Watercooling obsolete?

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

ChucuSCAD
Patron of SPCR
Posts: 255
Joined: Sun Jan 05, 2003 12:40 pm
Location: Phoenix, AZ, USA
Contact:

Post by ChucuSCAD » Fri Oct 29, 2004 7:28 pm

To all the overclockers out there.........

overclocking is a HOBBY. Plain and simple. There is NO PLACE for overclocking in the professional world.

I agree 100% with Rusty on all his points.....

I myself am an Architect and produce high end renderings and animations.

Now lets say my client requires a rendering at a certain time. Well an OCed system would infact render the final product faster. HOWEVER if the machine takes a dive at 3AM and I am expecting it to be done at 8AM for a 9AM printing I am SOL. The client isn't going to accept "my overclocked computer took a dive overnight and I won't have that poster printed for your opening this afternoon" We as professionals charge alot of money for renderings. We as professionals have lots of redundent systems to ensure ontime delivery. We as professionals can not hang our hats on computers that might be a tiny bit unstable.

And I can hear it now.."well if something is that important you shouldn't be waiting till the last minute to do it" well I hate to break it to you kids but thats not the way the professional world works. Yes I would love to have lots of extra time to get a project done. Everything is always needed yesterday.

Bottom line I would rather hang my reputation for ontime delivery on non overclocked systems then loose a client. For you kids out there a professional animation can cost upwards of $50,000 - $200,000 bucks (that isn't a typo). That's alot of money to be fooling with for a marginal gain in time.

chucuSCAD

Edward Ng
SPCR Reviewer
Posts: 2696
Joined: Thu Dec 11, 2003 9:53 pm
Location: Scarsdale, NY
Contact:

Post by Edward Ng » Fri Oct 29, 2004 8:19 pm

ecto wrote:
Edward Ng wrote: I'm still trying to figure out what sort of professional requirement would involve overclocking.
Seriously? There's a lot. Think raytracing, video editing, or some kind of encoding, which all has been mentioned in this thread I believe. They all require raw CPU power by the megahertz's. But as someone mentioned, if one needs it so much they should have the fundings to buy the necessary computing power. Well, maybe they don't. There is actually people sitting at home rendering more or less complete movies, even if it isn't Toy Story or Monster's Inc.
Russ already responded to this.
ecto wrote:I think it's kind of silly being a computer professional, which you are when writing for SPCR, and saying you can't figure out what professional requirement would involve overclocking. I also understand that Pixar probably don't run overclocked CPU's because they don't have to, but not everyone is Pixar or have their fundings. For a lot of people, a 5% or 10% increase in CPU processing power will be a significant change.
Charlie responded to this one.
ecto wrote:And for all of you saying there's no difference between 100 and 120 FPS - sorry to tell you this, but you are so wrong. Sorry to "push you down", but you obviously haven't played FPS games on a really high level, and you shouldn't talk about what you don't have enough experience to talk about. I get so upset when I read the things you write.

A couple of years ago I played in a top ~8 Quake 3 clan in Sweden. Considering how big Q3 was then, and the fact that Sweden has a history of being really good at FPS gaming, I'd say I really know what FPS gaming is all about and what it requires, and that I know what I'm talking about.

First of all, I'm perfectly aware that if I stare at a blank screen, my eyes can't discerne the difference between 300 and 200 FPS, even though you can easily notice the difference between 85 and 120 FPS. Many people tend to think like when thinking about TV's.. "I can't tell it's really just pictures showing at 24 FPS, so my eyes can't see more than 24 FPS". Everybody knows how easy it is to tell the difference between 60 and 85 FPS. But that's not what you talk about when talking about FPS.

1) You talk about minimum FPS. That means, when you say you have "100 FPS", that means you have 100 FPS at the worst moment, when your whole screen is filled with moving stuff. Everybody can have 500 FPS staring at a wall - the point is maintaining that FPS while in the heaviest fight, with multiple enemies, rockets and fire shooting everywhere. Translate this into benchmarks, this most often means that you have to have hundreds of FPS to be sure that you won't suffer from too low FPS while in the heaviest fights.

2) Another point with high FPS is that while you may have 200 FPS while just runnning around, you also have to have those FPS when you make a 180 degree turn in a split second. Having 85 average FPS just won't do it in those situations. You have to have a lot more to be able to compensate for the losses you face when turning fast/being in those resource-hogging moments.

Ok, ranting over.. sorry if I come about as being uptight - I just don't like when I think that people talk about stuff they only have artificial experience with. Try playing a FPS game at a national/international level (granted that it's a fast-paced game like Q3 for example) at a benchmarked average 100 FPS and see if you don't notice the difference when playing at a benchmarked 200 FPS. Trust me, you will notice the difference, if you play long enough.
Russ, when he's talking about frames per second, he's directing it towards myself.

ecto, you're right...

