Core i7 980x and UD7 woes.

simnol

Distinguished
Apr 7, 2010
4
0
18,510
Background:

I got an i7 980x for my birthday (in place of a car) and I instantly went out and bought an ASUS motherboard to go with it, this was a big mistake and after only booting into the OS a couple of time it developed a problem with the memtest function and I had to return it, being sick of ASUS at this point I went with a recommended UD7.

Current Situation:
Nothing... that is to say when i boot the computer nothing is displayed on my screen.

I've tried breadboarding it and I get a couple of errors when booting,

Just CPU: 29, also note that the fan doesn't spin up quickly and my current thinking is it doesn't like the CPU.
With RAM: 89 (I think) and this is with 1 and three DIMMS in, yes I have mixed and matched and am confident it's not the RAM.
With GPU 89, And again, no output.

Spec:
CPU - Core i7 980x
GPU - ATI Radeon 5970 xfx
RAM - Tripple Channel Corsair 1600
Mobo - UD7
CPU Cooler - Fenrir


Current thinking:
I bought the board thinking that it would have the latest BIOS firmware on it, I believe 980x compatability was included in firmware 3, could it be that the no-display, no post and unhelpfull error codes are caused because it is an old revision?

If so then, ow, my pride.

Any help greatly appreciated and I don't intend to sleep until I've solved this.
 

bilbat

Splendid
Just CPU: 29, also note that the fan doesn't spin up quickly and my current thinking is it doesn't like the CPU.
...with no vidcard in, 29 would make sense - 29 is "initialize video interface after reading CMOS location 14h to find out type of video in use - initialize video adapter."

...unfortunately, 89 is one of the many "Reserved/Unused" entries (actually, 86h through 8Ah); last few functions before this are:
83 save CMOS Data to DRAM
84 enable NMI, parity checker, and cache
85 initialize additional onboard controllers and USB devices
...which means - could be anything?!? [:jaydeejohn:3] If the wrong BIOS is unable to init the CPU itself - could be it...
could it be '68' upside down? That, too, could be CPU init - I think 68 is power management init...
Can you possible borrow a 920 for an hour or two for testing? If it's a 'too old' BIOS, it's always safer to flash the replacement with a supported processor anyway [:isamuelson:8]
 

simnol

Distinguished
Apr 7, 2010
4
0
18,510
Thanks for the response bilbat,

I was praying that this wasn't going to be the case :(

So if the CPU wasn't supported that would cause no display to show?

Looks like a visit to Scan and a begging session with the guy behind to counter to ask him to test it.

Thanks again.

So tired :sleep:
 

bilbat

Splendid
The whole problem of 'unsupported steppings' is a total unknown. You find yourself in the territory that ancient cartographers used to mark "here be dragons!" [:bilbat:7] I have an E0 9550 around here that was shipped with a BIOS (which, just to tick me off, 'reverts' to the 'as-shipped' whenever there's a BIOS problem) that didn't support my stepping; just to be on the safe side, I used to keep an old 775 celeron around, only to flash the BIOS back with a 'known-to-be-supported' CPU, until I had enough (good) experience with the 'dualBIOS' thing to trust it - and tried the E0 to do the flash - and found out it was no problem... BUT, you never really know, unless you can find someone in the same exact situation!

It sounds like you are sophisticated enough to have tried everything I would suggest - and it often boils down to the fact that, once you're down to GPU, CPU, PSU, (and it's damned hard to believe you could get several bad DIMMs!), you are at the point where all that remains is substitution - and if you don't have the spares, you're pretty much screwed [:fixitbil:1]

One thing I didn't see was any mention of 'beep codes' - if you don't have a case speaker, it's the best, most useful two dollars you could ever spend!
 

simnol

Distinguished
Apr 7, 2010
4
0
18,510
Yeh a short fall of mine was never getting round to getting a case speaker, looks like I'm adding that to the list for my merry jaunt tomorrow, going to be a full day :(

I wish I had the money spare to have even a 920 lying around but i've been saving for months just to get this little lot :)

Night, cheers for the insight.
 

bilbat

Splendid
Ahhh - you have pointed out a portion of the "early adopter's blues" that the average user never considers, but the 'server guys' know all too well (as they keep shelves full of 'spares' sitting around!) - a really cheap 775 is, well, really cheap - but a really cheap 1366 is, well, not-so-cheap! [:lorbat:6]
 

simnol

Distinguished
Apr 7, 2010
4
0
18,510
I'm actually a Datacentre Engineer and the shelves of spares are usually quite full as you point out :)

(Strangely the Dell R200 are some of the most unreliable I work with, silly Xeons)

I just expect things to work more often than not, guess that's my mistake.
 

bilbat

Splendid
I just expect things to work more often than not, guess that's my mistake.
Ahhh - not a mistake - "more often than not" they do work - it's just that when they don't, and you're down to the fatal three, substitution is about it! More like 'just plain bad luck'!

