscar Wilde was witty, but me, I'm just a lawnmower...
Several days ago RB gave me a good build-up but I'm afraid I failed to deliver and this late catch-up is not going to be a 'Burt Reynolds in a canoe' scenario. Right, meaningless cinematic allusions over I can settle down to doing justice to all those names and numbers so foully forgotten over the last couple of weeks. Well I did have an excuse, but even the dog in the alley didn't believe me. Really this is not the place to explain my minor calamities...though I can be persuaded by personal calls and donations of cash!
So far a few things stand out as far as times are concerned. Anything with a high clock speed/small cache has benefited from v3. Anything with a big L2 cache has lost ground as v3 is nowhere near as cache dependent as previously. Wildly overclocked Durons and Celerons have gained a lot, Thunderbirds and Athlons have also regained a shedful of credibility. The previously highly efficient Mac G4's and G3's have tarnished some of their lustre to the high clock speeds of 'PC' processors. Xeons have lost considerably as have the slower more exotic SPARCs, HP-PAs, Alphas & MIPS whose advantages of scale and cache have pretty much deserted them with v3. As the rollout of clients continues expect some gnashing of teeth as the IT guru's are contemplating how to upgrade their fleets to v3 and possibly some stopping altogether rather than pursue competition. As far as completion times are concerned the Berkeley boys and girls have effectively given the edge to overclockers, SOHOs and home enthusiasts and greatly reduced the large corporates clout. Though admins' with buildings full of PIIIs are still sitting pretty and laughing all the way to the WU bank, the balance has changed towards the smaller (fleet & farm) owner. Amongst the K6 and older Pentium fraternity there are sour looks and glum faces. The new client has saddled these processors with a lot more work while being unable to gain from the smaller v3 cache footprint. They have reported mixed results with v3 - generally similar or slower than v2.
On the OS front little has changed Windows ME has made no great difference, which is no surprise as it's just the manicured bastard offspring from the 98/98SE stable. If anything seems to have benefited it's Windows 2000 with dual CPUs and Linux also with multiple processors. As yet the jury is still out on NT4 and Mac flavours but doubtless as you carve and shape you systems to optimise v3 the picture will become clearer. As other, rarer, operating systems surface in the results we'll get an idea of any inherent advantages but I don't expect any major shocks or changes from the present hierarchy. 
Onto the past...Mac G3 stalwart OooklaTheMok tried out Mac OSX beta for 12:24 ('horrible') and then reverted back to OS 9 and went through various hoops (7:21, 7:08, 7:02) to end up at 6:55. Obviously there are some unresolved issues with OSX! In the Pentium arena we have Moller (19:46, not sure what's happening there), Jeffrey Ottie (a commendable 14:54 at 233MHz), Lightning (12:26) and Mournblade (12:13) both at 300MHz vying for some rays of light. While in this area ASF has a Red Hat/Pentium Pro combination that managed 13:22 but I'll be impressed if you get that much further up the table. First single digit time appears with Maek07 and a 300 PII/NT4 at 9:01.  As expected with the early state of the results table there are going to be sizeable gaps so next mention goes to gary running two clients on Win2000 that came in at 7:07 on a Celeron at 528Mhz. These times would have been good with a single client in v2. BP6 users are going to be very happy. Another Abit board (BH6) powered zouo to 7:07 with a 350 PII and NT4 - CAS2 RAM is the obvious next step. Zouo appears with a quicker time at 6:10 with 464MHz Celeron hand in hand with Linux.  Note: I'm trying not too place much emphasis on comparison v2 times as come November 18 v2 will be dead and buried (so we are told) and our only interest will be how various v3 flavours and system combination hold up. Good to see Win95 still hanging in there as OS of choice for many including Mex's 6:54 on the GUI which dropped to 6:28 for the CLI using a flipchip Celery clocked to 713Mhz (75 x9.5 multiplier). He thinks he can do better - hope so! By contrast lower clock higher bus (500, 83 FSB) gives Jim W similar results of 6:54 and his comments reflect some enthusiasm for v3. Three times from jonluck all on a PII Xeon/W98, first a 6:25 GUI effort (4.5x100) that became 5:52 (3.5x133) and finally with the CLI ending up at 5:11 (3.5x133) - a tasty little improvement. Another impressive Celeron (67MHz bus) time of 6:17 from Jeffrey Ottie who seems to have quite a way with hardware even Toshiba's. Dropping into the fives with 5:52 is another dual setup of Win2000 allied to 667MHz PIIIs (you know...the beast plus one). Haps should be reasonably pleased with that i.e. 2 completions every six hours. A little grouping of Athlons now and 5:49 from Gianfranco is first running at a gentle 500MHz, but as we'll see there is a long way to go with these re-enlivened CPUs. Simon ups the pace to 5:23 (ME, GUI and 750MHz) while jimN creeps nearer the four with a 5:13 on a 600MHz and W98. First Athlon under the 5 (4:59) belongs to betsy with another well set up 600.