I do only play games for fun. To me, that's what it is. A game. On the same token, this is why I aim for playable framerates with maximum eye candy, rather than absolutely high framerates allowing me to compete on a national or international level.

I never said there's no difference between 100fps and 200fps--I can also see an obvious difference between 100 and 200 fps--I said between 100fps and 120fps. On the other hand, I'm only talking about maximum frames per second, while you're talking about minimum; my mistake for not adding in, "maximum." This alone is proof enough that your aim for overclocking your 3D hardware is completely different than my aim.

Taking my two previous statements there, my point is that I'd easily sacrifice 20fps off my top end for higher visual quality when I play my games. Your stance is obviously the precise opposite; however, I honestly don't think there's more than a handful of professional caliber gamers at the current time here at SPCR; as a matter of fact, judging by the hardware people are using in their silent rigs that can still play games, they're aiming for the same thing as I do; decent eye candy at playable framerates, at silent or near silent acoustics, not extreme framerates for professional competition gaming performance. To get the sort of framerates you're talking about would simply not be possible in the newer games like Doom 3 without pushing the graphic detail down a few notches; Anandtech's own SLI preview today was proof alone; it takes dual 6800GTs to break into the triple digits with extremely fine eye candy. I'd imagine even using four compressor coolers on a dual Opteron dual 6800 Ultra SLI system overclocked to the very hilt still won't result in satisfactory competitive framerates for you, when the graphic detail is absolutely maxed out (Ultra Detail, maximum aniso and maximum FSAA at a minimum resolution of 1280x960). On the other hand, it would probably be playable enough for myself and moreover--look gorgeous, and that is what I am looking for in my o/c--to raise my eye candy as much as possible while maintaining decent framerates (decent, not stellar, but decent).

As a matter of fact, I don't think a silent rig is particularly appreciable at a place like, say, Quakecon...just a bit noisy to notice. :wink:

Sorry if I pushed your gamer buttons the wrong way there, buddy; no offense intended, whatsoever! :lol:

And just as Russ said, nobody here said low framerate is better; what I'm saying is, higher eye candy is better--imho--happy? :)

-Ed
Last edited by Edward Ng on Sat Oct 30, 2004 7:11 am, edited 1 time in total.

ecto
Posts: 106
Joined: Mon Nov 24, 2003 10:22 pm
Location: Uppsala, Sweden
Contact:

Post by ecto » Sat Oct 30, 2004 2:21 am

Rusty075 wrote:Ecto, you skipped the word "professional" in Ed's post. "Professional" implies in business; "sitting at home" is not professional. As someone who works in a field where raytracing is done professionally, I can assure you that no one is running OC'd machines at their business (no one that I've ever heard of at least). Feel free to insert some a comment about "I just don't like when I think that people talk about stuff they only have artificial experience with", here.
You are of course right. I must somehow have misunderstood Ed. My brain thought not in the lines of a professional business, but in the line of "does he say you can't be professional (good) just because you're at home?" I guess the language played a little plart there, too - "professional" here hasn't got anything to do with working - it's only a skill level. And I'm sorry about that comment! :oops:
Rusty075 wrote: As entertaining as your rant about gaming FPS' was, I'm not exactly clear who you were directing it towards. Was it DrCR, who said, "Oh, I agree 100% that overclocking to get a 100+ extra fps on a benchmark that is well over a gazillionfps already is nuts. "? or me where I said, "Gaming is perhaps the only area where a significant OC can affect performance"?
Actually, my post wasn't really directed to anyone specific, it was more of a reaction against the whole tone of the thread.

I believe DrCR's comment "Quake 3 anyone?" triggered me, as that is the game I've spent countless hours tweaking and overclocking to get maximum FPS out of - and it was necessary. Of course, with today's computers it's not necessary, but it was back then. But the same thing applies with today's games - even if you have a high-end rig, if you play a very demanding game at a high level those 5 extra FPS might actually be needed for you to be able to perform at your best.

It bothered me that people seemed to think that these few extra FPS is just for show, and that "noone actually needs them." I reacted to Ed saying "no need at all, just pure want". This is what I meant above with the tone of the thread - I got the feeling that people hadn't understood that there are people who need that little extra speed, not just wanting it. When it comes to these people's needs (like me a couple of years ago), it's no longer a subjective matter.

I'm sorry if I came off as harsh, it was absolutely not my intention. Like Ed said, I guess my gamer buttons got a little pushed and I overreacted. Sorry for some of my comments (I was the one being silly, not you Ed), no offense intended against anyone!

