790 corruption issues?

[Roger_D25]

There I go again speaking when I have no idea what I'm talking about (in regards to relationship with Asus), sorry guys! I also had no idea Dinos22 that you were just getting back from long journey, next time I'm give you a few minutes before berating you with so many questions With that said thanks a bunch Zowner for explaining those things, much appreciated! I believe it was already posted (by yourself Zowner) but its good to know I'm not the only one having no luck with tRFC (been trying for weeks to lower it with no luck what-so-ever)! I assumed it was because of some "other" setting but it looks indeed to be a bug?

[BenchZowner]

It is a bug.
With the very first BIOSes it was working fine.
But then again... things can go bad in the BIOS Eng. department right ?
For the time being I'm stuck with MemSet... but this BIOS bug has ruined some of my benches.
Forgot to adjust tRFC with MemSet before running some benches

I will be giving Asus some feedback soon

[dinos22]

i was at Nvidia editor's day in the last few days and i spoke to a few online news site editors >> i did not have a single one confirm that they had any abnormal corruption issues so it's probably not as common or they just have no idea what they're talking about

every manufacturer i've been involved with has the same procedure when it comes to hardware problems like this......they need confirmation from a few sources that it is withspread and then they need to be able to replicate those problems in their test labs so that they can first understand and second fix and retest the boards

[gmcg]

i was at Nvidia editor's day in the last few days and i spoke to a few online news site editors >> i did not have a single one confirm that they had any abnormal corruption issues so it's probably not as common or they just have no idea what they're talking about

every manufacturer i've been involved with has the same procedure when it comes to hardware problems like this......they need confirmation from a few sources that it is withspread and then they need to be able to replicate those problems in their test labs so that they can first understand and second fix and retest the boards

While they do it, competitors don't sleep BTW where did you get F1 EE?

[saaya]

saaya pls let ppl that actually have a problem test the board and report results.....

how am i keeping anyone from testing their boards or posting results?

if it has not been resolved and i can and will also confirm this when i get a chance....it will be reported to nvidia MCP guys direct from my end

thanks for your efforts man

very nice post bill!
so the corruption problem is completely fixed now for you with 0701?
and performance? is it a lot slower than other bios versions?

could you give me some tips for tweaking/configuring the 0701?
i could add that in the first post of this thread

[hifiking]

I have experienced the same as BenchZowner, I had corruption and SLI problems in the start, but after pushing things to the max they are gone. No corruption for over a week now, and SLI is rock stable. EVGA P04-05

[slim142]

Over here, 100% trouble with P1 and P2 enabled. However, when in Auto, 100% rock estable w/ SLI enabled.

[BenchZowner]

Will update with more details later on.
P.S. Flashed to 0703 just before I hit the bed, gonna give it a try later on.

[slim142]

Will update with more details later on.
P.S. Flashed to 0703 just before I hit the bed, gonna give it a try later on.

Hey BenchZ
Any trouble while running with P1 and P2 enabled when you were on 0701?

[BenchZowner]

Nope, P1 & P2 always enabled.
But with P1 & P2 enabled, depending on your FSB you need more chipset voltage.

[nachthymnen]

E8500 @ LN2: no corruption problems




http://service.futuremark.com/compare?3dm05=4284144



And my best

http://service.futuremark.com/compare?2k1=9380307

[slim142]

Nope, P1 & P2 always enabled.
But with P1 & P2 enabled, depending on your FSB you need more chipset voltage.

For 450mhz FSB, what would you recommend?
Right now I have 1.44v in BIOS and according to PC Probe II it is 1.46 Real.

Im running out of ideas on how to make this thing estable with P1&P2 enabled.

[BenchZowner]

Can't really help you out this time.
I haven't bothered testing the SB & NB voltage "requirements" yet.
Since I'm pushing it to the limits I'm using high voltages which will more than likely be more than needed ( currently using 1.65V for SB & NB ) at 442MHz FSB

[gmcg]

E8500 @ LN2: no corruption problems
And my best

You know, nachthymnen, I have exactly doubled all other guy's setting in my system, he runs 9650 24/7 at 4.5, as he said, and I can barely boot in Windows and get cpu-z validation, all tests failed. It looks that stability issue is different on different 790i motherboards, and even more, it is random - for example I got my computer frozen after 2 hours, someone else reported the same after 14 hours!!! or running Prime. Today you get it as soon as you turn on your PC, tomorrow the whole day it runs without any problem. I moved back to default settings to be stable enough for now waiting for Foxconn Blackops.
You are a lucky one with your board ... Thats why people with stable 790i based systems couldn't understand others.
With 1T on memory my systems was absolutely unstable (BSOD, restarts, etc), with 2T I got it more or less stable with random freezes only.

