Quad Extreme project

[DaMulta]

i am 3dmark stable at 955 core at 1.45vgpu @ 62c load, as long as your volt regs/ram is sufficiently cooled, i see no reason why you cannot get those speeds, no reason at all.

yeah, the full coverage is a drawback in that sense, I was quite tempted by crossfire 3870's. but meh, I lub my card , and i doubt id get anywhere near what i paid for the card + waterblock. damn this habit is expensive.....

however dx10 is a bit more picky, I cant run much above 900 in dx10 without small artifacts, but that was only tested with the bios's 1.35v, maybe more vgpu would help.

my ram voltage is veried atm, still trying to work out what the hell it likes. I have ran 1200 on my other cpu @ 1.4volts, been running around 1.35ish atm just to make sure that theres no instability. 4.6GHz at that fsb is dissapointing tho, i just cant seem to get a high fsb without crashes, im sure im doing something wrong... AMD overclocking was SOOOOO much easier

Yea running at 4.6 stable must really suck.......:shadedshu
O I wish i could hit 4.6

For the HD
I found that 1.25 was better on AIR at 900Mhz. I still had artifacts, but less of them. I seen some of the highest OC bios have lower volts.

Have you tried 1.25-1.30v yet with your card?

 

[Judas]

Good idea sticking the rad on the side never thought of that one
Blooming big job how long did it take you?

 

[mandelore]

Update

length of time, a good full weekend, plus waiting for parts, which im still doing.

I have got my new peltier, took apart the Arctic Web and installed. Just waiting on my new pump for the TEC loop. Been in touch with Todd quite abit (the nice guy who makes them)

When i took out the Arctic Web module I reasembled with secure worm gear 15mm clamps, so every tube now is 100% secure against leakage. Tested the TEC unit and it works. (unless fate has another idea)

I am, quite frankly going bonkers not being able to use my pc, currently typing from my folks pc, a ( ) quality Dell E3010 machine.

I aim to have some nice reports, pretty damn soon...

Believe my pump will arrive within the week since they were out of stock.

will keep everyone updated, getting restless not being able to bench!!!

Mand

 

[Irish_PXzyan]

holly smokes
ive never seen anything like that before!
You got a sexy QX9650 and overclocked to 4.6GHZ .....my god....I....I feel really pissed off

iam going to find your house and take all your stuff

 

[DaMulta]

update?

 

[mandelore]

update?

sort of....

after much waiting i still did not recieve the pump, emailed the website and cancelled my order. i got a Liang D5 Vario with a 3/8" adapter kit, so that will be here within a couple of days and i can start getting my pc working, at frikkin last....

I know the TEC module works, well, as in that there is continuity within the module unlike the one I replaced, as that was a certifiable dud.

I think the new pump will be better anyway, as I can get upto 1500L/hour with it, my old pump did 1200L/h (D4) and the other one i was ordering was a modded DDC to do 900-1000L/h.

Once it arrives, i simply need to refill the loop with coolent, then switch it on

and at last I will be able to get some benching done and check for load temperatures at various clock speeds and voltages. wewt.

 

[mandelore]

not happy, and extremely confused....

emailed the guy who makes the arctic web block.

new pump arrived, everything is set up correctly, correctly powered etc..

getting similar reported temperatures to when the pelt block was not working at idle.. as in no single figure temperatures or very low idles. similar shitty temperatures under load.

NOT HAPPY, NOT FRIKKIN HAPPY.

so far what a bleedin waste of money, time, effort and just frikkin everything. Wish i had just went for an Apogee GTX or something, just sick of this

 

[Random Murderer]

not happy, and extremely confused....

emailed the guy who makes the arctic web block.

new pump arrived, everything is set up correctly, correctly powered etc..

getting similar reported temperatures to when the pelt block was not working at idle.. as in no single figure temperatures or very low idles. similar shitty temperatures under load.

NOT HAPPY, NOT FRIKKIN HAPPY.

so far what a bleedin waste of money, time, effort and just frikkin everything. Wish i had just went for an Apogee GTX or something, just sick of this

that sucks... you're SURE the pelt is working?
and not to add insult to injury, but i told you so....

 

[DaMulta]

not happy, and extremely confused....

emailed the guy who makes the arctic web block.

new pump arrived, everything is set up correctly, correctly powered etc..

getting similar reported temperatures to when the pelt block was not working at idle.. as in no single figure temperatures or very low idles. similar shitty temperatures under load.

NOT HAPPY, NOT FRIKKIN HAPPY.

so far what a bleedin waste of money, time, effort and just frikkin everything. Wish i had just went for an Apogee GTX or something, just sick of this

What happens if you just run the block off of the cpu?

Does it freeze?

 

[CDdude55]

Did you set the Motherboard on the motherboards box? Or use the anti static bag to put the motherboard on?

 

[ShadowFold]

Did you set the Motherboard on the motherboards box? Or use the anti static bag to put the motherboard on?

Im pretty sure you have to put it on the anti static bag

 

[asb2106]

alright after reading every post here Im kinda confused, Ive not been happy because Im hitting a wall with my q6600 at about 3.6~3.7 with an apogee GT, and my temps are about 46 C and about 55 on all the cores. Those are temps after 100% load for 24 hours. Would I benefit from doing a TEC. I really really want to but it almost seems not to be worth it

 

[DaMulta]

His isn't working right still. His last TEC system(225watt) stayed in the -s on a hot chip.

