Dead/Deadly PSU? OCZ 700 Watt

[hobo]eclipse

Long story short, i stayed with the OCZ GamerXstreme 700 watt month or so back, knowing i wanted a newer supply in order to put the OCZ with my prior AMD rig for a HTPC setup. Well i got the overclock bug the last week or so and was trying to push my Xeon to 475-500 FSB. Getting it up there things seemed a lil squirrely, geting seemingly stable at one moment and acting iratically the next min. From a slugish feel and errors in benchmarks/tests that were just successfully ran, to bandwidth tests in everest showing half their bandwidth from one run to the next on the L2 cache.

Feeling the PSU may be holding me back i went ahead and ordered the CORSAIR CMPSU-650TX 650W ATX12V / EPS12V Power Supply

seems a great deal @ $90.00 after rebate.

setup an OC last night as i was messing with the router, 460x8 @ stock volts, +.05 to FSB, +.15 to MCH and RAM @ 2.1v with +.3 (rated)

Ran for 3.5 hrs with 2 instances of prime going using 3.7 GB RAM. After i got the router going i manually shutdown prime and was going to check the router. Imediatly after shutdown of prime the mouse was very slugish and skipped across the screen. I decide to reboot and set all to default and not mess around till i get the new PSU. reboot and a loud sound that sounds like a speaker alarm with a lot of rough buzzing happen for a second and the rig doesnt power back on.

after looking over the inside for anything significant i decide to push the on button. the psu kicks on and the fans begin to spin, but it only stays on for like 3 seconds before it shuts down, the board doesnt seem to get a chance to post? i hope that is the case and the PSU is shutting itself off because a problem is detected. I tried 3 times and all 3 it did the same, power on with fans then turns off.

What you folks think? Hopefully the rest of the hardware is OK.

[hobo]eclipse

so any opinions on this? i sure wanna hope the PSU is bad and rest of the hardware is OK!

Lowfat

Well a PSU will not cause a game to be sluggish. It will either work, or it won't. As for it shutting off like that, it could be the PSU, but I have my doubts. It does sound as it is it s a motherboard problem to me. Did you try resetting the cmos? Or turn the power supply switch off for 30 seconds. It sounds to me as it trying to post but is failing an overclock continuously. This issue has plagued almost all Gigabyte Core 2 boards.

__________________
27" iMac: i5 750 * 80GB X25-M G1 * 4GB DDR3 * HD4850
Stacker: 2600k * P8P67 Pro * 4GB DDR3 * 80GB ioXtreme * 30GB X25-E * (2) HD6950 * Asus D2 * Enermax Revo85+
HWD: (2) Xeon E5649 * EVGA SR-2 * 6GB DDR3 * 8800GTS * 1000W Antec
FT01BW: (2) Xeon E5645 * EVGA SR-2 * 6GB DDR3 * GTS210 * Gigabyte 800W Odin GT
Storage: (2) Sans Digital TR8M-B * 29TB

[hobo]eclipse

that maybe it, i haven't reset the CMOS as its in a kinda weird spot under my GPU, and have been busy,

this board has handled many reboots and always(before this issue) if the OC was unstable it would reboot itself upto 3 times, not posting but powering down and powering back up on its own, posting usually on the 3rd time and having the FSB reset to the CPU default

when i push the pwr button and the fans pwr on then it powers down, all seems the same it just refuses to repowerup.

If there was a mobo problem or other hardware i was expecting a bios beep code or something.

the loud buzzing noise when i shutdown for the last time is a clue i believe

o ya the psu has been powered off by the rear w/ cord pulled out for about a day now

Necrosis

Do you still have the OCZ to see if it's the PSU?

[hobo]eclipse

its all still together but unplugged, the new psu should be here tuesday, prolly gonna try the cmos reset later, but based on the way the board has acted up to now i dont think the issue is an unstable OC, like i said it usually powers off and and on itself when the OC is unstable and it puts the FSB and RAM multi at default to post. it even has a dual bios feature which i seen in action one of the 1st few days i had the board.

it stores an extra bios and reflashes itself it the bios is corrupted. I had flashed to the F3 when i recieved the board and upon playn around the first few days i did something bad and the bios was automatically refalshed to the F2, so i know that was working.

swimtech

Oh bummer...

Yeah, my EX38-DS4 powers itself off, then back on by itself in 2 or 3 seconds when it resets after certain bios changes. If the video card initializes properly with low fan speed, then I know the settings have taken ok. If the video card fan is back on running full blast, I know the settings are unstable. I have to to turn it off with the power button (held 5 seconds...) then back on and the board resets itself to stable settings again.

