Overclock: P35C-DS3R fsb setting ignored. Weird!

[imported_daveG]

OK, so I've read about a lot of folks having problems with gigabyte boards similar to what I am seeing but none exact. Anyone out there got any ideas?

GA-P35C-DS3R v1.1 with E6750 cpu, Corsair HX620 power, 2GB (2x1GB) DDRII 6400 memory in XP pro. F10 bios. Everything works fine at stock settings and speed. CPU-z, PC Wizard and another temp utilities used to gauge speeds.

Disabled all the usual suspects and tried to overclock. Nothing. 333mhz normal FSB but won't go to 334. Settings allowed and kept (nothing lost on power cycles) but the setting is ignored. There is no instability, just no change. Set FSB to 350, no change once in XP. The weird part is that this also applies to UNDERCLOCKING the fsb. A setting of 300 isn't applied either.

I can change memory timing or multiplier and they take. FSB frequency doesn't. Obviously the softFSB utility couldn't change things either. SMBus error I believe was the error.

I know about the Ctrl+F1 trick, it doesn't look like it changes things in the later bios versions.

Yes, I tried clearing the CMOS three times (removing the battery for > 20 minutes) with no change in behavior. Settings were cleared each time but the problem never let up. I flashed the bios - no change. Flashed the oldest bios I could get (F4 is earlies supporting E6750), reset to optimal defaults, reflashed to latest (F10) and again, reset to optimal defaults before tweaking. No change.

I give up. Temps are normal. Everything works at 333mhz FSB. Power management is not the cause (set "Always On" so speedstep shouldn't be an issue).


If anyone has clues (other than just an RMA), let me know!

 

[watek]

This problem seems to be power supply and the intel p35 chipset. Quite a few have this problem as well.

I just read something about certain power supply start up circuits and the p35 chipset don't get along. But who knows!

 

[Big Lar]

This is just a shot in the dark, but first off, is the cmos clear jumper/s in the correct position, secondly, do you have the dram multiplier set to 2?
There you have the extent of my knowledge until I get my board up and running.

Larry

 

[imported_daveG]

I don't have another power supply handy so I will have to arrange that. I did unplug the other drives before doing any of the tests so I doubt I am taxing the 620w supply. These are very popular components. There should be a lot more folks in this boat were this a compatability issue.

The startup circuit though is a possibility. In-rush current draw maybe? I hadn't thought of that. The supply is serious overkill though and my expectation is that it is not the issue especially if trying to underclock. Never the less, I will try another manufacturer's power supply before trying to RMA this board.

As for the cmos jumper, if it were wrong, the date and time would be lost. I never saw this happen. I can successfully change memory timing and the fsb multiplier. Also, the FSB setting that is ignored is always kept (through multiple power cycles). If the cmos jumper were the deal, it would reset the FSB setting before ignoring me.

The dram multiplier was in fact set to 2 for most tests. Never <with this board> have I had XP boot into a memory speed other than 333 or 400 (for my DDRII-6400/800), nor has XP ever failed to boot for me. If I were unintentionally overclocking my memory, I should get failures on boot or instability or at least a funky showing in CPUz.

All are really good ideas though. Thanks for chiming in.



I think this is a bad flash where the BIOS keeps its settings or maybe a bad communications interface between the BIOS and the system timer.


Now I am gonna go digging in my parts heap for another power supply. Yea! something to try until Gigabyte tech support opens Monday. I really don't want to return my motherboard and be without my system for a few weeks.

 

[VirtualLarry]

I sense an RMA in your future...

 

[imported_daveG]

Get this: Gigabyte support "doesn't guarantee overclocking". OK, that is valid. They can't guarantee I'll get the best overclock, but they say they won't guarantee the ability overclock. Even their own tool "Easy Tune 5" won't adjust this board (grayed out buttons).

If it works on the automated test bench at stock speeds, it will come back (still broken) with a repair charge 'cause they won't find anything wrong. I talked to a tech after over 33 minutes on hold and was told they may test to a slight overclock spec but there was no promise, then I was put back on hold. I gave up.

2-3 weeks + shipping time there and back without my computer plus probably another trip there and back when they don't check it over or under-clocked. I am looking at over two months without a computer.