:oops: :oops:

Next time I won't rant at 4 am while being drunk *bangs head against wall*

snutten
Posts: 341
Joined: Mon Nov 25, 2002 1:27 pm
Location: Sweden

Post by snutten » Sat Oct 30, 2004 4:20 am

Some people might need o/c, however in my opinion there are plentiful other sites out there catering to this and the SPCR forums are better off sticking to the silent business, save the "Not silent" thread.

Btw, professional mean exactly the same thing in swedish:
professionell yrkessmässig, skicklig som yrkesman (Svenska akademiens ordlista över svenska språket 1999)

ChucuSCAD
Patron of SPCR
Posts: 255
Joined: Sun Jan 05, 2003 12:40 pm
Location: Phoenix, AZ, USA
Contact:

Post by ChucuSCAD » Sat Oct 30, 2004 7:04 am

Edward Ng wrote:
Charlie responded to this one.
I don't think they know who charlie is.... :D :D


chucuSCAD
aka Charlie

ecto
Posts: 106
Joined: Mon Nov 24, 2003 10:22 pm
Location: Uppsala, Sweden
Contact:

Post by ecto » Sat Oct 30, 2004 9:49 am

OT, sorry.

Snutten, in english when you use the word "professional" you often mean "a person who works with something professionally". You never (ok, hardly ever) use the swedish word "professionell" in that context. Have you ever heard a swede say "jag är professionell" (I'm a professional) and mean that he does it for a living? I'm sure I haven't. Most swedes (at least where I'm from and the ones I've met) use the word simply for describing a level of skill. You are right about what the word means, but in reality (according to my experiences) people don't use it that way.

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Sat Oct 30, 2004 3:58 pm

I'm busy writing my university thesis at the moment so i'll have to skip over all the responses to this post thus far - apologies if what i'm about to say has already been said.

I agree - watercooling is largely redundant. About half a year ago i was seriously considering the idea of setting up a watercooling company (as some of you know) and i had a full business plan and everything setup, and a legal advisor on hand to handle all of that mumbo-jumbo. one of the main reasons i didn't go ahead with it is that i began to see watercooling as a method of 'dampening' a problem rather than a method of fixing one. The problem is cpu heat which needs to be gotten rid of in the most efficient manner possible. This has already been achieved with laptops, making most (if not all) non-high-end desktop systems seem like oversized wastes of space.

this i concluded by looking at my own computer which is much much bigger than it needs to be, and is nearly outperformed by my girlfriend's just-as-quiet 100x-as-convenient laptop.

but it's interesting to study the other side of the argument. years and years ago, cars used to be aircooled, now they're watercooled for noise and efficiency reasons. also, another plus to watercooling is that with a decent system, temperature no longer becomes an issue. my cpu hasn't overheated ever since i started watercooling, and i can turn my radiator fans down to 5v whenever i want. my overclocking is limited by my particular chip, not my temperatures.

aircooling doesn't afford you that - aircooling is always a compromise between noise and stability. with a decent watercooling system, stability is never an issue, giving one much more freedom with which to tune the noise levels.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sat Oct 30, 2004 5:50 pm

How did I miss this fun thread all this time? 8)

Getting back to the orig question / comment... as people have already said, WC will not go away, it's well entrenched in the enthusiast community. However, the likelihood of it becoming "mainstream" is about on par with silent PCs becoming mainstream. It's certainly less likely now than it was a year ago, despite Press-hot and gradual increases in heat ouput of AMD A64s. The main reason, I believe, can be summed up in one word:

Heatpipes.

They have arrived & been a factor in virtually every aspect of component cooling in PCs. The industry has become increasingly more sophisticated with the implementation of heatpipes for CPU and other component cooling. Most of the gains in HS technology have invoved heatpipes in the past year, combined with better utilization of heat conduction. Heatpipes can be described crudely as passive WC, but much less complex, much cheaper and w/o any maintenance issues. Beyond smarter implementation, the heatpipes themselves are actually getting better as well.

I have a pet project that I am hoping someone will take up around here: An OC'd gamer system using heatpipes & a massive HS -- bigger than fmah's-- using one or two nexus 120mm fans at ~6V. I'd like to see if this can challenge the "best" WC OC rigs. I just don't have the time & SPCR's Vancouver lab assistants appear to be too busy to take up this project... :? :(

I already have 2 of these massive HS (13" x 18" x 1.5", ~15 lbs); just waiting for someone to take on this project. I guess it would end up looking something like a mini-sized fanless Zalman TNN-500 case. Perhaps I need to post this somewhere else...

