Brisbane Testing in progress...

Printable View

Show 100 post(s) from this thread on one page

12-13-2006, 09:07 PM
FghtinIrshNvrDi

We know, but we always like to test clock/clock what it looks like.

Ryan
12-13-2006, 10:04 PM
Fred_Pohl

http://img172.imageshack.us/img172/8...ench31gkz5.jpg
12-13-2006, 10:58 PM
Epsilon84

Quote:

Originally Posted by FghtinIrshNvrDi

We know, but we always like to test clock/clock what it looks like.

Ryan

We already know C2D is ~20% faster clock for clock. Are you really expecting anything different compared to previous C2D vs X2 comparisons?
12-13-2006, 11:15 PM
Nanometer

Yes, because it seems AMD changed more than the processing size.
12-14-2006, 12:13 AM
RimRam

Quote:

Originally Posted by Fred_Pohl

http://img172.imageshack.us/img172/8...ench31gkz5.jpg

Is that Cinebech in winxp 64?
12-14-2006, 12:41 AM
FLMJIGGY

Nice that fact it can undervolt soo much. sweet!
12-14-2006, 01:03 AM
Lightman

Quote:

Originally Posted by Nedjo

In my book this has the same importance like overclocking:
Lowest stable voltage for the stock frequency (2.6 GHz):

http://img293.imageshack.us/img293/1...voltyo0.th.jpg

I’m off to bed

Can you confirm BIOS vCore settings??
I suspect it's 1.2V, but who cares if temperatures under load are so low!! Barely 10C more than ilde :clap: .

@FrankR if you want 3.6GHz on air from K8 then you need to wait for G2 rev. AMD promised new transistor design on 65nm after 3 months from initial production. I think most of initial G1 shipments anyway will go to OEMs. Now we should chase for 14 week older chip than 0635 which should be 065x.
12-14-2006, 01:06 AM
freeloader

Quote:

Originally Posted by Epsilon84

We already know C2D is ~20% faster clock for clock. Are you really expecting anything different compared to previous C2D vs X2 comparisons?

Well in the Cinebench it's only 13%. Is this just an abnomally?
12-14-2006, 01:20 AM
Lightman

Quote:

Originally Posted by freeloader

Well in the Cinebench it's only 13%. Is this just an abnomally?

No! Cinebench is intensively using x87 FPU for computations and AMD FPU is very strong at this ;]. Besides I suspect Fred did his bench in XP x64, in x32 difference should be even smaller!!
12-14-2006, 01:36 AM
w0mbat

I would like to see a competition between a 3GHz 65nm K8 and a 3GHz 90nm K8. Maybe on the same Board and with the same RAM. Then we can see if its more than just a Die shrink.
12-14-2006, 02:02 AM
kl0012

Quote:

Originally Posted by Lightman

Besides I suspect Fred did his bench in XP x64, in x32 difference should be even smaller!!

I suppose it was Cinebench 9.5 x32 edition. x64 edition should look like this:
http://www.xtremesystems.org/forums/...86&postcount=4
Look at the title.
12-14-2006, 02:20 AM
breakfromyou

oh geez. thats crazy. 2.6 GHz at just under 1.1v!

and to think i need 1.55v for 2.6 GHz :\
12-14-2006, 02:22 AM
Lightman

Quote:

Originally Posted by kl0012

I suppose it was Cinebench 9.5 x32 edition. x64 edition should look like this:
http://www.xtremesystems.org/forums/...86&postcount=4
Look at the title.

Thanks for correction :toast: .
12-14-2006, 02:36 AM
wonderwall

Did you test SPI 1M?
12-14-2006, 02:39 AM
FLMJIGGY

he's sleeping. tmrw....ssshhhhhh don't wake him up ;)
12-14-2006, 02:42 AM
w0mbat

Quote:

Originally Posted by brentpresley

You got it backwards. C2D is stronger in 32-bit, so the difference would be larger (this is because macro-ops fusion is not available for C2D under 64-bit mode and allows the gap to narrow).

The C2D was tested in 32Bit!
12-14-2006, 02:45 AM
Lightman

Quote:

Originally Posted by brentpresley

You got it backwards. C2D is stronger in 32-bit, so the difference would be larger (this is because macro-ops fusion is not available for C2D under 64-bit mode and allows the gap to narrow).

Yes and no! If you compare C2D 32bit vs 64bit latter still will be faster but speedup is smaller than for A64 going from 32bit to 64bit.
My point (wrongly assumed) was that Fred used 64bit when Nedjo used 32bit version. Now if both have done tests in 64bit then gap would be smaller ;)
12-14-2006, 02:48 AM
w0mbat

I think ur writing skills are bad, not my reading skills. Aber hallo!
12-14-2006, 03:12 AM
Lightman

Quote:

Originally Posted by brentpresley

Actually, that's simply not the case:
http://www.xbitlabs.com/articles/cpu...o-64bit_6.html

It depends on the application. Some gain performance under 64-bit and some lose some. Cinebench is faster under 32-bit for C2D in some parts of the test and slower in others.

My own experience with an E6600 @ 3.2GHz is that there is no signficant difference under 32-bit and 64-bit with that application (less than 5% measurable).

Yep! C4D Shading is always slower on 64bit due to some reasons (probably drivers not fully optimized), but we was comparing CPU speed which means RENDERING and for this I was right. C2D is faster in 64bit mode when rendering!
I put my first post you quoted wrongly, so I was thinking right but writing it down not so good :p: .
On the other hand your assumptions about C2D performance are correct as well.

Forget about small things, here are more important things to discuss.

WHERE is NEDJO?? We want more benches!:woot:
12-14-2006, 03:19 AM
w0mbat

Quote:

Originally Posted by brentpresley

Yeah, sure there Mark Twain.

Ok, lets see how good u can write german :stick:
12-14-2006, 04:04 AM
informal

Brent,please leave Conroe talk out of this nice discussion.It's about Brisbane and overclocking,not about 64 bit Conroe performance.And ,no ,I don't care who started it.

Nedjo used FX74 (air)cooler for this bench so the temperatures are very good(low that is).The vent spins at 6K.If only he had some other mobo for this test.But it's good we have this review,nice to see such a low volts at 2.6Ghz from first gen. of 65nm process.
12-14-2006, 04:17 AM
w0mbat

Quote:

Originally Posted by brentpresley

You don't want to go there. How many post-doctorate degrees (PhDs and MDs) do you have?

In a pissing contest, I guarantee you that I can piss further. :D

Ok, u can piss further :slap:

@Topic: I want benches!
12-14-2006, 04:30 AM
Nedjo

OK!
Here are Sandra results @ 3100 MHz
CPU
http://img221.imageshack.us/img221/2...0mhzkd3.th.jpg
MMedia:
http://img209.imageshack.us/img209/9...0mhzjq2.th.jpg
Memory:
http://img209.imageshack.us/img209/4...0mhzup0.th.jpg
Memory is working on 1033 MHz.
12-14-2006, 04:36 AM
Fred_Pohl

Quote:

Originally Posted by RimRam

Is that Cinebech in winxp 64?

WinXP32

BTW, it bears mention that multi-threaded Cinebench is one of K8's strongest BMs vs C2D (accounting for the relatively small difference) and that air-cooled CPUs do not idle at 10C lower than ambient air. Brisbane certainly appears to be just a dumb shrink and there is nothing surprising about the performance or temps as measured thus far. IIRC the 2.6GHz screen shot at 1.1Vcore was just that, a screen shot. It in no way implies that one could actually run a Brisbane at such low Vcore and 1.5V was needed for ~3GHz.

Show 100 post(s) from this thread on one page