nikhsub1
Thanks much for the great photos and in depth analysis. It is VERY much appreciated and well done mate.
Printable View
nikhsub1
Thanks much for the great photos and in depth analysis. It is VERY much appreciated and well done mate.
good pointQuote:
Originally Posted by D_o_S
i am yet to do mine and that is a good question
maybe someone can address it before i do mine lol
Still looking for a reply to my question...
I think you tighten the screws all the same. That way it creates the most equal bow that will have the greatest effect.
Quote:
Originally Posted by D_o_S
that's what i didQuote:
Originally Posted by sick_g4m3r
quarter of a turn each until i could not tighten any more :eek: :D
APOGEE BOWED :eek:
http://img257.imageshack.us/img257/5...eebowedcd2.jpg
Obviously it works, but it just looks and feels wrong on so many levels....
looks like it should leak, IMHOQuote:
Originally Posted by Angry_Steel
don't jinx me dude :D
It won't leak. The O-ring seal is now tighter than ever.Quote:
Originally Posted by nealh
Here's how I did it:
http://www.pcmoddingmy.com/e107_plug...?content.408.3
With the Apogee GT (not GTX) I think it makes absolutely no difference which way you attempt to bow the direction to. The side with the mounting plate sticking out will have a larger bow than the side without because the mounting plate in that direction is stronger. The mounting plate side parallel to the barbs (the side with no mounting plate edge sticking out) will have a lesser bow because the mounting plate actually bows in the opposite direction of the baseplate bow, thus lessening the actual bow of the base plate in that particular direction. The acetal top will also bow in the opposite direction following the mounting plate on top.
The Apogee GTX i believe 'might' have a more even bow though, because the top is solid aluminium and its strength is more evenly distributed in both bowing directions. Still then, the drawback of the LGA775 socket retaining design still creates some degree of IHS deformation which may affect the end result of how the bowed baseplate is mated to the IHS.
Sorry:DQuote:
Originally Posted by dinos22
I realize the nice think O-ring will prevent leaks...
I tried the EK o-ring on my FuZion..... SWEET!! Dropped me a good 3-4C right on rebooting. Ceramique is not even cured yet!!
EK O-ring on top of the inlet of the fuzion, right??
3-4C is amazing!!! Is your CPU lapped?
will bowing work for the q6600 in the same fashion?
Probably between the top of the the block and the base, to give it more of a bow.Quote:
Originally Posted by sick_g4m3r
well no leaks to report and performing very wellQuote:
Originally Posted by dinos22
:D
in the loop i have GTS240, DDC2, Microres, Apogee, MCW60, 3/8" tubing..........chugging away at 3.8GHz with E6700 with 1.43V and GFX card not breaking a sweat past 43C :shocked: .....max core OC added only a single degree to the GPU >> 44C max temp :)
CPU ~62C (i have a hot LxxxBxxx stepping)
http://i4memory.com/showthread.php?t=4090
Your CPU is loading at 44C???? With a GFX card and an E6700 both on a GTS240??? Holy smokes, bowing must do wonders!!!Quote:
Originally Posted by dinos22
Just a question for a GTX what diameter O-ring does one need? I see alot of people saying go to HD to get one but I don't know the size and rather not tear apart my whole setup and bring my O-ring to find they don't have it at my HD.
With photos and links like that, the design of the new AC "flow killer" makes sense. Now, i know a lot of you guys see 1/4" or "small bore" cooling as :eek: :rolleyes: :rofl: :ROTF: :shakes: , but think about it - a correctly mounted block shoots cold water (okay, sorta cold after chamber one) DIRECTLY onto each core, thus making it (theoretically) superior to any other competitor.Quote:
Originally Posted by NEvolution
I suppose the idea of having an exact "contact area" - that being the entire base of the core milled out bar a small strip/square (C2Q/C2D respective) would provide the best force, but be off by a few MM, and you face failure.
So is the general consensus that bowing simply boosts performance by boosting pressure?
GPU i meant to say there of courseQuote:
Originally Posted by sick_g4m3r
you can always look at the link i provided
here is my post at i4memory for the lazy ones and if i back down to 3.6 1.32v and stess both GPU and CPU i get roughly the same loaded temps ~43-45C on BOTH CPU and GPU but if i only load the CPU and quick GFX stress program i get a true CPU stress temp which is another 5C higher or thereabouts
ok so i tested both GPU and CPU now and i am more than impressed.........look at the temps on the GPU :eek:
3.6Ghz 1.32V
8800GTS 320M stock clocks
http://img405.imageshack.us/img405/1...llfft30qo2.png
CPU on its own another 5C on top
650MHz on CORE
http://img389.imageshack.us/img389/2...tesatitrg1.png
3DMARK06 TEMPS
http://img507.imageshack.us/img507/4...43cmaxlba7.png
compare those to my air cooled results
if ambient temps were linear according to that ambient sensor on the GFX card there would be something like 30C drop in load temps between air and water :eek:Quote:
another thing i couldn't do with Nautilus 500
3.8GHz 1.43v loaded smallFFT 1 Hour Stable
20C ambient with 8800GTS 100+Watt GFX in the loop :)
closed case
i will try 1.41v....should be able to do it
http://img112.imageshack.us/img112/7...allfft1ge7.png
my e6600 already has a convex IHS, will bowing my fuzion be necessary? will it give me better temps? Out of Bowed, unblowed fuzion or bowed, unbowed Apeogee GT, which would give me the best temps?
