will do im starting at 401 and moving my way on up to see where it's stable.Quote:
Originally Posted by raju
Printable View
will do im starting at 401 and moving my way on up to see where it's stable.Quote:
Originally Posted by raju
so where is all the noise this board was supposed to be making i dont see hardly anyone talking about this baord now..
hmm makes ya wonder... :rolleyes:
hi lestat,
yeah it's a curious board in some regards, it can bench and boot up quite well at reasonable fsb's over 420fsb. Stability however is in the eye of the beholder and each to their own. The problem I'm having is that my usual testing methods of 8 hours of Orthos (blend test) are not working too well with this board. I've used all the bios tricks and XP methods I know.
Many users are happy with super pi stability as long as it allows them to run their systems for everyday use (which is fair enough and I never argue with that).
For the larger audience here the board is fine as a lot of people have different demands and XS is really about pushing to the max.
Some of the people I deal with have different demands which is why I have to be a little more stringent in my testing, it's horses for courses to be honest.
The bios is definatley jerky in comparison to some other boards I have tested with same methods....
you often find that the real core user results start to flood in when the hype and dust of a product settles. A lot of people I know are happy with 3.6ghz overclocks that can be held unconditionally for 24 hours. There are no real hardware tests one can run to guarantee that success in all it's elements. However I find a broad range of tests - superpi (dual 32m), Orthos (8 Hours blend stress), S&M stress (100% load), 3d marks and gaming) help you get an idea.
It all depends on the audience and demands I guess..
My post here is just to confirm that my findings are not unique to my system, therefore pushing Abit a little for a better bios for the whole overclocking community and not just a single forum of users..
regards
Raja
OK so not so good for prolonged crushing tests... well hoping for a BIOS to take care of it. BTW, I have gone ahead and sold my P5B-DLX (430X9 done). Ordered ABIT from EWIZ.COM and it will take sometime for da board to fly few thousand miles and come to me. Gonna get dat water-cooling going once again.
i just ordered this board through my local supplier.. although i'm wondering whether it can beat ocability of my biostar 965p deluxe, but i hope this board is at least stable @ 400x8 coz i can run my E6400 at stock vcore at that speed with the biostar :D
Quote:
Originally Posted by irenic
It all depends on what you call stable my friend.
What we need is users to provide screenshots of dual orthos blend test if they can for 8 hours, just to see if it's specific ram, multipliers or even cpu's that do better or worse. I have contacted ABit and await a reply...
Like I said for the core users at XS it's no big deal as pushing superpi and other benchmarks may be the desired action, but in other circles people have different criteria and tests. So the board is still good no doubt. I just think it needs a better overall bios, which ABIT can provide us far more effectivley (hopefully) if we have lots of results from different users highlighting the issues....
It's the users that hold the power, if all the hype and bechmarking is in one direction, it takes a while before the issues start to surface. In the past, we have seen many boards not do so well in the long run, as people work against each other rather than together. I have my own forum which has only just launched, but as it is of small membership, there is no way I can compile results like I can at XS, that's why I'm asking everyone to chip in if they think it's worth trying, (i never work against other forums as seems to be the current worldwide fashion). Either way it at least presents a lesson to the manufacturer for future releases.. (though ultimatley I'm probably wasting my time and the better side of my nature by trying)...
regards
Raja
raju,
Its nice to see I'm not alone with higher demands of my hardware. Super PI is nice and all but my bare minimum is 4 hours orthos stable for me to be somewhat satisfied with an overclock.. My chip was 12 hours orthos stable with 1.475v set in bios & as much vdroop as 1.440v. I can't get the same chip to do 400 X 9 past 2hrs on the abit using 1.50vcore even with the board overvolting. My testing of this board tells me you have to throw out all your previous settings when overclocking this board because instead of the vdroop being vcore, on this board its the mch. Then as you up the mch the passive cooling becomes the problem. I've ordered the Microcool Northpole XE & Microcool ChipSinks, the Cooler Master Blue Ice Pro, & the Swiftech MCX159-CU Extreme Duty w/ 40mm fan. I also ordered the HSPC Techstation. I'm going to remove my board from the case, change the chipset cooling and remove heat as the limiting factor for overclocking this chipset.(I'm beginning to believe my asus faired a bit better due to the fan over the mosfets) Unfortunately I have but two kits of ram, one is a 1 GB kit of Corsair 5400C4 rated 4-4-4-12 that have no trouble running PC6400 @ same timings(Ram Guy @ Corsairs IC inquiry thread stated the sticks have Micron-ST DDR2-667CL4 Rev D Die's) & G. Skill HZ's which I believe are based on the same micron revison D dies.
