Smoked by a Packard-Bell!

Last night I ran a POV-Ray bench­mark avail­able at Have­land-Robin­son Asso­ciates’ site think­ing I was all hot stuff after get­ting this 300mHz K‑6 machine. My time was 2 min­utes 58 sec­onds. I went to look at the com­par­i­son to oth­er peo­ples’ results and was imme­di­ate­ly hum­bled by the best time report­ed: 3 sec­onds. Actu­al­ly, I think I can feel pret­ty good about the fact that my machine is only 60 times slow­er than a Cray T3E-900-AC64, which has 48 450 mHz DEC Alpha chips…

I have to admit that I did­n’t real­ly put it right up to the lim­it. I left some things run­ning in the background.

So the results say very inter­est­ing things about dif­fer­ent oper­at­ing sys­tems. I looked up my old machine, a 486dx2/66 , in the chart, and I found a few. The fastest run­time on a 486dx2/66 was 21:32, and it was run­ning Lin­ux. The next fastest was 22:05 on a machine run­ning DOS 6.2/QEMM 7.04, which prob­a­bly isn’t fair because DOS 6.2 is total­ly sin­gle­task­ing. Right behind the DOS bench was a NextStep run at 22:23. Then there were a few more DOS entries, and an OS/2 machine came in run­ning the native POV-Ray code at 23:18 and the DOS ver­sion in a DOS box at 23:34. The sec­ond fastest Lin­ux run I found on a 486dx2/66 came in at 23:43 and was fol­lowed imme­di­ate­ly by a Win95 box run­ning an AMD K5-133 at 23:56.

That’s where I see a very dis­turb­ing trend form­ing. A Lin­ux box run­ning a 486/66 out­paced a Win95 machine that by all rights should run about four times as fast?

So I scanned down the list for the fastest Win95 time on a 486dx2/66: 27:57. The next one was 45:08, about three and a half min­utes behind a Lin­ux machine run­ning a 486 DX33.

Ouch. What was this MS was say­ing about how Win95 would «make your pro­grams run faster»? Faster than what? I guess we for­got to ask. I guess if we had they would have said «Faster than Win­NT». Dou­ble Ouch.

So back to my sys­tem, which is run­ning Win98 on a K6-300. In com­par­i­son, the fastest time for a K6-300 was­n’t actu­al­ly for a K6-300, but for an over­clocked K6-266. Prob­a­bly exact­ly the same thing any­way. Run­ning Lin­ux. 1:55.

Found a won­der­ful typo: The Intel Pen­i­tum II.

I found a Win95 K6-266 that came in at 2:16 and a Lin­ux K6-233 at 2:21.

Pret­ty con­sis­tent­ly the Lin­ux machines ran faster than the Win­dows box­es some­times by a big mar­gin, some­times by a clos­er mar­gin. But what I real­ly found inter­est­ing were the num­bers clos­er to my time.

I don’t believe this to be true, but accord­ing to the list, a Packard-Bell P75 on Win95 beat my machine by four full sec­onds, also squeez­ing out a Sun Ultra­SPARC 30 at 267 mHz by a sec­ond. Hmmm… Per­haps the num­bers can lie.

My machine scored the same time as a Mac Pow­er­PC G3 233 MHz. I guess that is a mighty fast proces­sor, that G3.

So what does this dis­turb­ing trend mean? The most pop­u­lar oper­at­ing sys­tem is the one that crip­ples your proc­ce­sor the most. Should Microsoft­’s new slo­gan be «Do Less With More!»? What would Bill say if he knew how unfa­vor­ably his OS benchmarkes?

OK, well, I’m sure Bill already knows.

So the the­o­ries start­ed brew­ing in my head. Why would we, as a soci­ety want our machines to be crip­pled? We want fast machines, but we want soft­ware that slows them down. Some peo­ple would blame this on a con­spir­a­cy by Microsoft and Intel, but I think there must be a deep­er mean­ing. It’s too easy to hit the big targets.

Crim­i­nals have suc­cess­ful­ly and unsuc­cess­ful­ly tried the defense in court that they had no choice but to com­mit crimes because soci­ety made them do it. We see this hap­pen­ing again and again with every elec­tion. Our Elect­ed rep­re­sen­ta­tives are exact­ly the sort of spine­less weasels we want in pow­er. Why? Because we Amer­i­cans hate gov­ern­ment so much we make a con­cert­ed effort to rot it out from the inside. This is a plan I hope hurts them more than it hurts us. Nev­er­the­less, we can draw an inter­est­ing par­al­lel about this behav­ior with our choice of com­put­er software.

Could it be that, since we secret­ly fear com­put­ers that we are always keep­ing our eye on them and mak­ing sure that they don’t stab us in our sleep? Is it our fear that if we make these things too well that they will rise up and take our place, ala Ter­mi­na­tor? Per­haps so. And per­haps we, in our des­per­a­tion not to be enslaved by our own cre­ation are mak­ing sure that we put a leash on it.

And what a leash it is! Can you imag­ine what those Ter­mi­na­tor movies would have been like had the machines been run­ning Win­dows? There would­n’t even have been a movie! Can you imag­ine Arnold stand­ing there for two hours while our heroes have a bar­beque after decid­ing not to hit CTRL-ALT-DEL to restart the sys­tem? Full on! Crip­ple­ware is our best defense against a pos­si­ble com­put­er insur­rec­tion.

So we should be grate­ful to Bill Gates. He takes a lot of abuse for his mis­sion. And thank God he’s on our side, fight­ing day and night to keep those pesky com­put­ers in their place.

Won­der why the gov­ern­ment is going after Bill these days? Sim­ple. I think the politi­cians feel some com­radery with the com­put­ers, both being down­trod­den by the mass­es that way. Just think what could hap­pen if the politi­cians and the com­put­ers became unit­ed against us.

Once again, we’re blessed. Nat­ur­al selec­tion pre­vents any­one intel­li­gent enough to oper­ate a com­put­er from run­ning for pub­lic office. So the world is safe for yet anoth­er day.

BTW, I went after­wards and closed down almost all oth­er process­es and ran the bench again, and came in at 2:50… beat my «bur­dened» time by eight sec­onds. I felt pret­ty good about that. Faster, but not enough to be dangerous.