Re: product testing today – who is being served ?
 

Bill, I must have missed it in the other thread, is there factual data on the TIM joint reliability issues with AMD? is there actual data behind that?

And is there any factual data bout Intels TIM joint reliability? From what I understood most of this is all assumptions still.

Just looking to get more smart on that topic.

Re: product testing today – who is being served ?
 

Is the diode to be read known to be in the "hotspot"or is it still location unknown?
If the later; is the relation between it's temperature and IHS(Case) the relation which is sought
Think Incoherent's future Iron Age die may be more informative, it potentially being easier to model.

Re: product testing today – who is being served ?
 

uncollated - factual as in a chunk of folks with first hand tales of the issue, inc myself altho I seem to be one of few who has had CPUs with duff IHS contact from new. No figures... just "know" of it happening and it's existence.

Never heard of issues with P4's...

Re: product testing today – who is being served ?
 

Joe,
Marci had some good anecdotal AMD 'problem descriptions', a phase change guy (Eric K) said the AMD thermal cycling problem was well known, plus my observations on the (non)removal of the Intel IHS with the absence of reported issues - different mfgn process

the assumption that the Intel internal TIM joint is stable seems quite valid

Les
2 courses here, CPU and die
I agree that the die is more suitable for modeling (w/o a long ton of assumptions re the CPU, mobo, etc)
a die will always yield more and better data as that is its purpose,
but as an alternative to the cost and complexity of a die I wish to explore using a real CPU in some fashion

something that we can recommend to those scorned and despised review sites

agreed there is a temp relationship between the IHS and the silicon (lol), but as you say where ? and what is it ?
seems simple, assume a TIM joint value and add a bit for IHS conduction and calc a die temp

Re: product testing today – who is being served ?
 

If it is the "what" of interest, the power region,yes, should be simple.
But if not the high power then the so called compression is probably only known by the manufacturer. It will just add another temperature to the "unknown offset" list.

Re: product testing today – who is being served ?
 

"But if not the high power then the so called compression is probably only known by the manufacturer"

not understood

Re: product testing today – who is being served ?
 

Yes, can calc a die temperature but if not the temperature in "region of max power dissipation" then another offset is required. Our(pH's) temperature will only be the temperature that processor manufacturer allows us to see.
Still worth having - but I am not doing the work.
The difference in temperature of our(pH's) position of measurement and the temperature in the region of maximum power dissipation will be unknown to us.
I only guess that this is the temperature of "max power region"to which "well documented thermal characteristics of the TTV" applies (quote from memory).

Re: product testing today – who is being served ?
 

Why is it necessary to test with a CPU and not a calibrated heatsource as a simulated CPU?

Re: product testing today – who is being served ?
 

I think because that would cost money and Bill was trying to see what could be done by 'typical' review sites without laying out too much cash (which they would be unlikely to do).

Re: product testing today – who is being served ?
 

So what is the projected budget?

Re: product testing today – who is being served ?
 

undecided / unfixed. Budget will be whatever it costs once parts are decided. No matter how it's done, real CPU testing should still work out cheaper to produce than diesim etc... in theory...

and just for clarity to other readers...

In my case, aware of generalisation and lack of empiricism when statement made...

Re: product testing today – who is being served ?
 

as the use of a CPU with an IHS to test cooling solutions has been roundly denounced as unacceptable in the Apogee thread; it is time to close this topic, at least for procooling

it seems that while there is considerable interest in rather exotic testing hardware, this does not extend to consideration of a test method utilizing a CPU with an IHS (as bought)
by declining the challenge of defining a CPU based testing system, used for 99% (?) of all 'reviews', procooling forfeits the opportunity to shape, and assist those doing, that type of testing

is procooling only to address the absolute high-end ? (as seen from within procooling)

Re: product testing today – who is being served ?
 

Bill... the more I look at it the more it makes no sense to test with more unknown variables than are needed to do a simple, inexpensive (relatively) test. Since you cant prove that the Intel TTV's or CPU's are 100% consistent, since no one can prove for fact what kind of inconsistencies are seen with AMD cpu over a range of CPU's and different plants (same goes for Intel)...

I think in large the IHS is a liability to the testing no matter who makes the chip. Since we are discussing testing something that there are few facts about (just lots of assumptions and word of mouth observations from friends of friends of friends), and little confidence in getting real tangible results from on a consistent basis... It just seems to go against the logic that got us to the point of testing the way we are.

I mean look at the review industry as a whole, its all moving to better ways and methods. Think 5 years ago when most sites used mobo probes and MBM to do review testing with. Someones gotta move testing forward... or present different ideas and ways of doing it. Being happy with the variability in an IHS as "real world" gets us back to MBM days.