But one thing I can't seem to 'get', is why cases ceased to include a case speaker - if we're paying two bucks for 'em, how much can they cost a case manufacturer - thirty-five cents?!? Too many engineering-oriented companies, trying to provide engineering-oriented products, but run by sort-sighted (or completely blind) 'bean-counters' and 'money-shufflers'!!
 

dany191s

Distinguished
Apr 18, 2010
2
0
18,510
hello simnol, i have exactly the same problem as you, and with the same components (ga-x58a-ud7, core i7 980x, sapphire radeon hd 5970, g.skill 2133 mhz 6gb).
i also have this post code 89 and no beep from the case speaker !!
i think its caused by an old bios not supporting the 980x, but i have no other cpu to test !! i am totally screwed!!
please if you find a solution let me know !!!
i apologise for my english, but i am french and perhaps make some language mistakes !!
 

bilbat

Splendid
De toute façon, votre niveau d'anglais est très bon, mais mon français est très vieux et peu utilisé...

I think you have stated the problem exactly! While some CPU 'steppings' will initialize and run on a BIOS that doesn't directly support them, with merely an occasional hang or error, the major changes to 'hexacore' are probably just overwhelming. My guess is that you will have to find someone, somehow, from whom you can 'borrow' something like a 920 for an hour to flash the BIOS. You need at least F3 (released at the end of January), and I would suggest going directly to F6, which is the current release...
 

dany191s

Distinguished
Apr 18, 2010
2
0
18,510
thanks bilbat for your answer, your french is good too !!
i wiil try with a core i7 920 and will publish results here !!
 

Sevarin

Distinguished
Apr 20, 2010
1
0
18,510
I had the same issue with the exact same components. A BIOS update fixed it. Luckily I had an i7 920 laying around to actually perform the update.
 

FlashBimmer

Distinguished
May 15, 2010
46
0
18,530
I had problems similar to many of the posters above - it seems that Gigabyte went to market a bit early (or Intel didn't share enough with them until too late.) A 980X on a UD7 requires the F6 BIOS (this from Gigabyte's own tech support.) The catch-22 is that you can't do the flash if the board don't dash. For this purpose only I bought a 930 from Amazon and got the thing running, installed W7 and began moving in. I turned the machine off at the end of the day and it wouldn't start the next morning. I tried clearing the CMOS, changing out everything, removing 5 of the 6 2GB RAM chips, and still nothing. After much frustration I called Gigabyte again and was finally told to replace the UD7. Amazon's return/replacement department is nowhere near as streamlined as their normal shipping department and there was no way they could figure out how to send this for a Saturday delivery, so I have a UD5 on its way to me for Saturday delivery and the replacement UD7 will be here Tuesday or Wednesday. If anyone wants all that explained, please ask.

Was I wrong to go with Gigabyte? What does the forum-at-large think of ASUS boards?

I believe the error code mentioned at the beginning of the thread is actually 6B, not 89 (Your head tilts to the left, not the right to read the codes - why couldn't the codes have been set to read upright?) 6B was the error that seemed to clue the Gigabyte tech to the need for the new BIOS flash.

The system I'm building:

980X Extreme
UD7
Mushkin 2GB x 6 RAM @ 1600
Nvidia FX4800
Tuniq 120 Extreme CPU Cooler
Corsair Obsidian 800D Case (this thing's HUGE, but there's plenty of space to keep things cool.)

All of it to be returned if I don't start getting a smile on my face. The OS is W7 64 bit and I'll be running Adobe CS5, Revit Architecture, Revit MEP and NavisWorks on this thing. With files running upwards of 3GB each I expect this machine to be rockin n rollin.
 

bilbat

Splendid
I have spoken of this before. This is not a GIGABYTE problem, this is an industry problem, and a technology problem! Having the 'latest and grestest' 'new tech' bears with it 'teething problems' - and it has ever been thus!! The move from the first 'baby' 4004 to the 8008 was staggering; even more so to the 8080... The Z-80 was the first processor specifically designed for 'backward compatibility', and it still carried with it major obstacles, as did the Z80A - I can still remember that there were two instructions, back when we were 'hand-coding' ZCPR3, that were 'flawed', and had to be called through the BIOS, where the 'fixup' was done!