Sounds like about the best you can do is wait 'til Tuesday and try the new 650Watt Corsair in your machine, and I'll hope with you that it was your PSU and not your mainboard.

__________________
It's not so much getting your way that matters or not - what matters is how you go about getting it.

Necrosis

BTW, doh I thought you already had the Corsair...sorry. If it was the OCZ PSU you really bought it right on time then.

Lowfat

When the board shuts off after a few seconds does it remain off or try to start back up? If stays off, could be a pooched board. If it tries to boot up again, try out what I said above. Like I said tons of Gigabyte boards had issues like this. I highly doubt it is the PSU as the board is powering up and shutting down just like it does when trying to reset itself from a failed overclock.

__________________
27" iMac: i5 750 * 80GB X25-M G1 * 4GB DDR3 * HD4850
Stacker: 2600k * P8P67 Pro * 4GB DDR3 * 80GB ioXtreme * 30GB X25-E * (2) HD6950 * Asus D2 * Enermax Revo85+
HWD: (2) Xeon E5649 * EVGA SR-2 * 6GB DDR3 * 8800GTS * 1000W Antec
FT01BW: (2) Xeon E5645 * EVGA SR-2 * 6GB DDR3 * GTS210 * Gigabyte 800W Odin GT
Storage: (2) Sans Digital TR8M-B * 29TB

OldBuzzard

One "quick and dirty" way to see if the PSU is the problem is to leave everythng connected EXCEPT the 20/24 pin connector.

Make sure that the PSU in nplugged and/or turned off.

Holding the 20/24 pin connector horizontal, with the little 'locking barb' at the TOP, starting counting the holes in the TOP ROW starting at the left and moving right. You can 'jumper' the 4th and 6th pins with a paperclip, and then turn on the PSU. If the PSU powers up and stays running, then it's most likely OK.

The above is how Corsair tells you to check out a PSU, and also the way that Koolance uses for purging/leak testing their units. Koolance even supplies a little jumper wire or that purpose.

Page 46 of this PDF will show you a diagram on how to do that:

http://www.koolance.com/support/file..._v200eng_d.pdf

__________________
When looking for a reason as to why things go wrong, never rule out sheer STUPIDITY

PangingJr

that is something i can't explain, except you heard that sound when the system already booted to Windows, but Windows just couldn't boot and it restarts itself, this due to a lack of voltage in the overclocking setting and the system was not stable, however, i'm not sure why the system couldn't boot back, but, never mind the symptoms, when in doubt, clear your CMOS anyway.

come to think of it again about the loud sound you heard,
this (the loud sound) can happens, however, it's when the system already booted to Windows,
on some certain high FSB overclocking if the PCIe frequency was set to Auto and the PCIe voltage to Normal, then i used to get a loud sound that sounds like you mentioned.
and to fix this, i found the PCIe frequency will need to be at 100MHz (sometimes at Auto) and also i need to add some volts (+0.1 - +0.35V) to it.

Judas

PSU... i'm almost certain of it..

Pop it in another machine... hell just plug it into another motherboard with not much else in it and see if it does it again...

Pack it up and send it to ocz (RMA) and you could likely have it back within 2 weeks (at least for it they did)

__________________

PangingJr

IF the system keep powering up and shutting down (by itself) repeatedly, and you can suspect that, it keeps doing it more than it needs, then it's the system BIOS,
and this's normal behavior of Gigabyte board's BIOS from P965 up until now... it may happen often, or it may never happen again or at all, this depend on how you overclock your system.
if this happens a lot, then you'd better get used to clear CMOS values.

after the system takes 5-6 power up and down cycles and it won't boot up... switch off power supply, and clear CMOS values.
if you cannot reach the CLR_CMOS jumper pins and place a jumper cap on that pins, you can use a metal object like a long screwdriver to touch the two pins for a few seconds.
then switch on power supply, restart the computer and reset the BIOS setup settings.

[hobo]eclipse

Turns out it was the cmos needed clearing with the jumper,(, thanks lowfat and panging) this is the first time the board acted that way. and what really has me wandering was the loud screech i heard fom me rig after that shut down. the OCZ psu booted the rig normaly after the cmos clear. Looks like itll be ok for the HTPC setup.

Im pretty sure the problems i was seeing with the mouse skipping and some low L2 cahe bandwidth everst benchmarks is more related to the DPC latency issues on the board.