Okay I'm rambliing, what I'm tyring to say is, I'm not happy with the overall performance of this board, moving it to test bed and going back to the P5W DH as my gamer. I will get some results posted for you Raja if it passes prime for 4 hours.
Quote:
Originally Posted by The Nemesis
thank you, please try for 8 hours if you can, though 4 will be great if that's your thing (i'm flexible and understanding). Make sure it's blend test (stress cpu and ram) I use orthos as it seems to do everything neatly in one package...
At least you guys are willing to help for the benefit of all users...
respect
Raja
The Nemesis and Raju,
the same holds for me. couldn't get it past 3.5 hours double prime so far. I've increased the vmch to 1.8v and priming now. But I also loosen my ram timings to 3338 from 3334 at 2.45v. This board seems very sensitive to ram settings. Couldn't get it to run 4:5 unless under 900mhz for the rams; seems smoothest with 1:1.
The board was tuned for 1:1 performance via Gary Key (anandtech) and Abit for d9 series of ram, he did a lot of work with Abit on this board pre-release, even under his warning they decided to push it out into the market, knowing it was not completley ready for the max badge, other than some single superpi runs and a few dual super pi runs by users here and Gary, that he managed to get going quite well. He had to get some of his ram timings reset by the vendors and they worked a lot better after that.
After that the bios went out via a couple of forums and we all started to hit some nice super pi times with tight timings (myself included). In order to complete my review for Abit I needed to run Orthos stability tests too as there are people who need a broad range of tests rather than just a single type to buy a product and consider it's intrinsic worth. During the Orthos runs I found the max fsb was a lot lower than I had expected considering the super pi abilities, Gary had already made me aware of the 1:1 divider being the best. Despite the trend of bashing reviewers I always try to encourage them to share as much as they can with us, and I've always found Gary to be helpful in this regard.
sorry for my rants here, I just felt there was a lot to be cleared up...
regards
Raja
Thanks for the info, Raju. If I managed to prime for 8 hours, I'll post here again.
This may not help much, but hey, something is something.
Just ran X6800 @ stock 11x / 272fsb (this is what this board defaults to when all is on auto) with 1.32V (reads 1.35 in windows/uguru and which is the lowest option available in bios), mem Gskill 2GBHZ 2:3 with 2.05V for 8 hours prime95 stable. Ran 2 instances of superPI 32M and all was stable.
Reboot and set 13x multi for 3536MHz with 1.32V (1.35 in windows), boots straight into windows; run superPI 32M no worries, manual reboot back into bios set 14x multi 3808MHz with 1.32V, boots straight into windows then reboots to a black screen. Manual shutdown, restart and go back into bios and set 13x with 1.32V and boots back into windows. Run dual prime and second core fails after 2 minutes. Back to bios and set 1.35V (1.38V in windows), back to dual prime and core 2 fails after 15 minutes. Back to bios and 1.375V (1.4V in windows) and core 2 fails after 35 minutes. Restart dual prime and core 1 fails after 10 minutes :nono: Temps were hitting low 50's under full load (air cooling with 28C ambient temps) so I didn't want to put more Vcore through it till I get a better heatsink (still be air though, Iv'e had it with water, too much of a hassle).
Point is superPI means JACK in regards to stability. If an oc can't pass prime/dual prime or looped 3DMark for hours on end the it isn't a stable oc. My rigs for gaming and it needs to be stable.
