E6700 @4605mhz - Kayl Black Frozen SS

Finally playing with phase change for intel cpus

System:

• Intel Core 2 Duo E6700 QPGA B1 Step 5
• Kayl Black Frozen SS 1/4HP
• Asus P5W DH Deluxe 0801 bios
• 128MB Gainward FX5200 PCI card
• 2x1GB Gskill PC6400PHU2 2GBHZ Micron D9 (cooled by 120x25mm Thermaltake 81cfm fan)
• 74GB WD Raptor 8MB cache
• Pioneer DVD-RW
• FDD
• 700W OCZ GameXStream Psu

I spent all the time doing up my insulation around cpu socket and then I realised i forgot to install the 4x mounting threads for the evap through the motherboard LOL

Had to remove all the insulation and mount the threads and then put the insulation back on hehe.

As you can see from pics here http://i4memory.com/phasechange/kayl/insulation/p5wdh/html/, I still need to reinstall the Radiical chipset waterblock on MCH/NB etc so more to do for tomorrow

Performix Red LET


First top layer of insulation & dielectric grease in cpu socket only

 

Second top layer of insulation using thin sticky back foam (Aerotape) - whoops forgot to mount the threads to the board


Removed insulation, mounted threads and reinstalled insulation

 

First dry test mount - see gaps in middle pic below what should i fill it in with or do i need to ?


Interesting when setting up my phase change gear and transporting the chipset cooling loop over to the new position, I noticed the MCP350 pump outlet barb tubing had been kinked for a while now as you can see the tubing near the outlet barb had memory effect of a collapsed tubing... maybe this is a possible reason why i didn't gain much in max fsb on water cooled chipset ?? Seems now that i have fixed the kink, FSB scales better with vMCH but could this be due to phase change on cpu or both the unkinking of tube + phase change gave me better FSB scaling with vMCH ?

Very quick windows, Systool dual 32M PI (cpu benchmark) tests gave me:

10x multipler
4240mhz at 1.6v with 1.65v vMCH
4330mhz at 1.65v with 1.65v vMCH

9x multipler to test max FSB
1.75v vMCH gives me 452fsb max which is already 3fsb higher than my 1.85v vMCH before (due to tube kinking ?)

Since it could pass Systool dual 32M Pi test, it should be able to pass Super Pi 32M too
@9x452fsb


@10x424fsb


These are untweaked runs just to test limits of cpu on phase change cooling. Working on >4.3ghz right now seems to a tad flaky at 4360mhz at 1.65v. Seems i messed around with the mounting pressure too much and the max clock varies now a bit - so looks like I might need to remount a bit later but for now seems around 4300mhz at 1.65-1.675v is where I am at

Evap temps
@idle = -34C to -36C
@load = -32C to -34C

 

Let's start with 10x430fsb = 4300Mhz at 1.675v (result of bad mount pressure - loosened wing nuts and dropped from 4350mhz at 1.65v stable to this heh)



Not bad, but of course memory using upclock divider will have a much faster time

Max 1M so far still in same windows session as 4300mhz at 1.675v. Used clockgen to clock to 4524mhz - 10x452fsb

 

Now if ignore what Systool dual 32M Pi says my max is (4300mhz), and just run single 32M I can hit 4403mhz at 1.675v bios set



Just shy of <12min time

Bump vMCH from 1.75v to 1.85v in bios and can do 1M @4605mhz at 1.675v bios set vcore



11 seconds flat



Some async 32M times - 12min barrier has been broken

@4240mhz


@4282mhz


@4326mhz




hmm with 0801 bios still but 1.7v vcore in bios when set = 1.55v.. seems the max i have to work with is 1.675v bios option.

Any recommendations on bioses to try ?

 

shiiiiiiiiiiiiiiiiiit

that extra multi helps alot tooo

i think u need sub zero cooling on the nb too

 

10x sure does help

First sub 14min dual 32M time

 

Well still on 0801 bios and managed to squeeze out enough to move up hwbot & holicho rankings a bit more



All at 1.675v in bios

#6 on hwbot 32M http://www.hwbot.org//hallOfFame.do?type=result&applicationId=7
#8 on holicho 32M http://holicho.lib.net/pi3355/pi3355.htm

 

801 is de best????


i'm on 1305 or sum shit hehe

 

wow. 4 Ghz......

 

thanks

some say it's best my prob is 801 bios 1.6875v and 1.7v vcore = 1.45-1.55v vcore.. need to update my bios for those 2 vcore options to work.. hopefully get 100mhz more out of my cpu with higher vcore

 

Nice stuff Eva.. I been waiting to see if you would get a better Chip than the other ones you have tested so far..

thats an awesome speed for an awesome chip. well done.

 

Thanks

I flashed from 0801 to 1101 bios so that i could fix the 0801 vcore bug where 1.6875v and 1.7v vcore options resulted in 1.456v and 1.55v vcore in reality. Seems in bios with 1101 bios, 1.6875v and 1.7v are now same as 1.675v vcore setting, fluctuating between 1.648v - 1.656v - 1.662v vcore.

