Before=stable@aircooling - now=unstable@watercooling

System before:

Gigabyte GA-EP45-UD3P (bios f5b)
Q9650 E0 ES
4x2GB Corsair Dominator PC8500 DDR2
3xWD Raptor 150 ADFD Raid0
1xHitachi Deskstar 320
2xNEC 3550 DVD/RW
XFX Geforce 8800GTS (G92) 512MB
FSP Epsilon FX700 700W PSU
Thermalright Ultra Extreme 120 w/2xSharkoon Eagle 2000 push/pull
Thermalright HR05 w/Sharkoon Eagle 1000 on NB
Antec P180 w/3x 120mm case fans running at full
Vista Ultimate 64bit

Running stable@4.25GHz (8.5x500) @1.4V approx 65 degrees


Now with these upgrades:

Club3D 4870x2 (bought used) w/latest drivers
Thermaltake Toughpower 1200W CableManagement (brand new)
Swiftech 655 (new) @5
TFC 480 w/8xYate Loons DL12 push/pull @1000 RPM (new)
Swiftech GTZ (new)
XSPC X20 Delta Chipset (new)
EK 150ml res (new)
Masterkleer 11.1mm ID tubing (new)
EK ½" highflow barbs (new)
Antec P180 w/2x casefans running at low
2x 80mm Sharkoon Eagle 100 directed at memory running at full (new)

somewhat stable @stock running prime approx 45 degrees
always crashes sometime during HD id / windows bootup at e.g. 4.25GHz

I'm completely baffled that this rather expensive upgrade seems to have gone wrong.

Ambient is approx 21 degrees celsius

The watercooling setup has been bled and are completely airfree.
It's demineralized water w/feser base corrosion blocker.


Things I've noticed:

I can boot at stock settings and it seems to be stable running prime (at least for 15 mins).
Everest and HWmonitor both reports +12V at 0.15-0.9V, however BIOS reports it at close to 12V at boot.

I had placed one of the cushion mats for the nb block under the XSPC X2O Delta and thought it might be that it prevented proper contact so removed it - no change.

I had connected the 8pin mobo connector to the mobo, but doublechecked the mobo manual and found that it should be the 2x4pin mobo connector instead so changed to that (everything turned off naturally)
Can I have fried something on the mobo when booting with the 8pin?

Any ideas are VERY appreciated.

Thanks

Excluding any issues you might have with the introduction of new hardware(old nvidia drivers cleaned, good psu, etc) how is youre airflow in the cpu area?(over the voltage regulators)

Quote:

---

Originally Posted by njl

Running stable@4.25GHz (8.5x500) @1.4V approx 65 degrees

---

Question: What did you use to verify your system's stability originally? If that's fully stable, that's a very impressive fsb clock on a quad-core, and I'd say that the Vcore is below average for that clockspeed as well.

More to the point, I'm asking if you're absolutely certain that your system was stable previously? Prime can go a very long time without crashing, even if the system isn't entirely stable.

Edit: Also, in keeping with my first question, your old system had a single core GPU, your new one has a double-core card - you're doubtlessly moving a lot more data on the FSB now, which might have put it over the breaking point.

Never believe software monitoring for your PSU, buy a multimeter and test it right! I always test new psu's prior to installing and usually again with the system live, just make sure that when you test you do it at idle and the hardest load you can put on the pc for cpu+gpu's.

Btw, Thermaltake psu's tend to fluctuate a lot on the 3.3v under a hard load if they are near their max rms output.

How are your NB/Mosfets being cooled? If they're no longer benefiting from the same cooling (considering that air cooling keeps quite a bit of airflow over your motherboard) then you might be getting unstable because of that... Just a thought.

Quote:

---

Originally Posted by njl

I had connected the 8pin mobo connector to the mobo, but doublechecked the mobo manual and found that it should be the 2x4pin mobo connector instead so changed to that (everything turned off naturally)
Can I have fried something on the mobo when booting with the 8pin?

---

Not quite sure what you mean by this - are you saying you originally had a PCI-E 8-pin power plug in the motherboard? They're supposed to be keyed differently, because they're definitely not electrically compatible. If so, you might have borked something.