That was $150 wasted. I'll live with it at standard clock speeds until I get itchy for an upgrade rather than fight this. I cannot believe the level of customer support. I'll never buy another Gigabyte product again. It is a shame all those great reviews (and there were a lot of them) for this board failed to mention the company's reputation or that the great overclocking they promise in all the reviews was not a supported feature. I wish I'd known.

Wow.

:roll:

 

[Old Hippie]

Get this: Gigabyte support "doesn't guarantee overclocking".

AFAIK, nobody guarantees overclocking, but your problem is changing the FSB. Sounds like two different things to me. If the FSB settings are being ignored, isn't the board broken?

Or, am I way off base?

 

[mrfatboy]

Look here and make sure you are doing everything right.


Text

 

[imported_daveG]

Old Hippie: my point exactly.

mrfatboy: I used that link as my guide when things didn't go as I expected. I'll go through again and exactly copy the initial poster's setup.

New symptoms to add:

If I clear the CMOS and restore "Optimized defaults", my E6750 leads the bios to set 8x multiplier (correct) and a 100 MHz FSB (???). This should result in an 800 MHz cpu. I boot into XP like that and sure enough, CPUz thinks things are fine (2GHz 6x, speed step reduced). I run something to load down the CPU and the multiplier goes back to 8x and we are at the normal 2.66GHz. If I shutdown and go back into the bios, the 100MHz FSB is still there. Go figure.

Also if I then set the FSB to 450 (well beyond what it should do on air cooling and with the memory I have), it takes the value and upon saving, reboots a few times. When XP finally comes up, all is back to the normal speeds (2.0 6x / 2.66 8x). Here is the kicker though: In this case, the BIOS actually reset the FSB setting back. It saw 450 was too much and changed the saved value. In my regular trials, the saved BIOS value for FSB never changed.

Anyone make any sense of that? The fact that it rebooted a few times says the BIOS was listening to me and trying to set my unreasonable value. This leads me to believe my 335 test value is seen by the BIOS. What else could clear it without changing the BIOS saved value? Are there DOS based clocks that read actual FSB speeds? (not a simple CPU register lookup which lies)? Can the P35 chipset driver reset the FSB?

Any of you guys out there with this board overclocked, at POST when it says the CPU name, does it list it as an "E6750 2.66GHz (333 x8)" as a dumb string read from the CPU or does it reflect the actual overclocked speeds?

Thanks everyone.

 

[Old Hippie]

You have turned off these Intel power saving features (CPU EIST Functions... I believe there's two different values that need to be turned off), correct?

If not, do it and continue troubleshooting.

 

[imported_daveG]

The CIE, EIST, No Execute Protect and Thermal Monitor all seem to have no effect on the FSB problem. They kick down the multiplier but not the FSB as far as I can tell.

I tried with EVERYTHING disabled and no change. Virtualization, USB legacy support, etc.

OK, more info:
Set 300mhz, 333MHz or 350MHz CPU Host Frequency with 7x multiplier. Bios power on self test (POST) message, first line identifies the CPU as E6750 2.33GHz (333 x 7) <the same for all three tries>.

Set 300mhz, 333MHz or 350MHz CPU Host Frequency with 8x multiplier. Bios power on self test (POST) message identifies the CPU as E6750 2.66GHz (333 x 8) <the same for all three tries>.

The FSB is already ignored by the time the motherboard posts. It can't be XP or a driver.



NEXT:
I'll swap out memory and power supply. I will try a PCI graphics card with no hard drives at all. All this as soon as I get more time.

 

[imported_daveG]

FYI - Gigabyte's support response:

"If there are no issue operating at default specs there are no issue with the board. We only offer warranty service where you can send in the board first, if you wish to do so go ahead and submit your rma request"

 

[imported_daveG]

Final test:
Swapped out Ballistiks memory for another brand 1gb single ddrII-6400(800). Older PCI video card, not my normal PCI express video card. No hard drives at all. Nothing plugged into any SATA ports or USB ports. Just my motherboard, cpu, one memory stick and a low end video card (and a keyboard).

Same problem. No change.

I should be grateful it isn't resetting like many of the other GA-P35C-DS3R boards out there.

I found this link: http://forums.tweaktown.com/showthread.php?t=25051
Official support folks at Gigabyte used to try and help there. I think lately the mood isn't so favorable toward Gigabyte so they've been a bit quiet.
*** If you own a Gigabyte P35 motherboard, you should check the link out. ***

Anyway, thanks guys for the advice.

