Overclock.net › Forums › Case Mods & Cases › Builds & Case Mods › Case Mods › Case Mod Work Logs › [Build Log] The Switched Switch (A Reverse ATX Switch 810)
New Posts  All Forums:Forum Nav:

[Build Log] The Switched Switch (A Reverse ATX Switch 810) - Page 3  

post #21 of 86
Thread Starter 
Now for some numbers.

I still want to keep things "relatively" safe until I get a couple days under the belt to see just how well the cooling will be. For now, she is sitting in my office, next to my desk. Once I am happy with her, I will put her in the closet, where there is an AC unit running 24/7 at 68F/20C, that will keep things even cooler.

With the stock Intel cooler, I had to disable "Turbo" mode and keep the CPU locked at 3.5GHz in order to keep the 24/7 Folding temperatures at about 70-72C. I was able to bump up the memory overclock to 2133MHz, but kept the default CL11 timings.

After putting the water cooling loop in (again, many thanks to the wonderful people at XSPC for a well thought out design, and for the people at FrozenCPU for a fast delivery and low price), the temperatures dropped from 70-72C down to 45-48C!!! Again, that is still while Folding! Yes that was still at 3.5GHz on the CPU and 2133MHz on the memory, but after 45 minutes of Folding, I was utterly amazed that temps that remained 25C LOWER.

So I decided to bump up the overclock a bit. Still running at CL11 timings on the memory at 2133MHz, but this time, I bumped the CPU up to 4.5GHz. That's a full 1GHz bump from stock, and 600MHz over "Turbo" mode. To my surprise, the temperatures have been stable for the past hour at between 66C and 71C depending on the core.

I'll have to keep an eye on the PPD, but depending on the jobs she will be doing, I can already notice a huge jump in them. They will only improve with lower CL timings and going possibly up to 4.8GHz or more.



My HFM with White Ranger OC'd to 4.5GHz.
post #22 of 86
Thread Starter 
Well, I booted up in Windows 7 this morning and ran some tests. Seems that the temps in Windows 7 and Linux aren't even close. Not sure which to trust. I'm guessing the Windows 7 utilities are more reliable, since they have been tested by more people and better optimized.

Anyway, I fired up CPU-Z and took a screen shot of what I was running for about 18 hours now. Basically "defaults", except for the multiplier set to 45. CPU-Z shows 1.476V. Temps are really low, even for just running for like 5 minutes (RealTemp was just started, I had to download it).




Then I did a quick poke around in that Sandy Bridge guide (since it showed screen shots from my motherboard) and lowered my CL timings to 10-10-10-28 and upped the voltage to 1.55 to push the memory overclock even a little more.

Here are the numbers and screen shots after tweaking things a bit. Granted it's only been running for about 15 minutes now, hardly what I would call 24/7 stable YET, but so far, so good. Voltage shows 1.428V (I set it in BIOS at 1.45) and temps are actually LOWER by 3-6C!




The temps at 4.8GHz look excellent, and as long as the voltages and memory timings agree with the system, this will be a very good setup for 24/7 Folding. I'll let this run for a day, then push it to 4.9GHz, then after that, who knows, maybe 5GHz!


[Edit]



After over an hour of Folding and putting all the cores at 100%, the water is most likely completely saturated and shouldn't go any higher. The room I have it sitting in now is set at 74F/23.3C and here are the temps at what seems to be a stable 4.8GHz.


[Edit 2]





Well, after 4 hours of running 100% load in Windows and keeping temperatures in the 58C-64C range (23.3C ambient), I think it is stable enough to return to Linux. I guess I was wrong about the temperature monitors being vastly different between the two. i7z is reporting similar numbers now, only about 2-3C different.
post #23 of 86
Thread Starter 


Well, she made it through the night. Temps are just a hair better (I'm guessing the Arctic Silver Céramique 2 thermal compound slowly setting has something to do with that).
post #24 of 86
Can we say 5ghz...
post #25 of 86
Thread Starter 
Well, I've decided to rework this build a bit. I'm going to re-purpose the Switch 810 case (and eventually maybe even the whole machine and turn this rig into a 3930K based Folding@Home machine) and I've ordered a new custom case to be fabricated for "White Ranger".

Sadly, it won't be ready for 3-4 weeks, so stay tuned.
post #26 of 86
Thread Starter 
Well, just a little update. I've started to tear down the Switch 810 case today and I will be doing a "Reverse ATX mod" to it and and the 2700K and motherboard will soon be this new system.
post #27 of 86
Thread Starter 
Here are a couple shots of the tear down from this morning. Basically just took the side, front, and top panels off and removed all the screws that held the top metal on. After work I'll power it down and pull the guts out and set the Mobo, PSU, HDD, and water cooling system on a bench and power it back up so I can continue to do Folding@Home with it while I do the "Reverse ATX mod" to it (basically flipping the motherboard tray from the right side of the case to the left side of the case).





post #28 of 86
Thread Starter 
That poor Switch 810 case didn't stand a chance against my Jensen screwdriver. biggrin.gif




But fear not, in less than 15 minutes I had the core components all hooked back up and sitting on a piece of cardboard running again an chewing on a Folding@Home Work Unit. Since I did change the orientation of the radiator so radically, I did have to bleed the air out a little bit, but I did manage to remove the motherboard/CPU and the entire water cooling loop WITHOUT having to disconnect anything and drain/refill the loop.



Now to focus on doing the mod to the case while I can keep Folding. thumb.gif
post #29 of 86
Thread Starter 
Ok, now to start the tear down of the Switch 810 in earnest so it can be "Reversed".



First thing is to remove all the screws around the top of the case. There are about 18 of them.




Another view of the case with all the fans, cables, and I/O panel removed.




Finally it's time to pull the top off the case.




Next thing is to flip the case upside down so you can get to the bottom parts of the case.




After a couple of screws, off come the feet.




You might want to remove the little "push lock" on the back of the case so as to not damage it. Also, those 3 black pop rivets will eventually be drilled out so you can remove the bottom.




Likewise, you will be removing the 3 black rivets from the bottom front of the case.




For now, you will want to leave the 4 black rivets that hold in the middle 5.25" drive bay, but you will want to remove the top drive bay, so unscrew those lower 2 screws.




Mr. DeWalt will be your friend for this project. As you can see, I've already started to remove the rivets holding the bottom on. Simply just use a 1/4" drill bit and drill out the center of the rivet.




You will also have to drill out the rivets that hold the motherboard tray to the bottom.




Now that you have all the rivets out, remove the bottom panel. Be careful because the remaining parts have lost a lot of it's rigidity.




Another shot of the case with the top and bottom removed.




A couple more screws and the front comes off from the motherboard tray.
post #30 of 86
Thread Starter 
Now that it's torn apart, it's time to put the parts back together with mid section (ie Front panel, motherboard tray, and rear panel) flipped upside down.



Put the bottom panel down on the table and flip the motherboard tray over and set it in, same with the front panel. Then you can put the top panel on.



To keep the case from "racking", I put a 3/8" thick piece of aluminum stock under the edge of the motherboard tray and put some screws in the corner holes.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Case Mod Work Logs
This thread is locked  
Overclock.net › Forums › Case Mods & Cases › Builds & Case Mods › Case Mods › Case Mod Work Logs › [Build Log] The Switched Switch (A Reverse ATX Switch 810)