I have found my max speed at default vcore

[Duvie]

I basically left everything stock settings (1.396-1.408v actual) and just set a divider of 133 and auto setting with 2.7v....all at 1t

I did at least 10 passes of memtest windows (sometimes more) 50 passes of test #5 and #6 in Dos Memtest...and at least 12 hours of prime95...

I need to run a 100 divider (forgot about it......

133 divider (which were 2.5-3-3-6) = 355ddr

9x276=2484mhz was the best I could do....

166 divider (which were 2.5-4-4-7) = 446ddr

9x273=2457mhz

200 divider (which were 2.5-4-4-8) = 544ddr

9x272=2448mhz

I was able to run 200 divider pass the memtest at 274 but it failed prime in 3 hours and looks like it needs a touch more vcore to run at those speeds (vdimm to 2.8v didn't help as these TCCD chips like less vdimm and seem to overheat with 2.8v, IMO..no heatspreaders!!!) so I dropped the HTT down and so far I am passed the 3 hours so it may have helped but I am 4 hours away from 12 hour wall.....


this was all done with 1.41mod bios.....I want to try 1.62 next....


So basically I had to come down 36mhz in clock speed to be able to run 200 divider...What I will be showing as I find the limit for the next plateau 1.45v is that this spread will become larger...

Stay tuned....I will try to run some benches (a limited suite of test) at each of the above speeds to see if the trade-offs are worth it....



-------------------------2448 (200)---------2457 (166)--------2484 (133)-----xxxx (100)

Superpi 8mb------------6:05-----------------6:15---------------6:23------------
Cinebench 2003---------346------------------345----------------349------------
Winrar-------------------2:26-----------------2:27---------------2:25------------
ADT2004------------------1:16-----------------1:23---------------1:26------------
TMPGenc----------------2:18-----------------2:21---------------2:21------------

All test were ran in same order and with one fresh reboot at start.....x:xx is time format and lower is better...


UPDATE #1

200 divider all passed extensive memtest, 12 hours of prime, and benchmarks

9x272=2448 (544ddr) cas 2.5-4-4-8, 1t 2.75v (1.39v actual)

9x279=2511 (558ddr) cas 3-4-4-11, 1t 2.8v (1.44v actual)

9x285=2565 (570ddr) cas 3-6-6-12, 1t 2.85v (1.49v actual)

I couldn't get any further regardless of 2t or vcore...It just either needs more vdimm and/or better ram cooling... Edit: vcore didn't make a diff....all the way to 1.6v....


-------------------------2448 (200)---------2511 (200)--------2565 (200)-------2583 (183)------2619 (183)

Superpi 8mb------------6:05-----------------5:57---------------5:57---------------5:56-------------5:47
Cinebench 2003---------346------------------354----------------361----------------361--------------369
Winrar-------------------2:26-----------------2:23---------------2:19---------------2:19-------------2:17
ADT2004-----------------1:16-----------------1:16---------------1:20--------------1:16-------------1:15
TMPGenc----------------2:18-----------------2:15---------------2:14---------------2:14-------------2:11



Update #2 (183 divider using A64tweaker)

183 divider[/b] all passed extensive memtest, 12 hours of prime, and benchmarks

9x287=2583 (516ddr) cas 2.5-4-4-10, 1t 2.8v (1.49v actual)

9x291=2619 (522ddr) cas 3-4-4-11, 1t 2.85v (1.54v actual)

9x29x=testing know cas 3-6-6-12, 1t 2.85v (1.57v actual)

 

[qbackin]

:thumbsup:

 

[theMan]

:beer:

 

[NINaudio]

Nice....looking forward to the updates.

 

[Duvie]

updated...testing 100 divider now....auto has my memory timings at 2.5-2-2-5...interesting for sub 300ddr.....

Update: 100 divider would not even run at same as 133 divider but mainly due to what I believe was poor auto settings of 2.5-2-2-5....It is rumored with this board that when the trcd is 2 at bootup it actually undervolts the vdimm by .1v...So I would possibly have to rerun it and take the vdimm up to 2.8-2.85 to really test it.....Never mind I will test it more when I start to peter out in the ocing...

