Radical, that's the best view of rating waterblocks I've ever seen. My hat is off to you.
Is there any news to report yet?
Printable View
Radical, that's the best view of rating waterblocks I've ever seen. My hat is off to you.
Is there any news to report yet?
Thanks ;) That's something that came with the experience of testing blocks for the last few years. I absolutely lost the illusion that one person, or one institution, could judge which is absolutely best.
Also, what's the consumer searching for? As a consumer, you see a bunch of shiny blocks. Each advertised with the shiniest words marketing has to offer. Now what? Believe what they say? Then every block would be the best thing since sliced bread.
So what I think, the consumer wants information. Good and valuable information, some smart people do even read the text that comes with a review ;) Like with my last review, water blocks on a 775 quad core, I had 5 blocks that got an award. Yes, five, out of 16 blocks tested. All good blocks, all with their own strenghts and weaknesses. It's not up to me to judge which is the absolutely best block, it doesn' even necessarily mean that I'd use the block with the lowest temperatures for my own rig.
As long as a review can give a rough guideline, some valuable information, some basics that people can use for their own decision, I'm happy ;)
To me it seems that the biggest problem comes with consumers and/or companies that just want the #1 block, the best there is. To me, there's too many options, too many variations out there right now to really make that happen.
Any news to report yet?
I'm rather interested, this looked like it was gonna be a rather good block, but I know the ambient issue had to be taken into effect as well...
With the weather coming up out of the 60's now are we going to see some more testing with this block?
Yes Rick, just got back from vaction this week, I plan to resume next week with the conclusion of the DI and then on to the EK Supreme.Quote:
Originally Posted by RickCain
:) Hope all goes well and we get some good weather. I am ready for summer already.
Woohoo, now can you just get the actors and writers back to work so I can have my '24'? :D
Any luck getting those results into a graph? Can't wait to see your findings.
I have one more run to do - with any luck it will be today.
I've been waiting for you to update this with results. This isn't my style of block but it looks very nice.
Sorry to keep bringing this back to the top but how is the last run going?
u guys also need to be aware that i think scott is using the RD-30 on this test bed.
Unless you have a 2 x DDC-2, or possibly 3 x DDC-3.2 theres no chance you'll ever meet his headpressure. And if this block is like the EK Supreme, it loves head pressure.
Yes RD-30 at 18v which is MUCH like an 18w DDC-2 with a custom top... anyhow, testing of the AC DI is DONE! Martin has all my results so far so he can make me a pretty graph... the EK block is next, but it don't fit with the stock mount bracket... gotta get the dremmel out.
Without going into details, how did the AC DI fair in your comparison tests? Generally speaking of course.Quote:
Originally Posted by nikhsub1
Off the top of my head, I think it performed the best so far of FLAT blocks (non bowed, non stepped).
Thanks for the information. Looking forward to your EK Supreme results as well.
The weather should be to your liking as it is supposed to be in the high 60's/low 70's later this week.
I'm out of town so this old excel version will have to do until I get home.Quote:
Originally Posted by nikhsub1
First showing a line graph which displays all 5 mounts! yes that's all FIVE MOUNTS, great work!!
http://img137.imageshack.us/img137/8759/niksub1mh5.png
And averaging all 5 mounts bar style:
http://img138.imageshack.us/img138/977/niksub2le5.png
:up:
Yeah, it's awesome. Three mounts would get you close, but it looks like 5 really ties it down right.
I was also suprised to see the double impact up there a little ways. We'll see how the EK does, I'll have a try at that one too pretty soon.
I'm going to make a new thread... but it is with a 6600 dual core, single die. And Martin, YOU ROCK! One day when you have time maybe you can tutor me on how to make those graphs!Quote:
Originally Posted by mcoffey
It's too bad you don't have all the time in the world so you could do say 10-20 mounts; some of the variances are interesting. Five mounts definitely gives a good idea though...
e: on a side note, has anyone seen 45nm parts without a lid yet?
I wonder how much the different core shape/size/heat density pattern might change these results.
I also am interested in how the newer CPUs' heat topography will change the performance of waterblocks. Specifically, will a native quad-core Nehalem bring back a use for the Storm since it will be all one die with a more concentrated heat source? I guess we can only speculate now, but I would love to hear peoples' thoughts.
@iandh: If you mount a newer block 20 times, the performance would suffer pretty badly before you're done.
