More P5B secrets uncovered

[Tony]

http://www.thetechrepository.com/showthread.php?t=30

From the looks of 600fsb with C2 the new 965 chipset may have the 1600 strap, issue is performace may be poor. As soon as i get a board I will test.

For now let me know if theis helps you understand how Asus are gettting the board to clock

[FUGGER]

Thanks for the update Tony.

[EnJoY]

Very cool find Tony. I was curious as to how Asus would have made the board clock this well otherwise. How sneaky of them.

[Lestat]

hhmm interesting.
not what i would have done but if i was intel.. but .. i dont make those decisions.
higher ram and fsb but its still slower if compared to a higher multi same speeds.
kind of silly if ya ask me..

[ScottFern]

I don't see a graph?

[bachus_anonym]

400MHz "switch" up to 46ish you're talking about, Tony, explains a lot the results I was getting here and few posts down...

No wonder 400x9 DDR2-1200 is very close in SPi32M compared to 500x8 DDR2-1000, despite 400Mhz disadvantage on a CPU speed...

Thanks a lot for your investigative work, Tony

[The Batman]

Originally Posted by ScottFern

I don't see a graph?

I didn't see one either, but view the page in IE and you'll see it. Probably something to do with FF and adblocking (I have adblock+). Doesn't matter anyway, I don't get any of that mumbo jumbo

[HKPolice]

Any way to manually set strap settings?

[Tony]

Originally Posted by HKPolice

Any way to manually set strap settings?

No thats the issue

[cadaveca]

So, uh, 965 sucks. Conroe is great, but supporting chipsets suck. Wonder if ATI's got it figured out.

[FUGGER]

The chipsets dont suck, not even close. We are moving in the right direction and this is just the start of 600+ FSB in sync and that could equate to monster bandwidth.

[afireinside]

600 fsb 1:1 is great but if the NB settings are so loose that is murders performance compared to 975, who cares.

[situman]

higher fsb = high cpu clock speeds = better overall system performance

Besides, as frequencies climb the looser latencies are pretty much a non-issue. Before calling the chipsets as crap, think about how limited you are on the 975 chipset due to its fsb limitations. It gets to a point where raw bandwidth will not improve performance all that much, but high cpu speeds will always give you greater performance. So, yes bandwidth might suffer at certain fsb speeds but I'll give up a few points in bandwidth in order to max out my cpu.

[pcdoc1]

I didn't see anything here that says 965 sucks...

Ya just gotta love it... The damn board is fast, stable, reliable, and provides consistently reliable performance... Then someone explains how that's accomplished and all of a sudden it's crap...

Hard to believe... Now we're looking at a GPU manufacturer to change the chipset world. Guess I need another beer.

[cadaveca]

Originally Posted by FUGGER

The chipsets dont suck, not even close. We are moving in the right direction and this is just the start of 600+ FSB in sync and that could equate to monster bandwidth.

Ok, so the bios sucks. But from what I'm seeing, it will be a new chipset that does this, and not 975 or 965. They don't want us using those FSB speeds yet for a reason...next gen Conroe, maybe? 3ghz+ stock speeds? Next year Summer?


The whole "could" thing scares me...

[pcdoc1]

Probably not a good thread for conspiracy theories…

[Tony]

Ok guys we have more, infact a LOT more. Im off to bed but my brainstorming buddy FCG will carry on posting and hopefully enlighten you a little more.

I asked him to post on the BE thread but he will link here for you.

We found how to work out NB clock speed and where to see it in sandra, this should help you guys overclock a little better.

[pcdoc1]

Once again Tony, Thanks-

EDIT: Damn, hit a 1000 post after all these years... No better way to spend it than on "Tony AKA BigToe" help...

[aggybong]

Thanks for the results, Tony. I was considering doing 7x458 instead of 8x400, but now it seems rather pointless. I guess this explains why people have success skipping the 300s :p

[HKPolice]

Originally Posted by Tony

No thats the issue

What about using a prog like memset? Or editing the BIOS?

[gundamit]

Great read. Thanks for taking us to class Tony. Performance scaling on 965 is a little strange, but its a fun chipset.

[cadaveca]

Originally Posted by HKPolice

Or editing the BIOS?

Maybe I got this wrong, but i thought that was what this was all about?

Originally Posted by pcdoc1

Probably not a good thread for conspiracy theories…

???


I'm talking about timings, in case you misunderstood. That's really what the "strap" is all about. The "strap" support running this speed is probable due to the added latency support in the new revision, as the graph Tony put up kinda shows.

What I'm refering to is the actual performance at these speeds and latencies. From what we've seen already, I'm concerned it will take a few chip revisions of both cpu and chipset before we get the performance I'm really looking for.

But maybe I'll get proved wrong...I darn sure hope so.

[pcdoc1]

Originally Posted by cadaveca

I'm talking about timings, in case you misunderstood.

I don't think I misunderstood anything... Your exact statement was:

They don't want us using those FSB speeds yet for a reason...next gen Conroe, maybe? 3ghz+ stock speeds? Next year Summer?

The implication was that someone was intentionally holding back new tech for financial or other reasons... No mention of strap/timings... No problem though, carry on!

[freecableguy]

More to read:
http://www.bleedinedge.com/forum/sho...63&postcount=3

-FCG

[NO1B4ME]

Damn this is a pretty cool topic. Thanks Tony .