Showdown at High Noon: Clearing the Streets

[jayhall0315]

This thread has been amended to (twice in fact), in the middle of the article to address some concerns by certain forum members. Please take a look there if you have already read this before.

Showdown at High Noon: Clearing the Streets

Well, its almost time to empty those saloons and get the gunfighters out on the street. But before we get to the chipset showdown (and because it was fastest to type up this section), I thought it would be fun to look at a few myths that I continually see pop up here at XS and in other enthusiast forums. I also need to think about how much info I am going to put out in the next few weeks and how best to space it out (and how forthcoming to be about my custom die simulator without turning the results into a giant standards argument with all the associated flaming).

But before we start, if you wonder who I am or why I am doing this, then head to my earlier post here http://www.xtremesystems.org/forums/...d.php?t=151720
Basically, we all could use some more info on who is shooting straight and nothing beats real world testing.

And now, as Jedda might say, on to the meat and potatoes:



The Myths of Watercooling Put to the Test!

Watercooling Myth #1 – Commonly, you will see posts in the forums with something that goes like this “If my pump fails (and it may if you own a DDC-2 18 watt version) my CPU and system will fry, wont they? Help! Please tell me what to do blah, blah, blah ..... (Hell, the 680i motherboard will do this for you for free, even if your pump DOES NOT fail). In conjunction with this concern we also have the numerous posts of “Well, I guess I need to spend not only $75 to $250 or more for a quality pump, but I need to buy two and set them in series, in case one fails, so that my CPU and system do not burn up. It gives me the same peace of mind that my RAID 1 array provides blah, blah, blah .......”. Time to put this to the test once and for all.

Test of Myth #1 – Hey, we are at Xtreme Systems, right? - so we might as well go extreme. Heck, lets forget the pumps for a minute. Waterblocks, you say? We dont need no stinking waterblocks! Or heatsinks for that matter either. We'll just throw in the processor and hit the 'On' button at the front of the case!

Test 1 - E6400 with no heatsink, no waterblock and no TIM (in a Gigabyte GA-965P-DS3 board), while hitting the power switch.

Result: Hitting the power switch does nothing. No screen, no images, no BIOS – nadda, followed by immediate power shut off. This all took about one second. Repeating this three more times produced the same results. Taking the E6400 out of the motherboard and slapping it in the 680i with the standard Intel heatsink/fan produced immediate boot-up and went on to run the system just dandy. Looks like Intel's internal thermal management algorithms work. I decided to use a spare lab 965 board for this test since my 680i would probably never suffer this punishment. It only works correctly at the conjunction of the alignment of the nine planets and the arrival of my $10 million dollar check from Publisher's Clearing House.

One CPU wasn't enough however, so now I really have a desire to get on good terms with the folks at Intel's RMA department, so:

Test 2 – Q6600 with no heatsink, waterblock or TIM while hitting the power switch (dont try this at home, kiddies)

Result 1: Hitting the power switch flickered the led on the 965 and then nothing. No screen, no BIOS no visual input and then the power shut off. This all took about two seconds. Second and third attempt, produced no LED flickering and no response whatsoever. Once again, taking the CPU out and installing it in the 680i showed that it was fine and went on to run everything for the rest of the day.

Conclusion: As long as you have a Core 2 (and I suspect an AMD X2 performs the same) chip, you should be okay, even if you make the biggest of blunders. Just remember to attach the TIM and heatsink or waterblock.


Watercooling Myth #2 – “If my pump fails, my system will heat up and kill the CPU, destroying my ability to get to :banana::banana::banana::banana: at the rate at which I am accustomed.”

Test 1 – E6400 in the Gigabyte 965 with Fuzion (unbowed) block, MCR 220 Radiator, Swiftech Micro Reservoir ¾ full and MP655 Pump that is NOT plugged in, and then hitting the power switch. (the E6400 is rated at 65 watts) (this was done with 1/2 inch ID clearflex tubing with nylon snap grips at all the metal barbs).