DrCR
Posts: 538
Joined: Sun Jun 13, 2004 11:55 am

Post by DrCR » Sat Oct 30, 2004 11:06 pm

ecto wrote: I believe DrCR's comment "Quake 3 anyone?" triggered me, as that is the game I've spent countless hours tweaking and overclocking to get maximum FPS out of - and it was necessary. Of course, with today's computers it's not necessary, but it was back then. But the same thing applies with today's games - even if you have a high-end rig, if you play a very demanding game at a high level those 5 extra FPS might actually be needed for you to be able to perform at your best.
Yeah, I was talking about Quake3 w/ todays rigs. ;)
ecto wrote: I'm sorry if I came off as harsh, it was absolutely not my intention. Like Ed said, I guess my gamer buttons got a little pushed and I overreacted. Sorry for some of my comments (I was the one being silly, not you Ed), no offense intended against anyone!

:oops: :oops:

Next time I won't rant at 4 am while being drunk *bangs head against wall*
NP ;) (lol with respect to that last sentence. :lol: )
MikeC wrote:How did I miss this fun thread all this time? 8)
...
I have a pet project that I am hoping someone will take up around here: An OC'd gamer system using heatpipes & a massive HS -- bigger than fmah's-- using one or two nexus 120mm fans at ~6V. I'd like to see if this can challenge the "best" WC OC rigs. I just don't have the time & SPCR's Vancouver lab assistants appear to be too busy to take up this project...
lol welcome to the party.

Hum, I really don't think you'd really be able to do it. Perhaps yes if you don't mind your computer sounding like a leafblower. (Ah, memories of the 5800U lol). I mean, think about it. Cather's 2600+M (1.9GHz stock) was run at 3.0GHz with 2.7GHz is a more common high-end 2600+ OC (at least it is at pcper). I really doubt anyone will deny Cathar is the conventional watercooling king, and his latest Storm blocks are really the limit of conventional watercooling.

Imagine a 2600+M at 3.0GHz (at I don't even know if that was his max) on air! :shock: I mean, wow, I'm about to spaz out just thinking about it! I can not imagine that being quiet. Perhaps if you used a plethora of heatpipes connected of a huge heatsink, like 1+meters^2. Then again, as you said, heatpipes are quasi-'water' cooling as it is.

On the side, don't forget, us watercooler aren't the fringe. I mean, there are more than a couple of assembled packages for sale on the market. The phase coolers are the fringe guys! Perhaps they're the fringe of fringe taking liquid cooling to its logical conclusion (evaporation). The guys over at www.phase-change.com are an inspiration to me to just drop conventional watercooling and go phase cooling, but, well, I just way too intimidated! lol :lol: Of couse, one thing nice about phase cooling is that you get your max overclock at <=default voltages so there's no scare with running your Vcore at 2+V.

Hum, man if my next build didn't have to be a somewhat 'family' rig I probably would indeed go the phase cooling route. After all, the Storm design is probably as good as conventional watercooling is going to get. (Yes, I know you can add more rads, etc, I'm talking about max OC here).
From that article you linked wrote: I spent a fair amount of time trying different CPU speeds with various overclocking settings. This was done to simulate different CPU heat loads and see whether the CPU temperature would go out of control. The highest speed Althon XP is a 3200+ at 2.2 GHz. I was able to run the system for around 10-15 minutes at that speed, but the system was unstable, because I received a system error in Windows. The temperature was in the low 60s °C when the error occurred. I suspect the CPU/RAM combination is the main factor in the overclock success/failure, not overheating. The system seemed to be fairly stable at 2 GHz. Perhaps PC3200 RAM instead of PC2100 Kingston "value ram" would make a difference in the maximum overclocking tests.
A 1700+ at 2.2GHz and only stable for 10-15 min. in windows is IMHO very bad. Were talking zero load here. You can do better on air (standard heatsink/fan), much less water.

DrCR

________

Edward Ng
SPCR Reviewer
Posts: 2696
Joined: Thu Dec 11, 2003 9:53 pm
Location: Scarsdale, NY
Contact:

Post by Edward Ng » Sat Oct 30, 2004 11:39 pm

DrCR wrote:On the side, don't forget, us watercooler aren't the fringe. I mean, there are more than a couple of assembled packages for sale on the market. The phase coolers are the fringe guys! Perhaps they're the fringe of fringe taking liquid cooling to its logical conclusion (evaporation). The guys over at www.phase-change.com are an inspiration to me to just drop conventional watercooling and go phase cooling, but, well, I just way too intimidated! lol :lol: Of couse, one thing nice about phase cooling is that you get your max overclock at <=default voltages so there's no scare with running your Vcore at 2+V.
Any time you go below ambient, you're asking for trouble. If you're capable of dealing with that situation, great! Personally, I am not toying with the risk of condensation. I'm just personally not a fan of gooping the living crap out of my mainboard and/or my video card PCB just to prevent condensation from forming. To those who are fine with this, I salute you...