Intermission (popcorn and cola in the foyer now).
Little time out here - one for a rest from the numbers and two to give betsy a plug as he reminded me of a little known facet of SETI spy. Have a look at the first paragraph on his SETI Spy explanation page.  If you fancy your farm making occasional odd noises beyond the shrill cries of fans fusing bearings this is for you. 

Part the second...
Back to times and fsgray pops up with an 850 Celeron II that smokes a 4:34 with the CLI on Win2000. Think that's quick, just wait for the good news! Celerons have really made gains with v3 as you will see soon. Karstenv slots in a pleasant 795MHZ Athlon time of 4:32 while just ahead are a brace of Thunderbirds, jerrybaker (800/Win2000) at 4:11and Trombone8vb (850/ Mandrake 7.1) with a 4:07. We're creeping up on three hours now which if you cast your minds back to v2 days (seem a long way off now) was a staggering achievement for almost any CPU/OS combination. Manny and johndotjohn43 are just the wrong side of 4 with 4:07 & 4:03 respectively on PIIIs and Win2k. Cleanly in under the four hours (3:45) is bobgoblin on a Gigahertz PIII (133 bus). Are you beginning to notice a trend of raw clock speed here, which although not an absolute factor is definitely in the ascendant. Stand back from the results table and just look how the reduction in time correlates with an increase in clock speed. A useful offering from surfshop Jeffrey Ottie with a Dell Dimension (not best known for optmisation) PIII 866 & ME slapping down a 3:43 with the GUI and a fairly raging 3:18 with the CLI. That near half hour difference echoing v2 CLI/GUI comparisons. Harking back to Celerons, here's three staggering times from these cheap and cheerful contenders. A 3:40 and 3:37 from Mo where he upped the bus from 103 (927MHz) to 112 (1007MHz) to improve the time but had to drop from CAS2 to CAS3 to get the thing to run! Net effect was a drop of 3 minutes for all that work. Sounds just like old times! Just to cement that comment a little more how about another 566 Celery overclocked to 952 on a BX chipped Asus board that brings home a 3:34 for jschner. A slight retrograde step here in clock speed for Tomslik that's made up for by running his 600EB at a 166FSB (roasting in hell) to produce a snarling 3:32. An hour faster than his v2 time on this system. Another high end GUI/CLI comparison from Redbeard whose 1070MHz Thunderbird simmered a bench WU for 3:31 and 3:10 respectively. A 21 minute advantage to the CLI is plenty of reason to ignore the GUI, but I think you all know that anyway. Yes that's heading for sub 3 territory and I'm impressed, even if you've all seen it before.
MadMac's 3:31 is rather gorgeous once you realise it's from a Win2000 dual machine. This WU shredding outrage runs PIIIs at 903MHz on another exceptional dual mobo, an MSI694D. What will it do in single mode I wonder? Chad delivers (now there's a slogan for you) a GUI 3:30 with the now ubiquitous Thunderbird (100MHz) and though knocking a chunk off with the CLI asked me not to mention the time until he's happy with the improvement. Shame, but I can respect your desire to post something really outrageous rather than merely exceptional. Our first and only Duron appears at 3:29 courtesy of officeboy, a nicely understated title. It's presently overclocked from 600 to 1080MHz (!) and this would be the SETI crunching CPU of choice except available steppings are not dual enabled. Now if you could use a graphite pencil to SMP them you'd really be on to a winner! MattC's Thunderbird spat out a 3:27 (990/110 Win98) while a minute ahead but with GUI TuffGuy's Tbird (ME, 1135/103 - an 11 multiplier, what times we live in) slapped out a 3:26. You seem to have entered my vocabulary Hanser! So what does the CLI do...3:05!! Approaching the sub three zone. Well impressive. Amongst this Thunderbird carnage there are a couple of PIIIs, Jeffrey Ottie's 3:18 mentioned earlier and regular abuser of CPU's, neo1999 at 3:08 (933/133/W2K). Just one to go now for this rather long and overdue honours list - MadMac, instead of benching his dual 3:31 setup in mono mode he gently turned the screws on a Gigahertz TBird to 1176MHz and out popped a 2:53. When I first saw this result I nearly threw up now I just get mild stomach cramps. Peltiers, CAS2 top-notch memory, a good board and a bit of luck - easy when you know how. I'm off to update the results sheet with all your hot little submissions, hope to catch up with you soon. Thanks to Colin (dancing emails) Duane, Ignacio and Bonkers (still giggling) for their comments and look forward to the dodecahedral Alpha bench your going to run with v3 (hint, hint). Errors, ommissions, fibs and porkies can all be corrected if you let me know. Bench for fun, crunch for results.
Max out.