 

[GoaGas]

@ daveG - Hi, I had problems getting my X38-DS5 to accept the memory voltage - it kept either resetting them. or ignoring them. Sometimes it would multi-boot & then I realised it was resetting the BIOS to previous version it liked. I think I eventually cured it in the BIOS MIT settings by manually setting my volts then reset everything on that page to Auto (I expected that to revert my volts); re-booted; my volts were right. Re-booted again & reset the other Bios MIT settings to how I wanted them. Re-booted & everything has been fine since.
I know this isn't a replica of your problem; but I thought it might help. I find that Gigabyte boards often need several boots for them to learn a new Bios setting.

@ Old Hippy - Thanks! you might have just solved an unrelated problem. I'd noticed that some really small low-CPU tasks were taking much longer when the computer was idle than when it was busy! Maybe the EIST function is cutting in & slowing everything down. I know there is 1 Bios switch - but you mention 2. When I next re-boot I'll have look. It would be nice to get it sorted & know that it's not a symptom of a hardware fault.
This shows the unexpected benefits of reading these threads.

 

[ukhazard]

DaveG: I have the EXACT same problem with an E8400 and GA-P35-DS3L. My system shows as 900 MHz (9x100 fsb!) but boots as 3 GHz with memory ratio set to 2.0. Won't go any higher and I have flashed and reflashed, jumpered the CMOS, everything.

My system was working stably at 4.3GHz and then just began not accepting my manual FSB setting after a regular reboot. Like you Easytune greys out the CIA MIB and Overclocking buttons.

When the overclock WAS working, the POST screen did read as the actual value (4302 MHz, 478 x 9).

I hope you find an answer that isn't mobo replacement (I strongly suspect that it the only way).

EDIT: Scratch this reply. I fixed the problem by removing the CMOS battery for 10 minutes. I had previously flashed and set the reset jumper, but not actually removed the battery with power removed.

 

[imported_daveG]

Reading the history of Gigabyte's RMA procedures, I am more afraid to send it back. They don't replace with new as often as they repair or send refurbs. With the frequency (no pun intended) of failures, I feel my problem is better than the odds I face on an RMA.

That said, I am hopeful from ukhazard's words. I will try that again but leave it overnight with no battery or power cord. I will also try to power up in this state to assure caps are drained. If this is it <and I can't help think it is possible> it would make me very happy.

Let me just add that anandtech rocks and these forums are full of sharp, patient, helpful people. You have all been great, not just in this little problem but every time I looked up some crazy obscure problem. My hat is off to you all.

 

[Old Hippie]

Originally posted by: daveG
Reading the history of Gigabyte's RMA procedures, I am more afraid to send it back. They don't replace with new as often as they repair or send refurbs.

AFAIK, that's standard operating procedure for any computer component. There's no reason to single out Gigabyte for this.

If Ukhazard's suggestion doesn't work, an RMA maybe your only other option.

Good Luck!

 

[imported_daveG]

I tried ukhazard's idea. I left it without bios power for an afternoon. When powered back up, it detected the frequencies of the E6750 wrong but once in XP, it was back to business as usual - default speeds.

Old Hippie: I wasn't singling Gigabyte out for trying to repair, just for the impression I get from forums on their success rate (true or not).

An RMA probably won't cut it as it does operate at the factory frequency. Overclocking is not a supported function - nor is setting the supplied adjustments in the bios apparently. I am not about to send it in only to wait and get it back with "checked out ok" prognosis on a diagnostic/repair bill.

I notice that when everything is unchanged, I enter into the bios and change nothing, save & exit, the system restarts normally. If I change the frequency of the FSB up or down by 1mhz, save & exit, the reboot includes one funny clunk sound. I am guessing this is the m.b. trying the new settings, finding something wrong, restoring defaults and starting again. The thing is that nothing is actually restored, just ignored. Wish I knew the steps so I could better tell what it was seeing.

"uncle"

 

[Old Hippie]

If I change the frequency of the FSB up or down by 1mhz, save & exit, the reboot includes one funny clunk sound.

Funny clunk sound? From your speakers or case? I'm just curious. A clunk sound from inside the case is a mechanical sound. I don't recall any mechanical devices on a bare MB. If I were you, I'd be checking that out. If your MB is making a clunk sound, maybe there's something else wrong that would warrant an RMA.