I don't buy that testing without an IHS is an absolute high end, to me if you are an enthusiast, you probably are going to remove an IHS to get your moneys worth out of the cooling system. I just see us as catering to our audience: enthusiasts.

If procooling stands on its own to test in a way that is logically more sound than other sites, I guess thats what happens. We've been there before. If other sites want to join in and start testing sans IHS, then so be it.

Re: product testing today – who is being served ?
 

I've no real objection Joe, I just think we are walking away from something that is potentially useful - even with limitations
no mechanic discards a tool

Re: product testing today – who is being served ?
 

... but his main tool is still a hammer . ;)

Re: product testing today – who is being served ?
 

yup, a bfh is almost as useful as a heat wrench

Re: product testing today – who is being served ?
 

Re: product testing today – who is being served ?
 

It's not an easy discussion.

...and...
http://overclockers.com/articles638/

If we break down the above, we can list the issues:

1) temperature measurement

So we fix the temp measurement issue with pH's solution, but it's "too high end", so we switch to the IHS groove, but not all like it.:bawling: The IHS issue is bypassed this way.

2) Power

pH has been sucessfull in repeating a steady heatload, using the appropriate program.

It's understood that the power isn't measured, just repeated. The IHS issue does not affect the power applied.

3) need to measure flow

Use a Swissflow flowmeter.


Am I just beating a dead horse here, or did we have something all along?:hammer:

Who's approval does this need?:uhh:

Re: product testing today – who is being served ?
 

I must disagree, the primary tool, in any endeavour, is the mind.

Re: product testing today – who is being served ?
 

You haven't had to work with the design engineers I have to work with. You might change your "mind" on that. :)

AS for this test bench I would suggest this to start with:

-Intel Soldered on IHS CPU and compatible mobo with overclocking options.

-Use the onboard probe (bare with me)

-Use a swissflow flowmeter

-ignore all variables

-Forget dP for now

Take a series of tests with 3 known different performing blocks. Like a Maze 4, TDX and MP-05SP.

Compare these results with a more sophisticated test bench at same flow rates and see how the order of blocks are performance wise. If the order at each flow rate is consistent between both benches then you have a bench capable of usable results.

If the order is way off you can start changing things starting with a better temp monitoring system and start to figure out what exactly is needed to be done and work out how to do it.

It would be preferable that the tests were done by the same person at the same place I would imagine.

Anyway that is my simplistic way of looking at it. Start simple and move more complex until you get your desired result.

Re: product testing today – who is being served ?
 

dP can be done the old fashion way; vinyl tubing in a manometer arrangement. Very easy to do (I've done it, ask Thykingdomecome).

Re: product testing today – who is being served ?
 

Not really. I tried it once. Don't have high enough wall height for all that tubing. You need like 138" of vertical rise to get 5PSI worth of measurment. That is 11.5 feet. After my bench was setup I only had about 3.5 feet of height for tubing as standard walls are not much higher than 8 feet.

Unless I am doing something wrong.

Re: product testing today – who is being served ?
 

No, you're right, but many block's dP fall "near" 3 feet.

The dP test doesn't have to be run while testing; set it on the ground for a bit more height. ;)

Re: product testing today – who is being served ?
 

Do you want to take dP at the same flow rates you tested the block at? If so you would need the flow meter and valve still.

Re: product testing today – who is being served ?
 

Yeah, but you're measuring the dP across the block, right?

Re: product testing today – who is being served ?
 

I would hope so.

Re: product testing today – who is being served ?
 

Right. So you run the benchmarks while measuring flow then when you're done, take the block off, put it on the ground and run a simple loop, and measure both flow and pressure (nothing else).

You only need the flow or the pressure, while benchmarking; not both.

You then need to run the pressure and flow test on the side (off the bench). With the data that you collect, you can translate your benchmark results to dP instead of flow (or vice versa). ;)

Re: product testing today – who is being served ?
 

To much of a pain in the ass. Now you have to unplumb the flow meter and pump, plumb it into the manometer, unplumb it and plumb it back into the test bench. Or buy another pump an flowmeter that would cost more than a digital manometer like Robo uses.

Anyway the options are there.

Re: product testing today – who is being served ?
 

So how do we setup for these Intel Preslers? :hammer:

Re: product testing today – who is being served ?
 

no idea jd (not true)
the only co that was providing such you guys just vilified for so doing
congrats

the die sim folks gonna make dual independant sources ? (not too hard; 2 sizes, 1 big and 1 small)
lol