Yes, EK O-ring on the top inlet...... IHS is lapped down to 1500 grit with w/d paper. It's not a mirror finish, but it doesn't need to be. It just needs to be flat ;)Quote:
Originally Posted by sick_g4m3r
Idle is now 34-35C on both cores... down from 37-39C. Ambient is 75-76F. I'll do some more load testing in a few days, but after 9 hours of dual Orthos she was running at 54C-55C (coretemp & Intel TAT) on both cores w/ 1.41v.
You can't have the O-ring against the base directly. It will be interfering with flow through the pins. It must go between the inlet plate and the mid plate. The bow will transfer through the mid plate to the base.Quote:
Originally Posted by [XC] DragonOrta
OK, so I'm running a MP-05 LE right now. I have a G0 q6600 on pre-order. Were you in my shoes, would you change over to one of the 3 "bowable" WB's?:hm:
Unless you want to run ~5C hotter with the MP-05 LE for some strange reason :shrug:Quote:
Originally Posted by AMDfreak
That's what I did.Quote:
Originally Posted by AMDfreak
In that case, I'll be waiting (im)patiently for your forthcoming numbers, nikhsub1.:up:
anybody help me?
Personally I'd lap the 6600 and bow your choice of WB's. The goal is to keep the contact patch right over the core.Quote:
Originally Posted by zhangmaster12
bump... would also like to see results of the bow now from the OP
RESULTS ARE HERE: http://www.xtremesystems.org/forums/...d.php?t=151781
Very intersting. So where can you get all of the supplies to mod the Fuzion.
Well, u need a nozzle, I used the ones supplied by dtek, they were prototypes at the time. 4.5mm is the way to go but IDK where you would get it at this point, they should have been out long ago. For the bow of the fuzion, I used an ek barb oring between top and mid plate, see first post for pics.Quote:
Originally Posted by MaxxxRacer
I made a few nozzles for mine also, first a smaller diameter circle, then a slot, and finally a showerhead. I just made them out of some nylon kitchen cabinet inserts that came with some hinges...I just had to sand them down a little bit to get them to fit in then I used a drill press and dremel to make the nozzle types.
I ended up liking the showerhead style best...lower restriction and it had the best performance from what I could tell on my few tests. I think I gained about 1C with the small circular nozzle and 2C with the showerhead style.
http://www.overclock.net/gallery/dat...ithNozzleb.jpg
Slot
http://www.overclock.net/gallery/dat...NozzleSlot.jpg
Showerhead
http://www.overclock.net/gallery/dat...Nozzle7noz.jpg
I also bowed my fuzion with really good results, but I made my washer with some thin plastic I had lying around:
http://www.overclock.net/gallery/dat...BowWasher2.jpg
Here it is with the washer in place, I just used a small amount of "Yamalube" to adhere it in place before installing the top. Overall I think I gained at least 3C with the bow...very much worth the effort.
http://www.overclock.net/gallery/dat...nBowWasher.jpg
wow! :clap: :clap: :clap:Quote:
Originally Posted by Martinm210
martin rulz!!!!!!!!!!!!
hey martin..I would bet others would pay for your handy work....
1. Create nozzle for Fuzion.Quote:
Originally Posted by nealh
2. Mass produce it.
3. Sell it on Petra's site.
4. Become known as a WC god.
5. PROFIT! :bounces:
How the feck do you mount the bigger O-ring on the Apogee GT?
I tried today and utterly failed, I had to assemble it with the original O-ring instead :(
The ring pops out of the groove, and I do not have 25 fingers :(
Yeh martin, im gona have to agree with these guys here :D Id be willing to hook you up with some online help too if you need it. Web hosting etcQuote:
Originally Posted by AMDfreak
I put the bigger o-ring on the Apogee this weekend. It really is a pain in the arse. You kind of have to force it by screwing in a corner, then another corner, then another until you have all the corners screwed in and then you tighten each corner slowly so there's even pressure on all sides. Anyway, that's what I had to do.
Quote:
Originally Posted by TheOriginalTB
I rubbed the included thermal paste all along the o-ring, and that got it to stay in the groove long enough to get the top on.Quote:
Originally Posted by TheOriginalTB
Okay... Already used a Fusion on my gaming rig... Just purchased a GTX for my server --- so:
Where do I buy this larger oRing??? It's not in the box like stated earlier... Need one asap so who do I give my money too>>>?
Please Help Thanx!!!!
