Just look at the bottom of the page:
http://translate.google.com/translat...st&sl=ru&tl=en
I've just bought those 4 of Shin-Etsu MPU 3.7 from:
http://www.thermalfx.com/merchant.mv...P&Store_Code=T
They're really hard to find..
Just look at the bottom of the page:
http://translate.google.com/translat...st&sl=ru&tl=en
I've just bought those 4 of Shin-Etsu MPU 3.7 from:
http://www.thermalfx.com/merchant.mv...P&Store_Code=T
They're really hard to find..
For some reason Google translates only half of that page for me. I can read though they only use a 100W heater, and that's simply not enough nowadays to really show the differences. The bigger the heatload, the bigger the differences are gonna be and measurement uncertainty becomes less important then.
Also they don't show cure times, Liquid Pro starts to perform significantly better after 24-48h, and they don't tell how many mounts they did (at least they don't in the part of the review that Google does translate).
Shin-Etsu X23-7783D wasn't even more effective then MPU 3.7?
And about hard to find .. i saw shin-etsu pastes at petrastech shop / chilledpc.co.uk / chilternconnections.co.uk / chillblast.com and probably many more .. just use google.
A little better review then:
http://benchmarkreviews.com/index.ph...1&limitstart=7
Seems "Shin-Etsu MicroSi G751" is a new champion, so to speak.
And about hard to find..as i said earlier, Shin-Etsu MicroSi MPU-3.7 (this one exactly) is sold only at one store.
You can even check it here:
http://www.google.com/products?q=Shi...i+MPU-3.7&aq=f
The benchmark reviews roundup is largely useless, since they did not do enough mounts, and the differences are too small and measurement uncertainty too high. I mean the room temperature in any normal room easiyl varies by +/-1C every day unless you have a very expensive AC. You need to log the intake temperature of the CPU fan, and they did not do that, so accuracy jsut isn't high enough to claim that one TIM is 1C better than another.
What's very useful though I think is his experiences with different application methods.
Liquid Pro should also be the top performer here, without a question, yet it only gets a "B+" (being only 0.8C worse than the best A+ accoding to their numbers). They also say it needs no curing time, while the manufacturer clearly states that it does...
The review is defnitely a good attempt, but in the end it was just too much to handle.
Testing TIMs is incredibly hard, and the only good recent review I know of is Vapor's here.
Thanks for this review Vapor. After much thought I just got done changing from Arctic Silver Ceramique to Indigo Xtreme a few hours ago and am XTREMEly happy with the results. I found the application and reflow incredibly easy to do and will definitely use it again in the future unless something obviously better comes along.
My temps weren't bad before at all, but after running some benchmarks they've dropped 6-7c under load. Maybe I just did a better application this time, but personally I don't know of a cheaper or easier way to get those kinds of results.
Thanks again Vapor.
John
Roundup #2 is in the final countdown before testing begins :cool:
Going to be testing on an AMD processor this go-around and using a CPU-360 as the block.
In this test, barring issues: Indigo Xtreme, Coolaboratory Liquid Ultra, Coolaboratory Liquid MetalPad, Arctic Cooling MX-2, Arctic Silver Matrix, Shin-Etsu X23-7783D, Shin-Etsu G751, Gelid GC-Extreme, Zalman STG1, OCZ Freeze, and generic silicone paste.
AS5, MX-3, IceFusion (200g for $16...clearly a new value leader), GC-2, and a couple others *might* make it in should there be issues with one of the planned 11.
Did you just say 200g for 16 bucks? Holy christ that's alot for the money.
My memory's been faltering recently...it's actually $16.99 for 200g :p:
Exciteaments. I know you do testing on water blocks, so you may not know about this, but I think it would be helpful to mention that Indigo Xtreme reflow with top flight air coolers (like NH-D14) require you to actually heat up the heatsink manually in addition to the traditional reflow process. I had to put a halogen work light about 5 inches away from my NH-D14 to get the reflow to work.