Get yer times!
Yep, with titles like that, it's obvious why they don't let me write anymore.  In anycase, for those interested, there are already a total of 94 results (submitted, not tabled) for the new version 3 client. Damn you guys are fast. I'm sure Max will have lots of witty things to say. As for me, I'm just glad to have B)Gotten Donuts. -Rat

I'm not dead
Really I'm not. I am indeed doing some things behind the scenes here, although honestly not too much. I do find that I've got more free time now that I've moved and settled in. However, this last weekend was spent installing some new drives in my servers, along with a motherboard upgrade in one, along with 2 case changes, one of which required mods to add some extra fans. :) 
Been busier at work as well, as American Express has me training with IBM Global Services at the moment. Boy- sure sounds good on paper, doesn't it? In reality, the highlight is either that I A)Get out of working, or B)Get donuts. I'm kinda leaning toward A, as I could always bring donuts with to work.  Anyways, enjoy everyone, and I'll chime in as I'm able. -Rat

If it isn't one it the other...
After the recent Ober update it had to be SETI spy next, which has duly marched on to 3.0.1 and been picked up by ZDNet as well! 
Some observations from Roelof that make for educational reading...points 1 to 5 are based on his experiences with the TLC benchmarks (20 in all) carried out on his own machines (covering combinations of Pentium, PII, PIII, PIII Xeon, Win95, Win98 and NT4):
"1. The v3 GUI is generally faster than the v2 CLI (4% to 14%), but is slower in a few cases (-3% to -5%).
2. In all cases, the v3 CLI is faster than the v2 CLI, from 6% to 25%.
3. The v3 CLI seems to be consistently about 12% faster than the v3 GUI, for all processors tested.
4. There is very little difference in processing efficiency for the P2/P3 processors, with the CpF ranging from 4.31 to 4.72 (only a 9% variation). [Explanation of CpF, Cycles / FLOP ratio half way down SETI Spy page.]
5. Since the V3 client does 65% more work on the benchmark unit than the v2 client, the 6% to 25% decrease in WU time equates to a 76% to 120% increase in processing efficiency.  For Pentiums, P2s and P3s, the v3 CLI is about twice as efficient as the v2 CLI and for Athlons and Durons it will be even more."
After I updated the results table to give him enough to reasonably work on Roelof came up with a little more...
"This graph is based on the latest (limited) TLC benchmark results, comparing clock-for-clock performance of S@H 3.0 on different processors. For the x86 family, only the text client results were used. These are based on the reciprocal of the average CpF value, where the performance of a P II/III = 100%. It was possible to create three general groups, PII/PIII/PIII CuMine, Celeron/Celeron II and Athlon/Thunderbird. Their members could be grouped together because there were no significant differences between them. It looks like the Pentium II/III still holds a 20% performance advantage over the Athlon, in other words, an 800MHz P3 and a 1GHz Athlon have similar performance" (quoted with minor editing)...more benchmarks might change these slightly but the overall trend seems valid.

Although Xeons still come out comfortably ahead in the efficiency stakes it is nowhere near enough to offset the MHz advantage that many less efficient processors have. Those wonderful 1 & 2MB L2 caches are just not that important any more.  
I asked for news of dual boxes recently and Joey was kind enough to give me some averaged non-bench times for his kit. General observations are that his dual PII 450s (on an Asus P2B-DS) have remained pretty much the same, v2.4 to v3. However the dual PIII 700s at 933 (also P2B-DS) and dual PIII 700 at 966 (MSI 694D) have both dropped their times from low sevens to mid fours. So as hoped it seems that the v3 client has greatly reduced the disadvantage on running 2 clients on one bus (lower contention), in effect an extra speed up for duals beyond that the v3 has bestowed on most single processors. Recommendation is to update ASAP. I imagine that bench times are in the works...or I might have to do some myself!
Max out.