-Ed

Straker
Posts: 657
Joined: Fri Jul 23, 2004 11:10 pm
Location: AB, Canada
Contact:

Post by Straker » Sun Oct 31, 2004 12:19 am

you could just use a cooler with your existing water cooling, with the evaporator in your wc reservoir, i think that's a nice intermediate - don't have to go to as much effort, and you're not (basically) limited to only cooling your CPU.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Oct 31, 2004 12:25 am

A 1700+ at 2.2GHz and only stable for 10-15 min. in windows is IMHO very bad.
This was not due to poor cooling; fmah thought it was RAM. In any case, it was a completely fanless system.

The system I am taking about is with an even larger HS (13" x 18" x 1.5", ~15 lbs) -- maybe 2 of these -- along with dual Nexus 120s, which can blow >60cfm at <20 dBA/1m. I think with the right combo of multiple heatpipes well secured to the HS/CPU/GPU, this could be one hellova system with tremendous cooling power.

But it's going to have to be done by someone else, not me. I don't have the time...

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Post by sthayashi » Sun Oct 31, 2004 12:16 pm

MikeC wrote:I have a pet project that I am hoping someone will take up around here: An OC'd gamer system using heatpipes & a massive HS -- bigger than fmah's-- using one or two nexus 120mm fans at ~6V. I'd like to see if this can challenge the "best" WC OC rigs. I just don't have the time & SPCR's Vancouver lab assistants appear to be too busy to take up this project... :? :(

I already have 2 of these massive HS (13" x 18" x 1.5", ~15 lbs); just waiting for someone to take on this project. I guess it would end up looking something like a mini-sized fanless Zalman TNN-500 case. Perhaps I need to post this somewhere else...
Have you asked Bluefront? Of all the users here on SPCR, I'm convinced Bluefront spends more time in his shop modding computers than he does sitting in front of them (Excepting Bill Owen of course).

DryFire
Posts: 1076
Joined: Sun May 25, 2003 8:29 am
Location: USA

Post by DryFire » Sun Oct 31, 2004 12:34 pm

sthayashi wrote:Have you asked Bluefront? Of all the users here on SPCR, I'm convinced Bluefront spends more time in his shop modding computers than he does sitting in front of them (Excepting Bill Owen of course).
I'd have to agree with you there.
but tinkering is usually more fun then actual use, IMO.

DrCR
Posts: 538
Joined: Sun Jun 13, 2004 11:55 am

Post by DrCR » Sun Oct 31, 2004 6:49 pm

MikeC wrote:
A 1700+ at 2.2GHz and only stable for 10-15 min. in windows is IMHO very bad.
This was not due to poor cooling; fmah thought it was RAM. In any case, it was a completely fanless system.
No, I don' think that was the RAM (that 10-15min. stabilty that is), it was purely overheating. Perhaps with better ram he could get 2.2+GHz, but it would just overheat all the sooner.

A good overclock for the 1700+ would be 2.4-2.5GHz on air and significantly higher with water (with good chips, of course). One of the best I've heard was a little under 3.0GHz (high 2s, can remember exactly) on water and just over 3.0GHz on phase cooling (anyone heard higher on either?)

That's true, that was a 100% passive setup (and a really nice, professional looking job too I might add) and even just a little bit of CFM, as I'm confident you are aware, can make all the difference. Example, the XP120's passive cooling power vs. XP120 with an undervolted Nexus or Globe.
MikeC wrote: The system I am taking about is with an even larger HS (13" x 18" x 1.5", ~15 lbs) -- maybe 2 of these -- along with dual Nexus 120s, which can blow >60cfm at <20 dBA/1m. I think with the right combo of multiple heatpipes well secured to the HS/CPU/GPU, this could be one hellova system with tremendous cooling power.
Oh yeah, I'm sure it would be an excellent air-based cooling setup! :) I just don't think it "can challenge the "best" WC OC rigs." ;)


That said, I'm tempted to try it. :D Hum, maybe not for the CPU, but perhaps for mosets or something (if only to look extra geeky :lol:). From what vendors would I get the heatpipes and the large passive heatsinks? I'd probably have to mill my own copper blocks though right?

I like the concept. After all, why not utilize the side of your computer to aid in cooling? Innovatek has use the idea with their "Konvert-o-matic" (link) (now just give it to us in copper).

