Quad Crazy!! WARNING ALOT of pics

[GTFouts]

Ever thought of replacing those GPU's by 9800X2's? (awfully expensive but with your rig... LOL)

I know exactly why the video cards run a little hot. The EK full coverage blocks do not have screw hole standoffs. So when you tighten down the block, ever so carefully, it does warp the board a little, a very little, and it doesnt make perfect contact with the memory chips. This I discovered when I took the blocks off once and saw that I did not have nice AS5 squares where all the mem chips were, 2 of them to be exact. So I started putting it on and off to see if it was my fault. I decided it was not my fault because the blocks will just not seat perfectly on the memory. Either the blocks are not machined properly or the mem chips are not seated properly on the cards. Go figure. Both GPU's and the interface chip seat nicely tho, as does the rest of the memory.

 

[QuietIce]

It is a quad core, and I read where they run hotter than dual cores when you get above 1.4 volts, maybe thats why, I don't know. Like I said, I am not happy with it, but if you compare it to an air cooled system, it is cooler, or would be if this was an air setup. Can't really speak for others who have air. I would sure like to know what other people's temps are who have the QX9650.

My Q6600 in my other PC runs alot cooler than this chip and its OC'd to 3.6GHz on a p5k deluxe. But its not an extreme chip. Does this make a difference? Again, I do not know, evidently it does...shrug. Water is flowing like crazy, pumps are all cranking at 4400 rpm's, water starts out in the low 20's and ends up in the mid 30's after heat saturation, so there is definetly some heat. I am as confused about it as you and can not discover an apparent reason. All I can deduce is perhaps I have a hot chip.

I'm also running a Q6600 and though it's not as fast as yours, only 3.4 GHz, I've really got the vCore cranked up. It's at 1.52v and has been for over a week now but SETI/F@H load temps are only 40-45°C on a 120x2 heater core and relatively quiet fans. Your speed is ~20% faster but your vCore is ~20% lower so even with an extreme quad I just don't see those temps.

Right now I have all the fans turned down to about 1/3 speed. They are pulling a decent stream of air, if you put your hand in front of the screens, while staying pretty quiet. There are 11 fans pulling air into the case and 1 pulling it out. Definetly have positive pressure for sure and it is not very warm air inside either. Cooler than I would think with these temps.

Typically there's only a 2-4°C difference in intake to exhaust air and if anything yours should be on the low side with all the fin area you have. How big a CPU temp difference do you see with the fans cranked all the way up for 15-20 minutes?

You've got the fans turned down but they're pushing against a lot of resistance since your intake to exhaust is 1:12? Have you taken the cover off for 15-20 minutes to see if that resolves the temp issue ...?

 

[GTFouts]

Hmmmm....not too encouraging at all, 30 min each test time, at Idle.

Covers on, all fans full, 1 to 2 degree difference which is nothing.

Covers off, all fans full, 3 to 4 degree difference on Core 0, 1 to 2 on everything else, which again, is nothing.

It did, however, lower the water temps by 5 degrees, from 30 when I started to 25 right now. But that still did not lower any chip temps.....interesting.....depressing, but interesting.....

 

[Dawgdoc]

I agree those CPU temps are not looking good.

Even though is a MONSTROUS PITA Im sure, I would try a remount and see how it goes. Also I would check each and every fan to make sure its blowing air in the correct direction, and that you like the direction that your air is flowing. If you have fans blowing hot air in the wrong direction it can mess up your temps more than you may think......

Also, post some 3D scores with default resolution. Thats what everyone uses to compare. Noone really runs the higher than default resolutions except for playing around.

Nice work so far! Dont expect your overclock to just "gel" right away with a beast of this size. It will take weeks (or more) of tweaking BIOS settings, remounting, hardware/driver tweaking, yada yada yada.

But....when you finally get it all set for your 24/7 overclock you are going to be LOVING ANY game you play.