Result 1: Computer BIOS starts up, Windows XP loads, I log in and reach the desktop for one minute and ten seconds before a BSOD appears. Everest records 87 deg C at shutdown (do not feel confident in this measurement though)
Result 2: Five minute break then hit power switch - Computer BIOS starts up, XP loads for 24 seconds, then the welcome BSOD strangely indicating memory problems appears.
Result 3: Five minute break, then hit the power switch – BIOS loads and then Windows asks if I want to go into Safe Mode because of a problem, I click “Start Windows normally”, log in and am on the desktop for 37 seconds before the BSOD hits at 91 deg C. Removing the E6400 and slapping it into the 680i with the Intel heatsink/fan produced no problems and seemed to run everything correctly for the next three hours until testing had to proceed onward, and I removed it. Putting the E6400 back in the Gigabyte board with the Intel heatsink also works like a charm.

Test 2 – Q6600 in the same Gigabyte board with the same Fuzion/MCR/.... setup, and then hitting the power button. (Remember, this chip is rated at 105 TDP)
Result 1: I dont know yet because as I was getting this setup, I got sidetracked by demands in my lab so maybe I can get this out in the next week or so.

Conclusion: If your pump fails you will not lose your system. Although for the first few BSODs you may wonder what is up if your pump is very quiet and you do not notice that it has failed. It takes a great deal of heat to break water's hydrogen bonding even in a small 1 ~2 liter closed system. The thermal agitation of those bonds in still water is governed approximately by every engineer's chew toy, the Heat Equation (parabolic PDE) and flows spherically from the heat source. Your CPU will shut itself off or throttle down well before large thermal transfer in the water has taken place. Simply replace your pump or make sure that it's three pin/molex connector was plugged in, in the first place, and you should be good to go.



(I have been receiving some traffic from various forum members with much experience suggesting that 'Blow Outs' or ruptures do happen. But respectfully, if you post or PM me, telling me that this not true for your watercooled Commodore 64 from 1984 that has been mated to a special sever board that runs code developed by the Air Force at Area 51 (just joking), I can have no reply. I am only one man, who did these tests for about two months with about 150 hours in the project, not tens of thousands. I simply picked components and completed tests that I thought would be of the most interest to forum members as a whole (and to answer my own questions). Please just give me a chance to get the results out and then whack the experimental method if you find it wanting. I do welcome any criticism and actually this helps to hold me to more rigorous standards, so that is great. And I welcome other forum members, repeating any and/or all of my work (to not only make sure that I am speaking truthfully) to hold these companies up to their claims.

Since this was originally posted, several members still believe that if your system is left unattended while the pump fails, your tubing will 'blow out' (that mental image is just too funny) or your CPU will quietly overheat if you are not nearby to notice the big blue screen that is now on your monitor, killing it. With respect, this is simply false, at least for Core 2 chips (Notice, I am NOT commenting on your 1999 chip here or your GPU or chipset). Although, I have not listed all the results here due to space constraints and finger pain, several rounds of testing showed no such results (at least with Core 2 processors when using nylon snap grips on your barbs, cannot speak for everyone else here, but I suspect most chips made in the last 3 ~5 years will function about the same).

The Core 2 chips simply shutdown at prescribed points (which seems to be roughly 93 ~ 95 C, but I am working on this (Ntune, Everest and Speedfan all give conflicting info and die results indicate even higher temps) or throttle down, if you have the thermal management option set in the BIOS (but this is sacrilege to most overclockers). For a mental comparison, imagine placing 1.5 liters of water in a pot and then placing it on your stove with a hot gas flame. Obviously, that gas flame is much hotter than your CPU (several hundred watts or more depending upon your stove model), and yet if you leave your finger in the water (or place a diode down in the water, attached to the pan near where the gas flame is touching), you will notice that it will still be several minutes before you start to feel any pain or notice very hot water temps. Once again, this is due to the hydrogen bonding characteristics of water and is the reason why your CPU will long overheat and reach its breakpoint before you suffer a Jim Bob style 'Blow Out' ! (too funny)!

