QX6700 ES up and running...

[Duvie]

I just got it....PLopped it in and forgot to set the bios back to default....so it booted up at 8x424 = 3.4ghz...

It is a 0624 and B0 revision stepping 4....but it is a reviewer chip...perhaps cherry picked...

vcore was set to 1.48v and it droops to 1.39v under load...the biggest droop I have seen so far on this board...

Running with a 600watt OCZ powerstream...

+12v = 12.11v
+3.3v = 3.36v

everything looks good...

Been folding 4 instances for the last 30 minutes or so....


Temps

easytune says 55-57c
coretemp says 71-74c....tjunction on this chips is 100c not 85c like the core 2 duo....

 

[Duvie]

Gonna try lower vcore later and try to get temps to 50c or less on the easytune anmd 65c or less on coretemp....May put a bit faster 120mm fan on the ninja..

 

[Markfw]

Go Duvie !!! Looking good ! Besides, I need my 6300 from ya !

 

[Duvie]

I just scored 1795 in cinebench using 1 stick of memory in single channel..i am getting much better performance out of the ballistics...That is at 848ddr and cas 4....gonna try cas 5 and 2.5x multi

 

[Duvie]

It definitely may be the sign of the DS3 mobo but reboot issues at 3.4ghz and above...big v droop you dont see with C2D...


3.2ghz 8x400 w/ 1000ddr2 has been running for 1hr......vcore drops to 1.32v though 1.4v set in bios....

core temps are 66-69c...so that is better...I think I was perhaps throttle or the sign I was on the edge at 3.4ghz as running cinebench I was getting 1250 when I should ave been hitting 1790's....yes F@H was running but with 3.2ghz and 4 instances running my cinebench scores are 1680-1700 consistently....

I dont know if the power supply is the issue but the 12v rail is strong and little movement...

I am going to let it run at 3.2ghz over nioght and maybe try to drop vcore to 1.375v in bios...I would be happy with 3.2ghz...runs too hot IMO at 3.4ghz if coretemp is right....It is running almost 15-17c higher then Gigabytes easytune is reporting. When the E6400 was in there easytune and core temp were within 10c...


Edit: I notice my F@H times are running faster as well..on my WU for cpu#3 it took 17min to get 1% done and now it only took 13min with the cpu runnig 200mhz slower...cpu#2 I saw a similar 3min decrease...

Maybe this was a sign of throttling or a power issue from the board....

sandra scores and cinebench scores are looking right on target with reviews now.

 

[Duvie]

Looking to pick up a 750watt PC Power and cooling right after xmas.....Maybe even look at a water cooling case for this bad boy

 

[The-Noid]

the problem is the stepping. Earlier versions of the Kentsfield couldn't handle high fsb. They tend to do best under 400, so your goal of 4x800 is a good one. Ninja is made to dissipate 130w with the stock fan, so you need to get something with higher voltage. That processor is dissipating about 210w at your current settings. If you like I could give you a C2D Quad stock fan, I actaully had very good luck with it. Highest I ever saw at 8x400 was 57.

 

[Duvie]

Yoxxy is the stock quad cooler better then a Ninja or should maybe I try to reapply? I think I might. I am not sure how the quads layout under the IHS is and I did use the line method. maybe with the quads I need to coat the entire top for better assurance?

From 1 core to another I have up to 4c difference...All my core temps are below 70c now with 10 hours+ of F@H under its belt...Gigabytes temp guage is saying 52-53c


Any comment on why the tjunction is higher, and if you think I was throttling or just was it a power issue.

 

[The-Noid]

I believe it was just a power issue. The quadcore heatsink is a bit better, becuase to my knowledge it is the only one tuned for the high heat load. The others are tuned for getting a low heat load out fast (keeping cpu cooler). The quad cores sit under the LGA IHS about 4 mm apart and about 7mm from each edge. The best bet is to put two vertical strips on the beginning of Intel and at the end of Confidential. Make the strips approximately 1/2 inch in length.

Tjunction is higher just because the chips are binned a bit differently. For the same reason that the best overclocks come from X6800, the QX's are made to run a bit hotter. There is a program called throttlewatch that will tell if your CPU is throttling, but you have to be pushing 100's on the Quad before it will start.

PM me though, you can literally have that heatsink for shipping costs . I have very little use for it.

 

[cmdrdredd]

People over an XS are saying the P5B handles the new quad chips better than other boards right now. I'm assuming because of the vdroop issues and other concerns you expressed. Plus People mod the hell out of their P5B boards anyway.

 

[The-Noid]

He has a old stepping though. B0 had serious FSB problems on a lot of boards. It is multiplier unlocked so not really a big problem.

 

[Duvie]

I will try some higher multiplier lower fsb later once I get 24 hours under my belt at this speed....

 

[Technonut]

Quad action.... Sweet....