What batch # is your QX9650 BTW? If you are running your ram at 1800 you may be pushing the limit on your ram instead of your CPU. Just to see what happens, I would use only 2 sticks and run them 1:1 FSB:Ram and see how high you can get the CPU then.

 

[GTFouts]

I already posted all the different scores with default 3dmark06 settings, one or 2 pages back.

I only have 2 sticks of memory. Had 4 to begin with but I returned 2 thinking they were bad and they refunded me instead of sending me replacements.

Here are all the numbers off of the CPU box:

Prod Code: BX80569QX9650
S-SPEC: SLAN3
MM#: 894493
FPO/BATCH#: L739A646T
Version #: E23438-002 1.25V max.
Pack Date: 12/20/07
EAN: 5 032037 008020
UPC: 7 35858 19757 1

1.25V MAX? Good Lord, perhaps all this reading about not going over 1.4 has some merit to it. But if thats the case, just how are you supposed to OC this chip which was designed for OC'ing? Thought thats what extreme chips were all about, I just don't get it some times. I truely do believe its this MB, have no proof, just a gut feeling.

I'm only having this trouble with a 450 FSB setting. 400 lets me do about anything I want with no trouble, but its only DDR3 1600 then. Now I know that really wont make much difference in the end game, but I paid for DDR3 1800 memory, and the specs of this MB say it will do it, only I am finding out it sure doesnt like too. If I drop the FSB down to 400, my temps and voltages drop drastically, to where they should be. Maybe I'll try that again and see just how high I can get it running at 1600 instead of 1800. I'll almost bet I can go 11x or 12x before I get back to these temps and voltages. well, maybe not 12, but 11 for sure, 12 would be pushing it I think.

I guess I am just disappointed that I cant get a stable 4.5GHz with this MB, chip and WC setup while keeping my temps down to normal or a little above normal. I originally thought I could do that easy and that getting too 5GHz would be the pain. Boy was I totally off base on that one!

Just in case you're curious:

Ai Overclock Tuner....................Manual
OC From Memory Level Up.......Auto
CPU Ratio Setting.....................9.0
FSB Frequency..........................450
FSB Strap to NB........................333
PCIE Frequency........................110

DRAM Frequency.......................DDR3-1801
DRAM Command Rate...............2T
DRAM Timing.............................Manual

..............Primary Info.....7-7-7-20-7-60-10-5

CAS# Latency.................................7
RAS# to CAS# Delay.......................7
RAS# Precharge..............................7
RAS# Active Time............................20
RAS# to RAS# Delay.......................7
Row Refresh Cycle Time..................60
Write Recovery Time.......................10
Read to Precharge Time..................5

..............Secondary Info.....8-5-6-4-7-4-7

Read to Write Delay (S/D)...............8
Write to Read Delay (S)...................5
Write to Read Delay (D)...................6
Read to Read Delay (S)....................4
Read to Read Delay (D)....................7
Write to Write Delay (S)...................4
Write to Write Delay (D)...................7
DRAM Static Read Control.................Disabled
DRAM Dynamic Write Control............Disabled
Ai Clock Twister................................Strong
Ai Clock Skew for Channel A.............Auto
Ai Clock Skew for Channel B.............Auto
Ai Transaction Booster......................Auto


CPU Voltage.....................................1.43125
CPU PLL Voltage...............................1.62
NB Voltage........................................1.57
DRAM Voltage...................................2.04
FSB Termination Voltage...................1.54
SB Voltage.........................................Auto
Loadline Calibration..........................Enabled
CPU GTL Reference...........................0.63
NB GTL Reference.............................0.67
DDR3 Channel A REF Voltage............Auto
DDR3 Channel B REF Voltage............Auto
DDR3 Controller REF Voltage............Auto
SB 1.5v Voltage ...............................Auto

NB LED Selection..................NB Volt
SB LED Selection..................SB Volt
CPU LED Selection ...............CPU Volt
Frequency LED.....................Enabled
Voltminder LED....................Enabled

CPU Spread Spectrum...............Auto
PCIE Spread Spectrum..............Auto

 

[QuietIce]

Just for grins you might try 4.2 GHz or so with the lid off - maybe even a little active cooling across the board, like an external fan. I still think it's the air-flow, partially supported by the drop in temps you did see. Hmmm, maybe flip the fan directions on the rear rads ...???

