Printable View
Marci, QFTQuote:
Originally Posted by Marci
It'a exactly why rookies like me have so many questions and confusion about the right decisions. Glad this thread turned to something useful.
Thx Maxxx for cleaning it up, it woke me up from thinking there is only ONE absolute possibility, but I feel more enlightened and really think differently about things that for me were recently simple un-argueable facts ...
S'all about what conclusions folks draw from the evidence presented.
In IHS Situations, all are equal.
In IHS-Less situations, Storm appears to win.
This says to me personally that the Storm is the better block of the two... it's just it's advantages can no longer be seen/demonstrated on today's processors and it appears to perform the same as everything else. Should the day come when CPUs go back to being IHS-Less from the manufacturer (no-one can say whether this is likely or not or under what timeframe), imo the Storm would reclaim it's crown providing the area of heatload falls within the Storm's cooling area.
In the meantime, open yer wallett, see what cash is in there, decide how much you wanna spend, and buy whatever suits that figure.
Performance ranking can only be done on a setup where ALL variables are controlled. In a physical PC setup, NO variables can realistically be controlled.
No testbench can adequately (in the eyes of the masses) simulate an IHS'd CPU at this time. The bond between diesim and IHS cannot be replicated nor controlled nor it's c/w measured or defined. Testbenches become useless. PC's were useless for this purpose from day one (minute variances in voltage regulation across two identical boards means the actual wattage heatload can't be defined). Intel and AMD TTVs are NOT easily available to anyone... I've tried with both companies and neither will part with one, therefore, the same loadsimulator cannot be used by all testers, so no-ones results are comparable. There literally is no answer.
Split the testing groups down into 3 categories...
CPU-based
Diesim-based
TTV-based
Any proclamation made by any one of the 3 groups can easily be shunned by any of the remaining 2, until all 3 groups produce identical results via their varying methods.
The fact remains that the only true way to empyrically rank anything is via it's c/w and associated curves, with all figures derived via environmentally controlled variables. No-one can currently do this due to the inability to replicate the IHS > Die joint repeatedly. Even the IHS itself cannot be controlled. Multiple mounts and dismounts eventually deform the IHS, and it must be replaced, however, then the joint between IHS and Die has changed and all previous results become invalid once more. It's a spiral. TTV is the sure answer. Availability (or lack of) to independants prevents it from being the solution.
References... (or former arguments that all influence the above)
http://forums.procooling.com/vbb/showthread.php?t=12864 - Define the point of temperature measurement
http://forums.procooling.com/vbb/showthread.php?t=12610 - Controlling airflow
http://forums.procooling.com/vbb/showthread.php?t=12585 - Ensuring Calibration of all temperature monitoring devices
http://forums.procooling.com/vbb/showthread.php?t=12888 - Ensuring surface flatness
http://forums.procooling.com/vbb/showthread.php?t=12431 - Defining what testing is necessary
http://forums.procooling.com/vbb/sho...highlight=tim1 - Definition of TIM C/W
http://forums.procooling.com/vbb/showthread.php?t=12306 - Definition of measurements
http://forums.procooling.com/vbb/showthread.php?t=12304 - Diesim design
http://forums.procooling.com/vbb/showthread.php?t=12376 -|
http://forums.procooling.com/vbb/showthread.php?t=12475 -- The original argument
That lot's for starters... the average user will never understand the content of the majority of the above, nor want to... however, it is the average user who will question the results derived by those referenced... mainly due to lack of understanding of the methodologies / lengths that must be gone to simply to characterize a waterblock.
I don't think there will EVER be an agreeable solution. All decisions hence forth become a matter of conjecture/heresay. Simple as.
Neither is better. Let fanboyism / wealth / perception decide, without outside interference.
No-one will ever be able to conclusively say "X is better than Y".
Give it up. *shrug*
I'm just glad radiator testing doesn't require any involvement of TIM's, IHS etc etc, just a "controlled" heatload dumped into the water via whatever method, a "controlled" liquid flow rate, a "controlled" airflow rate, and a "controlled" ambient environment. "Just" however still equals lots of expensive equipment. Accuracy of measurements must be to minimum 3 decimal places across the board... variances are tiny.
Folks don't realise / appreciate how hard it is to empyrically say anything is better than anything else... and the lengths that people voluntarily go to and the amounts they spend just to allow others to recommend one piece of hardware over another.