Aside from putting your PC back to the way it was aircooled and testing the new hardware that way, it's hard to judge what is causing the instability with multiple changes made.

This happened to me several times so far. One time, I found a leak, fixed it, and the instability was gone. Don't know how that fixed it though...

$5 says it is the PSU. I had a similiar issue using a OCZ Gamestream 700 watt PSU, great PSU, but I was having weird instability issues. Swapped out for a Gigabyte Odin 800 watt PSU and the instability issues went away. I swapped out my 2 HD4850's for an HD4870x2 and started having problems with the new card. Connected it to it's own PSU and sure enough the issues went away. This was of course after ruling out other possible problems, such as drivers, etc. I have a lot of components connected and I am pretty sure I am pushing the limit of these PSU's. I will probably eventually upgrade to a 1KW PSU, but that's for later.

Quote:

---

Originally Posted by Utnorris

$5 says it is the PSU. I had a similiar issue using a OCZ Gamestream 700 watt PSU, great PSU, but I was having weird instability issues. Swapped out for a Gigabyte Odin 800 watt PSU and the instability issues went away. I swapped out my 2 HD4850's for an HD4870x2 and started having problems with the new card. Connected it to it's own PSU and sure enough the issues went away. This was of course after ruling out other possible problems, such as drivers, etc. I have a lot of components connected and I am pretty sure I am pushing the limit of these PSU's. I will probably eventually upgrade to a 1KW PSU, but that's for later.

---

i would also go with it being a power problem.

Looks like either your windows is the culprit and got bodged somewhere along the upgrade.
Or as stated before it's the PSU that cannot handle the new load.
Be it your rails aren't steady/powerfull enough, or you have reached the max to power your bits.

Pretty sure it's one of these things or both.

i had the same PSU as well .. watercooling too ..
but when u plug too much HDs .. the power fluxation is gonna cause random restarts ..

sad to say that FSP is one of the known "good" PSU manufacturers .. but i do not know why it has huge power fluxations ..

solution: use lesser power consumption devices (random restarts only occur when i plug in my 4th HD :()

Quote:

---

Originally Posted by funk32

i would also go with it being a power problem.

---

It's a 1200W power supply from a good company, I doubt that it is a problem with the PS

Quote:

---

Originally Posted by cobra_kai

It's a 1200W power supply from a good company, I doubt that it is a problem with the PS

---

I didn't even see he had a 1200w, still doesn't mean his 12v rail can't be borked though.
Although it's highly doubtful, I would go with a reinstall of windows.
Just make a ghost of his OS in the state it is now, format and reinstall.
If it isn't working, he can always put back the image and try to figure out what really the problem is.

Some truth to that. Going from a Nvidia to an ATI card, easy for driver/register remnants to be screwing things up.

Quote:

---

Originally Posted by cobra_kai

It's a 1200W power supply from a good company, I doubt that it is a problem with the PS

---

Well the original post stated 700 watt, not 1200. If he is running a 12oo watt, then yeah could be drivers.

Edit: Now I see where the 12oo watt is in the upgrade section. My bad. :D

Quote:

---

Originally Posted by MpG

Question: What did you use to verify your system's stability originally? If that's fully stable, that's a very impressive fsb clock on a quad-core, and I'd say that the Vcore is below average for that clockspeed as well.

More to the point, I'm asking if you're absolutely certain that your system was stable previously? Prime can go a very long time without crashing, even if the system isn't entirely stable.

Edit: Also, in keeping with my first question, your old system had a single core GPU, your new one has a double-core card - you're doubtlessly moving a lot more data on the FSB now, which might have put it over the breaking point.

---

I was using Prime and it crashed after about 5hrs, but I think the main reason for that was I couldn't get rid of the heat - it was reaching about 70 degrees so it was pumping out a lot of heat - which raised the ambient, which raised the heat in the cabinet etc.

It may or may not be be stable, but using better cooling, I should still be able to boot to windows - right?

Do you have another suggestion for stability testing?

I agree that the new GPU can affect stability, but shouldn't such stability issues (based on the GPU) *only* show itself running games or other GPU-intensive applications?