DrCR

_________

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Sun Oct 31, 2004 7:51 pm

DrCR wrote:No, I don' think that was the RAM (that 10-15min. stabilty that is), it was purely overheating. Perhaps with better ram he could get 2.2+GHz, but it would just overheat all the sooner.
200Mhz FSB on cheap PC2100 RAM.

The fact that it occured several minutes in makes it even more likely that it was RAM and not the CPU, not less.

DrCR wrote:A good overclock for the 1700+ would be 2.4-2.5GHz on air and significantly higher with water (with good chips, of course). One of the best I've heard was a little under 3.0GHz (high 2s, can remember exactly) on water and just over 3.0GHz on phase cooling (anyone heard higher on either?)

You cannot use the max overclocks on completely different systems to compare completely different cooling methods. There's simply too many variables involved.


DrCR wrote:That's true, that was a 100% passive setup (and a really nice, professional looking job too I might add) and even just a little bit of CFM, as I'm confident you are aware, can make all the difference. Example, the XP120's passive cooling power vs. XP120 with an undervolted Nexus or Globe.
MikeC wrote:.....along with dual Nexus 120s....
That's, um, exactly what Mike already described. :roll: :wink:




Mike, I'm glad to see this idea is still kicking around in your head. I seem remember us discussing this thing waaay back.

If you look at the performance numbers, Fred was getting about 0.6°C/W passive. The Zalman TNN-500 performs at about 0.44° passively.

Considering what we already know about how much impact even a little airflow makes on a heatsink, I think there's a lot of performance potential in such a system. And 60CFM is not just a little airflow. :lol:

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Oct 31, 2004 9:34 pm

Rusty075 wrote:Mike, I'm glad to see this idea is still kicking around in your head. I seem remember us discussing this thing waaay back.

If you look at the performance numbers, Fred was getting about 0.6°C/W passive. The Zalman TNN-500 performs at about 0.44° passively.

Considering what we already know about how much impact even a little airflow makes on a heatsink, I think there's a lot of performance potential in such a system. And 60CFM is not just a little airflow. :lol:
Uh, I am reminded of it daily now that I moved some stuff out of the garage -- the heatsinks are leaning against one of the shelving units in the lab threatening to crush my toes in a passing accident. :roll: :lol:

Don't know if you saw my post a few days ago about the Nexus 120 in the The Top SPCR Quiet 120mm Axial Fans Compared + 120mmAC Fans thread. Here is the core info again. These are my measurements averaged from 8 samples:

12V : 38~41 cfm : 22~23 dBA/1m
9V : 31~33 cfm : 19~20 dBA/1m
7V : 23~24cfm : <18 dBA/1m
5V : 15 cfm : <17 dBA/1m

Two of these could be run at 9V and you'd probably rarely be able to hear them. 22 dBA/1m (for 2) with their smooth sonic signature is extremely unobtrusive.

Anyway, I guess I should contact Bluefront -- or perhaps he'll see this post & respond. Anyone else who is interested, just give me the word -- I will supply the HS & as many of the components I can get my hands on. A high efficiency high power PSU is no problem, along with Nexus 120 fans, and a slew of design concepts and encouragement. All you have to do is make it work and then take pics & write about it. :lol: :lol:

Bluefront
*Lifetime Patron*
Posts: 5316
Joined: Sat Jan 18, 2003 2:19 pm
Location: St Louis (county) Missouri USA

Post by Bluefront » Mon Nov 01, 2004 2:54 pm

Humm....I never saw this thread before MikeC contacted me about it. That's probably because I've no interest at all in water-cooling a computer. I deal with enough water-cooling problems with autos as it is. :P

MikeC's idea about a super air-cooled setup using massive heatsinks is interesting, maybe even possible.....but I bet not with two Nexus fans @ 6v.

You run into a design problem with this arrangement. The motherboard would have to go between the two heatsinks, which would require a much wider case...maybe a server case. The thing would be a one-of- a kind build....of not much use as an experiment. I've already looked into a flexable connection between the CPU and a large heatsink. It's possible....and would make assembly of such a system easier.

Would I be interested in such a project? Maybe after I retire from my working career. Not right now.....it sounds like a project for someone with access to more tools....more expertise than I possess, and more free time. :cry:

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Nov 01, 2004 4:02 pm

Hey Carl,

Forget putting this in a case: The HS is the case, much like in fmah's design. I am getting some flat heapipes from Scythe/Heatlane for this project. The idea is to use a motherboard with the socket at the very edge, then turn it upside down. A U-profile flat heatpipe clamps over the CPU on one end of the U and the other end folds under the motherboard and clamps against the flat side of the HS. The motherboard actually mounts directly on the HS on threaded spacers. Pretty much like the design of the Scythe e-Otanashi M-ITX fanless case. (see pics on pages 2-3)

Like fmah's design, this thing would stand upright, and the bottom would be completely open. The PSU would be sandwiched in some fashion between this HS and the other side panel. A HDD suspension moduel could also be clamped in there, and all the cooling flow would be like a chimney -- maybe two, from bottom to straight out the top. I guess I have to post a drawing somewhere... Later.