Endusers demand a performance scale to answer the "what is best" question... even if the differences are MINISCULE, via empyrical testing, one is always better than another. Whether that translates to realworld use, who cares? Scientifically one beats another, reality one beats another... s'just the hardware we're using isn't precise enough in it's measurements to reflect any of it. But folks will still always ask... what is best?
End of the day, those involved in all the above links, and all those involved in any form of waterblock testing couldn't derive a satisfactory answer that could please everyone with results that were indesputable... this includes the folks producing the graphs and figures referenced thru-out this thread. What makes anyone else think THEY can derive a satisfactory conclusion from any of it?
I appreciate the challenge with doing scientific waterblock testing... but I, and probably most people, don't need indesputible evidence that one block is "the best"... Just knowing how a couple of different blocks perform in someone's system is enough for me to make a purchase decision. I encourage folks in this forum to continue to provide their adhoc informal test results for exactly this reason even if they are highly debatable and cause flame wars... what else do we have?
Quote:
Originally Posted by virtualrain
Given the flamewars about and the ensuing crap subsequent to a well-intentioned person submitting the results of their own "ad hoc infomal testing," what incentive does one have in providing one's own test results?
Clearly, in my own thread, the moment I posted my own test observations, there quickly came individuals who found it necessary to launch insults. Among them was the insinuation that I don't know what the hell I am doing and that I am just fueling controversy... I even saw someone calling someone else an "idiot," just because the posting individual did not share an opinion or conclusion.
I have learnt to keep my opinions and test results to myself. After all, they are only for my own benefit. Apparently, sharing information only serves as an open invitation for flaming.
I guess part of the elitist ego comes with the territory of calling/regarding oneself as a membe of the "el33t" Xtremesystems.org. However, that is absolute nonsense and no one has a monopoly on knowledge and skill, and that includes the gurus of this forum and their haloed heros (I don't need name names.. these individuals are readily identifiable). I don't see a general willingness to learn here, just a rapid desire to look down on everyone with contempt and call them "n00bs" and "idiots"
ps: i was thinking, some of this teasting does need to be cpu loop only with a SMALL midrange rad. like a 80mm rad. the reason being, if you test 5 waterblocks with a PA 120.3 and yates @ 12v for example, it would seem to me that the rad would do a lot of the work and bring results very close to each other. whereas with a much more simple, low performing rad, but with constant heat load, ambient temps and only variable flow, maybe then we could see some separation in results?
anyway just some musings.
I think we can all agree the extra money would be better spent on a thermochill rad!!
Like the difference between a BIP 120.3 and a PA 120.3 is larger than between an apogee and a storm (no, let's not discuss this:D)
Jochenp
Oh boy, haha. All I'll say is Marci's link in his sig shows the PA120.3 clearly being a superior product. Same can't be said for waterblocks, especially with IHS still on.Quote:
Originally Posted by Jochenp
Ofcourse the PA is cooler, more silent and has less restriction.
I know the facts, I've read the reports, it's just another case of "is it worth it?".
If you asked me, I would say it is, but thats just my opinion.
to me: asking if a more expensive PA rad especially the triple is "worth it" is asking if the brand name super powerful PSU is "worth it".
the answer for me and i think should be for others: HELL YES.
why? cuz the problems saved by having those 2 firm foundations pays for itself day in day out. throw anything else in the system together and at least you have a solid deep foundation like what they building skyscrapers in NYC with hahah
All that would do is increase the coolant temp... which wouldn't cause the difference between blocks to get any larger (since you're working with dT (core/slug/whatever temp minus coolant temp)). To increase the spread, you'd need to increase the heat load applied to the block.Quote:
Originally Posted by el rolio
Example: The thermal resistance of waterblock "A" is 0.12*C/W... which means that its use would result in temps that are .12*C over the coolant-in temp, per watt of heat applied. Let's say that waterblock "B" has a thermal resistance of 0.13*C/W. With 60W applied to the blocks, block A would be running 7.2*C above the coolant temp and block B would be running 7.8*C above the coolant temp--a very tiny difference. However, with, say, a 160W load applied, block A would run 19.2*C above the coolant temp and block B would be 20.8*C above the coolant temp. You can see that the gap gets larger as the heat load increases.