Having said that, I will show some results next week of letting the pump fail while cooling an older GPU in which a leak did take place (although no nylon snap grips were used). Please remember that the specific heat capacity of water is incredibly high at 4.18 J/g.C compared to say Copper at 0.386 J/g.C, or Aluminum at 0.901 J/g.C or even Silicon at 0.705 J/g.C and this is the technical reason for the very poor conduction of heat by water (as opposed to copper, but more on that later). This is also reason why water is good at holding heat, which makes it perfect for us watercoolers)


Interestingly, since PCBs are layered, it would seem that additional lithographed copper layers could easily be added to the manufacturing runs on motherboards, spacing them between layers and increasing surface area by several hundred percent (perhaps roughly equivalent to the very high end airsinks out now and greatly distributing the hot spots you now see if you image a board in infared)(remember with airsinks, it is all about surface area, but for us watercoolers, it is all volume and flow) with taps to redesigned new form factor case walls that could easily dissipate several hundred watts safely (you would need a new embedded copper latch though for the CPU socket). No need for CPUs with integrated channels. Or huge ten pound heatsinks that bang into your side case door and bow the motherboard.

As an aside, if you are extremely religious or suspicious of Darwinian evolution theory, and really believe that the Earth was made in 4006 BC as some Kentucky fundamentalists believe, I suggest possibly looking up Joseph Fourier and his attempts to see if this can really be proved or disproved thru heat conduction from the Earth's core (if you are a creationist, then the result may surprise you). Some accounts even credit Fourier with creating his infinite trig series (the Fourier series) to solve this very problem. Or the ideas of some fundamentalists in the late 19th century that the Sun might be powered by coal or liquid gasoline (to account for the 4006 BC creation date) leading us to understand that something truly special was taking place within the sun because the density of coal or gasoline would not be adequate to withstand the gravitational pressure of an object of this size under it's own mass (even though fission and fusion would not be understood until 40 years down the road, first approximation results from the spherical integration of the spring ODE showed that something odd was clearly up).


Watercooling Myth #3 - “If one pump is good, then two or more in series will kick serious butt. And will let me outclass, all those jokers with that expensive Iwaki stuff. Braggerts!”

Test 1 – Q6600 cpu (with 680i at home) with Liang/DD/Swiftech MCP655 vario pump (set on #5, 4800 rpm at 12v) powering a loop as follows: Swiftech Micro Res ¾ full -> 10 inches of ½ inch ID Clearflex tubing -> the aforementioned D5 -> 20 inches ½ inch ID clearflex -> MCR 220 -> 14 inches ...tubing -> Fuzion -> 8 inches ... -> MCW60 GPU block -> 10 inches .... -> back to reservoir.

Result 1: 1.83 gal/min with a temperature of 41 degrees C (at home) (43.9 C on the die simulator at 105 watts at the lab but more on that later)

Test 2 – Same loop as above, only now there are two 655 pumps, one after the other in series with six inches of ½ inch ID clearflex tubing between the two pumps
Result 2: 2.08 gal/min with a temp of 40 C (at home) (43.4 C on the die simulator at 105 watts)

Test 3 – Same as loop in Test 1 except the second 655 Pump comes after the MCR 220.
Result 3: 2.21 gal/min with a temp of 38 C (at home) (42.1 C on the die at 105 watts)

Conclusion: It probably ain't worth it. Eighty some bucks for a 655 at Petras for very little gain and more noise and less free space in the case. You decide. As for how it compares to an Iwaki, ask someone who owns one (maybe Cathar, Petra or Nikhsub1). $240 at Petras is too rich for my blood but I am happy to take one way donations


Well, hope you enjoyed this first installment and perhaps there will be a few less noob questions along these lines in the future. If you happen to run across this Maxxracer, maybe you could kindly add it to your Beginners WC FAQ sticky. . The Showdown at High Noon: Part I for the chipset blocks and pumps should be out tomorrow or on Friday morning. Also, not having any luck getting my 2 MB pics uploaded (duh, I guess the 200kB limit is the factor), so could someone kindly lay out the steps I should do to best get them inserted at the correct points (Digital photos and compression formats are not my bag) with the best resolution.