Not sure whether the other HS would be needed/useful. Depends on how / whether the GPU could be heatpiped to it.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Mon Nov 01, 2004 4:25 pm

MikeC wrote:Not sure whether the other HS would be needed/useful. Depends on how / whether the GPU could be heatpiped to it.
Use a 90° bend in a second flat heatpipe to connect the GPU to the other heatsink to be the other side of the case.

Or, mount the VGA on a 90°riser card, and then use another U shaped heatpipe. That would let you keep the case very thin. More of an aesthetic issue, I guess.

Now if only I had a machine shop.

Bluefront
*Lifetime Patron*
Posts: 5316
Joined: Sat Jan 18, 2003 2:19 pm
Location: St Louis (county) Missouri USA

Post by Bluefront » Mon Nov 01, 2004 4:59 pm

If the heatsink is the outer case wall, presumably with fins on the outside, just how do you intend to use the Nexus fans? Looping the heat-pipes down, around the bottom edge of the board, than back up to attach to the big heatsink, would certainly degrade the cooling ability of the setup. I'd rather see the pipes remain level, cross the case to attach to a heatsink that faced the board, rather than behind it.

Lot of different ways to do this, of course. You know when refridigeration people install a new outside compressor.....it comes already evacuated, charged with coolant. There are quick-connect couplings on the hoses, so it just snaps into the evaporator inside the house.

Some quick-connect coupling setup for CPU heatpipes would make this whole heatpipe project a lot easier. I've got a supply of copper tubing, and a bunch of R134 coolant, and a vacuum pump. But not much extra time. :)

Oh....You also should know that automotive AC systems.....much like a CPU heat-pipe system, uses rubber hose couplings in places, mostly by the compressor. So you could build a system employing a length of flexable hose, which would make everything line up and install easier.

And...automotive AC test equiptment on newer cars that employ R134, just snap into connectors on the lines, with no loss of refridgerant. These sorts of couplers can be bought cheaply at auto parts stores.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Nov 01, 2004 5:48 pm

OK, here are some pics & crude visuals.

The PSU is there just for scale. The HS is huge... and filthy right now.
Image

It's very thick, too.
Image

This is the layout concept. The fans are mean to be at the bottom -- and the bottom is completely open, supported by 4 struts or whatever.
Image

Here's a perspective shot... the arrow shows the up direction.
Image

So you can think of the HS & all the components being wrapped in a rectagular tube with top and bottom wide open. The details of how all the bits fit... well that will come as you start wrapping your mind around this.

DrCR
Posts: 538
Joined: Sun Jun 13, 2004 11:55 am

Post by DrCR » Tue Nov 02, 2004 6:26 pm

Rusty075 wrote: 200Mhz FSB on cheap PC2100 RAM.

The fact that it occured several minutes in makes it even more likely that it was RAM and not the CPU, not less.

My bad.
Rusty075 wrote:
DrCR wrote:A good overclock for the 1700+ would be 2.4-2.5GHz on air and significantly higher with water (with good chips, of course). One of the best I've heard was a little under 3.0GHz (high 2s, can remember exactly) on water and just over 3.0GHz on phase cooling (anyone heard higher on either?)

You cannot use the max overclocks on completely different systems to compare completely different cooling methods. There's simply too many variables involved.
Well, the goal was to have it good enough to "challenge the "best" WC OC rigs." Am I missing something? Comparison seemed to be kind of the whole point...

Rusty075 wrote:
DrCR wrote:That's true, that was a 100% passive setup (and a really nice, professional looking job too I might add) and even just a little bit of CFM, as I'm confident you are aware, can make all the difference. Example, the XP120's passive cooling power vs. XP120 with an undervolted Nexus or Globe.
MikeC wrote:.....along with dual Nexus 120s....
That's, um, exactly what Mike already described. :roll: :wink:
Um, exactly. ;) Please look at the context to see where these extractions fit in my position as a whole. Just giving credit to the concept where credit is due. ;) :)
Rusty075 wrote: Considering what we already know about how much impact even a little airflow makes on a heatsink, I think there's a lot of performance potential in such a system. And 60CFM is not just a little airflow. :lol:
Again, that was my point. Please see the previous response. My apologies if the nuances of my word choice did not properly convey my intented dissemination.


