Stren's R9-290/290x Water Block Testing

[stren]

You guys may have done this already but I wonder.... Where there are thick pads in use, would there be any mileage in copper shimming the gap in order to have thin pads?

I don't see why not though you'd have to be careful and a thinner pad would always be better as you'd still have to have TIM between the shim and the blokc.

Thick pads are used for VRMs I think because they are not always exactly the same height. So you need the thick pads to accomodate for those VRMs that end up a little higher or lower than average. Also, I think you're not really limited by VRM or RAM temps, so you wouldn't notice difference even if you managed better heat transfer. Someone with more recent video card overclocking experience correct me on that if I'm wrong

This is true - some manufacturers have said this is why they do it, and we've seen problems in the past with certain boards when block manufacturers haven't taken this into account. E.G. 7970 Lightning. It depends what you want really, safety vs cooling. At this point it would be rare for any full cover waterblock to throttle the card based on temps so if that's your metric then any is acceptable I would say. Personally I like "the best" which to me means cooling everything as well as can be in the hope that it gives longer lives to the components etc etc

Awesome work. With GPU being such a close match, I'd personally put more emphasis on the VRM results.

AC has my vote for both vrm performance and looks but I'd have to sand down that palm tree graphic. Not sure about cost though, I'm cheap too and would probably weigh cost heavily as well..perhaps more.

Thanks Martin - yes I would agree with you. The titan core results were close already and I think these will be even closer! Yes the AC active backplate is really cool and performs well but it certainly isn't cheap.

[Justintoxicated]

Is the aquacomputer backplate even available in the US yet? I just had the EK waterblock in my cart, it matches the rest of my system and is affordable even with the backplate.

The one thing I do not like about aquacomputer is acrylic, will there be a non acrylic version?

I can't decide between the EK and the Aquacomputer but leaning towards the EK due to price, and the fact that aquacomputer only has an acrylic option

Help me decide Stren!

Oh and thanks for the amazing reviews you keep pumping out.

edit: Oh I don't think my GPU is much of an overclocker either, anything over stock speeds has resulted in screen going crazy or blue screens when I tried to mine a litecoin. Even when setting temp to 70's and allowing fan to spin up full speed. I really need to get it under water though, it sounds like a hair dryer. So I have to wonder if the AC backplate and block are worth the additional ~$55.

Would I be crazy for going EK copper/ acytal? The koolance looks good too, would the EK backplate fit on it? That might be kind of goofy haha.

[stren]

Is the aquacomputer backplate even available in the US yet? I just had the EK waterblock in my cart, it matches the rest of my system and is affordable even with the backplate.

The one thing I do not like about aquacomputer is acrylic, will there be a non acrylic version?

There is normally one - though I don't see it on the website for the 290x. So maybe not this time - maybe Shoggy can answer?

I can't decide between the EK and the Aquacomputer but leaning towards the EK due to price, and the fact that aquacomputer only has an acrylic option

Help me decide Stren!

Oh and thanks for the amazing reviews you keep pumping out.

edit: Oh I don't think my GPU is much of an overclocker either, anything over stock speeds has resulted in screen going crazy or blue screens when I tried to mine a litecoin. Even when setting temp to 70's and allowing fan to spin up full speed. I really need to get it under water though, it sounds like a hair dryer. So I have to wonder if the AC backplate and block are worth the additional ~$55.

Would I be crazy for going EK copper/ acytal? The koolance looks good too, would the EK backplate fit on it? That might be kind of goofy haha.

[/QUOTE]

Well it's always hard to define whether something is worth it to someone. The difference in core temps between the two is negligible. The VRM difference is real, but VRMs are designed with heat in mind so running them hot shouldn't mean a card that degrades quickly. However cooler is always better. I would avoid mixing and matching backplates simply because the screws may not be the same thread and the EK backplate expects a screw to come in from the block side where the Koolance top would block that. Kinda surprised Koolance haven't done a full backplate yet.