[slim142]

Can't really help you out this time.
I haven't bothered testing the SB & NB voltage "requirements" yet.
Since I'm pushing it to the limits I'm using high voltages which will more than likely be more than needed ( currently using 1.65V for SB & NB ) at 442MHz FSB

Oh well thanks anyways
Are you planning on running the system on air for personal use? You might be able to help us...

[BenchZowner]

Will do m8.
Just gotta get some things done first.
When I start the "normal" measurements for the review I'll be examining stuff thoroughly, like the required voltages for every "configuration", etc.
When I'm done with the review it will be used as my gaming rig, so 24/7 stuff will come.

[BenchZowner]

PM me known bugs (at least other than tRFC, which worked for me iirc :|).

That's weird.
tRFC doesn't work for me and 2 people ( haven't asked any1 else ) with most BIOSes ( 0601, 0602, 0603, 0701, 0702, 0703 for sure ).
Try setting tRFC to 36, 48, 52, 60, 72, 80 and you'll find out that it sets random numbers each time.
The same symptoms occur to any other number that I've tested, just told you these ones because I've used those for sure.

[slim142]

Will do m8.
Just gotta get some things done first.
When I start the "normal" measurements for the review I'll be examining stuff thoroughly, like the required voltages for every "configuration", etc.
When I'm done with the review it will be used as my gaming rig, so 24/7 stuff will come.

And hopefully that will be soon

[dinos22]

guys i cannot test this bios because my board is dead
i killed it on the last LN2 session it seems as i have a stuck FF code
i did have that before and the board revived but not this time
replacement is DOA as well...i'm really unlucky here
i'll change PSUs maybe that's what's causing it donno as i'm using X900 seasonic atm as it was handy(in case of incompatibility of some kind)......will try HX620 sometime soon

[slim142]

That sucks dinos
hopefully you can get a replacement soon

[bingo13]

i really really dont like that FAQ...
it sounds like only anandtech had this problem, and it sounds like it only happens if you try really strange fsb/dram combinations that dont really make sense to run... and they say 1600 is out of spec? but at the same time they advertize 790 as ddr3 2000 capable?

and they say a bios was released that fixed the problem.
thats not true... they make it sound like the issue was fixed a long time ago, when its still happening! the situation has improved, but the problem is not solved!

That text is for the P05 BIOS that was released last week. We had it earlier than that for testing and it resolved the majority of data corruption problems we could replicate on a consistent basis (400/1600 was repeatable). I am still testing it, but for the data corruption problems that were consistent in our testing, they are fixed. I am sure there can be other instances where data corruption can still occur because the GTLRef/VTT tables are still not quite correct.

ASUS, starting with BIOS 0703 has implemented some of the base code from NVIDIA, a full update will not occur until next week from all indications. With 0703, increasing SPP and MCP voltages slightly while manually tuning GTLRef helped solve about 95% of the repeatable corruption problems we had.

As far as the first line of the text, NVIDIA knows other users have had this problem, our name is on it as we seemed to be the only "website" that reported our findings to them. It appears that telling the emperor he is not wearing any clothes is difficult for some. I find it funny that certain websites dismissed our claims and a few hours later, NVIDIA admits there is a problem and releases a BIOS update for it.

[OBR]

i am playing many days with stability and GTL voltages, thousand of restars but NO CORRUPTION! With 0703 BIOS on SiiEX

[bingo13]

You missed the relationship a bit
Dinos is "helping" Gigabyte

That might explain why their board is three months late. Sorry, but that was one large opening you left for us.

[dinos22]

That might explain why their board is three months late.

now now Gary

[saaya]

thanks for the feedback gary!
i apologize for my rant about nvidias faq then...
it seems they really did a good job and fixed almost all of the corruption issues, at least at the common speeds.

finding the random, hard to repeat problems is a real btch...
i hope nvidia doesnt give up and keeps working on it, if they can iron out those problems 790 or a chipset refresh might go into the books as the best ddr3 chipset before it all went IMC

That might explain why their board is three months late. Sorry, but that was one large opening you left for us.

LOL