 

[GTFouts]

I've thought about that, but dont like it much. The rear rads are the CPU and NB rads. Then I would be pulling all the hot air from inside the case across them. But I'm still thinking about it. I am going to wait until I get the blackops and see what it does. If I still have this issue, then I will give it a go. Hell, can't dance, might as well flip some fans.


I may end up having to put some fans in the covers. Sure hate to cut holes in the acrylic panels tho. We shall see.......

 

[Dawgdoc]

A few more suggestions GTFouts.....

Put both of your spread spectrum (CPU annd GPU) to DISabled.

Play with your GTLRef settings. I see you have them on something other than auto. What I typically do is keep those at auto, find my max overclock, and THEN play with them to see if I can gain a few extra mhz. (thats just what I do, others may do differently).

Thats a decent batch QX9650 of week L739A. The final #s of the batch (646) are a bit sketchier, but I think you should get CONSIDERABLY more with an L739A batch of QX9650.

1.4v max is nonsense IMHO. I would put that CPU higher under CONSIDERABLY more volts, but thats just me. Under good water (which you have) once you get your mounts/fans all the way you like them I would personally max this chip at 1.6v for 24/7 usage, but thats just me and I am known to really like the volts . I would only use 1.6v if it gains me more overclock, and if it doesnt I would use the least volts possible to achieve the max stable overclock I am looking for.

It is VERY strange IMHO that your temps drop when you decrease your FSB. Temps should be strictly a function of volts, cooling, and ambient temp. FSB and amount of overclock shouldnt factor into temps in any way at all. But hey...Im not there you are so you would know more than I about your specifics.

Check your secondary RAM timings with the manufacturer of your ram, and check on the ram manufacturers forum to make sure your subtimings are not hindering you. For me personally, unless I am benching and looking to eek every last mhz or 1/100th of a second on a bench I keep all subtimings at auto for my 24/7 rigs.

Have you tried a max FSB overclock yet? Drop your multi as low as it will go, or at some reasonably low multi such as 6X or 7X. Then see what your max FSB that you can acheive with the CPU mildly overclocked. At 6X multi you can get 500fsb and still have that CPU at stock so it shouldnt be a limiting factor. That will allow you to see if it is your CPU, or your mobo holding you back.

Check your ram volts. 2.04v looks low to me but I havnt really used much DDR3 so I could be way off. Start with stock volts as per manufacturers recommendations on your ram, and then if you are going to overclock it at all bump it "a bit".

I would also personally set your FSB termination voltage to auto, but I dont think that is likely causing any problems at the low volts you are currently using.

Good luck man!

 

[QuietIce]

I've thought about that, but dont like it much. The rear rads are the CPU and NB rads. Then I would be pulling all the hot air from inside the case across them. But I'm still thinking about it. I am going to wait until I get the blackops and see what it does. If I still have this issue, then I will give it a go. Hell, can't dance, might as well flip some fans.


I may end up having to put some fans in the covers. Sure hate to cut holes in the acrylic panels tho. We shall see.......

You could always go reverse of "normal" and put the rear fans on intake and the front on exhaust. My fans are that way ...

 

[GTFouts]

I am finding out, thru playing with this pig of a MB, that it has issues, bad ones. When I cant go above 450fsb using 6x no matter if I set things to auto or set them way high, then to me that points right to the MB. It wont work for anything set above 450, no matter what multiplier I use. This mb should be laughing at 6x-anything. I mean, it wont do squat at 6x460 and thats BELOW stock speed for this chip...pathetic.

If 450 is the ceiling for this MB, then trust me, its going to get drop kicked right out the back door. Especially when I have read numerous reviews where they were topping out in the 500's for the max FSB test. Truely is frustrating I tell ya.

