QX9650ES L737A under 2-stage cascade

[tiborrr]

Initial tests by tiborrr:
I have recieved another Yorkfield engineering sample from Intel Slovenia Channel, batched L737A in order to have some fun and try some hardcore overclocking. First i have done some preliminar preparations on the singlestage cooling on my own. I haven't run 32M SPi nor PiFast since i did not have any high-MHz running memory at the hand:

System specs:
Intel C2Q QX9650ES L737A
DFI Blood Iron (P35)
LCPower LC6550 v2.0 550W
2x512MB Corsair XMS5400 v1.5 (D9DQW Fatbody)
X1950XT Sapphire 256MB PCIe
Tambient = 24.5°C

Singlestage unit:
- Electrolux MP12FB 3/8HP compressor
- 0.63kW condenser
- chilly1 clone evaporator
- R404a gased
- ghetto aluminum L-profile case :)


SuperPI 1M @ 5076MHz (9.125s, 11x multi, 5:6 RAM divider) | Wprime 1024M @ 4796MHz (281.425s):
http://shrani.si/t//oy/20Oh8IAo/1/dscn0734.jpg (http://shrani.si/?/oy/20Oh8IAo/1/dscn0734.jpg) http://shrani.si/t/3y/ii/1ZjUp3aD/1/dscn0738.jpg (http://shrani.si/?3y/ii/1ZjUp3aD/1/dscn0738.jpg) http://shrani.si/t/1J/U/ZEnLdof/dscn0739.jpg (http://shrani.si/?1J/U/ZEnLdof/dscn0739.jpg)

Basement subzero subcooling tests on SS unit:
I moved into our basement / occasional workshop, opened the windows and let the chilled air inside. Thermal probe on the evaporator stopped working after hitting the -63°C marker.

SuperPI 1M: 9.047s @ 5120MHz (RAM 4:5 divider) | SuperPI 32M: 10m 3.844s @ 4976MHz (RAM 4:5 divider)
http://shrani.si/t/2C/wo/22A5ErSd/dscn0756.jpg (http://shrani.si/?2C/wo/22A5ErSd/dscn0756.jpg) http://shrani.si/t/g/sD/46qUpYkl/dscn0765.jpg (http://shrani.si/?g/sD/46qUpYkl/dscn0765.jpg) http://shrani.si/t/3F/gx/1mqv0awJ/qx96505120ss9047s.jpg (http://shrani.si/?3F/gx/1mqv0awJ/qx96505120ss9047s.jpg) http://shrani.si/t/A/Z1/3fRlAc3f/qx96504972sspi32m.jpg (http://shrani.si/?A/Z1/3fRlAc3f/qx96504972sspi32m.jpg)

Some observations on the DFI Blood Iron board (running latest official BIOS) on SS cooling:
- Yorkfield has no cold boot bug (CBB)
- If the board needs to hard-reboot itself (via shutting down), it will never come back alive
- Solution: plug out the battery (CMOS reset via jumper does not work) for 30sec
- BIOS settings order: 1) Disable CPU features, set DRAM timings & voltages - reboot - 2) Set mem. divider - reboot - 3) Set multiplier - reboot
- If you need to change the mem divider, set multiplier on default.
- If you set mem divider after setting the mutliplier, boot will fail :rolleyes:

2-stage cascade overclocking with Moonman & tiborrr:

Khmmm, first of all - DFI Blood Iron died after all the stressed put through :( One phase had no voltage output and when we tried to disconnect the phase the whole PWM blew out. After dissasembly of valuable components (solid caps, inductors, coolers etc...) i kicked the hell out of the board :mad:. It was time for Moonman's plain P5K vanilla to come back into action.
http://shrani.si/t/3L/sC/1rvtwhkM/dsc00346.jpg (http://shrani.si/?3L/sC/1rvtwhkM/dsc00346.jpg)

System specifications:
QX9650 L737A
ASUS P5K vanilla (Vdroop & Vcore mod, PLL mod removed)
2x1GB Crucial Balistix Tracer PC6400 (D9GMH)
GeForce 6600GT 128MB PCIe
Enermax Liberty 620W PSU

Sharp's 2-stage cascade:
- Electrolux MP14FB 1/2HP compressor on both stages
- 1.1kW condenser /w Fan & two extra 120mm Deltas
- coil desuperheater on the 2nd stage, cooled by two 120mm Deltas
- coil interstage HX (a bit underpowered)
- R404a 1st stage / R1150 2nd stage
- MXV on the 2nd stage for CB/CBB tuning

http://img184.imageshack.us/img184/8386/dsc00264dg4.th.jpg (http://img184.imageshack.us/my.php?image=dsc00264dg4.jpg) http://img296.imageshack.us/img296/6914/dsc00266bi5.th.jpg (http://img296.imageshack.us/my.php?image=dsc00266bi5.jpg) http://img183.imageshack.us/img183/8602/dsc00275ht4.th.jpg (http://img183.imageshack.us/my.php?image=dsc00275ht4.jpg) http://img181.imageshack.us/img181/922/dsc00280hu6.th.jpg (http://img181.imageshack.us/my.php?image=dsc00280hu6.jpg) http://img242.imageshack.us/img242/4852/dsc00283si0.th.jpg (http://img242.imageshack.us/my.php?image=dsc00283si0.jpg) http://img247.imageshack.us/img247/7728/dsc00284cm1.th.jpg (http://img247.imageshack.us/my.php?image=dsc00284cm1.jpg) http://img341.imageshack.us/img341/7825/dsc00286ii8.th.jpg (http://img341.imageshack.us/my.php?image=dsc00286ii8.jpg) http://img186.imageshack.us/img186/7922/dsc00288ld9.th.jpg (http://img186.imageshack.us/my.php?image=dsc00288ld9.jpg) http://img183.imageshack.us/img183/2161/dsc00290hq6.th.jpg (http://img183.imageshack.us/my.php?image=dsc00290hq6.jpg)