The numbers of the beasts...
Some early v3 results up on the v3 results link courtesy of RB. As Rat would say "it's not pretty but it works". Will produce some decent words soon but right now I'm in the grip of the cold from hell (almost funny). Better thank Roelof as usual who's name appears staggeringly often on the list! Also SetiDriver now at 1.5.0.3. Main thing to notice from benchmarks is the general speed increase for most systems and configurations. More table results soon, as I know there will be a shedful waiting when I check!
Update: Another 10 or so results added to the v3 list. Perhaps I should mention that though times have dropped for many of you  there is less happiness in the Xeon & G3/G4 camps. Their large L2 caches no longer protect them from the clock speed onslaught that Thunderbirds and PIII's are obviously benfitting from with the v3 client. Would really like to know how dual boards running Win2000 are doing, any BP6ers out there? Are Celeron II and Durons edging towards equality? Without those benchmarks, I can't be sure.
In memoriam: Bench submissions for v2 have just about dried up which is convenient as Rat and I were wondering at what point to consign them to history (glorious though it was). So no more v2 times, please, though I will eventually catch up with the ones already in and your efforts will be recognised. 
Max out.

Real speed and red herrings...
I mentioned a while ago (September 10) that Mike (Ober) and Roelof (Engelbrecht) might be creating a bastard offspring - the mother of all SETI utility software! However, it seems the comments I picked up on referred specifically to making SetiSpy and SETIdriver work symbiotically in bug free harmony. Unfortunately they were not aimed  towards the creation of a single 'Spydriver'. Thanks to Roelof for enlightening me on that one and puncturing my rumour balloon.
I'm pleased to see that v3 (including command line client now) has rolled out fairly cleanly and with slightly less fuss than the upgrade to v2. Perhaps this has a lot to do with  the Berkeley bunch learning from experience and end users (us) being pretty much in with the flow of what to expect. Far more is known about how WU crunching is going to be affected by the new version than was the case with v2. Just look at the info from Lawrence Kirby and Roelof put into zAmboni's article on the front page. However, there is still a warm buzz of expectation as to how exactly your times are going to be enhanced/compromised. This whole enterprise is science driven so no matter how wrapped up you get in squeezing minutes off your result times, you are still involved in something a little bit bigger than 'which team is top of the pile'? Speed and competition are fun and worthy, but the underlying data sieving/searching is ultimately far more important.
An interesting thread on alt.sci.seti (seems to be improving in read-worthiness lately) entitled "I'm way to fast - Help!" (7 Oct on) concerns the staggering speed Gary Harris is mangling his way through WU's. He's running v 2.04 on Win2000 with a K6-2 at 450MHz.You could expect around 10 hours for a lean setup, with the best K6-2/Win2000 in at 8:25 on the table. But this guy has disembowelled 130 units at an average of 2:02! There's more to come...because some of his early WU's were on a P133 and 'normal' the average time taken is inflated. At the moment WU's are being crunched seriously sub two... the follow up post mentioned that the average had dropped to 1:57 in the course of the morning!! To get an idea of the numbers being returned he put up a screenshot of a 'Spy progress window over the output log file. The first column is process time and the average of those 21 times is 13 minutes...oops! One of the first posted responses (tongue in cheek I hope) was that it must have 8MB on die cache. I don't care if it had the whole of Lawrence Livermore on the chip it still wouldn't be that fast. Well it makes interesting reading and I'm going to be a little saddened when the truth emerges, whether duff cpu or more likely corrupt client. The idea of a 13 min K6-2 system is really rather sweet. I will keep you posted on the outcome. 
I guess most of you just want to know how bench times for the v3 GUI (client benchmarks soon) are shaping up and have waded through the above with hardly a glance, fair enough...time to put you out of your misery in a paragraph or so as there are a few v2 submissions to mention first.