MikeC, where did you get the heatsink? That is indeed a nice heatsink.

And is it an issue that the airflow is parallel to the heatsink with it open like that? What about an angular approach with a shield over the fins (thinking of video card heatsink designs with the 'shield' part).

Image

The pic is wrong. The shroud should be attached to the base of the HS (the upper left line should be going to the upper left corner of the HS). The issue with this design is no air is going over your motherboard, which looks like something you would like.


DrCR

________
I know you are all coveting my graphics. lol

DrCR
Posts: 538
Joined: Sun Jun 13, 2004 11:55 am

Post by DrCR » Tue Nov 02, 2004 6:41 pm

Another thought:

Image

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Tue Nov 02, 2004 7:17 pm

DrCR --

You don't seem to have grapsed my basic concept. The HS is standing on its edge with the fins running up and down. Neither of your drawings would take advantage of convection.

As I wrote before, think of the whole thing being wrapped around a 4-sided rectagular tube or sleeve, with the bottom and the top open, the fans at the bottom blowing straight up.

Here is another rough drawing, front profile. The pale blue over everything, and the darker blue vertical lines on either side is the "tube" or sleeve -- the casing. It is open and the top and bottom. The draw blow arrows at bottom show air moving in. It would be totally open, as much as possible. It's like the original Apple fanless G3 Cube. (I have left off VGA cards & the like for clarity.)

The Blue U-shape is the flat heatpipe, but it should have been extended a lot further up along the HS.

Image

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Tue Nov 02, 2004 8:34 pm

I wonder if the outer casing really serves any purpose. Why not just use the heatsinks themselves as the casing?

Here's an equally rough idea of what I'm talking about:

Image

Talz
Posts: 223
Joined: Mon Jul 07, 2003 9:29 pm

Post by Talz » Wed Nov 03, 2004 11:47 am

Water cooling allows my system very near silence under load, and that was not possible on air, even using the best performing heatsinks etc reviewed here. I am overclocking a P4 and NV6800, and I'm doing so mainly to make my games look as nice as possible without unacceptable slowdowns.

But water cooling if anything is slowly moving closer to the mainstream, though I don't know if it will ever get there. But cpu's and gpu's are getting ever hotter and it's taking more and more elaborate (and expensive/bulky/loud) designs to keep adequate temps with air cooling.

Now clever use of and improvements with heatpipes and other clever tricks with heatsink design will help keep air cooling dominant for a long time to come. BUT water cooling is much easier to adapt to a variety of system designs, especially if it is a setup with an external radiator. There are some very serious advantages to moving so much heat far out of the case through a small water path that just isn't practical with air cooling. For this reason I see water cooling continuing to grow on the desktop/enthusiast front for awhile. While heatpipes and careful airflow management etc will continue to dominate on the SFF and OEM pre-built designs.

So yes it's not mainstream, and I'm not mainstream. But there's a big market out there with room for lots of different niches.

pangit
Patron of SPCR
Posts: 445
Joined: Wed Feb 05, 2003 1:48 pm
Location: Sydney, Australia

Post by pangit » Wed Nov 03, 2004 2:10 pm

Rusty, your picture is what I thought Mike meant at first, but his last picture makes more sense to me, as it doesn't rely purely on convection to cool the large heatsink. The fans blow up both sides, through a channel between the HS and outer case.

Which one is better, and whether you need one or two HSs, I'm not sure. But until somebody takes up the gauntlet it's all speculation and guesswork!

DrCR
Posts: 538
Joined: Sun Jun 13, 2004 11:55 am

Post by DrCR » Wed Nov 03, 2004 8:21 pm

Yeah, my bad MikeC, turn the first pic up on its side to allow for natural convection. But then there would be airflow through the interior like your and Rusty's prototype layouts.
pangit wrote:Rusty, your picture is what I thought Mike meant at first, but his last picture makes more sense to me, as it doesn't rely purely on convection to cool the large heatsink. The fans blow up both sides, through a channel between the HS and outer case.
Hum, a combination of Rusty's dual HS + MikeC's in-case idea (to keep the air flow passing over the HS fins) sounds like a good approach.

Oh, and what software are your guys using to create those sketches? @Last, FH, PS?

DrCR

________

ascii
Posts: 93
Joined: Tue Sep 14, 2004 1:10 am
Location: Sweden

Post by ascii » Thu Nov 04, 2004 4:37 am

My brother and I talked about implementing something similar to what MikeC is proposing, i.e. making a Zalman TNN-clone. In the end I got myself a Reserator instead, because it is so much less work.

Post Reply