Turion ml-40@nf3 250gb

[z3scott]

Keep testing. If pi doesn't run though you are at a limit. May need more voltage to ram or proc. Remember to memtest too!

 

If pi won't complete you need to watch things and tweak some more or back off a little. You can read in that overclocking link too. You should try to work out the max of the cpu and max of the ram seperately.

 

To work on the cpu run a divider like 100 and LDT/FSB Frequency Ratio 2.5x and keep raising HTT/FSB and voltage.

 

Test ram by running 200 1:1 with your multi at 7x LDT/FSB Frequency Ratio 2.5x and raise your HTT/FSB and memtest with test #5. Keep raising HTT/FSB til you start getting any error in memtest. Then adjust with more vdimm or loosen timings and test again. Repeat again and again...

 

http://www.dfi-street.com/forum/showpost.p...167&postcount=3

 

Thunda has detailed this very very well.

 

I'd like to see if you can get that ML to 2900-3000. :nod:

 

What do you have your LDT/FSB Frequency Ratio set at now?

[Guest Niki]

Current LDT is 3x, it seems to be stable, so i didn't loose it up. At weekend (again) I'm testing the absolute max of cpu and memorys. I hope to hit atleast 22mhz more to get 2.8ghz. Max cpu speed is at 1.375+123% ~1.69v 2788mhz (310x9). I'll try and write to you pm, or write here.

[z3scott]

Current LDT is 3x, it seems to be stable, so i didn't loose it up. At weekend (again)

 

You are overcooking your LDT@3x320. Spec is 800 on our NF3 boards you are at 960 which may or may not work ok but the benefit of running your LDT high is nil and it can lead to stabilty problems and interfere with the important cpu and ram speed gains you can make.

[Guest Niki]

You are overcooking your LDT@3x320. Spec is 800 on our NF3 boards you are at 960 which may or may not work ok but the benefit of running your LDT high is nil and it can lead to stabilty problems and interfere with the important cpu and ram speed gains you can make.

 

 

Should i put 2.5x instead of 3x? Or would 2x be even better? 2x320 is 640, so that would be better for my mobo?

[z3scott]

If you are going to push on past 320 drop it to 2x. That will give you headroom. It really won't hurt you performance at all but will keep you from having problems getting more speed. from your rig.

[Guest Niki]

Scott: My new PSU:s 3.3v line is "only" 3.25v @ idle. At full load, 32mPi, OCCT, Prime it's same or 3.24v. My PSU is adjustable from out side, should I rise that 3.3v somewhere between 3.3-3.8v, that is the adjust range. Would I gain more volts for my memorys?

 

P.S. I HITTED OVER 2.7ghz! 318x9 @ 1.72v, 2.5-3-7-3 @ DDR 440mhz, 3.1v. :nod:

[z3scott]

Scott: My new PSU:s 3.3v line is "only" 3.25v @ idle. At full load, 32mPi, OCCT, Prime it's same or 3.24v. My PSU is adjustable from out side, should I rise that 3.3v somewhere between 3.3-3.8v, that is the adjust range. Would I gain more volts for my memorys?

 

P.S. I HITTED OVER 2.7ghz! 318x9 @ 1.72v, 2.5-3-7-3 @ DDR 440mhz, 3.1v. :nod:

9x318@2862! :shake: Good Show!

 

I like running my psu 3.3v rail .15v-.2v over the vdimm. It won't hurt to bump the adjustment to 3.4v when at 3.1v though.

 

EVA knows ram well but please note post #11 from OCZ rep

http://www.xtremesystems.org/forums/showth...04&page=1&pp=25

 

The ram warranty is to 3.0v just so you know. Also if you run high volts 2.9v+ you should add a fan right over the ram to help cool it.

 

Here AG is running same ram but on NF4 with 3.5v vdimm:http://www.dfi-street.com/forum/showpost.p...187&postcount=3

 

This ram should run 2.0-2-2-x in the 215-230 range up w/3-3.1v. Shouldn't need more volts with those timings 2.5-3-7-3 I don't think any way.

