[QUOTE=T.Rex;176813]This old Itanium2 machine has several Numa factors... meaning that 16 cores is not 2 times faster than 8 cores, not at all... And Glucas is written in C, no assembly. I played a few with Prime95 on 8x Nehalem, and scalability was not very good (due to the fact that Prime95 is already very efficient on 1 or 2 cores).
T.[/QUOTE]

Still wouldn't mind having a machine like that running a bunch of tests in parallel for me...

[QUOTE=Kevin;176814]Still wouldn't mind having a machine like that running a bunch of tests in parallel for me...[/QUOTE]Sure. But it consumes a lot of power: it uses triple-phase... Good for winter time. But not for summer ! And do you have rooms tall enough for a little more than 2m50 ? And it weights a lot. And fans are so noisy ! An old machine... Old good times... I remember the DEC PDP/11 I used in 1980... 768 KBytes of memory.
Tony

[QUOTE=T.Rex;176815]Sure. But it consumes a lot of power: it uses triple-phase... Good for winter time. But not for summer ! And do you have rooms tall enough for a little more than 2m50 ? And it weights a lot. And fans are so noisy ! An old machine... Old good times... I remember the DEC PDP/11 I used in 1980... 768 KBytes of memory.
Tony[/QUOTE]

I was talking about the 8x Nehalem :smile:. I also wouldn't mind having something to keep me warm, since apparently Minneapolis gets even colder than Ann Arbor in the wintertime.

[QUOTE=Kevin;176788]All 4 cores of a Q9550 at stock speed (2.83ghz) on 64-bit Ubuntu, and I'm getting around .018 seconds per iteration.[/QUOTE]

What would be very useful for comparison is if (once you have finished this run) you would try your four cores against an exponent that uses a FFT length of 4096K (something around 77mil).

I get 18.3ms per iter at that FFT but it would be good to compare against Prime95/x64. I believe I should get around double the speed (using four times the cores).

As this exponent has shown the benefit in bringing out the big hardware in terms of speed has been minimal this time around. Tony as his hardware is aging and me as the software can't efficiently use the right FFT length (ATM).

Thanks, Rob.

[quote=rgiltrap;176820]As this exponent has shown the benefit in bringing out the big hardware in terms of speed has been minimal this time around. Tony as his hardware is aging and me as the software can't efficiently use the right FFT length (ATM).[/quote]
Yeah, but verification on more than one software/hardware system is still important...good thing Tony's will be finishing at the same time Kevin's will, so we won't have to wait around to find out if it's really prime or just some bug in Prime95.

[QUOTE=rgiltrap;176820]What would be very useful for comparison is if (once you have finished this run) you would try your four cores against an exponent that uses a FFT length of 4096K (something around 77mil).[/QUOTE]

I'll try and remember to do that Saturday night/Sunday morning.

[QUOTE]Nope, the urban legend that sewers have overflowed from so many people watching the Super Bowl on TV simultaneously using the bathroom at half-time.[/QUOTE]

Ah. I had not heard that before.

[QUOTE]As this exponent has shown the benefit in bringing out the big hardware in terms of speed has been minimal this time around. Tony as his hardware is aging and me as the software can't efficiently use the right FFT length (ATM).[/QUOTE]

Wouldn't the double check on these more powerful machines have been faster if the right FFT had been used as opposed to one that was far larger than necessary?

[quote=Kevin;176812]Nope, the urban legend that sewers have overflowed from so many people watching the Super Bowl on TV simultaneously using the bathroom at half-time.[/quote]

A few weeks ago I saw the a prog on TV about the UK equivalent of this effect. A guy sits at a control panel and, when the commercials come on in the middle of a popular soap, flips a switch to bring online a power station in France. To power all the kettles.

All this talk of the Superbowl gives me a great idea. We should send out mass emails with a subject something along the lines of free/discount tickets to the Superbowl. Of course, this email would contain a virus that then installed Prime95 and started putting their computers to work. Of course, the program would have to be hidden :smile:

[QUOTE=Kevin;176788]All 4 cores of a Q9550 at stock speed (2.83ghz) on 64-bit Ubuntu, and I'm getting around .018 seconds per iteration.[/QUOTE]

I'm getting only 25.5ms/ite compared to your 18ms on a Q9450 (2.66ghz), but I'm running Windows XP x64, probably slower than Linux.

[quote=Primeinator;176828]All this talk of the Superbowl gives me a great idea. We should send out mass emails with a subject something along the lines of free/discount tickets to the Superbowl. Of course, this email would contain a virus that then installed Prime95 and started putting their computers to work. Of course, the program would have to be hidden :smile:[/quote]

Somebody did something similar with BOINC a few years ago. He distributed Winows updates, which installed the BOINC-client, that crunched then on his account. ;)