After a CMOS clear last night i set all to stock, 333*9, ddr2 1066 5.5.5.15, +0.3v DDR2. Watching an x264 720p HDTV mkv movie i experianced random lockups of the movie and DPC latency went through the roof (600,000+) at that time in DPC latency checker. Problem resolved after minimizing the movie and maximizing again (kept playing). I also noticed during the movie lockups the mouse did the same as before, skipping and lagging in an inconsistant manner across the screen.

I let the CPU set on today at all stock settings downloading from usenet and DPC latency checker on and there is 1-3 samples consistantly over 1000 micro sec. every 10 samples or so.

Also i do recall a few time while playing oblivion after I pieced this rig together @ 400*9, DDR2 1066 5.5.5.15 having what I 1st thought was lockup but would then pull itself through, very similair to the the way the movie acted.

this is a screen shot i took of the OC settings i thought were going to work out great that night. Right after i took this i believe the DPC latency went haywire and the mouse was almost imobile. Shut down and then the screech.

Round off checking was enabled in p95.



the DPC latency is one thing (not a good thing in its own right) but something had to set it off, i believe either it was the low voltage OC or the only 2 secondary RAM timings i had set in the bios, Refesh time cyle @ 68 and Performance level @ 7.

The new PSU is here and ill prolly get it in tonight.

The DPC latency thing is not good and i see various people on other forums saying they contacted gigabyte support and was given a new beta bios to try. but no official betas or new bios' in awhile so looks kinda dim.

Kudos to oldbuzzard, though i did not need to try that trick that seems some very useful knowledge that i'm sure will come in handy later, thank you.

Judas

hmm


checking your voltages and see how much flux there is

__________________

[hobo]eclipse

load line calibration enabled, acording to CPU Z lowest it dips is 1.168V to the CPU

RAM was incorrectly reported in MEMSET

RAM in pic = 460*2.4= DDR2 1100

PangingJr

i know about the DPC latency issues in Gigabyte X38's as well, but mine and a couple of my friend's boards (a few DS5's and a DQ6 old model) do not get that bad.
i'm sure Gigabyte will issue a new BIOS that also includes a fix for that.

about you had to clear CMOS values, like i said before it's normal behavior of the board's BIOS and don't worry about that,
lol, once you try overclocking to 550 FSB and up, you'll see more of that, probably.

swimtech

Just for comparison [hobo]eclipse, I saw the DPC latency checker too - downloaded and installed on my new machine with the same board as yours with F3 bios, and it showed some occassional spikes to nearly 700 or so about every sixth bar, but otherwise it was well within the software's determination qualifiers that I could stream media ok.

I've noticed no hanging or stuttering problems at all so far, but I do have a clean build - only Civ4 and MS Office 2007 installed as applications so far. The rest of the software is benching and monitoring utilities running Vista Ultimate x64. Could it be some corrupt software or odd conflict otherwise???

__________________
It's not so much getting your way that matters or not - what matters is how you go about getting it.

[hobo]eclipse

panging u mentioned u had a r.1.1 board. I cant seem to find any info printed on the board aboput mine, any tips on how to aquire this info?

Do u guys see any secret options when using the cntl f1 trick in the bios? specifcally in the MIT and the PCI IRQ area?

and the screeching noise? any opinions? it was a scary noise!

swimtech

I hadn't tried that trick yet, though I did hear it was necessary to get to some of the bios goodies on the previous chipset boards...

__________________
It's not so much getting your way that matters or not - what matters is how you go about getting it.

PangingJr

the PCB rev. number is printed on the corner of the board.
this board is sold out all the time in my area, so mine is a very new shipment board.

but i know for sure that rev.1.1 has been out in retail stores for some time now.
i heard from a friend (but cant confirm) that the PCB in use is the same as X48-DS4 rev.1.3

ah, the rev.1.1 also printed on the box as well.

[hobo]eclipse

on the bottom left in the last PCI slot i have the eSATA lead mounted, above it but below my sound card i see a 1.1, ty, relieved as id want a r1.1 i think.

out of curiosity panging, do you use a lot of periferals? usb devices. the eSATA bracket, sound card? how many HDD do you connect? i seen you low DCP latency screen in the other thread

i hope gigabyte is on that DPC latency issue, this board is my first gigabyte product and furthur purchasing/recomendations are hanging in balance!

