8000UL CAS3 testing (WR included)

[TaPaKaH]

System specs:
C2D E6400
ASUS P5B
2x512MB handpicked 8000ULs (92mm + 120mm + chilled air)
2MB PCI video
500W BlueStorm

Vcore = 1.35V
Vmem = 2.65V (Vmem mod)

sry for low results : running CPU under BOXed cooler and using old OS w/o tweaks

3-2-2:
500MHz - 4M stable
507MHz - validation (sry no link - validator reports "corrupt file")
3-3-3:
537MHz - 4M stable
542MHz - validation

and now ... WR for CAS3 : 573.8MHz

[Exedy]

[ozzimark]

extremely impressive, even if they're handpicked, very very nice


[ramble]
though to me, it's just another reason why memory WRs should be sorted by the primary 3 timings. let's say someone pulls off 550mhz at 3-3-3. to me, that is much "better" than 574mhz at 3-6-6, but won't be listed anywhere because the records are listed by cas latency only.
[/ramble]

[Exedy]

TaPaKaH,

could you try max oc with 4-4-4-5 ?

[TaPaKaH]

620MHz - 4M stable eeeeeeasy
640MHz - suicide validation (quick run, not pushing it)

Vmem = 2.63V

[Micutzu]

Awesome runs

[Exedy]

hehe this was nice

but 2.63 is hard pushing for 24/7 right?

even with customized sinks on each memory ( not "single" heatsink ) and 92mm on it?

thanks for the test

[kiwi]

Chipset volts?

[jmax_oc]

Why do you use the strange 3:5 ratio divider for your CAS3 tests ?

[Fr3ak]

Nice results. 2.63V is indeed a bit too much for 24/7 usage if you ask me.
But then again we dont have that much information about long-term usage.
I will give my sticks 2.4V max for 24/7 usage.
Thats already 0.2V more than the sticks are rated for and 0.6V over jedec specs.
Maybe 2.3V will do it as well.
Gah who knows, no risk, no fun

[SMa]

Impressive!
especially that 3-2-2!

[TaPaKaH]

Chipset volts?

1.6V I think

Why do you use the strange 3:5 ratio divider for your CAS3 tests ?

cuz, my CPU doesn`t go 2.8GHz+ very much under BOXed cooler ... so I have had to use 3:5 and 1:2 dividers

2.63V is indeed a bit too much for 24/7 usage if you ask me.

24/7 I use 2.45V ... 2.63V was for benching

[SMa]

Nice results. 2.63V is indeed a bit too much for 24/7 usage if you ask me.
But then again we dont have that much information about long-term usage.
I will give my sticks 2.4V max for 24/7 usage.
Thats already 0.2V more than the sticks are rated for and 0.6V over jedec specs.
Maybe 2.3V will do it as well.
Gah who knows, no risk, no fun

Tell me... what's the use of running your ram at eg 800 3-3-3 for 24/7?
Personally I can't feel any difference with 667MHz 3-3-3 or even 667MHz 4-4-4...
It's senseless to overvolt your ram 24/7. You're killing it, your max overclock will degrade over time.

:edit:
I just looked, currently mine is running at 533MHz 3-3-3-9 (1.8v).
If someone told me it's at 800MHz 3-2-2-1 I would have believed him, just can't feel the difference.
Altough, high numbers still look great

[Pt1t]

http://valid.x86-secret.com/show_oc.php?id=125192

http://valid.x86-secret.com/show_oc.php?id=125198

P5B dlx bugged with cas 3 .... look My gskill @ 2.4v

[Kunaak]

http://valid.x86-secret.com/show_oc.php?id=125192

P5B dlx bugged with cas 3 .... look My gskill @ 2.4v

how do you get the bug to show up?

I have this board, and about 5-6 sets of ram, and I tried most of it, on this board, and I have spent tons of time tweaking ram with this board, and I have never had a bug show up yet, in any form.

I know the Gigabyte DS3, is exceptionally prone to crazy latencies (800 mhz at cas 3-1-1-4 come to mind... which I do have screen shots of) however, I have not been able to stump this P5B deluxe at all to get it to show any bug or even a minor glitch...

[Pt1t]

http://valid.x86-secret.com/show_oc.php?id=125204

and WR cas 3 @ 2.45v ....

in 1:2 that would hit 600.

585 also ... http://valid.x86-secret.com/show_oc.php?id=125205

[bachus_anonym]

how do you get the bug to show up?

It's not a *bug* per se, but you can e.g. boot @ 600MHz 8-4-4-4.0 and use Memset to change it to 600MHz 8-3-4-3.0... CPU-Z picks it up and that "8-3-4-3.0" is as stable as what you booted at

I'm not gonna comment on above results, though... Athough, they seem a bit high. Besides, as ozzimark said, 3-6-6-8 should NOT be considered CAS3 WR. In my mind, and probably many others, CAS3 record is 3-3-3-5 one...

[Pt1t]

It's not a *bug* per se, but you can e.g. boot @ 600MHz 8-4-4-4.0 and use Memset to change it to 600MHz 8-3-4-3.0... CPU-Z picks it up and that "8-3-4-3.0" is as stable as what you booted at

I'm not gonna comment on above results, though... Athough, they seem a bit high. Besides, as ozzimark said, 3-6-6-8 should NOT be considered CAS3 WR. In my mind, and probably many others, CAS3 record is 3-3-3-5 one...

it does only this when p5b dlx boot @ spd , when cas 4 is set manualy , changed cas 4 by cas 3 to high gave freeze.

[Pt1t]

System specs:
C2D E6400
ASUS P5B
2x512MB handpicked 8000ULs (92mm + 120mm + chilled air)
2MB PCI video
500W BlueStorm

Vcore = 1.35V
Vmem = 2.65V (Vmem mod)

sry for low results : running CPU under BOXed cooler and using old OS w/o tweaks

3-2-2:
500MHz - 4M stable
507MHz - validation (sry no link - validator reports "corrupt file")
3-3-3:
537MHz - 4M stable
542MHz - validation

and now ... WR for CAS3 : 573.8MHz

and now ... WR for CAS3 : 585MHz

Not fake but Not real

[bachus_anonym]

it does only this when p5b dlx boot @ spd , when cas 4 is set manualy , changed cas 4 by cas 3 to high gave freeze.

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

This is just to ilustrate the problem and I'm not saying this is the case here! TaPaKaH's CAS3 results seem very high but they are not neccesairly bugged or faked.



... and a bit higher, and on a different ratio...

[TaPaKaH]

my result is not bugged !!!
you can try P5B if you want - changing the timing by MemSet doesn`t affect CAS !!!

Besides, as ozzimark said, 3-6-6-8 should NOT be considered CAS3 WR. In my mind, and probably many others, CAS3 record is 3-3-3-5 one...

Why ???

[overcrash86]

and now ... WR for CAS3 : 585MHz

Not fake but Not real

TaPaKaH OWN3D by bug

[kiwi]

If you boot at cas 3 on this board then it should be real i.e. don't use memset

[Pt1t]

If you boot at cas 3 on this board then it should be real i.e. don't use memset

but , how to proof it

edit : a video

[ozzimark]

my result is not bugged !!!
you can try P5B if you want - changing the timing by MemSet doesn`t affect CAS !!!Why ???

you can always run benches to prove it
extraordinary results require extraordinary evidence was it?

for the why? it's because from my own experience with micron chips on AM2, 3-3-3 and 3-6-6 and everything inbetween clock very very similarly. like within 10-15mhz close