A smaller radiator would increase the coolant temperature which, yes, would raise the temperature of whatever you're cooling but it wouldn't change the temp differences between blocks. An increase in coolant temperature won't change the thermal resistance of a waterblock.
...and someone needs to make it so that ° works... :stick:
I understand... however, no one should let a bunch of flaming and name calling deter them from posting their experiences here... if product X works better for you than product Y... let us know! That's the whole purpose of these forums!Quote:
Originally Posted by IanY
Everyone can then make up their own mind whether your results add to their body of knowledge or whether they discard it.
Having said that, it's not unreasonable for people to question results and to push for more thorough testing methodology, etc... that's fine and makes us all think a bit more about how we test. However, they need to accept that testing by forum members is what it is... acronyms like FWIW and YMMV were invented for a reason!
ahh, thanks man, i was missing the fundamental understanding of block c/w properties..... like other noobs :O!
Thanks Marci: Lotsa' stuff to read.
@IanY: I'm sorry poeple got on you. but it's usually not a good idea to post temp diffrences between blocks as conditions between test can change dramatically.(ambients,humidity,paste consistancey,mounting pressure. And on and on.)
Even screenshots like this helps no-one. And could be misleading.(I could get flamed LOL)
http://img172.imageshack.us/img172/8...2x10rb4.th.jpg
Full load prime 1hr
Too much missing information.(Ambient 20.6c) Even if someone has the same model parts. Under the same conditions. The results would be diffrent. Plus I suspect the board sensor reads abit shy.LOL
Edit: It's not my intent to misslead anyone. But by the nature of the beast itself results like this tend to be missleading.(In otherwords thats how it turned out) Not too bad eh?
I see that over the two weeks I spent on my first vacation in 5 years that WSP has been up to his usual tricks of rehashing a tired old pointless debate filled with invective vitriol.
I don't believe that it was the Apogee that caused this, but rather the inability to decide on just how to assess performance once IHS CPU's became the unavailable reality. The Apogee's release just more-or-less co-incided with the height of the struggle of the enthusiasts to deal with what to do about IHS capped CPU's in terms of assessing wb performance anyway.Quote:
Originally Posted by Marci
Back when things were bare-die, it was far easier. Nowadays with the IHS in the way there is no way to repeatably, reliably and adequately assess waterblock performance. I also fall into the camp where I think that a TTV without a Tjunction diode is also next to useless as it doesn't allow the tester to track variations in die-IHS, which btw is what internal testing in Intel and AMD do, but none of the "public" validation testbed TTV's provide. The heatsink validation TTV's are just that, validation tools, not intended as testbeds to measure and account for all TIM variations from the heatsink back to the heat-die, so I don't believe that even getting a hold of a TTV is the answer unless you can get hold of an Intel internal testing TTV that has the additional thermal diodes that allow for full characterisation of the various TIM interfaces, and they'd be impossible to get a hold of.
Of course, a number of artificial die models were proposed in an attempt to solve the issue of testbed unavailability, but it seems that for some people their professional careers rested on the defense of the validation TTV's as a valid assessment tool, even though the validation TTV's were never designed to wholly fulfill that role.
I still believe that the "copper mushroom" artificial die can be a valid test tool, since after-all, the TTV is also just an artificial test die itself, and the copper mushroom approach removes die-IHS variations, and so brings us back to the days of the repeatability and variability issues that we were faced with bare-dies, meaning not perfect, but at least there's less variables.
Of course there will be those that will cry foul and say that a copper mushroom doesn't simulate the real world, but the real world CPU's are full of per-cpu variations themselves, dependent upon manufacturing variations in how well an IHS mates between both the CPU and the heatsink, and the quality of the CPU-IHS TIM, as well as the pressure and angle of pressure. A validation TTV doesn't allow us to measure those variances. Some will argue that those variances are unimportant, although I still fail to understand why.
This is the issue. With IHS's there is now ALWAYS going to be a much larger disjoint between an artificial test-bed and the real-world, but the problem here is that the real-world CPU's introduce a rack of variables that cannot be measured or controlled, so what do we do?
I for one believe that the best that we can do is to attempt to level out those variables (i.e. the copper mushroom die-sim) and explain what is going on, what the assumptions are, where additional variables lie, why what the user sees on their own CPU may differ to what the die-sim is showing, and so on.