Prime time to start the system, -78°C is a cold boot bug (CBB) temperature on the P5K board, we did not experience any CB up to -107°C (the lowest temperature we've seen with this cascade). I guess its nitrogen time :up:
http://img181.imageshack.us/img181/3615/dsc00291jg4.th.jpg (http://img181.imageshack.us/my.php?image=dsc00291jg4.jpg) http://img175.imageshack.us/img175/2195/dsc00293ln4.th.jpg (http://img175.imageshack.us/my.php?image=dsc00293ln4.jpg)

Evaporator temperature during SuperPI 32M benchmark, 1.85Vcore (just a hair below sustained OVP limit) & two empty coffee mugs :p:
http://img169.imageshack.us/img169/5681/dsc00327wm2.th.jpg (http://img169.imageshack.us/my.php?image=dsc00327wm2.jpg)

Results:
SuperPI 1M = 8.610s @ 5363MHz | Wprime 1024M @ 5031MHz (253.953s) | Wprime 32M @ 5171MHz (7.922s) | PiFast 17.98s @ 5215MHz (RAM 4:5 @ 592mhz CL5.0 4-4-15 2T)
http://shrani.si/t/3Q/aL/2xAZVLv3/dsc00321.jpg (http://shrani.si/?3Q/aL/2xAZVLv3/dsc00321.jpg) http://shrani.si/t/1U/n6/2JMFccFw/dsc00331.jpg (http://shrani.si/?1U/n6/2JMFccFw/dsc00331.jpg) http://shrani.si/t/j/E1/4xMBxAut/dsc00335.jpg (http://shrani.si/?j/E1/4xMBxAut/dsc00335.jpg) http://shrani.si/t/3v/aQ/13PJzUUw/signatureimg.jpg (http://shrani.si/?3v/aQ/13PJzUUw/signatureimg.jpg)

The first chip we had, L734A was a bit better (we could run Wprime 32M higher, achieving 7.68s) but had coldbug therefor had no potential under LN2 cooling. 3D numbers & SuperPI 32M is coming this weekend!

:cool:

Sham, supply us with those X48 Blackops already, so we can start posting as a QFS team :D

[Swarm dude]

Well done :up:
But I must admit, the LCPower PSU kind of, doesn't fit in there. But it seems like it is very much up to the task.

[tiborrr]

Well, i was in serious doubt too, so i gave him the ultimate test: Wprime on all four cores at 1.84V real. 12V line drooped from 12.15V to 11.84V, but held it rock solid there. I was amazed, it was a really good performance show for a 40€ PSU :D Much better than most 100€ Tagan's i have seen.

[lukas69]

Nice Rur Tiborrr

[Solarfall]

great numbers there tiborrr .. lol i wish i could afford a QX9650 too.
btw cool pics mate :up:

[tiborrr]

Thank you fellas! :yepp:

My buddy Moonman & I did also perform a trick and now could calculate Hexus PiFast 10M @ 5293MHz (~80MHz higher than before), achieving 17.7s:

http://shrani.si/t/1N/121/39FsIMTk/11.jpg (http://shrani.si/?1N/121/39FsIMTk/11.jpg)
* Just found out that the image is screwed - the other CPUz should have shown the memory tab (4:5 divider 5-4-4-15 ~ 601MHz)

We have also tried some 3D benchmark in order to find the maximal 3D stable frequency with the old trusty aircooled 6600GT. VGA board was not Vgpu voltmoded, only Vddr (one day i was in the need of 20kOhm resistor, so i took it off the card ;)) @ 2.47V. Seriously GPU underpowered runs - every 5MHz GPU clock gains 1000 points in AQ3. Next time we will do it properly with singlestage cooling on the GPU, Vgpu vmod & geometric delta clock mod.

3DMark2001SE: 41056 | 3DMark06: 2235 | Aquamark3: 103814
http://shrani.si/t/3q/w7/8NV4Glv/3dm012.jpg (http://shrani.si/?3q/w7/8NV4Glv/3dm012.jpg) http://shrani.si/t/38/CH/3kf9tJPT/3dm061.jpg (http://shrani.si/?38/CH/3kf9tJPT/3dm061.jpg) http://shrani.si/t/2R/fg/1lZGPMdI/aq31.jpg (http://shrani.si/?2R/fg/1lZGPMdI/aq31.jpg)

[LardArse]

[B][SIZE="3"]
Sham, supply us with those X48 Blackops already, so we can start posting as a QFS team :D

yes as fast as is possible need to get hold of some chipsets first....

[T_M]

X48 supply seems to be the only hold up for most brands now (except Asus :P)

[tiborrr]

http://www.youtube.com/watch?v=0TURqnM3kLw

Here's a video showing operating of sharp's oldskool R404a/R1150 cascade. :p:

The 6600GT is getting ready for it's debute subzero run. Here's a photo of a re-usable (hehe, we're cheapskates) cap mod done by Moonman ;)
http://img148.imageshack.us/img148/8730/6600gtmodym5.th.jpg (http://img148.imageshack.us/my.php?image=6600gtmodym5.jpg)