[hobo]eclipse

here is my DCP latency while surfing around....and then i decided to open CPU-Z, still stock settings

PangingJr

[hobo]eclipse,

you mean the one i posted here - http://www.hardwareheaven.com/motherbo...ml#post1142959

that's from my P35-DS4 and with the said BIOS version "and" the test was running in Windows Server 2008 64-bit,
it never come out that good when i was running the test in Vista's.

as for about my experience in using Gigabyte's, well, you probably didn't ask me this, but anyway, Gigabyte boards have problem here and there, just like the other brands,
so i'm not going to say that my Gigabyte board is the best board out there, cause as far as i know it's just not possible.

however, i already know a few things about Gigabyte boards, so, it's just better and easier for me to use it (this includes, buy it, return it for a replacement one if needed, and then later on when i give it to someone).
anyway, i find it hard to believe myself, and just have to admit it to myself, that in terms of overclocking, my P35-DS4 rev.1.0 is still about 50 percent (or so) better then this new EX38-DS4,
the P35-DS4 could go on and on overclocking both processor and RAM, and its clock generator chips is very quick in processing my overclocks.

PangingJr

this is from EX38-DS4, i just ran the DPC Latency checker... also, while surfing around and some applications running in background.

original board BIOS F3, 450 FSB overclocks, and under Vista 32-bit,
2 wireless mouses and a keyboard using both USB and PS/2 ports,
4 USB ports, 2 LAN ports, 6 SATA ports (4 drives online, the other 2 ports are for Gigabyte eSATA bracket) in use,
Printer and IDE ports are disabled in BIOS.

i think this should be better in other OS's like an XP or the Server 2008.
and it differently or hopefully should get better with a new BIOS.

PangingJr

[hobo]eclipse,

here's a EX38-DS4 F2 BIOS board and its out of the box test for max FSB on the board,
the system ran with a 8800GT with an Enermax 400W PSU, but it was tested on an open testbed.


(found and taken from the web)




that looks like it should make overclocking quite a bit more fun than with 45-nm C2Q processor,
i think i have to try this board with a C2D processor.

[hobo]eclipse

flashed to an f4a bios that was hosted by gigabyte a few months back, now its was gone from their site. Downloaded it elsewhere. DPC latency seems a lil better but i,ve also disabled a few more unneeded services(windows defender....) and disabling things in the bios(HDD smart....), a;so unplugged my esata bracket from the mobo, this made the DPC latency drop the most, barely any past 750 microseconds after that(at idle), ive also read of not using the IDE controller helping but as of now my only DVD ROM is IDE.

whats your opinion on the HPET setting and HDD settings, ive been using RAID, native mode enabled.

Curious to see what would happen with ACHI mode or leaving it all at default, i used raid in order to maybe set up some drives in RAID sometime but never have yet.

changing these HDD settings, would that require a re-install of windows?

also, running the everst cache bandwidth bench, i am getting bad performance with latency and bandwidth in the l2 cache bench mark, about 8000-10000 below some screens i've taken and about +3.0ns latency

FSB voltage on gigabyte = VTT ????

ill post a screenie in a bit, gotta run now, prolly try the F2 bios in abit as well.

PangingJr

i'm running my OS's pretty much at default, i don't disable any services in Vista at this moment.

both P35-DS4 and EX38-DS4 have the eSATA bracket plugged in at most of the time,
the HPET is also enabled in BIOS, 32-bit mode (for Vista), 64-bit (for Vista and Windows Server 64-bit OS's).

anyway, i just had an image of a Server Stardard SP1 from other computer to install on the EX38-DS4 based computer,
but due to the image was made from a system that uses different motherboard's chipset, so i need to clear out some old drivers from the Windows installation, and then i'll check whether or not the DPC Latency gets a better result.

btw, the BIOS F4a did not include fixes for the DPC Latency.
the other Gigabyte X38's and P35's have had a BETA BIOS to try with, but for the EX38-DS4, not available yet.


-----------------

just about the same with Vista's one, or just a wee bit better right now,
the F4a BIOS is also in use this time.
i think, it will be like this, up until a new BIOS arrive.



the last time i checked on my P35-DS4 with the BETA BIOS i had quite different results on both Windows OS's.

[hobo]eclipse

gosh i hope they dont forget about the ex38-ds4

i see the same basic DCP latency step pattern pretty much what ever i do. same as yours. just reinstalled vista x64 u and my sound card seems to give me a bit more over head keeping the latency at 200 or so, but the step pattern is there. thing is it seems so irradic and i dont know what really sets it off specifically when overclocking.

[hobo]eclipse

thats what is kind of troubling, the board has been around plenty long enough. I have a support queston opened up on the gigabyte site. according to their site no one has looked at it in the 2 days since i posted it.

Emailed Janus from tweaktown gigabyte forum with screenshots and an everst HW report. hopefully i'll here back soon, its been less than a day. havent posted there yet.

i called gigabyte support in california and after being told there was no beta bios i finally got the f4a bios sent to me but was told there were no others

can this problem be related to IRQ assignments?