 

[Wile E]

not happy, and extremely confused....

emailed the guy who makes the arctic web block.

new pump arrived, everything is set up correctly, correctly powered etc..

getting similar reported temperatures to when the pelt block was not working at idle.. as in no single figure temperatures or very low idles. similar shitty temperatures under load.

NOT HAPPY, NOT FRIKKIN HAPPY.

so far what a bleedin waste of money, time, effort and just frikkin everything. Wish i had just went for an Apogee GTX or something, just sick of this

Look in the Wolfdale/Yorkfield performance thread. Everybody is getting bad temp readings on these 45nm chips. You're gonna have to verify the setup with a probe. Also, try what dm said, and run the pelt loop with the block off of the cpu, to see if it actually freezes.

Also, in my research, I have found that most quads won't stay stable much over a 400fsb. I know my quad doesn't like a fsb over 400. That's why you see the 3dMark06 record runs done with high multis and low fsb. I say stick with an even 400fsb (give or take a few Mhz), and use the multis from there.

I'm eagerly awaiting for you to work out the bugs. I feel the TEC bug starting to bite again. lol. Need a better quad tho. I'm gonna try to get a QX9650 with my tax return, but I have to buy another vehicle first. If I can't swing the 9650, I think I'll try for a Q6700.

 

[DaMulta]

I'm really wondering if the meanwell is working right. If this 2ed TEC isn't freezeing it makes you wonder.

 

[mandelore]

Temperature sensor issues to blame?

Right, I noticed something interesting. My bios was showing the same as core temps lowest temp reading until i reflashed to the latest bios (907)

it then showed 9C in the bios, the core temp readings did not change at all, but, speedfan showed 9C for the cpu temperature also, as well as the "cpu" temp from occt on its results graph.

I did some poking around and found a few people with MY motherboard and the QX9650 that have had temperature sensor problems, but only with the QX9650.

So... I decided to run OCCT , speedfan and core temp and do some testing.

I have made three screenshots:

1) Idle with the TEC unit powered up:

2) Load with TEC powered up:

3) Load with TEC unpowered:

And I actually got some interesting results:

The speedfan reported temperature matches the initial bios reading, which incidentally matches nothing shown in core temp, the latest speedfan does not require temperature offsets to be made on core2 processors. and the load temperatures on the "core" readings are so drastically different to the cpu temperature, which does scale with load, and when the TEC is unpowered. (everest apparently doesnt make sense of either correctly)

15C 100% load on each core looks far better than 40+

what do you guys think?

 

[asb2106]

that sounds like a good result to me. and ~30 looks great on the cores! Try to do some OCing and see how it goes up, I am suprised though that your temps didnt go below 0 during IDLE, I thought TEC was supposed to freeze up when at idle

 

[mandelore]

that sounds like a good result to me. and ~30 looks great on the cores! Try to do some OCing and see how it goes up, I am suprised though that your temps didnt go below 0 during IDLE, I thought TEC was supposed to freeze up when at idle

31C was with the TEC upowered, it got to a max of 15C under full load on each core.

but i still need to know if i can trust that cpu temp, and that coretemp is indeed innacurately detecting temperatures

 

[asb2106]

31C was with the TEC upowered, it got to a max of 15C under full load on each core.

but i still need to know if i can trust that cpu temp, and that coretemp is indeed innacurately detecting temperatures

the 15C was the CPU, the 30's were the CORES, the temps on the cores are always higher and they are they number that us really important

in everest you will see 14 on CPU and 30's on CORES, be sure to check the cores

 

[asb2106]

the 15C was the CPU, the 30's were the CORES, the temps on the cores are always higher and they are they number that us really important

in everest you will see 14 on CPU and 30's on CORES, be sure to check the cores

notice

 

[mandelore]

the whole point is that i think the "core" readings are bugged. the unpowered idle/powered idle are the same, the numbers ALWAYS reset to the EXACT numbers seen in coretemp, eveerest numbers just dont budge as it dont read anything correctly

Edit:

I believe it may be a problem with the motherboard/cpu combo, as on a copuple of other sites peeps have reported similar occurances.

im just trying to find a set of numbers i can trust, as at 4ghz, the "core" values are similar to air cooling!

 

[asb2106]

the whole point is that i think the "core" readings are bugged. the unpowered idle/powered idle are the same, the numbers ALWAYS reset to the EXACT numbers seen in coretemp, eveerest numbers just dont budge as it dont read anything correctly

Edit:

I believe it may be a problem with the motherboard/cpu combo, as on a copuple of other sites peeps have reported similar occurances.

im just trying to find a set of numbers i can trust, as at 4ghz, the "core" values are similar to air cooling!

core temps are always higher than the CPU temp, and a probe will tell you surface temp, the cores are internal, who knows, maybe there is a bug with the qx but every dual/quad proc Ive ever had the cores are always more than the surface CPU temp

 

[DaMulta]

Seems high for the wattage of that TEC to me

 

[Random Murderer]

Seems high for the wattage of that TEC to me

me too. i still think you should pull that sucker off and see if it freezes...

 

[pt]

me too. i still think you should pull that sucker off and see if it freezes...

second that