[quote author=macca22au link=1225569427/90#104 date=1226318935]Those of us who have owned off-shore yacht racers wouldn't call that thing of Nicks a yacht.  Gin palace would  be a better description.  But let me tell you, motor or sail, that any boat is a hole in the water into which you throw money.

quote]

As I said, I am not a seaman     thought it a tad too big to be called a boat so yacht seemed appropriate.

But I LIKE "Gin palace"!! 

Ouch.   I presume the 8800 can be clocked so I can only hope that the tech can.

Vic

I just got word that BIOS 0801 has solved some of the memory chip noise issues and is much more stable than previous releases

Its in BETA and not available at the Asus site through normal means

It may be some weeks before it is released

Until I hear back from my Asus contact I am going to assume at this point the voltages I gave you for VTT and GTL is correct based on correct electronic enginnering formulas and the fact you are stable with those settings in everything and in testing..  except 3D


Here is what we want to do next... This is the real pain in the ass part becasue there is no set value to work with. Even identical memory modules can display 2 different tweaking combos for skew so what you need to do is get it set back up as we had it booting and stable for DDR3 1800 in your OCCT and MEMTEST results and start with the following

CPU CLOCK SKEW

NB CLOCK SKEW

Here is the way to do this... Leave CPU SKEW alone and start with a 100ps on the NB. If that fails go to 200ps, then 300ps

Once you reach 300 on the NB and the stability has not changed then set NB to AUTO and then move to CPU SKEW. same deal.. 100, 200, 300

If you see no change then set NB to 100 (LOCK IT THERE) and then change only CPU to 200, then 300 then 400 and last 500... I dont see any reason to go above 500

If you see no change then set CPU to 100 (LOCK IT THERE) and then change ony NB to 200, then 300, 400, 500

If during any of those tests you start to see elongated stability times then assume the last setting made a difference and instead of moving to the next test.. increase the value of the SKEW you were working with again so in example.. if you see that a 100ps NB SKEW with a 300ps CPU SKEW definitely increased the stability, raise CPU to 400ps. If its worse then return CPU to 300 and increase NB to 200


After you pull your hair out and go insane with all that testing, and assuming you do not come across anything that makes any difference return the CPU and NB SKEW to AUTO for both and we switch to DRAM SKEW CONTROL

your targets will be slots A2 and B2 only all others on AUTO

DRAM CTL Skew on DIMM A2:

DRAM CTL Skew on DIMM B2:

Start out with quick and dirty changes... set BOTH equally to DELAY 50, 75, 100, 125, 150, 175, 200, 250

If no result next you are going to do the same thing with these two you did with NB and CPU except smaller steps

Start with A2 at AUTO and B2 at DELAY 25, then DELAY 50, then DELAY 75 ... all the way up to 250ps

If no result then swap and set B2 to AUTO and do the same with A2

If no result then set A2 to DELAY 50 and start working B2 from DELAY 25 and up and reverse that if no result

Vic this is pure trial and error and there is no science behind it simply because they have not invented a oscilloscope to come with motherboards so the user can see the noise and trim in real time






Once you go through all that if you see absolutely no change then there is nothing else I can suggest except wait for 0801 BIOS



or avoid all of this and buy DDR3 1800 modules

But do be aware 1T at 1800 is hit and miss on those Corsairs and at that speed you really do not need 1T

Not sure if what they are passing around is the latest beta of 801 or not.. but use it for now

2 1/2 hours FSX with no problems!

Think I'll leave it this way and see what happens.

Fingers, eyes and toes crossed!  

I can google for various ways but is there a particular way you would clock the 8800GTS 640?

As usual - mucho gracias!

No prob Nick. I agree and I spent the time reading a LOT of stuff on how the parts tie together. Can't say  I understand technically but I have a heluva better overview of the process than I had before. Sort of like a lock - all the tumblers have to fit just right or it's no go. I'll give it a rest for tonight and get started on the card tomorrow.

2 1/2 hours in FSX to me is a good bench since that's the program I'm mainly interested in. Didn't bother with prime95 blend.

Now we wait and see for a bit.

BTW, what do you use to stress the vic core? 3d Mark?

macca22au wrote on Nov 10^th, 2008 at 7:08am:


im assuming you mean GTX 280??

well you can get one of those for $550 in aus. http://www.staticice.com.au/cgi-bin/search.cgi?q=gtx+280  whenever finding out a price in aus, go to staticice, its good .

Ha! I figured you were right.  Could easily tell the 0801 BIOS was very different - everything came together as planned.

So far so good - I've noted a few anomalies but think they were due to my clocking the video card. After testing with ATI Tool, I've dropped it down considerably from where I had it originally.

Thanx for the formula - got it jotted down for future reference.

I don't normally fly FSX up at 30K but I am amazed at the clarity I'm getting. I don't pay attention to FPS but it is smooooooth!

BTW, if you ever get time to fly FSX for fun, there is a freeware a/c that is second to none. The SectionF8 F-86 Sabre (for FS9) - easily ported to FSX.