If I did not already have this X6800, I would be all over the QX6700.

Duvie: Once you get your OC settled, make some AutoGK runs and post the results...

 

[lopri]

What issue are you exactly having? (lower performance due to throttling?)

It's amazing that DS3 can handle Kentsfield up to 424FSB.

 

[Duvie]

Originally posted by: lopri
What issue are you exactly having? (lower performance due to throttling?)

It's amazing that DS3 can handle Kentsfield up to 424FSB.

I have not verified if there is actually any throttling....It feels like diminished power with the cold restart issue above 400fsb....

Until I try lower multipliers and get back up to a higher speed I cannot say for certain the issue is power, heat or a combination of both....

I think if it is a power issue it is a DS3 issue...This is not a c2 revision...

 

[tersome]

Originally posted by: Duvie
It is a 0624 and B0 revision stepping 4....

Ouch... Stepping 4 qx6700... bet it's a hot mofo.

I remember people having trouble getting these working properly at high fsbs. Hopefully a ds3 can handle it.

 

[Duvie]

24 hours so far so good....full load.....I need to reapply AS5...temps are under 66-69c with core temp under load for 24 hours...gigabytes temp guage says 52c

Yoxxy was kind enough to show me that the cores are running opposite direction of the C2D...To try applying 2 lines....

The guy who had it ran it on a P5B and said they were able to get a lot out of it....

 

[Duvie]

reapplication helped idle temps by about 3-4c in core temp using Yoxxy's method....Load temps still around 68-66c in limited time...

Howver that was now at 10x330 = 3.33ghz with 1.4v set in bios...

ran the benches fine, but we will see...Rumor has it this chip booted much higher on an Asus mobo.....

 

[Duvie]

pretty much anything with an 11x multi doesn't work, or least wont work at the same speed I can get 10x to work...

I have tested 10x330 and it needed at least 1.425v....I tried 10x335 w/ 1.45v and I noticed similar slow downs aas described above...I downloaded Throttlewatch and it shows no throttling. Problem is multiple apps slow each other down. FH work units are taking significant 30% longer then done at 150mhz slower speed. Sandra is off by 50% if ran while F@H cores are running. They are low priority and should relinquish to sandra. Also similar to cinebench testing which can drop by as much as 25%. TMPGenc drops by 25-30%...

NOw if I run it at 10x325 with 1.40625v I can do tjose apps fine while folding and get scores that are right for that speed...

 

[The-Noid]

10x330 is nothing to sneeze at, especially on that board. What bios are you using as well.

It is possible that your board is maxing out the bus on MHZ. No matter what the cpu/multiplier is. That is what would happen on my older stepping. It would not pass 3.3 ghz regardless of what I tried for multiplier. I simply thought it was the motherboard but it could be in fact that the stepping could not pass a certain amount of mhz. No matter what multiplier and ratio I would try it would always die around 3.4 ghz.

 

[Markfw]

Originally posted by: Yoxxy
10x330 is nothing to sneeze at, especially on that board. What bios are you using as well.

It is possible that your board is maxing out the bus on MHZ. No matter what the cpu/multiplier is. That is what would happen on my older stepping. It would not pass 3.3 ghz regardless of what I tried for multiplier. I simply thought it was the motherboard but it could be in fact that the stepping could not pass a certain amount of mhz. No matter what multiplier and ratio I would try it would always die around 3.4 ghz.

Nope, he has run that one at 500x7 I am pretty sure. 429 in his sig even.

 

[The-Noid]

Quad though not the C2D. For some reason Quad handles bus differently because of the huge amount of data. On p5wdh Quad could do 330 while an e6700 could do 420. Somewhat weird deal they have.

 

[Markfw]

They need hypertransport and an IMC. Add that to what Conroe does already, and you have a killer chip ! Since AMD has those already, if K8L is similar to Conroe in efficiency, it will be killer !

 

[The-Noid]

To be perfectly honest, I think the implementation of the QuadFX is superior to the Quadcore. It is pricey, Yes, but for people purchasing either of these setups in most cases money is not an object. Having 2 Conroes with indendant FSB/Memory Controller/Memory would allow for higher overclocking potential, and IMO better speeds. Logistically its not as great of an idea but with 2 cores strapped together the way it is now, there is no reason why they couldn't do it similiar to the woodcrest. In the end a single quadcore with 12 megs of cache coming q3 2007 is the only real quadcore people should be buying for a while.

Right now Quadcore is a novelty, not a neccessity. It is great for F@H, other DC projects, specialized uses, or 3dMark06. Otherwise you are still better off using a dual-core setup.

 

[sdsdv10]

Duvie,

You noted previously that you have Pinnacle Studio 10. Is this program multi-threaded enough to use all four cores when encoding DV to MPEG-2 (for DVD generation)?