I just recieved some indigo extreme from Masska and was wondering if it is essential to go through the reflow process straight away? Or would say doing a windows install before hand be okay?
Great stuff Vapor! I know it's probably too late to ask, but would the Prolimatech PK1 be included on this roundup? I'm really curious to see how this matches up against the other pasts, since newegg sells it for $20, I would have high expectations out of this paste.Quote:
Originally Posted by Vapor
I would strongly suggest you DO NOT do this. Get Windows set up first and running fine with the stress testing and temperature monitoring software before using Indigo Xtreme.Quote:
Originally Posted by converge
Vapor: btw, what is the reason for next roundup to be tested on AMD? imho o/c-ed i7 is able to dump way more heat per area (bad from consumer POV, but imho better for tests) then it's amd counterparts?
+1.Quote:
Originally Posted by johnnymossville
I'd much rather be using an i7 for a lot of reasons. Temp sensor output scaling is linear and 1:1 on all the chips I have, I know how they OC a lot better, I know how they behave at TJMax (important for the MetalPad and IX applications), I have all the necessary hardware to run them, they do run hot at safe voltages (I consider up to 1.55v safe for the typical i7), etc.Quote:
Originally Posted by churchy
I'm using an AMD because that's the kind of IX I have....in late 2009 I was offered to do a review of IX for AMD, I had a 965BE lying around and figured why not. Enerdyne sent me 5 mounts of it and I set out to finish up my waterblock tests and gave my word that I would do a second TIM review. Between vacations, some testing burn out, extreme frustration with my testbed flowmeters (the Dwyers were so loud), I had some downtime. And during the downtime my 965BE was run under LN2 briefly, only to realize it was a frigid 'witch' of a chip. It's pretty easy to gauge how hot a chip runs when you're on LN2 since the hotter chips are simply heavy drinkers, and if you know what you're doing, it's easy to sense. My 965BE was unusually cold, so in addition to my averseness with my obnoxiously loud testbed, I was just discouraged further by having a chip that wouldn't show much of a difference between TIMs.
Months passed and as my queue of waterblocks went down and as I started planning a testbed upgrade (now complete), I still had the 5 mounts of AMD IX ready to go and due for a test. So the wheels started turning to get the TIM review in gear....I bought a 790XTA-UD4, I bought the board blocks for it (two big parts of my new testbed are a focus on silence and the ability to add blocks for secondary components without effecting the CPU subloop), and I started scouting and collecting TIMs. I knew my 965BE was a frigid witch so I bought a 1055T as well--hoping it'd be a hotter chip.
I get to mapping out the temp sensor output of the 1055T and it's pretty obvious it's also a cool running chip. So I run some calorimetric tests on all my CPUs at pretty aggressive settings....and yeah, my AMD chips are very poorly suited for this (though I don't know how the overall population of Intel/AMD chips are for this, just the chips I tested).
n.b., All the following figures are heat absorbed by the CPU block, not total power consumed by the chip! There's a lot of secondary losses with CPU cooling...i.e., a lot of heat dissipates through the board.
My old i7 920 (R.I.P.) did ~190W at 4GHz, 1.5v, my new i7 930 does ~215W at the same settings.....the C2 965BE does ~155W at 3600MHz, 1.6v and the 1055T does ~175W at 3850MHz, 1.6v. Despair set in pretty quickly when I saw those numbers....if I were going to spend 45 days testing TIM, I'd at least want the end results to be meaningful and discernible (additionally, I think the AMD temp sensors aren't as adept as the Intel i7 ones are). I don't think the Thuban would survive 45 days of 24hr loading at 1.6v, either....so the real amount of heat through the TIM would be even lower since I'd back down on the voltage.
Behind closed doors, I was complaining to some friends how upset I was about the upcoming TIM test because these CPUs weren't suited for it. One friend suggested I knock on some digital doors and ask around to see what kind of chips other people had available....there were two people in particular that we thought were a good idea but I didn't know either of them at all. Nevertheless, I contacted one of them and explained my situation with the TIM tests and my current CPUs and asked if there were a way he could help me out.