Quote:

---

Originally Posted by hellcamino

Never believe software monitoring for your PSU, buy a multimeter and test it right! I always test new psu's prior to installing and usually again with the system live, just make sure that when you test you do it at idle and the hardest load you can put on the pc for cpu+gpu's.

Btw, Thermaltake psu's tend to fluctuate a lot on the 3.3v under a hard load if they are near their max rms output.

---

I will test it using a DMM tonight.

The reviews I've read at jonnygurus.com regarding the Thermaltake I bought did not indicate a lot of fluctuation, so if you can document what you say I'd be happy to read it.

Quote:

---

Originally Posted by Bobly

How are your NB/Mosfets being cooled? If they're no longer benefiting from the same cooling (considering that air cooling keeps quite a bit of airflow over your motherboard) then you might be getting unstable because of that... Just a thought.

---

The NB is being watercooled with the XSPC X2O Delta Chipset block, so I hope it's getting the cooling it needs. Unfortunately there's no temperature sensor on the mobo for the NB, so I cannot verify this.

The Mosfets are using the stock cooling blocks and yes there's less airflow across them, but also a lot less heat being generated in the cabinet.
I will try tonight to add more fans directed at the mosfets and see if that helps.
If that's the case I will have to get more waterblocks since my reason for going watercooled was to get rid of the fan noise...:rolleyes:

However - if there's a lack of airflow across the mosfets - will that really show it so soon in the startup process? I mean I cannot even boot into windows - it restarts almost instantly. Sometimes it can begin to boot windows, but it never shows the startup logo - always restarts before that.

Quote:

---

Originally Posted by MpG

Not quite sure what you mean by this - are you saying you originally had a PCI-E 8-pin power plug in the motherboard? They're supposed to be keyed differently, because they're definitely not electrically compatible. If so, you might have borked something.

---

I'm saying that when I got the new PSU, their manual stated that there was two choices for the additional 8pin plug on the mobo. One connector which was a 2x4pin (from the PSU) used for singlecore and dualcore and one connector which was a "true" 8pin connector used for "the latest quad cores".

I chose the latter and when I first saw the unstability I decided to doublecheck the psu manual and the mobo manual.

The mobo manual stated that I should use the 2x4pin with 4xGND and 4x12V

The PSU manual stated that the 2x4pin was 4xGND and 4x12, but the "true" 8pin (that I used initially) was 4xCOM and 4x12V.

So, I changed back to the 2x4pin and tried again - but with the same results.

I've now talked to one of my colleagues and he says that the COM is the same as GND, so I shouldn't have fried anything.

Is this correct?

Quote:

---

Originally Posted by Spankyfart

I didn't even see he had a 1200w, still doesn't mean his 12v rail can't be borked though.
Although it's highly doubtful, I would go with a reinstall of windows.
Just make a ghost of his OS in the state it is now, format and reinstall.
If it isn't working, he can always put back the image and try to figure out what really the problem is.

---

I don't see what good a reinstall would do me.

I can boot perfectly at stock settings, so why would a reinstall help me boot at OC'ed settings?

I stated in the OP that the machine always restarts *BEFORE* the machine begins to boot windows - sometimes right after checking memory, sometimes during HD detection, and sometimes when it's just beginning to boot windows.

Sometimes I can get into BIOS and it restarts there!

Again - I seriously doubt a reinstall will help this at all.

Quote:

---

Originally Posted by njl

I will test it using a DMM tonight.

The reviews I've read at jonnygurus.com regarding the Thermaltake I bought did not indicate a lot of fluctuation, so if you can document what you say I'd be happy to read it.

---

I'm sorry I never made a video :P but I make it a habit to test the psu on any pc I build. The TT PSU's are generally pretty decent actually but they do seem to be weak on the 3.3v stability, the voltage will stay in spec but fluctuate wildly across the entire range. I retired a TT toughpower 750 of my own for this reason, I watched the 3.3v on my multimeter while someone else played games or I ran 3d benches and after a while the 3.3v would start to go nuts and the pc would lock-up. As I said I test psu's on any machine I build so I've gotten to see characteristics of many and a fluctuating 3.3v seems to be common but doesn't neccesarily cause problems in all machines.

