John_VanKirk

Distinguished
Hi everyone,

I just finished building a brand new rig, using a Gigabyte GA-P55-UD7 MB, (ver 1.0), an Intel i7-875-K CPU (which is on their approved CPU list), Corsair Dominator 8GB (4X2GB matched 1600 C8 RAM, which is on their approved DDR3 list), a new XFX Radeon XFX 5850 video card, & Silverstone 1000 W PSU.
On POST, get an error code 26 h. No beep code, and since it does not get to the System Setup, I don't know the Award BIOS Ver #.
Sent an email to Gigabyte Tech Support but no reply so far.

Indicator LEDs OK on Soft OFF. High Utilization MP, VP, CPU Phase LEDs on power ON.
Gigabyte web site says i7-875 requires the BIOS F5, which was posted 5/14/10. I suspect there is an earlier BIOS ver F1-F3 on this MB requiring a lowly i3-530. i5-650, or i7-860.

Board purchased from NewEgg 5/15/10. Any ideas?
Any way to get to the Sys SetUp to flash the BIOS? Any other tricks to troubleshoot the MB?

Debug Code 26h says "1) If Early_Init_Onboard_Generator is not defined, Onboard clock initialization. Disable respective clock resource to empty PCI & DIMM slots. 2) Init onboard PWM. 3) Init onboard H/W monitor devices." No idea what these mean!

The next error code 27h (which ? actually cause the stop says) "Initialize INT 09 buffer".

Demeaning options include buying an i5-530 for $100 to see if it POSTS, take it to a shop and beg them to try an older CPU, or RMA back to NewEgg.

Or possibly there are other tricks I'm not aware of that could be tried at home.

Any next step suggestions would be greatly appreciated.
 
Solution
Well, if it's any consolation, I've only ever heard good things about XFX's support - should be a quick RMA... I do machines for quite a few people, and I always tell 'em: "Once your computer's been running a month without a failure, it'll likely be indestructable for years - but - getting that first month is can, sometimes, be hell itself!"

It's that old bugaboo, 'infant mortality', and there's just no way around it, short of 'burning in' each & every part, say, seven-hundred hours, with a large number of power cycles... Used to always be blamed on chip 'bonding'; i.e., the chip itself (the substrate) had to be connected to the pins/pads - and the connections were prone to early failure. Don't know who gets the blame currently...

bilbat

Splendid
While it is always possible that an 'unsupported' CPU will fail to start up, in practice, the majority of the cases I've actually see have involved 'hexacores', both from AMD and Intel. I'm kind of 'leaning away' from that as the source of your problem, as:

18h is the step where the BIOS 'queries' the processor ID string, and initializes its caches/cache management...
1Dh initializes the PMI for the CPU...

26h looks like it's the initialization for the on-board LPCIO (low pin count input/output) chip (in your case, an iTC 8720 - no 'newcomer'!) - the guy that 'does' the fan sensing/fan control, the floppy interface, the temp sensors, PS2 kbd/rodent...

27h is intializing the BIOS' keyboard buffer - where it 'puts' keystokes as they arrive, via its interrupt service routine...

I don'y really know what's going on here (I've said it before {in fact, just yesterday [:isamuelson:8] !}, and I'm sure I'll say it again - I'd kill for source code to a modern BIOS!), but my overall 'feeling' is that these are not processor related indications...

What I'd start with is a 'strip-down'; most people pretty much just 'shovel in' all their parts at once, so finding a single problem becomes exceedingly complex... You always want to start with this:
0270n.jpg

...and only 'the basics' installed - the vidcard, one DIMM, the power plugs - once you know that's working, then, into the case - boot up again to see that was OK, then add parts, one at a time to be able to detect individual problems...


Mind you, there are two ways to do this: you can do it either in or out of the case. The advantages and drawbacks:in the case is easier and faster, but will not find case-related problems, like shorts from extra, mispositioned standoffs, or ground plane problems; out of the case takes longer, and you may run into 'reach' problems - power supply cables and front panel power switch headers may not be long enough; for the power supply, it's usually just a matter of removing four screws to temorarily relocate it; for the power switch, you can just do this (very carefully):
power2.jpg

You only need to short the pins momentarily - that's all the power switch does...Out of the case also affords you an easy opportunity to 'flip' the board to check your heatsink/fan attachment setup, to be sure all the pins are fully seated, locked, and not cracked... If you do the out of the case, you need to lay the board on a non-conductive surface: the box the MOBO came in is ideal, someone recently suggested a phone book which is excellent; but - the foam pad it came with, and the bag it was in are not - being 'antistat', they are somewhat conductive, and may induce problems...

Another item worth mention at this point is case speakers: if you haven't got one - get one!
http://www.cwc-group.com/casp.html
A lot of people operate under the misaprehension that the 'diagnostic beeps' should come through the speakers attached to their sound-card/chip - not so! Your three hundred dollar Altec-Lansings won't do you any good here - you have to have a case speaker attached to the front panel header, and, often by this point, it's the only diagnostic info you'll have to go on...