 

To really know what your voltage rails are at you need to test them with a digital multimeter like this:http://forums.extremeoverclocking.com/show...ad.php?t=137886

 

If you are going to run higher vdimm than the standard bios you will need a modded bios like the tic tac 4.0v. There is a forum member with a collection of modded bios here with a few to pick from:http://www.dfi-street.com/forum/showthread.php?t=22031 Please note as a disclaimer you are always on your own if you run modded bios. Not saying don't... just understand that. I have run the jess modded early on when running some OCZ VX 10x250 2.0-2-2-10 @3.4v in my rig.

 

Maybe drop your divider and see if you can take the ML40 higher :nod:

 

Test the ram by dropping your multi and LDT to remove them from the equation.

[Guest Niki]

Some thing went really baad, i can't get past 300 anymore. 303x9 is max, it run's 10h of prime, but 305 doesn't go past post, it gives detectin array screen allways. I did put my ldt to 2x, like u said, but something is wrong. I hitted 320x9, got to win, then it locked up totally. Nothing happened, I did powerdown atleast 5 times, from psus switch nothing. Over 2h without battery, and it started, after that 300x9/303x9 is max, not even 1mhz more. I don't know how i'm going to get past that array issue, please help me again scott...

 

P.S. Scott thx for helping me to get my rig running, can u help me again!!!

[z3scott]

Detecting Array means you have gone too far in the overclock and need to back down. Some setting is unhappy.

 

Did you test your memory first at that 320 HTT? What was your memory speed and timings?

 

At the edge on limits you need to work in small increments of change. You also have to be careful for bios corruption. The long cmos clear was a good thing. I would suggest clearing the cmos again for an hour the boot load optimized defaults reboot and the load your default timings and 11x200 voltages and reflash the bios a couple times to "steady" it back down. Then start testing again.

 

What were your settings, what settings did you change when you hit the wall? You may need to try different multi's, dividers and voltage combinations to push higher. Since we don't have any feedback on the Turion's it's charting new ground. It's a lower power design processor and voltage may not get you enough past a certain point. Sometimes with the regular procs people push too far on the voltage and find a bit lower setting on the vcore or a different combo of voltage and % very close works better. You'll have to tinker with it but first get yourself settled down and stable again.

[jAMBAZZ]

Detecting Array means you have gone too far in the overclock and need to back down. Some setting is unhappy.

 

Did you test your memory first at that 320 HTT? What was your memory speed and timings?

 

At the edge on limits you need to work in small increments of change. You also have to be careful for bios corruption. The long cmos clear was a good thing. I would suggest clearing the cmos again for an hour the boot load optimized defaults reboot and the load your default timings and 11x200 voltages and reflash the bios a couple times to "steady" it back down. Then start testing again.

 

HMM, I would really like some clarification by you and Niki on this;

I never heard of BIOS corruption before, can you elaborate how this can happen and why clearing the CMOS over a long period matters?

 

The reason I ask is because I get boot problems tons of times with my NF3 250Gb, Detecting Array especially, just like Niki. And this even happens if im running @ 5x200 mhz (=1 Ghz) so its clearly not because of overclocking. I am worried if it could be this BIOS corruption.

[jAMBAZZ]

On another note, Niki, why dont you use you 11x multipliers? You wont gain performance by increasing HTT on the Athlon64 platform, this has been thoroughly tested and documented.. You might as well use the high multiplier and avoid problems with high HTT.

[Guest Niki]

On another note, Niki, why dont you use you 11x multipliers? You wont gain performance by increasing HTT on the Athlon64 platform, this has been thoroughly tested and documented.. You might as well use the high multiplier and avoid problems with high HTT.

 

 

Actually I didn't know that. I thought that higher HTT means more performance, and I just kept increasing fsb untill i hitted the "wall". I just tried to go as far as this rig goes, by decreasing multiplier and increasing fsb. My rig runs steady at 300x9 (2.7ghz) @ 1.65v, and most i can get out of it at 11x is "only" 260 (2.6ghz). I haven't even tried to use ex. 9.5x or any other not full multiplier. I don't know does these .5 multipliers work, but some where I read that they don't work as well as ex. 8x,9x,10x multis.

 

So to get that 2.7g i need to use 9x multi and 300 fsb. LDT works only on 2.5x, 2x gives allways detecting array even if i use 11x200 and ldt 2x. Fsb doesn't matter on my rig. It seems that detecting array issue has something to do with LDT multiplier. Like I wrote earlier, I'm n00b with a64 rigs, this is my first and i'm trying to learn everything about a64/turion cpu's.