If ever one man resembled two dogs fighting over a bone it has to be Mournblade who is still clocking his Pentium (P55C) to within microns of its existence (presently at a coffee-spill-enhanced 300MHz/100 bus) and is justly proud of slapping down a 13:33 upon the altar of combat. It's only a personal best by 8 minutes but anything that gets him closer to neo1999's 10:49 will lessen the indignity of second best. Another silent assault from fflix has knocked over an hour off a recent submission to drop to 9:12. Obtained, it would appear, by upping the bus from 133 to 144 - welcome to the black arts of cpu cookery! Good improvement but there's still something holding you back there. Patry has a Dell XPS T600 that returned a 6:33 from the 2.4 client on NT4, a fair time on a branded beast, what more can I say except tweak that sucker. Finally in the old (now) client stakes a little bit of beauty in the form of a Compaq Alpha running Tru Unix 64. This happy chappie is an Alpha 21264 third generation 64-bit processor running at 667MHz, its monster memory bandwidth helps it return a very healthy 3:14. It might be rare for this page but there are enough of these beasties around to warrant their own ported client, though whether you'll get a v3 is less certain as Berkeley seem to be reducing the number of differentiated clients. Thanks to Roger W. for that. Always good to have something a little different.
Onto v3age at last...and I think that many are going to be very pleased by the drop in times that early submissions suggest. But real world WU's are very variable and the angle range (AR) will be decisive in determining what to expect. Just to be contrary OoklaTheMok is not a happy bunny. A 12:24 on his 500Mhz G3 running the OSX public beta that recently benched at 9:15 is not going to help him sleep at night. The "This is horrible" comment from OTM himself seems about right! But revenge is sweet and Ook also benched the 'standard' Mac GUI v3 for OS 9 and was rewarded with a 7:21 (9:20 same system v2.04) that must have put some of the bounce back into his life. Roelof also reports mixed results, though not quite the same variance as we've just unearthed. Firstly a stock Gateway (350 PII Deschutes) has gone from 8:29 (v2) to 8:54 (v3 GUI) for a 5% loss. Secondly his stock Dell Inspiron 3500 (366 PII Dixon) went the other way from 9:48 (v2) to 8:30 (v3) for a 13% gain. Both running Win98. Onto a Celeron II (a SETI under performer thus far) from neosupply. With an 8.5 multiplier, 112 bus for 952MHz running Win2000 it managed a 5:53 which is far superior to almost all Celery times for v2. The v3 needs less cache than v2 so Durons and Celerons are going to be in for rich pickings, while Athlons will probably end up on par or slightly ahead of similarly clocked  PIIIs. Next up with seriously low times is bobgoblin, a 3:45 from a Gigahertz PIII with Win98 and also jottie at 3:43 from a PIII (864/133) and Win ME. This last time comes from the same system that jottie benched two weeks ago with the 2.4 client and produced a 4:22 so he's dropped 15%. Sounds pretty pleasant so far. Finally a little bit of justice for all those Athlon owners - looks like v3 is your saviour. It's going to take the derisive sniggers out of the Intel owners delightful charm school manners. An Athlon 1100 under the relieved control of Chad posted a 3:30 after previously being in the deep blue doldrums of the high 5's. A serious decrease of over 2 hours. I wonder what delights the client holds? How much will Win2000 dual processor systems benefit from the reduced cache dependence? How many monkeys got sore digits trying to write this page? All for the future. Hunt those benchmarks aggressively.
Housekeeping:  Thanks for the extra submitted times that I haven't mentioned. I decided quite a while back not to include beta times in the benchmarking. Repeat them on the v3 client now it's out (especially TuffGuy, that's one v3 bench I want to see) and I'll be pleased to include them. Since we are into a new chapter of numerology I better reiterate that if there are any mistakes just let me know and I'll try and correct them. All benchmarks presented here rely on trust and good faith with a smidge inquisitive error weeding. Rat Bastard work/family (remember those) commitments prevented him from carrying out some planned maintennance so don't expect these figures to appear in the results tables yet. Also in the submissions form (to be updated soon, honest) any details that are not selectable include in the comments and I'll clean it all up. Famous last words! 
Max out.

 v2 epilogue? Or the best is yet to come...eventually.
I assume that you have traveled here by way of the front page where zAmboni has already disseminated the good news of V3age. Well, good news if you are not the kind of person who has settled into my "and-on-the-tenth-day-he-created-update" style (add "and-he-saw-that-it-was-good" for brownie points). Soon I am going to be snowed under with new benchmarks from the feverish hordes. I should be very happy but the greatly enlarged TLC team size means I'm about to drown in submissions. All very well being a Rat as they leave first and are good swimmers! Come back all is forgiven - I might even leave the lid off the garbage occasionally for you.
Squeaking of furry friends (fiends?) RB cobbled together some pages for the results tables which now come in 4 sections. They are not pretty but they work and the numbers are all there. Some semblance of normal service for you swarthy followers of the TLC flame. Not bad for a rodent who wants a life as well! 