The standard 'strip-down':

Power down at PSU switch
remove everything except
CPU and heatsink/fan (check carefully that the fan retaining pins are fully inserted, completely locked, and not cracked)
one stick of RAM, in slot designated for 'single slot' use
video card and monitor connector (if more than one PCIe slot, again, in slot closest to CPU)
all power plugs - 20+4 or 24, 2x2 or 2x4 ATX power, graphics card power
case speaker and power switch connectors
keyboard (don't need a mouse at this point)
place jumper on RST_CMOS pins
remove jumper from RST_CMOS pins
power up at PSU switch
power up by depressing case power switch (or shorting the 'power' pins...)
If you get video, enter BIOS with <DEL> (may need a <TAB> to get to POST screen, if 'splash' screen is enabled)
Select and execute "Load Optimized Defaults" - save and exit, reboot
power down
reinsert other components, one at a time, testing each time after addition...
 

John_VanKirk

Distinguished
Hi all,

Have done several things today to hone in on the problem. As suggested, I disconnected everything except the CPU, cooler, 2 sticks of RAM (4GB) and the video card.
However, I did take the XFX HD-5850 ZNDC video card out and tried an old PCI #9 VGA Video card (remember that company?), and it POSTed perfectly!!! So, I flashed the BIOS to the latest F6 (6/30/10) and that went well. Didn't have to touch the i7 875 CPU.

However, using the 5850 video card afterward still won't POST. It stops at error code 26h which XFX says is where the video card is initialized. Occurs in both PCI-e 16X slots.

I am going to take the high end HD-585A ZNDC down to our local Central Computer tommorrow and ask them to try it out on one of their MB's, and also buy an inexpensive 4350 or GForce 210 to have as a basic test PCI-e card. Hopefully that will answer the question as to whether the 5850 is flaky, or if the Gigabyte GA-P55A-UD7 has bad slots.

I really like all the features of the GA-P55A-UD7 board, but now see it is "unavailable" at NewEgg and Amazon. Very unusual.

Wonder if they have had problems with this high end MB and are redoing it or if it is just in such high demand.

Anyone heard of any problems with this latest Gigabyte GA-P55A-UD7?
 

bilbat

Splendid
Just thought of a couple things:

Do the pair of PCIeX6 power plugs on the 5850 'seat well'? i.e., do they feel solid, or when they're engaged, do any of the emerging wires seem loose in the plug?

Have you tried various settings of the "Init Display First" setting on the "Advanced BIOS Features" page of the BIOS? Have you done the (obligatory) "Load Optimized" since the new flash?
 

John_VanKirk

Distinguished
Hi everyone,

The only thing that feels "soft" is the insertion of the card into the PCI-e slot, a lot easier that inserting a regular PCI card. The is my first PCI-e card so possibly the weight of this card make it sink into the slot easier.

I took the 5850 OC card down to Central Computer, our high end Silicon Valley retailer and repair shop and had one of the certified techs check it out. He said it was definitely defective, couldn't get it to POST in several boards.

I bought an inexpensive ATI 5670 PCI-e card to have for testing, and for general lower end use if needed. It POSTs perfectly in both my PCI-e 16 slots, so this card is definitely bad.

Will call XFX this AM to send it back, since I just bought it 1 month ago and just tried to install it several days ago.
They said the other day the Gigabyte error code 26 h is where the Video card initializes, so that all makes sense.

Thanks for all your suggestions.

Anyone else have a problem with a new HD-585A-ZNDC OC, or ZNFC standard Video Card out of the box? My new rig is going to be really nice when running, but as many know, your heart sinks when you spend months acquiring just the perfect components for a great new computer, and then nothing happens (it won't POST) when you initially turn it on.

John VK
 

bilbat

Splendid
Well, if it's any consolation, I've only ever heard good things about XFX's support - should be a quick RMA... I do machines for quite a few people, and I always tell 'em: "Once your computer's been running a month without a failure, it'll likely be indestructable for years - but - getting that first month is can, sometimes, be hell itself!"

It's that old bugaboo, 'infant mortality', and there's just no way around it, short of 'burning in' each & every part, say, seven-hundred hours, with a large number of power cycles... Used to always be blamed on chip 'bonding'; i.e., the chip itself (the substrate) had to be connected to the pins/pads - and the connections were prone to early failure. Don't know who gets the blame currently, haven't been following the 'foundry end' that well; still trying to 'get my head around' the new HKMG - for all I know, the 'balls' on an LGA might be 'grown' right onto the substrate?

Anyhow - glad you found the culprit; have fun with the 'new toy'!

 
Solution

ronss

Distinguished
Jun 16, 2009
230
3
18,695

i also had a xfx card that was bad , and had to get it exchanged.....got me wondering about getting xfx again