Until tomorrow,
Jay

(Yeah, right now, Ubuntu is still a bit rough around the edges. It lacks needed drivers and Wine still cant play many games worth a hoot, but if we all contribute something small, then it will pay off down the road for all of us.)


(While taking a class on thermodynamics at UF (Go Gators - More opponents to chomp on this coming football season!) as an undergrad, a fellow student asked our pompous prof if heat radiates infinitely, even in a vacuum, due to quantum considerations, to which he responded with, “If the Earth were a globe of solid steel on which no one lived and only one single house fly alighted upon it's surface only once every one million years for only one second, then by the time that the earth was completely abraded away from the friction of that one fly's landings, infinity will not even have begun – so, you have your answer")


Here is a late pic of the E6400 undergoing testing without turning on the pump, just to see what failure mode will appear!

[phelan1777]

<hears an all to familiar tune whistled in the wind...............................

[Machinus]

I don't even use water but I am reading these. If you know about bernoulli and fourier there is probably interesting stuff going on!

[ColonelCain]

Nice work! Waiting for further installments.

[ranker]

I'm more amused with the color commentary than the results, although I do appreciate the results!

[[XC] junglemaster]

I am still laughing at reading all of these myths put forward by people without any idea previously...LOL. Good work. Look forward to future physics lessons

[virtualrain]

I'm more amused with the color commentary than the results, although I do appreciate the results!

I have to agree... the commentary is excellent and the results are interesting! Keep it up!

[jayhall0315]

Nice work! Waiting for further installments.

Thanks Colonel. One question, are you a gunnery sergeant or a Colonel? I guess there is about a $60,000 pay grade difference in that question.
Once in a while, I see forum members who have BF2 ranks. Having never played the game, is BF2 rank awarded based upon skill (so all the generals are real mean mofos you do not want to try and take down) or is it simply an indication of the amount of hours you have fought against others (for example, every twenty hours played is a new promotion)?

Maybe I'll get around to giving BF2 a spin later,
Jay

[conference]

So much for not trying to be dramatic...

Good work though. It feels a lot better knowing that my computers probably won't blow up if I do something stupid.

[phelan1777]

Thanks Colonel. One question, are you a gunnery sergeant or a Colonel? I guess there is about a $60,000 pay grade difference in that question.
Once in a while, I see forum members who have BF2 ranks. Having never played the game, is BF2 rank awarded based upon skill (so all the generals are real mean mofos you do not want to try and take down) or is it simply an indication of the amount of hours you have fought against others (for example, every twenty hours played is a new promotion)?

Maybe I'll get around to giving BF2 a spin later,
Jay

you could have Pmed him, because this is completely off topic, not that is completely unusual here, but its not even related to water cooling

[YugenM]

Thanks Colonel. One question, are you a gunnery sergeant or a Colonel? I guess there is about a $60,000 pay grade difference in that question.
Once in a while, I see forum members who have BF2 ranks. Having never played the game, is BF2 rank awarded based upon skill (so all the generals are real mean mofos you do not want to try and take down) or is it simply an indication of the amount of hours you have fought against others (for example, every twenty hours played is a new promotion)?

Maybe I'll get around to giving BF2 a spin later,
Jay

Points system.

Really turned me off from the game, since I found so many little kids camping to accumulate points.

Servers with mature players were few and far between

on topic: Awesome article!

[Jedda]

Yes, nicely spiced meat and potatos, at that.

[ranker]

Points system.

Really turned me off from the game, since I found so many little kids camping to accumulate points.

Servers with mature players were few and far between

on topic: Awesome article!

Camping doesn't reward you with any bonus points. You accelerate point gain by working together and doing things as ordered.

The only thing that might be considered "camping" to gain points is defending a vital location.

This isn't Quake. It's a team/squad based game focused on securing objectives!