Your PSU has 4 rails, move things around a bit to lean on the rails equally and see if it helps. I would still test your psu's volts under load and idle conditions, it's quick, easy and may just save you some time.

Just wondering if you installed a new NB or SB backplate on your mobo during the upgrade?

Odd thinks can happin if you have 2 contact points shorting out on a backplate.

Long shot I know. :)

In any case your problem is not because you now have watercooling, unless you have a leak or the mount is really bad.

You need to start eliminating causes. Try 2 sticks of RAM instead of 4, then try using your old videocard and disconnecting other non vital components.

Have you cleared the CMOS? You need to start clocking the system from scratch, 500fsb is quite the achievement. Try higher multi and lower fsb for the same speed. Also, 45nm chips do get worse over time with high VTT / vFSB. I even noticed this on my 65 quad, that would do 3.8Ghz 1.3v at first but then got a lot worse over time. This was with excellent watercooling.

Having a jiggawatt PSU doesnt mean anything, it can still have problems with that particular motherboard for example.

Quote:

---

Originally Posted by Talonman

Just wondering if you installed a new NB or SB backplate on your mobo during the upgrade?

Odd thinks can happin if you have 2 contact points shorting out on a backplate.

Long shot I know. :)

---

I did - there was a new backplate with the GTZ, however that came with an adhesive center pad. I doubt anything is touching the mobo outside the center pad, but anything is possible, so thanks for the heads-up.

The NB block is held down with two screws that are held with a nylon washer and two counter-locking nuts on the back, so no backplate there.

Quote:

---

Originally Posted by v0dka

In any case your problem is not because you now have watercooling, unless you have a leak or the mount is really bad.

You need to start eliminating causes. Try 2 sticks of RAM instead of 4, then try using your old videocard and disconnecting other non vital components.

Have you cleared the CMOS? You need to start clocking the system from scratch, 500fsb is quite the achievement. Try higher multi and lower fsb for the same speed. Also, 45nm chips do get worse over time with high VTT / vFSB. I even noticed this on my 65 quad, that would do 3.8Ghz 1.3v at first but then got a lot worse over time. This was with excellent watercooling.

Having a jiggawatt PSU doesnt mean anything, it can still have problems with that particular motherboard for example.

---

I will try 2 sticks of RAM instead of 4, but I cannot try my old videocard as I've already sold that...:rolleyes:

Yes, I've tried clearing the CMOS, but only to reload "old" settings.

I will probably try to overclock from scratch, I just think it's REALLY weird that it crashes that fast.
I've never seen that in the past - either it wouldn't post or it would simply freeze when booting windows - or it would boot just fine. Never seen these reboots during harddrive detection or inside the BIOS setup.

Problem solved - for now at least.

I disassembled everything from the case and it's now lying on a cardboard box on the floor looking like a bird's nest, BUT now it's running prime@4.25GHz.

I re-seated both waterblocks and disconnected everything but the harddrives, monitors, mouse, keyboard and sound.

I still haven't found the cause of the problems, but at least it wasn't the PSU.
If so - it must be a loose connection somewhere.

Thanks for all your suggestions!

Any suggestions for other software for stability testing besides Prime (that actually works)?

MemTest 3.8 - best RAM<-->NB stressing program
IntelBurnTest 1.8 - Burning CPU only like hell.

OCCT - not the most complete CPU/FSB tester, but it tends to find its errors pretty quickly. Once it's clear, you can run heavier-duty programs to make sure everything is stable.

Quote:

---

Originally Posted by MpG

OCCT - not the most complete CPU/FSB tester, but it tends to find its errors pretty quickly. Once it's clear, you can run heavier-duty programs to make sure everything is stable.

---

Thanks.

Are the heavier-duty programs the ones WaterFlex refers to?

If not - can you tell me which you had in mind?

Quote:

---

Originally Posted by njl

Thanks.

Are the heavier-duty programs the ones WaterFlex refers to?

If not - can you tell me which you had in mind?

---

Yeah. Prime is actually good too, just takes a long time to verify stability - I've gotten more than a few errors after the 10th or even 15th hour of running. But the IntelBurnTest will pretty much get your CPU hotter than anything else, so it's good for verifying maximum temps.