Core 2 Duo users...

[n7]

Basically, i need some E4300/E6300/E6400 vs. E6600/E67000/X6800 results for Super PI 32M.

To make the comparison at least somewhat close, i need a set speed to run this at.

Let's use 3003 MHz (7x429) 1:1 (DDR2-858 5-5-5-15)
If you can't lower your CPU multi to 7x, then just do as close as possible to 3 GHz (9x334 e.g.)

Use Super PI Mod 1.5: http://www.xtremesystems.com/pi/super_pi_mod-1.5.zip

n7's results: E6300 @ 3 GHz 1:1 (DDR2-858 5-5-5-15) - 18m 24.906s
WoodButcher's results: E6600 @ 3 GHz 1:1 (DDR2-667 4-4-4-10) - 17m 41.922s
gerwen's results: E6400 @ 3 GHz 1:1 (DDR2-750 4-4-4-12) - 18m 40.891s
Brunnis's results: E6600 @ 3006MHz 1:1 (DDR2-668 4-4-4-12) - 17m 28s

Thanx for running this for me guys.

 

[WoodButcher]

17m 41.922s
What is your argument? 2m vs 4m cache not worth the extra bux?

 

[n7]

Originally posted by: WoodButcher
17m 41.922s
What is your argument? 2m vs 4m cache not worth the extra bux?

Nah, just attempting to explain extra cache is something Super PI benefits from.

What RAM ratio/timings did you use for your time?
1:1 4-4-4-10?

 

[WoodButcher]

Originally posted by: n7
What RAM ratio/timings did you use for your time?
1:1 4-4-4-10?

Eggzactly! This speed / setup is the best happy medium I could find for this mobo. In the near future I plan to move to the asus commando and mushkin 8500 memory to see how fast this chip will go. I can get 3.2, barely,,,:disgust:
Your sig shows a nice OC, Where did you get the memory?

 

[gerwen]

Not 32M, but relevant:

From the last post on this page(Read the post, it explains the tables):

http://pics.bbzzdd.com/users/gerwen/time.jpg
http://pics.bbzzdd.com/users/gerwen/cycles.jpg

Clock for clock, the 4MB parts beat the pants off the 2MB.

 

[WoodButcher]

Originally posted by: n7

Originally posted by: WoodButcher
17m 41.922s
What is your argument? 2m vs 4m cache not worth the extra bux?

Nah, just attempting to explain extra cache is something Super PI benefits from.

Sorry about the double post, I meant to ask if you could explain or post a link. I bought the 6600 not fully understanding the cache, When I bought I had the extra $100 and the reveiws pointed to the 6600 as a top all round chip and still haven't spent the time to learn.
Is the extra cash worth 45 seconds on super pi? Not to me but I hate to get caught in a blizzard with only a teaspoon. Best reason a non-techie like me could come up with.

 

[Roguestar]

When I get home from work and finished my dinner I'll fire it up and give you my best .

Anything for an ATer.

 

[Shimmishim]

check out the hwbot.org webpage as well... as they have pi times out the wazoo over there...

 

[gerwen]

Oh, and 18:40.891 8x375 4-4-4-12 for 32M on an E6400

 

[lopri]

The 'Cycles' table is very interesting. I didn't think of it before. Very nice, gerwen.

I think we can do the same for 32M, which will obviously answer n7's original post, without actually running the calculation under the exact configuration he suggested.

 

[Brunnis]

17m 28s

E6600 @ 3006MHz (334*9)
2GB DDR2 @ 668MHz 4-4-4-12

 

[lopri]

I get 16m 20s with E6400 @3.0GHz (8x375) <-- 64-bit OS cheat

 

[lopri]

Oh and 1T cheat, too.

 

[WoodButcher]

Originally posted by: lopri

The 'Cycles' table is very interesting. I didn't think of it before. Very nice, gerwen.

Ditto!


I get 16m 20s with E6400 @3.0GHz (8x375) <-- 64-bit OS cheat

So is it the vista or the 1T making the diff?

This scares me,,, a little reading and I think I'm beginning to understand..... I'm going to lay down, this thinking stuff hurts,

 

[n7]

Originally posted by: WoodButcher
Your sig shows a nice OC, Where did you get the memory?

