Well, I think we located a few problems with the skew tables last night. I have BIOS 1102 now, will post it up once I make sure it does not melt the board down. ;)
Printable View
Well, I think we located a few problems with the skew tables last night. I have BIOS 1102 now, will post it up once I make sure it does not melt the board down. ;)
Lol, according to unspoken amd subforum rules you are required to give up the goods regardless of any restraints you are under or damage it may cause :D after all we are all responsible adults, If we blow up boards we will just rma them :shakes:
Now on a serious note bingo, all I did was set known good timings/clocks, fooled with dps , settled on normal. Havent touched drive strengths yet with this ram all auto for now, reserved for fine tuning.
Looking forward to it :)Quote:
Originally Posted by bingo13
The skews has acted quite strangly for me. If set to auto, on boot it will switch between normal/150 delay and 300/300 delay, on random :shrug:
(thats with 0903, will give the new one a try tonight)
BIOS 1102
I have not thoroughly tested it yet, so "Danger Danger Will Robinson" rules are in effect here.
Will give it a shot tommorow and report anything odd I find..
1. Still have a slight Core VID overvolt (+.0125) situation with Auto, but manual settings over 1.4V seem to be holding true again.Quote:
Originally Posted by chew*
2. Best to set manual settings, reboot, go back into the BIOS, save, and reboot again. I had the memory settings change to SPD settings a couple of times on me, not every time, but something to check. The same held true for HT link speeds, auto wanted to default to 1600 on the 720BE once NB speeds were raised above 2000, not every time, but enough for me to tag it.
3. Northbridge speeds and voltages are back to 0801/0903 levels.
4. Auto settings seem to be dialed in again on skew and delays.
5. Will test more later tonight.
Good to see you and Gary working together on this board here. It's only the enthusiasts who can get these changes put in to play. :up:
they should get paid:ROTF:
honestly they should use that info (tested by MoFo enthusiasts) in their marketing! It would be excellent example of truth telling marketing!Quote:
Originally Posted by EnJoY
imagine chew*'s avatar on ASUS box :D
Thanks for the tips Bingo. Im at 2585 NB with 1.4v cpu and cpuNBv on auto now. Ill get brave and try again soon but a few corrupted installs have me chillin for now. Im using 1002 bios atm. Cant get AOD to work worth a flip in Beta 7 :(.
Eh gary's the man here. I have patience but from having done it in the past running through multiple os's is grueling......I'm just reporting stuff thats hindering my path to ocing :p:, in all fairness he probably noticed all that stuff allready.
Bingo, have you put the board under sub zero with any of these bios's? If not I can probably tackle it this weekend see if any anomalies appear when cold.
Pick the chip, 720 or 940......you probably know which one is more likely to show issues.
gigabyte needs to add chew* tested mother approved to their box....that board is one tough cookie for a budget board....I really put the screws to that board to get it to perform up there with the more expensive options....Quote:
Originally Posted by Nedjo
So, overall, worth a try? Any idea what ACC related changes were made in 903? They seem to be no good for my chip, which is helped greatly by ACC on the 801 bios. My M3A78-T had a bios update that mentioned a similar fix, but didn't seem to harm my OCs.Quote:
Originally Posted by bingo13
--Matt
That would be great. I sent over a board and ummmm a special chip to my sub-zero guy for testing. I have no idea how the BIOS will react at this point but do know that 0801 was tested at sub zero in Taiwan. I will be tardy on updates today, spent most of the day getting the new MSI and DFI boards up to speed after a few BIOS changes last night.Quote:
Originally Posted by chew*
I have not tried ACC on 1102 yet, but I did request that it work again. ;)Quote:
Originally Posted by mattkosem
Excellent :-) I'll give it a shot tonight.Quote:
Originally Posted by bingo13
--Matt
allright, I will toss the 720 in as I have a feeling that will be the problem child of the 2 retail parts if any problems are going to show themselves.Quote:
Originally Posted by bingo13
Ok definitely having issues with prime and 1002, seems it was easily benchable, not so prime stable however not even with NB downclocked.
Having trouble getting prime stable with either bios, seems to need more volts than my 790gx and then temps become an issue, may need to ditch the scythe ultra kaze and dice some carrots.
http://www.heatsinkfactory.com/delta...igh-speed.html
*EDIT* Well I think I figured it out, this board has some preety wicked v droop......1.6125 = 1.5875 ( what I need for 3800+ and prime to pass ) Will verify with a multimeter, just need to find the read points........I think this is a volt mod candidate.
With almost twice the PWM phase of my 790 gx thats kinda :/
I can confirm the vdroop being rather significant. I wasnt sure if it was Heat or my PSU to blame. I "assumed" with 8 phase the board wouldnt even be an issue ! I cant go over 1.4v without issue atm. Bios 1002 atm, didnt notice this with 701.
Woot! My ram is stable at 1200 in the 1102 bios! The skew values that auto provided are still way out there, but setting them manually to 300/150 works on this one as well as it did on 801! 1102 seems to bring some extra memory bandwidth with it clock for clock and timing for timing as well. My everest read speed in unganged mode is up 200mb/sec over the 801 bios. Clocking tests next.
--Matt
Matt i'm wrapping up a prime stable 1 hour run. I will take a shot of bios to give you a ballpark...for 3800+ prime. Boards sort of finicky.
My luck seems to be consistent with yours. My clock is stable at the same level that it was under 801. I've passed 4 iterations of linpak at 3.7 so far, which is about 8x as far as I was able to get on 1002. Not a hint of memory instability at 1200 either.Quote:
Originally Posted by chew*
*pats Asus on the back for fixing ACC and memory clocking*
--Matt
Well you gotta wait for another hour, downclocked the NB, 2800+ just not gonna happen on this bios already test 2600 + and that was fine. Tried to compensate for vdroop and just no luck, benching seems solid but 24/7 stability just isn't up to snuff yet.Quote:
Originally Posted by mattkosem
I see .01v over requested under linpak. My board never drops down to or below the requested voltage as far as everest is reporting. What tool are you using to check voltages?Quote:
Originally Posted by chew*
--Matt