But I have gained an extra 43-49mhz at max Systool dual 32M Pi and single Super Pi 32M max

0801 bios
Systool dual 32M Pi max = 4300mhz at 1.675v bios set
Single Super Pi 32M max = 4403mhz at 1.675v bios set

vs

1101 bios
Systool dual 32M Pi max = 4349mhz at 1.7v set
Single Super Pi 32M max = 4443mhz at 1.7v set

New max cpu clock on 1101 bios for 32M.



evap temp
idle = -38.2C to -41.0C
load = -31.7C to -34.4C

Asus PC Probe2 temp
idle = 230C minus 256C = -26C
load = 253C to 256C minus 256C = -3C to 0C

 

4.6.. is that all?!

Nice work chief.

 

Good going - certainly seems the 6700's have somewhat more headroom than the 6600's.

And great to see the heatload the Conroe Core is subjecting the phasecooler to. Not bad, I think a VapoLS should do a commendable job with one of these cpu's?

Now all we need is better DDR-2 ram to help with FSB bottle necks on the Intel architecture, so we can run ram higher than fsb.

Should help a lot with SuperPI

 

thanks Phillip - seth filled me in

all i need is more vcore and vdimm to go higher hehe

 

Phase change is fun thanks to Kayl Black Frozen SS 1/4HP phase change unit and a nice E6700 cpu

Moved up to #5 on hwbot 32M rankings and moved to #1 non-X6800 ranked 32M time on hwbot

System:

• Intel Core 2 Duo E6700 QPGA B1 Step 5
• Kayl Black Frozen SS 1/4HP
• Asus P5W DH Deluxe 1101 bios
• 128MB Gainward FX5200 PCI card
• 2x1GB Gskill PC6400PHU2 2GBHZ Micron D9GMH (cooled by 120x25mm Thermaltake 81cfm fan)
• 74GB WD Raptor 8MB cache
• Pioneer DVD-RW
• FDD
• 700W OCZ GameXStream Psu

E6700 QPGA B1 @4416Mhz at 1.7 bios set = 1.648-1.656v vcore
2x1GB Gskill PC6400PHU2 2GBHZ Micron D9GMH @552Mhz 4-4-4-4 at 2.51v (max pencil modded vdimm).







evap idle = -38.6C to -38.2C
evap load = -34.4C to -31.7C
Asus Probe 2 idle = -24C to -26C
Asus Probe 2 load = 0C
room ambient = 26.8C

 

No problem, thanx for your input too, George!

Anyway great score to be no5, sure seems a nice overclocker you have.
If I can get a 4400mhz chip on phase, I'll be happy.

Need to raid swapmeets again, both my 6400 and 6600 hitting ceiling around 3600-3800 with water. Probably around 4gig with phase, so no good.

So, the search continues.....

Keep it up, looking excellent!

 

no #4 now

Today it's a slightly cooler day in Brisbane = good for benching

evap idle = -40.8C to -39.2C
evap load = -34.6C to -34.1C
Asus Probe 2 idle = -26C to -28C
Asus Probe 2 load = 0C
room ambient = 24.9C

System:

• Intel Core 2 Duo E6700 QPGA B1 Step 5
• Kayl Black Frozen SS 1/4HP
• Asus P5W DH Deluxe 1101 bios
• 128MB Gainward FX5200 PCI card
• 2x1GB Gskill PC6400PHU2 2GBHZ Micron D9GMH (cooled by 120x25mm Thermaltake 81cfm fan)
• 74GB WD Raptor 8MB cache
• Pioneer DVD-RW
• FDD
• 700W OCZ GameXStream Psu

Now to employ some extra tweaking for 32M pi runs that I have yet to apply

Super Pi 32M

Before tweaking:
E6700 @4416Mhz - 10x441fsb at 1.648v - 552mhz 4-4-4-4 at 2.51v = 11min 34.422s

After tweaking:
E6700 @4416Mhz - 10x441fsb at 1.648v - 552mhz 4-4-4-4 at 2.51v = 11min 31.047s

Push cpu and mem higher with tweaking:
E6700 @4425Mhz - 10x442fsb at 1.648v - 553mhz 4-4-4-4 at 2.51v = 11min 29.344s

E6700 @4434Mhz - 10x443fsb at 1.648v - 554mhz 4-4-4-4 at 2.51v = 11min 27.906s

E6700 @4443Mhz - 10x444fsb at 1.648v - 556mhz 4-4-4-4 at 2.51v = 11min 26.250s

E6700 @4452Mhz - 10x445fsb at 1.648v - 556mhz 4-4-4-4 at 2.51v = 11min 25.109s

E6700 @4483Mhz - 10x448fsb at 1.648v - 560mhz 4-4-4-4 at 2.51v = 11min 20.109s

Very good day for Super Pi benching



32M


16M


Put's me in #4 place on 32M hwbot http://www.hwbot.org//hallOfFame.do?type=result&applicationId=7 just 5.328 seconds behind Joe Camel's #3 X6800 time and 4.859 seconds behind Hideo's E6700/LN2 time http://holicho.lib.net/pi3355/pi3355.htm

 

Ah, so pushing voltage, and getting bit more from the core. Good going. Next step probably CO2/Acetone setup.

Should give a little better headroom. would be nice though to hit 4500 on that core - magical number.

Do you know of any retail 6700's anywhere in Oz?