BRY
larger Oring is not created yet, so be patient.Quote:
Originally Posted by sickboy8820
I dont mean to be a downer, but as time has shown, the profit side of thigns is a bit more elusive than it would seem at first glance.Quote:
Originally Posted by AMDfreak
my understanding from Gabe as of GTX purchased after July 16..they come bowed with the larger O-ring in place alreadyQuote:
Originally Posted by sickboy8820
check there swiftnets website...its under the news and updates on the GTX page
Thanks for the suggestions guys, but it wouldn't pencil out for money. As a hobby for fun it's all good, but I've also learned that everytime I've tried to turn a hobby into profit the fun dissappears..:shrug:
I probably spent over an hour making just one nozzle. If I figured how much money I could have made at work, I probably have more money invested in my nozzle than I do the block itself...but that's not what it's all about. It's all about customizing and tinkering, which is cheap recreation..:D
Thought I would resurrect the thread, has good info. :D
Utnorris
Hopefully someone with some knowledge still reads this thread. Anyway, I bowed my Apogee GT using the supplied o-ring, I am running into two problems currently and need some suggestions.
1 - I have remounted the water block twice and still I get 3 cores at one temp and 1 five degrees higher. I tried adjusting the screws to see if I could change the spread and that didn't work, so any thoughts?
2 - I noticed when I was remounting it the Artic Silver was for the most part spread down the middle from top to bottom with very little on the outside, it would almost seem as though my chip was bowed downward, but since I lapped it and verified it was truly flat I know this isn't the case. From the looks of it I would assume that the waterblock is bowed oppisite to what it should be. Is this possible? I didn't lap it before bowing it so it just has the same finish it came with which is hard to verify that it was bowed afterwards. The mount that came with the waterblock is not the type that has springs, instead it has screws that screw into the backplate. Should I swap this out with screws that mount from behind the MB and use springs with nuts to hold it down? If so, how? Or should I just get a different waterblock like the EK?
Here is a link to the mounting instructions so you can see what I am talking about when I say it doesn't use springs:
http://swiftech.com/products/install...75bckplate.pdf
Thanks,
Utnorris
Are you using coretemp or TAT?Quote:
Originally Posted by Utnorris
TAT seems to have a bigger differential between cores and I believe something is wierd with it. When I log temps with TAT I get an average of 3 degrees difference just between two cores....but notice how this 3 degrees difference swaps places in the middle of this test.
http://img116.imageshack.us/img116/1...lfontz7en1.jpg
I would hold much trust in differential between cores, actually I wouldn't trust TAT or Coretemp unless you log it and average at least 10 minutes worth of logged data....
Here is another test with coretemp and TAT logging at the same time with the same loading.
Full Test:
http://img401.imageshack.us/img401/1...retemp1tk6.png
Zoom on a small piece of time:
http://img401.imageshack.us/img401/2...retemp2wl6.png
Coretemp consistently had more consistency between cores, yet they are supposedly reading from the same sensors..:shrug:
This is why I want to build my own die simulator with more accurate sensors. I just don't trust in the internal processor sensors.. I went to all this effort to get .01C relative accuracy on my crystalfontz sensors only to see TAT and Coretemp bouncing around by 3 or more:(
Bowing probably does make mounting consistency lower especially for quads, but I wouldn't let the internal processor sensors worry you too much about differential between cores..
Thanks for the information, I am new to w/c so I wanted to just make sure I wasn't missing something. By the way, I am using the latest coretemp in Vista Home Premium x64, but I was getting the same results in XP Pro x32, so I don't think it is the OS. Anyway, I will probably fiddle with a little more, my temps aren't that high, idle is 40 and load is 55, so they are not that high anyway. I am just trying to squeeze every little drop of performance out of it.
Again, thank you for the info,
Utnorris
I realize this question is sacreligious in these forums but being a newb I'll just come right out and ask.
Why should I care about 1-3*C day in and day out? If I'm running 48*C why should I go to all this work to run at 46*C? :confused:
The only reason I can see is that you're trying to squeeze another few MHz out of an overclock, but if you're that hell bent on getting the temps down get an active cooling solution such as phase change.
Is it just "because it's there"?
Its something to do! And it is free. Why NOT?
There's no single magic bullet for watercooling. You gain a few degrees here, a few there, and the sum of all those improvements are what determines the final performance of the system. Bowing is just one of those potential areas for improvement. Certainly, you can ignore it, and still have a good system. You can also run smaller tubing, a weaker pump, a poor/smaller rad, bad fans, etc, etc. No single compromise will suddenly make your system "suck", but they all represent opportunities for improvement.
Yes, going to Phase/beyond will beat even the best watercooling hands-down, but when it comes to performance that you can set and forget, running it 24/7, watercooling is still pretty hard tro beat.
If u think like that u start loosing alot of C in the end. 2-3C on the block, 1-2c on the pump, 3-4c on radiator++ it adds up.
edit: And for me who cares alot about silence too, every C gained is a few less rpm on my fans.