[voigts]

As an aside, if you are extremely religious and really believe that the Earth was made in 4006 BC as some Kentucky fundamentalists believe, I suggest you look up Joseph Fourier and his attempts to see if this can really be proved or disproved thru heat conduction from the Earth's core (if you are a creationist, then the result may surprise you).

The testing information and much needed analysis and hard numbers on the watercooling stuff you are evaluating is seriously needed. I applaud you for the tremendous effort. And your humorous side comments really do add color to the conversation.

However, can we please keep this thread to topics somewhat related to the watercooling discussion at hand and not an opinionated observation as to the origin of the universe? There are plenty of scientists who would beg to differ with your stereotyped comment. I happen to like Kentucky (went camping there this weekend), even though I live in Tennessee. Thank you.

[jayhall0315]

The testing information and much needed analysis and hard numbers on the watercooling stuff you are evaluating is seriously needed. I applaud you for the tremendous effort. And your humorous side comments really do add color to the conversation.

However, can we please keep this thread to topics somewhat related to the watercooling discussion at hand and not an opinionated observation as to the origin of the universe? There are plenty of scientists who would beg to differ with your stereotyped comment. I happen to like Kentucky (went camping there this weekend), even though I live in Tennessee. Thank you.

I have amended the statement a bit and you are right about staying on topic. I still leave the two competing arguments for you and other readers to weigh and decide upon as you see fit. And yet it is interesting that the same laws that govern heat flow in our little water loops also lead to almost undeniable conclusions about other important origins (assuming I guess, that is, that you do not believe that the basic axioms of mathematics are flawed). And your right, my trip to Kentucky three years ago to the Mammoth Caverns was great but it is also home to the nation's only $30 million dollar creationist museum, is it not? And my Aunt resides there as well, although she claims that those other fundamentalists in her town are all wrong, and that the first day of the Universe, as described in the Book of Genesis, actually took place in 5037 BC (she arrives at this calculation from mixing the Torah and the Book of Genesis together). Everyone in my family agrees, that she is the most exciting dinner guest.

[jayhall0315]

I should mention to all those out there who are cooling their Northbridge or Southbridge with a waterblock, that depending upon the motherboard, those chips may not be nearly as resistant to pump failure as modern Core 2 CPUS. On the other hand, unless you own that colossal turd of all overheating products: the 680i motherboard (whether the EVGA or the Striker or the Abit IN9 or the....), your NB or SB are likely not putting out enough heat to fail quickly in the event of a pump failure anyway. Many of these chips may or may not have the thermal shutdown protocols of modern Intel chips, so pump failure COULD effect you more severely in this department.
This is something of interest to me and I may get these results out when I am finished with the 680i I currently have.

Also, if anyone has a Koolance aluminum radiator that they want to donate to the project (Santiago ?), please let me know. This aluminum versus copper/brass stuff has already been resolved in the engineering world but for average everyday watercooling, it is time to see who is telling the truth.

Hope this helps,
Jay

[lowfat]

Very nice. I am pretty sure almost every modern CPU out there will shut off once getting too hot.

Even back in the day I tried powering on a Althon XP 2600+ dozens of times and it would just shut off instantly, the chip continued to run fine for year and years overclocked until i sold it.

[alpha0ne]

interesting

[Jedda]

Hmmm, sorry Jay.
You're tilting at windmills, so far.

[IanY]

Why don't you try that with an AMD cpu, eh?

How about you buy a $400 AMD Tyan mobo and two FX-74 cpus. Cost? Aboyut $1,500. Now try it without a heatsink.

I'll supply your with the handkerchief that you would be crying on.


I'll tell you what happens when your pump fails... the water around your block will heat up faster than a kettle on the stove. Remember that you have PVC or some sort of vinyl tubing that expands.. and expands.. and pressurizes... guess what happens? Water will spill everywhere across your mobo. God forbid, you suddenly realize your mistake and turn on your pump now.. which would be even worse. Guess what happens next? The entire system is fried. I invite any of you to try it at your own risk.