Take the side panel off, and locate this clunk?

I guess this is just the old millwright comming out in me. When I went to troubleshoot a piece of machinery, I used all my senses to help. Sometimes you can't "see" the problem, but other factors ( smell, temperature, sound ), will set you on the right track.

Ya just never know......

Course, I could be waay off base!


PS. I would also try another PS, in case you haven't.

 

[imported_daveG]

The clunk is the power going off in a not-so-normal way. I believe it is the hard drives effectively being unplugged. It is consistent, repeatable and not much more than the normal power click.

If I go into the bios, change nothing, save and exit, it reboots without this sound.
If I change anything, it does this the first reboot but not again.

I haven't tried another p.s., still looking for a spare working unit. I did however, eliminate the power hungry video card (replaced by an antique PCI based, passively cooled Matrox Millenium) and all other devices removed with no changes. There was extremely little draw on the psu. Still, point taken and I'll keep looking for one to test with.

 

[mrfatboy]

my system makes the same sound. I hate it. It just all the fans, power supply fans, etc shutting down. It's wierd, but these boards do it.

 

[sebbolduc]

I have the same problem as DaveG i've flash my Bios with the new F8b but my board just won't start anymore.I've sent my board RMA and bought the same motherbaord and now have the same problem...

I just hope that gigabyte is going to solved the problems soon...

My previous board an Asus P5B could easily run my E2160 at 2997 mhz (9 X 333)

i'm not to happy with my recent "upgrade"

 

[imported_daveG]

Ok, found another working power supply. With everything unplugged and the older syle pci video card in place ('cause the other power supply didn't have the connector for the 8800 GT), I tried again. The corsair 620HX was not the culprit.

The clunk was a combo of fans and hard drives suddenly stopping as suspected. With the changed supply and all drives removed, the clunk is gone.

What I did notice was that when nothing related to FSB was changed, a power cycle would happen quickly. The power supply fan would never stop. If I changed the 333 manual FSB to 332 or 334, the reboot would stop the power supply fan and take about 5 seconds to start the system. This is consistent with the original power supply. The replacement has a louder fan so this is more apparent.

I believe the bios is seeing something as bad and restarting with defaults - hence the longer power cycle. What it is seeing, I'll probably never know.

A $60 solution was found and is being looked at: Abit IP35-E. Yea, it is $60 wasted but I've spent more than that in my time alone. I like Abit and have had good luck in the past.

Thanks everyone.

 

[rastamanfromjamaica]

Yo daveG!

Perhaps you will know what p35c is seeing!

Stay with it! It's great!!!
And it's working!!!

I'm an owner of the p35c ds3r Rev.1.x too (using F10 BIOS), and i had the same brainsqueezing problem like you. Same sounds, same behavior, same everything ;o)!
Yes! I had! Till this morning... (Thank you for all the good hints that lead me to the solution!!!)

Solution for me: USB-Devices, shut them down!! Or deactivate in Bios (detect legacy USB-Storage-Devices or something like that)
No problem anymore after that! Switch them on again-> clack (this ugly sound you know about!)
So deactivating is the way to go! if it doesn't work-> switch off or disconnect USB-devices!
I hope it works 4u2!!

I know that you tried to deactivate the USB-Stuff, but really everything?
Perhaps any other device is causing trouble!?!?!?!
For me it was an external USB-Drive(500GB)!

I wonder if anyone can tell me about the link between USB and this weird behavior.

P.S.: Sorry for my bad english, i'm german...
Other Hardware on P35C: E6850@3600, G.Skill 8500 HK, MSI 8800GTX OC

 

[imported_daveG]

I'd bet if the USB devices you have plugged in went to a powered usb hub, you may get different results. My guess is that the devices are drawing just enough juice at startup to tip the scales. The bios sees just too little power and the door slams shut on overclocking.

I've seen devices plugging into the front of the case that would reboot the machine. Never would this happen if they got plugged into the powered hub. Nowadays everything is USB and the MB has a lot of weight to pull. Think about it: printer, scanner, keyboard, mouse, ipod, camera, external hard drives, flash drives, etc.. It adds up.

That is just my guess.

Myself, I had only a USB trackball and a ps2 keyboard plugged in. Probably should have left off the mouse. I did have legacy USB support disabled. Possible retest if I get motivated again to take everything out.

Thanks for the idea though.