@headala: It's different with every die size and every IHS shape. Sometimes it makes a small difference, sometimes it shuffles the results completely. At least from what I've seen so far. Plus, the blocks aren't all the same quality. Sometimes you get a better match for the given IHS, sometimes it's worse. It doesn't always, but many times make a significant difference.
lolz welcome to the club, if your talking about the Supreme that is. I had a problem with a mosfet getting in the way, but nothing a dremmel wont fix ;) I see Eks has a 755 mounting plate available for it now :up:Quote:
Originally Posted by nikhsub1
Any ways I cant wait to see what the RD 30 will poke through this block :eek: It will be a good test. My D5-B will do 3.5LPM
I think 3 mounts is plenty, but 5 really ties it down. Anything more wouldn't be necessary, and this type of testing is already extremely time consuming.
Even the larger spread on the Fuzion 4.5mm nozzle only has a standard deviation of only .78C, so with a 1.85C spread to the GTX mean, you statistically have a clear winner...no need for more mounts.
I just wish the review sites would do this...:rolleyes:
What about putting the average delta with a spread like 23C ± 0.6C so if a block have a larger variation, the spread would be larger. This would give a better idea if block X perform better than block Y and if mounting is harder to be consistent with a certain block.
@martin: Well, as you said, it's very time consuming. If you ever had to do a review on a tight schedule, you'd see how much you can do in a given time period. That's one point.
The other point is that many manufacturers or shops are really late with their samples, so you can't put up a test that needs like at least a week for each block or part.
My last review had something like 16 blocks competing. All were measured at 2 flow rates and 2 CPUs. That's 64 measurements. Now if you say that each block should maybe should have the time to "sit" for a day, maybe to get the thermal paste into shape, and do that 5 times with each block? That's simply not working.
People do also want to have a test result when the stuff hits the market, they don't want to know that some months or half a year later. By then, there are new CPUs, new whatevers and they'd want to know results of this new combination, not the old one ;)
Definately, I was referring to the few reviews I've read that compare core temps and shrug off ambient(they looked at a wall thermometer and said,,,eh close enough) and multiple mounting. There are many good reviews, I just havn't seen anyone go to this level with 5 mounts before and want to applaud these efforts.
Some of the reviews I've read could have had a 2 degree mount error, 1-2 degrees of ambient temperature error. So their results of 1-2 degree difference in their conclusion wasn't very conclusive with the amount of error left in place.
More than anything because the difference between blocks is getting so small, it's just becoming more and more important to try and tie these things down.
Unfortunately that just make the workload extreme. I've made several runs, but I havn't done nearly the amount of testing you or Niksub1 have done. And before I do much more myself, I decided logging of temperatures will help ease the pain a little.
Hopefully great examples like this will encourage other review sites to do the same or take a step in the right direction.
That's a good idea, maybe the mean +- 1 standard deviation.:up:Quote:
Originally Posted by Xilikon
I wish that with the new logging of temps you got, it might be more bearable with only 20-30 mins of setup each day (unmount, clean and remount) with 1 hour at first for setting the test bench.
For a good testing, expect 1 week per block at least.
That's why there is this graph... Thanks to Martin of course! Shows you the deviation of each mount...Quote:
Originally Posted by Xilikon
http://www.anonforums.com/builds/tes...hsub1mount.png
@martin: Yes, you're absolutely right. I really like the level of this review here, it's an adorable result and yes, the effort definitely is worth an applause :)
People that put up articles that don't even use anything like a temperature probe, as you say a wall thermometer, well... I don't call that a review ;) You'd normally expect more from a consumer that would post some kind of result in a forum, at least if you want to find the information usefu at all.
I also like the reviews were someone does multiple different flow measurements so you can see how the block behaves with more and less flow. Sadly, that adds even more work to the whole thing.
So many people rush the results and want to give the masses some numbers just to appease them. I want to applaud nikhsub1 again for taking the time necessary to do the job right when so many were after him for some kind of data. :clap: :clap: :clap: :clap:
This one is for you Scott...:toast:
I concur :up:Quote:
Originally Posted by Radical_53
but seriously, fantastic job nik
Absolutely, thank you so much for taking the time/expense to do such a thorough job. And, thanks for putting up with us :yepp:Quote:
Originally Posted by SiGfever
Next time I'm in SoCal, I'll buy you a :toast:.
Guys not to throw a spanner in the works but are you keeping mounting pressure constant during these tests?
Intel recommend 17 thru 72LBS of pressure force for core2 and as we know applying 72 over 17 can have a massive impact on how well a block performs.
Im testing the new OCZ block here which is in reliability testing and I have the ability to set 55lbs, 75lbs and 100lbs pressure, I compare with the same mounting hardware each time on each block so I keep apples to apples.
I think you need to ask all the companies what pressure their mounting kits are applying and make sure all are at least within Intels spec's and second you report what applies more etc.