This all requires a level of mature reasoning and understanding from the enthusiast community to support such an approach, rather than mindlessly bash when someone's favored waterblock doesn't win out. Of course in the real world with lots of variables we are always going to be able to pull out some instances where one block does better than the test-bed can show. Here's where the reviewer must introduce their disclaimers and explain the variables and state that the "ranking" is merely more of a "trend" rather than a be-all, end-all statement of waterblock performance. i.e. sure block X may be better than block Y in the testing by 0.5C, but understand that in the real world the variances may be up to 2.0C, so on average the differences will be 0.5C, even if in your particular experience block Y is the better block, and hey, that's fine too, because on that CPU with the way the mounting works and the way the hosing hangs off and the way the IHS is deforming, it really is true that block Y is better than block X, even though block X is the higher ranked block in tests.
No need to get all emotional about it. Approach it with calm reasoning and logic and understand that in a modern CPU world with large degrees of greyness that perceiving everything in black and white is foolish. Instead, perceive differences in terms of trends rather than absolutes, and I reckon there'll be a lot more tolerance.
Just my 2c.
P.S. I really am tempted to do the copper mushroom die testbed anyway, but the only thing that really prevents me from doing so is the then perceived conflict of interest between me making blocks and providing test results, and then having to put up with those who don't understand the variables attempting to shoot down credibility based on their real-world results, which may very well be quite true for them, but don't understand that this is all quite reasonable for them to find empirical disagreeance in their instance. I'll just do the testbed anyway and keep my results private.
please do the copper mushroom testbed and please make the results public. It's better to have some public data rather than none at all. How many times in the past where people believed false theories or data only to be shown years later they were wrong? In order for things to progress people have to take chances. Please if you do this share your information.Quote:
Originally Posted by Cathar
Do it.
Put a disclaimer up. Be obtuse and verbose in it. Point anyone who is critical to it.
Don't let the narrow minded ppl stop u. It'll be a great waste for the rest of us...Quote:
Originally Posted by Cathar
Please do the mushroom testbed and make the results public!
@Cathar,
I don't see how tests from someone with your experience could ever be conceived as useless. You obviously bring a great deal to the table in "real world" design, testing and manufacture.
Personally I would not care if the flamers stood in a circle and set themselves on fire. We need people testing, evaluating, retesting, rethinking every design and possibility. Until the day the CPU manufactures can guarantee that EVERY CPU will be built to the exact specifications we will always see differences.
Off topic: I hope that you and the family got to spend some quality time together. The children grow up waaaaaay to quick, just try not to blink! ;)
A "little" more? LOL :DQuote:
Originally Posted by OCme
Now that I've read this I'm gonna have to lap my G5 and hope temps drop even further.
Has anyone seen a review of the performance of a Silver G5 block on a Kentsfield cpu?
After I have read through some of the posts in this thread, I can make some conclusions:
If you have a CPU which is IHS-condom equipped, it really wont matter which CPU block you take. I mean, Why grab a G4, which is very restrictive, hard to get and exspensive. When you can buy a Apogee that is equal-performing?
For the persons that want the high-end, I suppose the G4 is the choice. Just because it performs 1*C better at some conditions. Well, it also looks nicer.
But for me, who will have a IHS-equipped CPU,, wouldn't the apogee be a better choice? I am also a student, so grabbing a G4 seems to make no sence, thinking of my economy.
If we but it this way. Without the IHS, the G4 is said to perform about
1*C better. But does this really matter? I mean, how much more can you overclock? like 2 fsb or what?
Absolutely not. Go for the G4. The reason is that if money was really an object of consideration, you'd be buying a Freezer 64 from Newegg for 17 bucks over any water option and save yourself 150 bucks. So I say go for it!Quote:
Originally Posted by NysoO
Umm by that logic why would overclockers even use anything less than X6800's and QX6700s. I keep wondering why while the storm is a better performing WB in alot of situations everyone is putting down the apogee. I saw someone say looks like it was built with a hacksaw. So what? does it perform?, if you want beautiful then buy an air brush and paint it onto your case. If you want performance for cheap go with the apogee, if you want a little more performance and dont mind spending more cash for it then go with the storm.Quote:
Originally Posted by Graystar