It didn't go very well, the best option that brewed was a trade for an unknown/unproven C2 965BE (which should be a hot chip, but Lady Luck already denied me once with my current C2 965BE), I was ready to just decline the trade and begrudgingly use my 1055T. As it turned out, my friend, who originally suggested I ask around, went and knocked on virtual door #2 about a day later--without me knowing.
The person behind virtual door #2 was very quick to help, talking to the person behind virtual door #1 (who essentially turned me away when I approached him) and within a day, a known bloody-hot 955BE was on loan to me for the duration of these TIM tests. I haven't received it yet (almost all of this happened over this past weekend), but things are looking up for these tests and the despair is gone and I'm actually looking forward to it now. I'm extraordinarily grateful to the three people who helped put the wheels in motion to arrange for me to have a hot chip--it's been a tremendous help. The chip may not be a 400W scorcher (not that you could keep a 400W chip at functioning temperatures with the watercooling hardware we have today), but if it's the hottest AMD has to offer, then that's a huge aide for my testing.
The worst type of testing to do is the testing you don't want to do. When it's done, you're burned out and you stop...and you sometimes don't re-engage with the testing you DO want to do. At so many points along this timeline, I just didn't want to do a TIM test on an AMD processor--I didn't have all the hardware I needed, I wasn't familiar with the hardware, the hardware I eventually got wasn't suited for this kind of testing, etc. The cloud has all been lifted though; I'm now just a couple days away from hitting the "go" button on the first mount of many and trickling back results and finalizing with a big review :)
I realize this is an extraordinarily long answer to a simple question....but sometimes the whole backstory can lay a good foundation for the stories ahead. The long and the short of it all is that the frustration and despair of getting this testing together has worked itself out over time and I'm really looking forward to this testing now. It would have been easier/cheaper/faster to just buy >5 mounts of IX LGA1366 for a second test, but I gave my word months ago that I would do an AMD IX test and have put a lot into getting it together and making sure it's done right and that the results will be useful.
All that's really ahead now is the 45 days of testing :cool:
In a PM to me you once mentioned that you weren't sure about the accuracy of AMD CPU's internal sensor. Has anything lifted your concern about that?
You could also test the IX vs a known paste on the AMD setup and then test all the other pastes on the i7 930; I think a lot of people would be happy if you were to do the testing on an i7 - I personally would be more than happy to donate the money for the i7 IX kits if that's all that's keeping you from testing on the i7 :) Seems like the least one can do considering all the awesome testing you've performed and are still performing...
As far as the Liquid Metal stuff altering the block/heatspreaders surface, I think this can be shown to be the case/not be the case by re-testing a paste that you tested before the Liquid Metal stuff.
Yes and no. I can always 'remap' the output of the thermal sensor such that a 1C increase in water temps = 1C increase in CPU temps. If I measure a large enough range (to cover the entire spectrum of TIM performance), I'll know that when a thermal sensor says 31.3c, it really means 34.2c (or whatever). You pick one arbitrary temp sensor value to equal one arbitrary meaning and you collect enough corollary data and you have a full spectrum calibration.Quote:
Originally Posted by Nickel020
It's a pain to do and a pain to convince the average reader to trust what's done, that's for sure :eh: :(
(although it is important to note that this process is still done as a validation step with every other processor used for testing, just there's no correction factor needed for an i7)
As for precision and repeatability, those are other variables that show their true nature over time. I don't think repeatability will be an issue with AMD processors and I don't think precision is any worse than an Intel processor (resolution of the temp sensors is actually higher.... 1/8th of a degree on a Deneb vs. 1/4th of a degree on a Bloomfield).