Okay, on with the minor upgrade news, SETISpy to version 3 to match/celebrate/tread water with the recent trickle of v3's from Berkeley. Roelof has also rewritten the instructions page, very comprehensive and nearly exhaustive. This leads nicely on to SetiDriver, as Obers' opus is not to be outdone in the minor bug removal fest, moving up a tiny bedpost notch to 1.5.0.1. Possess an 800 series Intel chipset, Win98 or above and a shiny new disk? Then you probably need these updated Ultra ATA drivers (v6.03/Sept 19, 2000/1.8MB), am I good to you or what? Might just be useful for a few of you. If like me, you go for small footprint info utilities then here's a newish one called CPUID (200kb) that appeared recently on Overclockers. It bears a great resemblance to Oda's WCPUID but although the processor usage bars were fun I still think Oda is the man at present, sorry wasn't trying to ferment competition...what, here on the benchmark page, unthinkable!

A few v2 times to report (any in transit will be duly noted but I am expecting a tsunami of V3 numbers soon and they will get precedence). Opening shots come from MNfan11 who must be very patient waiting for a 500MHz K6-2 to dredge its way through to a 34:34. Hard to see any positives in that one. There's a long, long way to go to be respectable as most of your contemporaries are looking at around 11 hours! In similar sad vein bbbb (was aaaa already taken?) has a 400MHz Celery pumping them out at 30:00. Pathetic. The 66MHz bus might have a little to do with it!  But even so there are several similar clock speed 66MHz FSB Celeron systems doing sub tens. You probably need to weed out some background garbage, dig into the bios settings (re caching/shadowing/CAS) and smoke that sucker to make it sing. Moving up a ledge to perch in Linux Valhalla is fflix using Mandrake 7.1 on a 733PIII giving 10:30. A very mediocre starting point. You can knock at least 3 hours off that with a little work and perhaps even get into the mid to low sixes (5:00 is about the extreme limit). A welcome return to these august halls of fame for OoklaTheMok, a man who has had his fair share of problems. He's run the Mac 2.04 Cli.  using Mac OS 9.0.4 on a 500MHz G3 (for a 9:20) and then same machine using the public beta of Mac OS X for a 5 minute gain with 9:15. Not a ground breaker but I get the feeling that some of you would sell your souls for 5 minutes. Thanks Ook (if I may be so familiar, Mr. OTM to the rest of you) and keep them coming.  Next up is zouo with a brace of times from a PII 350 delivering 8:38 on a very intentionally underoptimised system (see relevant comments for reasons) and 7:52 from an old-stager CeleronA 300 thrashing along at 464MHz. Overclockers have a lot to thank the Intel processor development line but I'm not up to history lessons tonight. Getting into the fast zone we have JBL with an HP Vectra (PIII/7333) that slaps down a rusty gauntlet of 5:45. There are faster vanilla 733's around but this gives an indication of what can be achieved (i.e. half the time of fflix). Another occasional regular, Redbeard, stumps up a 1070MHz Athlon time that is quite respectable at 5:41 given their poor SETI performance. Same system also manages low 4hrs average with the 2.76 beta which is relevant as this version has just become v3. Athlon owners can expect some better numbers to come. the delightfully named Zippy the Pinhead has ramped his PIII bus from 100 to 133MHz (933) to breeze in with a pleasing 4:47 on Win ME (Win 98TE as I prefer to think of it). But this is overshadowed by jottie (of surfshop and setiathome@klx.com mail fame) whose Dell (!!) 4100, PIII 863/133 turned in a moon-howling 4:22 also on ME. Such is the competition for low 4hr PIII times that oil price fluctuations seem tame. And to top that Tomslik obtained troll rights to a 550E that eventually yielded up a 4:16 after the bus was mildly thrashed to 166MHz. So there is a use for PC150 after all! Eight whole minutes closer to the magic 4 hours for Tomslik and he's a understandably pleased. Shame that v3 is going to put a stop to all this nonsense. So that's about it for v2 as the next chapter in the SETI crunching story is just beginning. I will try and keep up with help from Rat Bastard.
Finally a call to neo1999 to get in touch as I have a man who's lusting enviously after some of your system specs!!
Terminally, finally for the odd one or two who can't live without the buzz of beta software DirectX8 has gone up a notch to build 183. I include this mainly for the poor souls who timed out on the previous betas and rather than reverting to their previous setup were informed that DX wouldn't work at all, got to love those cutesy MS jokers! Live dangerously or why bother?
Max out.