Yes Tony. I use the SAME mount kit for each block... I don't want the mount kit to be the deciding factor in performance, the block is. My mount pressure is roughly 50lbs or so. D-Tek sent me a kit which is what I am using for all blocks, in the springs are limiters so as soon as I feel the limiter, I stop tightening. This is also part of the reason why I do 5 mounts of each block.Quote:
Originally Posted by Tony
But then you still have different mounting plate heights that should give you different mounting pressures.
I use the same mounting hardware for each block too, but I measure the length of the spring with a caliper to apply the same force to each blocks plate.
I fail to see how this would make one bit of difference. If the springs are compressed the same amount on each block what matter does it make the height of the mount plate?Quote:
Originally Posted by Radical_53
I agree, the only thing hieght would do if you used the same backplate and springs/nuts is cause some different variabilities in mount variations. The lower the mounting plate the more likely you will be centered over the IHS correctly and also less likely you have torsion on the block where one side has more pressure than the other.
Good lord......
Lets discuss how well the block performed versus how spring limiters can change the results. :shakes:
3.6C going from 55lbs to 100lbs, 2c going from 55 to 75, it makes a HUGE difference.Quote:
Originally Posted by RickCain
You can buy plasi-guage from the local auto parts store, it is used to measure bearing gap when you rebuild engines.
It will give you a good idea if your mounting pressure between blocks is consistent.
Excellent work nikhsub1!
I've just read through the whole thread (only skimmed before) and it is very interesting to note all the conjecture over the different testing methods and variables at hand. Needless to say, I think you've done an admirable job. The results speak for themselves but as noted throughout, they must be taken in context.
I'd never fully appreciated the significance of numerous mounts until noting the deviations in your graphs.
On a different note, I was considering stocking some EnzoTech blocks, but looking at the results I think I'd only stock limited quantities for those that like the looks. ;) Ironically, the mounting consistency of the SCW-1 almost makes you think there was something wrong with the mounts when you look at the deviation of the others ;) But I gather this might also be related to the consistency of the block's heat transfer properties across the entire mounting surface? (Speculating with some ignorance)
Thanks for all your efforts!
Cheers
GAM
@nikhsub: I might have understood you wrong. If the limiter is above the mounting plate, indeed that would have the same result as measuring the spring length. My bad.
I had a picture of a limiting mechanism in my mind where the "limit" was given by the bolts, so different heights of the mounting plate meant different mounting pressures.
Any news on the EK Supr????
Don't mean to push, just curious. .
GAM, I was floored by the mount consistency of the Enzo block as well...Quote:
Originally Posted by GAM
No worries, I was pretty sure you misunderstood what I meant, and yes, the limiters go in the springs which are above the mount plate.Quote:
Originally Posted by Radical_53
None yet, gonna hopefully have the Supreme finished next week.Quote:
Originally Posted by LogAn'sRun
Tony no doubt about it... I can assure you there is no where near that difference in any of my mounts. If there was more than 1lb from mount to mount I'd be surprised. Again, this is part of the reason 5 mounts are performed. My methods are certainly not perfect but I do the best that I can.Quote:
Originally Posted by Tony
Has it been a week yet???Quote:
Originally Posted by nikhsub1
:D
Well there is a problem :/ The Supreme is performing really bad on my test setup... like the worst of any block yet. Perhaps i need to dremel the mount plate some more.... uggggggghhhQuote:
Originally Posted by LogAn'sRun
Hi people
I have installed Aqua Computer Double Impact block on my quad core on thermal paste Zalman ZM-STG1 and Zalman ZM-GWB8800 GTX in loop for my GF8800GTX with flow 1,5 liters per minute
Normally I run on 3,6GHz on 1,36V in bios, water temperature 23 Celsius
Screen with idle
http://images33.fotosik.pl/164/232ea43980ae82c7.jpg
And stress
http://images32.fotosik.pl/164/517ff3ee030644c4.jpg
My best result is 4140MHz with water temp 10 Celsius ( window was open ) with 1.65V inn bios
Idle
http://images28.fotosik.pl/169/396d03007a08c8d1.jpg
And stress
http://images28.fotosik.pl/169/b20f7fb8b9a0edd3.jpg
Good Job
Awesome q6600.
Are you using a chill ? With my qx9650 at 4200mhz with vcore = 1.45V, cpu is at 72°C on 8K and 10K iterations of prim95 (room temperature between 25 and 30°C I think).
Your temps are similar to mine. Q6600 @ 3.8GHz, 1.5 vcore....Quote:
Originally Posted by Repliquant67
42c stressed at 3600.. I think those temps are innacurate and not adjusted..