Also I feel this board has some ram issues, be they spd programing/compatibility or whatever. It needs more tweaking on the part of Abit's bios writers to iron out the bugs and get this thing running to it's MAX potential.
I am using the beta 13 bios which gives all the multies but atm I'm unfamiliar with Intel ocing and the 975 chipset and it's settings having just come back from 2 years witn AMD. Never been an extreme ocer anyway so just disgard all that I have said if in fact it's nothing more than hog wash.
thanks for all that info Lowdog, it's helpful and confirms my findings.
I think what has happened of late is that many of the guys here at XS who hold the drive for marketing and presenting new products, are such good overclockers that they have different criteria than some of the more relaxed guys (who still make up a very large part of the market outside XS).
What ultimatley happens in some cases is that products are brought based solely on single super pi runs and quick fsb shots, which does not always reflect what all of us need to see, but we all get excited by the results regardless.
I see the viabilty of both approaches in the market as they both sell motherboards. But we need to try and work together to get things up to a standard so that people get their money's worth... I hope I'm not offending anyone here with my very open minded views...
regards
Raja
Raja you hit the nail on the head!!! It definitly has memeory issues.here are my results with this board. My highest orthos stable(8hrs +) oc is 11 x 355 my highest 1:1 FSB stable is 392fsb. Now using OC guru I can boot into windows at 440 fsb 1:1 and play games run superPi , sandra for hours but this fails othos in 2 seconds every time. Anything over 400 fails othos in 2 seconds ,I have tried 3 sets of ram. LowDog where did you get beta 1.3 with the unlocked multi for the 6800?Quote:
Originally Posted by raju
Why is it that I can only run my ram at 1:1 at DDR553 with strap setting of 1066 giving me 272MHz. This ram is DDR800 rated for 400MHz :confused:
If I run it with the 2:3 divider it's 400MHz but gives worse bandwidth scores than 1:1 @ 272MHz.
Shouldn't 1:1 be the DDR2800 setting? Sory but it's all a bit beyond me atm. :(
AFIK it's the 1066 strap, to run 1:1 you have to use the ddr533 memory settting. to run ddr667 1:1 you need a 1333 strap like some of the badaxe boards. do you have a link to the 1.3 beta?
Tony has a guide somewhere which you can look up bro'. There are many things which decide the dividers and there overall abilities and there are obviously issues that both ram vendors and mobo makers need to work out on this new platform. The bios auto sets a lot of the chipset latencies and secondry timings based on fsb and divider, Tony and his friends put up a very good article on this over at bleeding edge forums...
regards
Raja
I'll check with who I got it from and if they say it's ok then I let you know. It was given to me in good faith to not disclose the source, so at this point in time I'll just have to wait till I get the ok as to whether I can divulge that info or not. Sorry for now.Quote:
Originally Posted by Bootsy
Thats cool, at least I know they are working on it and it close. Thanks lowdogQuote:
Originally Posted by lowdog
i had no problem running 10 hours small FFT and 10 hours blend at 400X9. Havent tried any higher though...
Hi Budwise
Could you please try blend test for me over 400fsb? for some reason it won't work as well (not for me anyway). Again I understand it's not your citeria for stability and respect that...
regards
Raja
If you mange to get a lot higher than 400fsb for 8 hours or so on blend can you please tell us what cpu, divider and ram you are using, so that we can see any patterns (should they exist)...
i'll give it a go when i have some time to mess with it. Today is my 8th day in a row to work! YAY
now can somebody tell me about those different clocks? is it a bug? or can't i run my e6700 with 6-9x O_o
http://i65.photobucket.com/albums/h2.../abit_11ef.jpg
thanks my friend, i hope someone can pop the 400fsb or so on blend, looking forwards to your screenshots.
thanks for taking the time in the first place...
regards
raja
Quote:
Originally Posted by Keiki
that is strange they are all reporting different speeds, is c1e and speedstep switched off in bios?
regards
Raja