Now that is an A+ idea....! I don't mind buying the i7 IX (I've already bought a 1055T, 790FXTA-UD4, EK GA AMD kit, etc for the AMD testing....nevermind the dozen or so other TIMs I have here), but I do like the idea of testing some on the AMD CPU (IX, MX-2, IceFusion me thinks) and then testing all of them back on the i7, where I have homefield advantage :DQuote:
Originally Posted by Nickel020
I think I'm going to run with this, thank you! :bows:
Also a great idea....though my hope is that the nickel plating on the CPU-360 alleviates any diffusion issues with the Liquid Ultra and MetalPad. The full truth of the matter is that if either CLU or CLMP do alter the surface, then it'll show as soon as mount #2 on the respective TIMs. And if I clean MX-2 off between mounts, it's only fair I clean CLU/CLMP off--but if that cleaning requires surface alteration, I guess that's what I can investgate :)Quote:
Originally Posted by Nickel020
Vapor: Hats off, thanks for sharing interesting background story. And good luck with your tests. There is big crowd that feeds by your tests/reviews in this world where vendors publish lies and not much independant tests made with clue exists. :)
BTW, maybe display of data can be presented in more minimalistic fashion? My idea being - usually people care only about max performance levels, but at idle load cooling temps should be good enough even for worst performing TIM pastes, so no need to display irrelevant data. I never understand, why idle temps are presented at all in most reviews :/. It's just doubling displayed data on charts/tables with numbers most probably don't care about.
No need for you to purchase additional kits mate. Just ask and we'll send you some more :up:Quote:
Originally Posted by Vapor
Your time and effort is always appreciated
You're welcome, I'm glad to have contributed something :)
As for the Liquid Metal stuff - a polished surface like that of the Koolance might withstand it, but the nickel plating of CPUs seems to be different from the nickel plating we know form heatsinks/water blocks (my AMD CPU I tested the Liquid Ultra on is full of blemishes, the Thermalright CPU heatsink only has one small spot). By the time you get to the Liquid Metal stuff we'll probably have enough results from people using it to know how it behaves when used for a short time.
@churchy
Vapor didn't show any idle temps anyway. And they're not only irrelevant because there's little difference there, but also irrelevant since the thermal sensors aren't anywhere close to accurate there. Most people (including many "professional" reviewers) simply don't know enough about cooling in general, minimizing measuring error and CPU temp probes to recognize/produce useful reviews though.
Nickel: partially i meant also results for lesser overclock or @stock speed, not just idle in plain meaning. Bigger heatload - bigger deltas, easier to see performance differences between TIM pastes, and imho one (max o/c) set of data enough.
Yeah, but the more important thing is that the sensors are probably way off at such low temps and there's no point measuring if you know your probes are very much inaccurate. And if you know something about cooling & CPUs you simply won't care about idle temps anyway...
I've never been able to figure out if you're Enerdyne or not :p: I'm considering the offer, but I am very hesitant to accept because I am way overdue on the AM2/AM2+/AM3 kit testing :shrug:Quote:
Originally Posted by masska
Anyway....running temp sensor validation my i7 930 right now and, lo and behold, at high temperatures (about 10C higher than I ever got with my 920), there's a different response curve :rolleyes: Looks like I'm spending the next few days remapping the 930 :p:
I'll be lucky if I'm not doing TIM tests for the next 90 days, but I decided I want to leave as few unturned stones as possible. So here's the current list of what's going to be tested....
AM3:
MX-2, AS5, Indigo Xtreme, and potentially IceFusion (I have questions about its stability/consistency based on prelim tests). This test will be released first.
Then the long haul will set in on LGA1366, where I will be testing, in batches, the following TIMs:
AS5, AS Matrix, OCZ Freeze, Shin-Etsu X23-7783D, Shin-Etsu G751, Zalman STG1, Zalman STG2, MX-2, MX-3, Gelid GC-Extreme, Tuniq TX-3, ThermalRight ChillFactor 3, Feser H-Bridge, Scythe Thermal Elixer, Coolaboratory MetalPad, Coolaboratory Liquid Ultra, IC Diamond (yes, again...I'm giving it one more shot), Gelid GC-2 (potentially), CM IceFusion (potentially), Indigo Xtreme (hopefully)
Looking at between 20-24 tests there :eek:
I kinda can't wait to see all the data roll in :cool: