X48 Rampage Formula Preview.

[Alien Grey]

Thanks for your reply.
O.K. I will give the G-Skills 9600 a try.
I questioned for "the right memory sluts" because one stick wouldn´t fit with their headspreder under my Thermalright IFX-14. It does not seem to be a question of the right sluts but of the right DRAM Clock Skew settings.

Once you know what DRAM Clock Fine Delays to use with the Memory it doesn't really matter what slots you use.

Another quest to MemSet V4.0. Do I get you right, that one time save setting plus load will change the tREFs after the next system boot and so on ?

Yes, MemSet V4.0 is an excellent program. I hope Felix adds Auto tREF in a next version that calculates tREF according to JEDEC specs for the Memory speed that is used.

To change tREF and automatically load it on the next boot all you have to do is set the tREF timing you want, apply it and save.

[Fritz_the_germ]

Thank you Alien Grey,
the memset4 job was very simple. I had to change my LCD settings though. MemSet does not like more than 96 dpi but it didn´t matter at all.
I made a cachemem test with Everest like you can see on the screenshot.
The "read" seems to be a little better by the 16383T tREFs now - not to mention the stability that I´ll have to check up now.
With the best regards
FRITZ


The left one is with 16383T. The right one with the 3120T. Both made on WinXP OS.

[Alien Grey]

Thank you Alien Grey,
the memset4 job was very simple. I had to change my LCD settings though. MemSet does not like more than 96 dpi but it didn´t matter at all.
I made a cachemem test with Everest like you can see on the screenshot.
The "read" seems to be a little better by the 16383T tREFs now - not to mention the stability that I´ll have to check up now.
With the best regards
FRITZ

Sorry, I can not handle this attachements: My file of 206.2 KB bytes exceeds the forum's limit of 200.0 KB for this filetype. 6.2 KB to much.

Just cut out the part you want like I did with the picture of MemSet or make it a little smaller and higher jpeg compression.

[ChaosMinionX]

Any particular reason that my L2 cache in the Everest benchmark is pretty slow in comparison to others?

[Alien Grey]

A copy of a post that I made on the ASUS Rampage Formula Forum.

It looks like the CPU and NB Clock Skew didn't help to get rid of that random reboot I'm experiencing in Ubuntu 8.10 64 bit edition.
What I know is that it isn't caused by the CPU. I would see that on the results in BOINC for the packages that I'm working on.

It's probably something with the memory timings in Linux but that's difficult to find out because of the lack of decent programs like EVEREST Ultimate Edition for Windows.

So, I asked again to the ASUS Global Tech Support to do something about tREF in the BIOS. I would be very happy with a tREF timing in the BIOS. Something like tREF on Auto (Disabled) and Enabled. On Auto (Disabled) 3120T like it is now and Enabled a tREF calculation according to JEDEC specs.

(7.8 x 1000) / (1000 / DRAM MHz)

Example: 1066MHz

(7.8 x 1000) / (1000 / 533) = 4157 -> tREF 4160T

They are going to look what the problem is with EIST showing to be enabled in EVEREST Ultimate Edition and what is wrong with tRTP. The tRTP timing in EVEREST Ultimate Edition is 5T and not 3T like in the BIOS. It's confusing because there is a timing in the BIOS tRTPD that is 5T.

[screwtech02]

Hey all, is there gonna be a noticible performance difference from going from a Q6600 to a X3360 on this board??? I know the 45nm stuff runs cooler, and also has 12mb cache, but other than that???? I'm tryin to hold out on the I7 stuff till it gets cheaper....

[neo_rtr]

By the way Greg got him self a new CPu and he is happy as a child in a store with free gooooodies. Just waiting for him to update with his results..

[greg.m]

By the way Greg got him self a new CPu and he is happy as a child in a store with free gooooodies. Just waiting for him to update with his results..

OMFG!!! I had for one hole mfyear this QX9650 piece of mfcrap and i was mftrying to find the right mfvolts, skews, mews, chews, and what the fcukever ..ew there is in the hole mfworld

And one fine day i found one fine guy who wanted to buy my mfcrapqx9650.......and he bought it AND then i went to a fine store and bought my Q9650. This took place yesterday and i have to say that my oclife begun, YESTERDAY just look here:

[greg.m]

By the way with 0803 i needed higher vcore than with 0701 so i got back to the second one because i was scared that i would damage my new cpu with setting vcore to 1.26750 instead of 1.2500

.......and all that with llc enabled without having any problem at the moment. Guys idont now what to say This chip is the best i ever had. I cant believe what its capable of. It

[Alien Grey]

By the way with 0803 i needed higher vcore than with 0701 so i got back to the second one because i was scared that i would damage my new cpu with setting vcore to 1.26750 instead of 1.2500

.......and all that with llc enabled without having any problem at the moment. Guys idont now what to say This chip is the best i ever had. I cant believe what its capable of. It

It's a wise decision from you to use BIOS 0701 instead of BIOS 0803. You can't believe how much faster CPU degradation would kick in with 0.01750V higher voltage.

When you are saying that you need more vcore with BIOS 0803, is that for Prime95 or IntelBurnTest?

Is there a reason that you use PL8 and not PL7?

[Spyrus]

You can try higher values. It is possible that it works and that your memory runs stable like that to but following JEDEC specs can never be wrong.

JEDEC is not wrong neither is the proposed tREF.
But are you running your memory timings in full JEDEC compatibility?
Unless you are not overclocking
Think that changing from 5-5-5-15 to 5-5-5-14 you are out of JEDEC.
Same goes to tREF .... try whatever is faster AND stable

[Alien Grey]

JEDEC is not wrong neither is the proposed tREF.
But are you running your memory timings in full JEDEC compatibility?
Unless you are not overclocking
Think that changing from 5-5-5-15 to 5-5-5-14 you are out of JEDEC.
Same goes to tREF .... try whatever is faster AND stable

What are you trying to say? That a static 3120T as tREF that we have now with the ASUS Rampage Formula is following JEDEC?
It should at least follow the spd timings and than tREF should be at 4160T for my memory at 1066MHz and not 3120T.

Try what ever is faster and stable isn't good because ASUS doesn't care if it's faster and stable. It's stable with 3120T and that is good enough for them.

So, if we can't have larger static tREF timing in the BIOS because 3120T and 800MHz is standard they should at least make it possible in the BIOS for us to have the tREF timing that we need for the memory speed that we are using.

When your still not satisfied with that there is always MemSet or Bar_Edit to change it to a higher value.

That's how I see it and perhaps I see it wrong.

[greg.m]

It's a wise decision from you to use BIOS 0701 instead of BIOS 0803. You can't believe how much faster CPU degradation would kick in with 0.01750V higher voltage.

When you are saying that you need more vcore with BIOS 0803, is that for Prime95 or IntelBurnTest?

Is there a reason that you use PL8 and not PL7?

It was with prime95 that i got errors. But here is something strange, with:

vcore:1.25
cpupll:1.5
fsb: 1.4
nb: 1.41
ram: 2.1
llc:enabled
everything else auto
and
475x8.5@1140, 5-5-5-15, pl7

i could do 5 passes @ intelburntest and when i tried to load and play crysi wars my system did freeze I tried different(higher) volts @ vcore, nb, fsb, ram, cpu-nb-gtls, but got the same freezes. Tried also with 470fsb, pl8, pl9 and it went a little bit better but in the end i got freezes again. After that i got back to 450x9@1080 and everything was running smoothe again.

Im not sure but i think the problem might be my memory or whatelse?
I have somewhere 2x1g 1200mhz kingston modules. Ill trie
to find them and give them a try.
Any suggestions?

[Alien Grey]

It was with prime95 that i got errors. But here is something strange, with:

vcore:1.25
cpupll:1.5
fsb: 1.4
nb: 1.41
ram: 2.1
llc:enabled
everything else auto
and
475x8.5@1140, 5-5-5-15, pl7

i could do 5 passes @ intelburntest and when i tried to load and play crysi wars my system did freeze I tried different(higher) volts @ vcore, nb, fsb, ram, cpu-nb-gtls, but got the same freezes. Tried also with 470fsb, pl8, pl9 and it went a little bit better but in the end i got freezes again. After that i got back to 450x9@1080 and everything was running smoothe again.

Im not sure but i think the problem might be my memory or whatelse?
I have somewhere 2x1g 1200mhz kingston modules. Ill trie
to find them and give them a try.
Any suggestions?

IntelBurnTest isn't a reliable source for stability. Especially with the FSB Termination Voltage.

Give the extreme stability test a try like I do. Download Unreal Tournament 2004 demo. Let it run in a window 1024x768 pixels (or something like that) with maximum quality. Choose Game rules Set goal score and time limit to 0 (infinity). Choose spectate and let it run for a few hours if you can with Prime95.

[Alien Grey]

The problem EIST showing up enabled in EVEREST Ultimate Edition isn't a problem with the BIOS but a software problem with EVEREST Ultimate Edition.

It's like I thought and ASUS confirms it.

[mikeyakame]

Was kinda bored and constructed this with my board Had the spare DFI sink lying about as I replaced it with a Thermalright HR05 SLI/IFX cooler and have a 80x25mm fan on it!

DFI LT X48-T2R Thermalright NB heat sink, Thermalright HR-09U VRM sink on IO side, modified RF VRM sink with heatpipe removed (temp till i pick up hr-09s for there), and SB sink replaced with copper SB sink off some base board.

[Alien Grey]

Was kinda bored and constructed this with my board Had the spare DFI sink lying about as I replaced it with a Thermalright HR05 SLI/IFX cooler and have a 80x25mm fan on it!

DFI LT X48-T2R Thermalright NB heat sink, Thermalright HR-09U VRM sink on IO side, modified RF VRM sink with heatpipe removed (temp till i pick up hr-09s for there), and SB sink replaced with copper SB sink off some base board.

Nice modification.

I've got a Thermalright HR-05 IFX to cool the SB with 80x80x25 fan and for the mosfets a Thermalright HR-09U and a HR-09S.

[mikeyakame]

Nice modification.

I've got a Thermalright HR-05 IFX to cool the SB with 80x80x25 fan and for the mosfets a Thermalright HR-09U and a HR-09S.

I was going to go the HR09U on the DFI board to replace OE vr sink but i ran into a hitch, my TRUE with push pull, ( 25mm push, 38mm pull ) doesnt clear the hr-09u when its secured! so im stuck with DFI black sink!

[Ghostleader]

@mikeyakame

Nice mod, have start it up yet and checked the NB temp?

I run my NB in the same waterloop as the CPU, NB got nice and cool temp with water even with high volt.



Got 8 pieces of Enzotech BMR-C1 on the mosfet and CNB-S1L on the SB, working really good.

[JoeBar]

Be careful with these heavy cooper ramsinks on the mosfets. In the past it destroyed one 8800GTX by falling off. Personally i would go with some light aluminum ones...

[RCG_Bex]

Need some help.

Got a Qx9650 and it's been able to boot up into Windows and run fairly stable with the following settings

420*10,
333 strap,
PCIE 101
cpu @ 1.375 (tried up to 1.4v)
pll @ 1.5
NB @ 1.39v
ram @ 2-2.10
vtt 1.32
0.63%
0.67%

- tried RAM with all sorts of settings etc

Trouble is, Prime95 either fails on core 3, or if I drop down prime95 just stops responding. Don't know if it's a Windows 7 bug or not.... not really sure what to tweak tbh. Tried OCCT too, and it does the same after ~15mins

~Bex

[Ghostleader]

Be careful with these heavy cooper ramsinks on the mosfets. In the past it destroyed one 8800GTX by falling off. Personally i would go with some light aluminum ones...

They have been there for more then 6 month and are still firm in place but thank you anyway, I´ve it mind though.

Need some help.

Got a Qx9650 and it's been able to boot up into Windows and run fairly stable with the following settings

420*10,
333 strap,
PCIE 101
cpu @ 1.375 (tried up to 1.4v)
pll @ 1.5
NB @ 1.39v
ram @ 2-2.10
vtt 1.32
0.63%
0.67%

- tried RAM with all sorts of settings etc

Trouble is, Prime95 either fails on core 3, or if I drop down prime95 just stops responding. Don't know if it's a Windows 7 bug or not.... not really sure what to tweak tbh. Tried OCCT too, and it does the same after ~15mins

~Bex

You probably need more nb and vtt juice, after all you run 4870X2 crossfire (if sig is right) and a quad.

What bios do you run?

I did some test run with my Q9650@450 4,05GHz 4x2 ram@1080 555 15 tRD7 Strong with bios 802, no problem.

Don´t remember my setting, I´m at work now, gonna check later when I get home and post my setup.

Check out Zucker2k post

[Fritz_the_germ]

@RCG_Bex,
first of all I would give 1,49v to the NB, 1,36vtt and run prime95 again.
Maybe the Ai Clock Twister should be cut down to moderate.
What are running with prime95, the large ffts, the small ones, blend or what the h***?

[Ghostleader]

Need some help.

Got a Qx9650 and it's been able to boot up into Windows and run fairly stable with the following settings

420*10,
333 strap,
PCIE 101
cpu @ 1.375 (tried up to 1.4v)
pll @ 1.5
NB @ 1.39v
ram @ 2-2.10
vtt 1.32
0.63%
0.67%

- tried RAM with all sorts of settings etc

Trouble is, Prime95 either fails on core 3, or if I drop down prime95 just stops responding. Don't know if it's a Windows 7 bug or not.... not really sure what to tweak tbh. Tried OCCT too, and it does the same after ~15mins

~Bex

Here´s my Q9650@450 4,05GHz 4x2 ram@1080 555 15 tRD7 Strong with bios 0802.

Q9650 E0 VID 1,2375 Batch L824B301
4x2 OCZ Reaper PC8500 ver3
Windows 7 beta 64bit

The rest as in my Gaming Rig


It´s stable with 50 loop LinX (5+hrs, max problem size and max memory use), 6hrs with four instances of P95 blend. Is it stable or what.

It´s not an over the hill OC but it´s a decent 24/7 OC setup in my opinion.


Important bios settings in the pics.

If you don´t push for low tRD you may lower your NBvolt.






Good luck

[greg.m]

Need some help.

Got a Qx9650 and it's been able to boot up into Windows and run fairly stable with the following settings

420*10,
333 strap,
PCIE 101
cpu @ 1.375 (tried up to 1.4v)
pll @ 1.5
NB @ 1.39v
ram @ 2-2.10
vtt 1.32
0.63%
0.67%


- tried RAM with all sorts of settings etc

Trouble is, Prime95 either fails on core 3, or if I drop down prime95 just stops responding. Don't know if it's a Windows 7 bug or not.... not really sure what to tweak tbh. Tried OCCT too, and it does the same after ~15mins

~Bex

Take a look here:

http://www.xtremesystems.org/forums/...postcount=1618

Zucker2k had completely right!! I couldnt believe that my expensive qx9650 was the real problem...I tried everything but i always had similar problems with you. I rmaed my old qx9650 and get a new one hopping that i could run easily @ 4.0 BUT..... NOTHING, the same problems were still there
Finally i took Zuckers2k advice and sold this mf piece of qx9650 and bought one Q9650
Now im in love with this......... i have no words i really couldnt believe that there would be such a HUGE difference

Now im in testing period but for the moment i run @ 9x450 @ 1080, 5-5-5-15, prime stable with vcore 1.256 under load , nb 1.35, fsb 1.32(bios) 1.23 real and my dominator @2.0v