This thing is going to the end and that may take a good week or 2 to get it all figured out....

starting now to find the 1.45v top speed for each 133-166-200....stil using auto settings on ram....

 

[Duvie]

major update 31....

Since defult vcore was done I tried to run out 200 divider...

1.41 and 1.62 bios seemed to have the 166 divider fubarred as I couldn't get it stable at 1 lowsy HTT over the highest 200 divider even though the ram was like 100mhz ddr slower....I falshed to 1.8 bios and I tested fine at 2592mhz with 183 divider (519ddr cas 2.5-4-4-10, 1T 2.8v)

More to come...Analysis at the end....

 

[n7]

Keep us posted

I have a feeling i will have a ton of questions for you when i get my MSI K8N NEO2 Platinum & Venice

 

[Duvie]

I can say that the temp problems do not seem to exist for me with the 1.8 bios...also the temps in the bios sem to be in line with windows now....

A new DRAM STRENGTH feature popped up and I am confused about its effects....Also I am finding that 2t has little effect for me whatsoever....just deosn't seem to add anything in terms of running a divider out further....

I think 1.8 seems to have better results for the 166 divider but really didn't add anything over the 1.62 bios in terms of the 200 divider...

 

[Dadofamunky]

Is that a new 1.8 BIOS? I recall the beta BIOS I tried out had the temp reporting bug.... Any linkies?

I find the 166 divider to run like a tank but the 133 crashes every time in otherwise similar settings....

Yer results with 2T are not a surprise, anyway....

but 2.5G stable on a 3000+ ain't shabby in any way or form.

 

[Duvie]

I am not getting any temp errors now though I could have sworn I did th efirst time....Same 1.8 bios I linked in a thread I created about 2 weeks ago if you search...do " Neo2 " and "Bios"...

The 2t just doesn't appear with my modules, and the 1.41, 1.62, and now 1.8 bios to give just one more HTT over what I can get with 1t....

133 is a tank but 166 was up until 1.8bios mirroring 200 divider. The 166 would go about as far as the 200 divider at same HTT....It seems inconceivable when I hit the end of the stability with the ram timings at a certain vcore that I could not drop down to 166 and get stability...I mean if 200 was enough to run some time of memtest, prime95 (hours before I stopped it)...by no means BSOD flaky... then I would assume a drop of near 100ddr should have been conceivable. I ruled out vcore and vdimm so it was clearly the ram divider and timings.....Now new bios and I have been able to find stability with 166 as well as 183 thru A64tweaker...

I have finished 9x288=2592mhz w/ 1.49v: 183 divider @ 518ddr 2.5-4-4-11, 1T (2.8v)...will do a fresh restart and benches at that speed later....

 

[mUcHiLuS]

hi, i'm curious, about your setting, i'm already oc my a64 3500+ venice core up to 255x11 with 1.600v good enough,. but did you try to reach 2.9? because for me is almost imposible and let me tell you that i have a extreme good cooling (water) custom made, and good memory ocz vx pc4000 running 255 2-2-2-6.

Let me know any sugestion!

 

[Duvie]

I can post into windows as high as 2.75ghz with 1.6v.....I can post into bios as far up to 2.8ghz....

 

[Duvie]

Update #2 partial.....Doing testing now at 293 with 183 divider cas 3-5-12-5, 1t 2.85v.....

The way things are going versus my original prime testing I think 1.57v will peter out buy 294HTT if I can hold 183 and that will likely be a 3-6-12-6 incarnation....


Once I am done finding my vcore limits with my highest ram clock I am at those same speeds going to try for 133 divider and then tighest timings...it appears to me that winrar likes bandwidth yes but it may actually like timings as well...All my sandra scores are in the 7000's and have been for most of the time ranging from 7000-7300....

Basically I am going to have a whole bunch of data and then maybe I will relook at how to present it so ppl can see what effects ram timngs had..ram bandwidth had, ram ddr speed on OCs, etc...As well as get a clearer picture of how a OC starts to slow down as one starts to overvolt....

I will be running to 1.61v for my testing...temps with 1.8 bios are about 1-2c lower at idle but 1-2c higher at load....So I am seeing highs of 52c right now....