Next, you doubled up on pumps but yet you failed to double up on radiators. MCP655/D5 isn't exactly the pump to double up on since the heat dump is substantial. Of course, you'll see a performance drop. Now try this. Put two PA120.3s together. Now double up on DDC2 w/Petra. Now tell me that the performance does not increase relative to one DDC and one PA120.3.

[jayhall0315]

Why don't you try that with an AMD cpu, eh?

How about you buy a $400 AMD Tyan mobo and two FX-74 cpus. Cost? Aboyut $1,500. Now try it without a heatsink.

I'll supply your with the handkerchief that you would be crying on.


I'll tell you what happens when your pump fails... the water around your block will heat up faster than a kettle on the stove. Remember that you have PVC or some sort of vinyl tubing that expands.. and expands.. and pressurizes... guess what happens? Water will spill everywhere across your mobo. God forbid, you suddenly realize your mistake and turn on your pump now.. which would be even worse. Guess what happens next? The entire system is fried. I invite any of you to try it at your own risk.

With respect, Ian, this is not true (at least not for Core 2 chips) nor did this happen in the testing. The Fuzion block was attached with 1/2 inch ID clearflex tubing and nylon snap grips and after every attempt (actually several more than I published in the post), the chips internal breaker shut off the chip (and the power) well before the conditions you describe could take place. This is also holding up for the Q6600 as well at this very moment.

As for distilled water leaks frying your sysem, hmmmmmmmm, sounds like another common Watercooling Myth that I should have got around to testing and publishing in the next few installments (and I did). You may be surprised.

As for AMD, I wish they would get their together soon, since competition drives innovation and Intel all bye itself would mean maybe $1000 quad core chips instead of the $266 they will be soon. I would have included more components in my testing (and certainly AMD) if only time had allowed without putting pressure on myself to drop out of grad school.


Hope this helps,
Jay

[Vapor]

I've had a pump 'fail' (well, I froze the liquid in the rad actually, but effectively the same thing) while doing some benching on an X800....it would crash part way through GT1 03 every single time, but would start back up like a champ and idle in Windows endlessly (core temps were ~12C when idle, but went way above 80 on just an ATiTool load, heavily vmodded and clocked....would take a few minutes to drop below 15C again, but they would each time, even with zero flow). Eventually pulled the dry ice off the rad, warmed the rad, and it was working just as well as before the stopped flow (it was windshield washer fluid if anybody was wondering, the stuff that freezes at -40).

I'd think if a GPU from a few generations ago can withstand a non-flowing system, that modern CPUs and chipsets could as well. AMD already had this 'fiasco' with K7 or K6, and they definitely fixed it with K8.

[Machinus]

There are plenty of scientists who would beg to differ with your stereotyped comment.

Just how old do you think the Earth is??? Getting your degree from a church doesn't make one a "scientist."

[[XC] DragonOrta]

I know that s939 A64s have an autoshutdown, or at least mine did.

I had an opty165 build that I forgot to plug the pump in for. I'd get a few minutes into Windows, and it would shutoff. Tried to turn it on another 3 times and it wouldn't even get into Windows before shutting off. Finally realized that my 50z wasn't making it normal whine and the plugged it in. Off it went just like nothing had ever happened.

[jayhall0315]

I know that s939 A64s have an autoshutdown, or at least mine did.

I had an opty165 build that I forgot to plug the pump in for. I'd get a few minutes into Windows, and it would shutoff. Tried to turn it on another 3 times and it wouldn't even get into Windows before shutting off. Finally realized that my 50z wasn't making it normal whine and the plugged it in. Off it went just like nothing had ever happened.

Hey Dragon,
Thanks for the validation of my results from the AMD side. I owned a s939 X2 4800 dual core chip for 14 months and it ran like Seabiscuit. Actually, I suspect that any chip made by AMD or Intel in the last three years or so would be fine, I just cant prove it. And at the end of the day, when the huffing and puffing is over, that is all that counts.