8000UL CAS3 testing (WR included)

[blind_ripper]

tested i can boot with cas 4 @ 500mhz mem and set in windows with mem set on p5b non deluxe , notting in time changes for super pi

[jmax_oc]

2 fakes
One with an extraordinary spi result
The other with a memset bug

Next will be on gpu or maybe an hdd ?
Keep me in touch

[kiwi]

tested i can boot with cas 4 @ 500mhz mem and set in windows with mem set on p5b non deluxe , notting in time changes for super pi

Yep, I noticed that a few days ago as well

[Kunaak]

Unfortunatly, this is a confirmed way to fake things...

so now the question is... can we really call any claim on a WR, a WR, if it's done on a 965 chipset?
cause I've known the DS3 and basically all gigabyte boards are prone to crazy stuff, and I can dismiss them without a second thought, after my experiences with the board.

but now to see even Asus isnt immune to this kinda bug, it makes me question all 965 chipset claims on WR's.
I've always been the most severe skeptic, which is why you will rarely see me in a WR post, being the first saying congrats. cause thats kinda what I do... be the skeptic...

I've seen some crazy claims at WR's with 975 boards too, but thats always when using the bad axe, so it's easy to dismiss them too... cause to me a WR, really doesnt mean anything, if it's just a high number, but slow as dirt.

would a 3DMark WR impress you, if it was done at 6 ghz conroe... but only 47,000 with a X1900?

but now seeing this weirdness with the cas 4 - 3 being easily manipulated... just makes me think, 965 boards just plain old should not be taken serious, if anyone wants to really claim a WR with one.
but thats just me...

heres my confimation that the bug exists, and is easy to do...

Corsair 5400 UL, I run straight from bios cas 4-3-2-8 everyday, at 510-520 depending on how much I feel like pushing the voltage that day on the CPU.
but always 2.41 volts regaurdless.
cas 3 from this ram, is not possible past 450, regaurdless of voltage, atleast, not from a cold boot.
so seeing cas 3 at over 500 mhz, in windows, is really just a joke really...

again, below, these are fakes.
just testing the bug, nothing more.





validated, to show, validation means nothing on this kinda bug.

http://valid.x86-secret.com/show_oc?id=125236

[Gautam]

Hehe, lemme show you how its done with CAS 3 on D9 fatties. :p



http://valid.x86-secret.com/show_oc?id=102872

CPU-Z has a lot of holes that seem to be able to be exploited pretty easily with today's mobos. One things for certain, it can't be trusted as the end-all and be-all anymore.

[Kunaak]

this isn't a CPUZ issue, and CPUZ was never intended to be the one and only absolute form of OC verification.
frank is about the only programmer, besides the guy that made XS Super PI, to actually take any of my advice when it came to trying to make cheating harder.
CPUZ alone, is only 1 tool.
still as good today, as any other day.

this just shows theres more flaws to take advantage of, with the 965 chipsets, thats all.

[Gautam]

Yes, I could have worded myself a bit better.

However, one issue I *think* is that CPU-Z seems to calculate the memory frequencies based on other parameters instead of reading it, such as Clockgen does. I'm sure it used to in the past with A64's, hence so much of the divider confusion, etc. Whether that algorithm has been changed or whether its even easily possible to be changed is beyond me. However, in the case of the Bad Axe, and the screenshot I showed above, it appeared to read the memory divider incorrectly, and then report an incorrect memory speed based on this.

Nothing against Franck or CPU-Z, just saying its always worth it to take any results with a grain of salt at first glance.

[corvus_corax]

nah, I can do this all day long and on at least 0507, 0614 and 0706 BIOS

With 0609 i could boot at 500MHz@3-4-3 with only 2.45V, max at 504Mhz (SP1M), but if i set 505Mhz i coulnd't enter to windows, just i get a reboot. BUT, actually, i think that was a bugged *strange*cas, cuz if i set 500Mhz@3-4-4 the system hangs in bios, i mean, even there was no boot.

Testing with 0507, 0614, 0701 or 0706 i can't repeat that. So, the "only way" is using Memset but as we know that new CAS is unreal, so, BS if you ask me .

[ozzimark]

so the moral of the story is that cas3 can only be trusted on AM2 setups

[erwinz]

hehehe

so the moral of the story is that cas3 can only be trusted on AM2 setups

[funkflix]

Don't call results a fake, if the guy doing this don't know about a "bug"..

But RAS# Precharge works for me with Memset, i bootet with 3 and set it to 2, and i got a boost in 1M.