 

[Dawgdoc]

450 is super low for FSB maxout.

Just as an example (not a good one) my Mars can do 600+. Of course, thats a P35 chipset and your Asus Maximus Extreme is an x38.

To be honest, the more and more I read, and the more and more people post results, I cant EVER see myself getting an x38 mobo. At least not when x48 is out there

Try increasing your NB volts an see if you can push further.

Also try increasing your PLL volts.

Additionally, some mobos have a "hole" between certain FSBs. Try jumping up 25 -40ish fsb and see if that works. Go for 475, and if that doesnt work try 500.

Ya never know...

 

[GTFouts]

Ah well....I had to try...kinda felt this MB was Da Debil! I have tried every combination I can think of. I even set all voltages to 1 notch below red and went thru 451 thru 550. It wont even show initial post video as soon as you go past 450 on any multiplier. 450 is this ones max. I hope it enjoys its very short life span. Kinda figures I'd get not 1, but 2 lemons. Figure the odds on that!!!! First time I have ever had problems with an Asus MB....Geesh.....

It works and will stay on 9x450 until foxconn decides to release the blackops. What more can I say......

 

[Perseus]

Looks nice! I am interested in the fittings used, and the converters used on the PA with teh fittings?

And guys, how much it costs is not of your business... would you rather he not post because you make a big deal about cost? I will delete those posts on site from now on, and, if you catch me in a bad mood I will ban you for 3 days.

Wow, things have changed a bit around here during my break! Well said nik, and DAMN fine box, GT.

 

[Vagabond102]

Awesome build man! I just read the entire thread. Sorry for your issues.

I did see something in your post #165. It shows your PCIE at 110. That sucker needs to be locked to 100, or in some cases, 101. There are very very little gains over default that I am aware of, and it can really cause havok. I don't know how this board handles it, but some the FSB will modify the PCIE frequency.

I could be wrong here, but just another Item for you to try.

 

[Hazaro]

Awesome build man! I just read the entire thread. Sorry for your issues.

I did see something in your post #165. It shows your PCIE at 110. That sucker needs to be locked to 100, or in some cases, 101. There are very very little gains over default that I am aware of, and it can really cause havok. I don't know how this board handles it, but some the FSB will modify the PCIE frequency.

I could be wrong here, but just another Item for you to try.

Clock it up to match his CPU speeds.
I forget where I heard that though.
Mine is running 110

 

[Lukee]

Wow, what a project.

I thought that the temps would be lower with all that cooling in there. Is there a point where more radiators is "bad"?

 

[GTFouts]

Has nothing to do with the radiators. I finally did find why the temps were so high at idle. Believe it or not, when running Everest it adds 8 to 10 degrees at idle. I couldnt believe it but its true. If I turn everest off and just have Coretemp running, BAM, temps drop down to the mid to low 30's....Go Figure....

Anyway, I have my BLACKOPS now...Muahahahahaha...going to try to install it tomorow. Man will I be glad to be away from this Maximus, so disappointed in it. But thats ok...live and learn. I will say this, it has hurt my confidence in ASUS, and I was a hard core Asus user....sigh.....

I dont see any reason I can have this monster cookin at 4.5 easy. I am hoping to get it to 5GHz on this water setup, but, that may be a bit much for it. The blackops is an amazing board from what I have read, so we will find out what its maxes are for water. Like I said before, I'll be quite satisfied with a stable 4.5, jumping for joy with 5! Hey, a man can dream, right? We shall see....

 

[Lukee]

Great. Good to know.

 

[Mpegger]

Everest is a resource hog for some reason. I've noticed fps gains in most games just by turning it off. And not just 1 or 2 fps, sometimes as much as 20+ fps depending on the game. Wish there was something better though cause most of the other programs I've tried are either outdated, crash prone, or do not/can not monitor as many different parts of my PC as Everest can.

 

[necrokiller]

May I ask how much money you've spent on this, so far?