There is no such thing as a 'perfect' processor. You can't toss a billion transistors onto a postage stamp without some incipient problems... [:isamuelson:8] 'Cores' (mistakenly used, instead of its normal def as 'an execution unit'), or 'steppings', are partially due to this, and, as well, to advances in a specific chip 'architecture'. New 'steps' will often overclock better, access memory faster, and consume less power at a given clocking, due to these advances - but - every processor still has an 'errata list' (Intel updates and publishes theirs regularly as 'Specification Updates'); as it is in the nature of every [url=http://en.wikipedia.org/wiki/Von_Neumann_architecture]'Von Neumann machine' to be able to 'emulate' any other 'Von Neumann machine' (albeit, at a cost in time...), they are also infinitely 'malleable'; errors can be 'fixed' by 'workarounds', usually in the BIOS.

'Backward compatibility' can only be allowed to go so far - without it crippling the actual advances made from architectural step to architectural step in the 'evolution'. The 'BIOS compatibility' issue has always existed. I used to keep an old, cheap Celeron lying about, just for use, a half-hour at a time, to flash BIOS for 'unsupported' 775 steppings. The problem here, as I have pointed out, is not so much whether initialization will 'work', and 'wake up' the CPU to run; it's whether an 'unsupported' stepping will correctly execute the entire instruction stream required to flash the 'supporting' BIOS successfully! This is, and always has been, a 'crap-shoot'!! While I have always had good luck with GB's 'dual-BIOS' mechanism's ability to 'recover' a bad flash, I well realize I risk an RMA every time I do it, and try to avoid it whenever possible... The big problem now, is that there is no 1366 equivalent of a Celeron, which I think cost me ~$35! A 920 is a pretty pricey 'spare' for most people to swing :??:

I've often thought consumer systems would be better off set up as servers are. For most servers, an 'unrecognized' processor simply will not even attempt to 'start-up'; this is because, if you have a data-set on your server farm that you conservatively value at, say, a quarter-billion dollars, you kinda don't want 20% of it corrupted by 'errata AAJ42: Incorrect TLB Translation May Occur After Exit From C6'!! The best suggestion I have, at this point, is to look up a local 'user group', and attend a few. I don't know for sure that these are the same geeks I used to rub elbows with, 'back in the day' [:bilbat:6], but when I used to go, there were oceans of the 'clueless', and a handful of guys carrying a logic probe, soldering iron, and a particularly fast Z80A in their pockets, just to test peripheral responses. I gotta guess you'll be able to find someone who can 'lend a processor' just for a flash; if not then and there, at least 'by arrangement' [:fixitbil:1]

Again, there is no GB solution to this on-going (ever on-going!) problem, nor can one be reasonably expected! The best they can possibly do is flash boards to the current BIOS at the point they leave manufacturing. The rest is up to the 'channel'; if you buy from someone who does high volume, like NewEgg, the odds are better you'll get a board with a 'late BIOS'; but, there aren't (and can't be) any guarantees!
 

FlashBimmer

Distinguished
May 15, 2010
46
0
18,530
thanks for the reply, bilbat. You're a wealth of experience and I appreciate your being willing to wade in there among those of us with less.

I don't really mind having to do things like troubleshoot, flash BIOS, etc. It's the poorly manufactured components that get to me. I did a lot of research and tried to select manufacturers that got high points for performance as well as quality and so far the track record for all of them, to put it plainly, sucks. (ok - Intel excluded, so far.)

I received a UD5 board today to use to test the rest of my components - it's looking like I'll be replacing every single major component before this system is up and running: the i7 980X has been replaced once, though it was because Amazon sent one some parked a truck on before they boxed it up; Gigabyte tech support has recommended replacing the UD7 (a new one is on its way), the Nvidia FX4800 may be bad (this came from Newegg); and two of the six 2GB Mushkins are confirmed as bad (these came from Amazon.) I know the resellers aren't necessarily the cause of the failures, but that makes it even more exasperating - I can't just look to a bad supplier and decide not to use them again. In these cases it looks like it's manufacturing defects. I also bought an i7 930, initially because I needed it to update the new BIOS for the 980X, but I might just keep it and make a second system - the way this is going I think a spare might be a good idea to have around.

Any opinion on Mushkins? I like the latency figures, but now I'm wondering about the quality. 2 out of 6 bad? That's bad.

Even though I don't get Dell's next-day repair/replacement, all-in-all I'm still happy to be building my own again.