Tank Guys back in early autumn before RAM prices skyrocketed.
For once, buying early [C2D (+ DDR2) not long after they hit Canada] turned out to be not so bad financially.

It's decent RAM.
On this mobo, it can do up to ~ DDR2-1180 5-5-5 @ 2.3V (within warranty), though only @ 1:2, which is useless for real use...

Originally posted by: WoodButcher
Sorry about the double post, I meant to ask if you could explain or post a link. I bought the 6600 not fully understanding the cache, When I bought I had the extra $100 and the reveiws pointed to the 6600 as a top all round chip and still haven't spent the time to learn.
Is the extra cash worth 45 seconds on super pi? Not to me but I hate to get caught in a blizzard with only a teaspoon. Best reason a non-techie like me could come up with.

Honestly, the E6600 is not worth the extra $$$ from a pure bang for buck point of view.
But it's the best high end option, since E6700/X6800/Q6600 pricing is a joke IMO.

 

[n7]

Oh, & thanx a lot guys for posting your results.

Much appreciated.

 

[lopri]

1T makes a big difference (750MHz/4-4-4-3T in this case) although it's not very stable on my system. I believe 64-bit OS can improve on Pi times, other things being equal, compared to 32-bit version. (but not major)

There is nothing scary about it, though. 4MB CPUs are definitely faster than 2MB CPUs for Pi calculation from my observation although the gap narrows as the calculation length gets longer.

 

[n7]

So i take it 64-bit Vista is still doing well for you?

Good to hear...i might just take that plunge when my Ultimate DVD arrives, if ever...

 

[lopri]

So far, yes. Only app that I couldn't get to work is Alcohol 120 (which installs virtual drives) and Superfetch is, while not perfect, great IMO. You don't have to enter a serial # for installation, and not until you have to activate (1 month) so if you're not certain you can always try it out first or dual-boot. But I do think $400 for Ultimate retail is way too expensive. (especially if you have more than one system)

 

[WoodButcher]

Originally posted by: lopri
1T makes a big difference (750MHz/4-4-4-3T in this case) although it's not very stable on my system. I believe 64-bit OS can improve on Pi times, other things being equal, compared to 32-bit version. (but not major)

There is nothing scary about it, though. 4MB CPUs are definitely faster than 2MB CPUs for Pi calculation from my observation although the gap narrows as the calculation length gets longer.

Very scary,,, old dog like me learning new tricks?

Honestly, the E6600 is not worth the extra $$$ from a pure bang for buck point of view. But it's the best high end option, since E6700/X6800/Q6600 pricing is a joke IMO.

Originally I had the 820D planning to upgade to extreme 840. Even the prices on ebay never justified the purchase of the 840EE over the 6600, Once in a while you might see them go cheap but there are still people asking $1,000 for that chip.
I agree totally, with the performance of intel's low end chips most of us would be hard pressed to justify the added expense.

 

[Roguestar]

Originally posted by: lopri
So far, yes. Only app that I couldn't get to work is Alcohol 120 (which installs virtual drives) and Superfetch is, while not perfect, great IMO. You don't have to enter a serial # for installation, and not until you have to activate (1 month) so if you're not certain you can always try it out first or dual-boot. But I do think $400 for Ultimate retail is way too expensive. (especially if you have more than one system)

Yeah, well here in the UK we have exactly the same prices as you guys have in dollars, except in pounds!

We are basically paying double for the exact same software as you guys just so Bill can subsidise the relative cheap price on your side of the globe. $400 != £400, Bill!

 

[slag]

I'll try this when I get home. I'm at work and cant change my cpu speed right now, but using superpi 1.1, I got 16 mins, 27 seconds with an E4300@ 3.25 ghz.

I'm going to retest with 1.5 and see what happens.

EDIT: 1.5 gave me 16 mins, 28 seconds. I'm remotely logged into my machine from work but there shouldnt be noticeable overhead. So basically no difference between 1.1 and 1.5.

I'm using 2 gb of cheap microcenter Vdata memory, ddr2-667 @ 362 mhz bus speed currently. 5-5-5-15 memory timings. I will tighten it up tonight. I know the memory will perform much better.