On a side note I stopped by Swiftech's suite at CES and Gabe confirmed they will take part in the roundup but they are not in production yet. Photos to come later. XSPC's full cover block, backplate and universal block arrived last night, I'm just going to re run the Aquacomputer block with the backplate on but without the active heatpipe attached first.

Watercool agreed to send the gpu-x3 core too. So for universal blocks we'll have at least three including the EK supremacy too.

[GliderHR]

Hi Stren,
I appreciate your reviews very much. Thank you for the work!
One question : Do you have the "standard" cooling temperatures for air cooling? The comparison between "standard", sophisticated air cooling and one block water-cooling would be nice.
Greetings from Leverkusen

[Justintoxicated]

Hi Stren,
I appreciate your reviews very much. Thank you for the work!
One question : Do you have the "standard" cooling temperatures for air cooling? The comparison between "standard", sophisticated air cooling and one block water-cooling would be nice.
Greetings from Leverkusen

It's going to be a night and day difference I can tell you that, and the graphs would have to be zoomed way out! Not to mention that at full tilt or even 75% the 290x fan sounds like a hair dryer.
I have mine mining litecoins as I write, set to maintain 80c core (currently at 77c with fan at 89%), and my VRMs are at 60 and 65c.

[stren]

Hi Stren,
I appreciate your reviews very much. Thank you for the work!
One question : Do you have the "standard" cooling temperatures for air cooling? The comparison between "standard", sophisticated air cooling and one block water-cooling would be nice.
Greetings from Leverkusen

It's going to be a night and day difference I can tell you that, and the graphs would have to be zoomed way out! Not to mention that at full tilt or even 75% the 290x fan sounds like a hair dryer.
I have mine mining litecoins as I write, set to maintain 80c core (currently at 77c with fan at 89%), and my VRMs are at 60 and 65c.

Yeah I was hoping one of the after market air coolers might send a sample, but no one yet. Justin has it right under any real load the fan has to spool up so much it sounds super nasty. Part of the problem with air cooling is knowing where to set the fan, I guess the default fan profile is the way to go on this card and then maybe the max setting also.

These are not cards you want to keep with the stock cooler that's for sure. However for the price of a 780 you can get a 290 + waterblock (assuming those miners aren't kicking the price up too much) and overclocked performance is comparable to an overclocked ref 780.

[GliderHR]

Thanks for the information. I thought so, but to get the feedback is good. Here you/we can show the real advantage (cooling power) of water cooling (which I think is not to discuss here, we have the same opinion).
I made the same experience with GTX cards (480/680/Titan) and I will not miss the fullcover cooler(s).

[stren]

So this is interesting - I retested the AquaComputer backplate, but removed the heatpipe and used the normal bridge, so the backplate was now just a "normal" style backplate and it still performs excellently. I now suspect this is because it has thermal pads in an area that no one else does and *not* because of the heatpipe as previously thought (though that still has a small effect - I'll take a photo of the thermal pad layout once I remove the backplate. For now here's the data:

[Justintoxicated]

So this is interesting - I retested the AquaComputer backplate, but removed the heatpipe and used the normal bridge, so the backplate was now just a "normal" style backplate and it still performs excellently. I now suspect this is because it has thermal pads in an area that no one else does not because of the heatpipe as previously thought (though that still has a small effect - I'll take a photo of the thermal pad layout once I remove the backplate. For now here's the data:

Well that's pretty cool, the non active backplate is cheaper too assuming you can buy one in the US someday soon. What happens if you put some thermal pads under one of the other backplates for the EK block in the same location? Which location is this anyways?

[Jakusonfire]

Now that is interesting! Glad you tested this, as I just could not understand how a small diameter, single way water pipe could have that much effect. I would expect some water flow in and out of the back plate section with passive circulation but it could surely only be a fraction of a proper two way in out flow design.

[Church]

There simply isn't much of heat dissipated on that side to begin with. Imho it might also enhance cooling by more mount pressure / evened out contact, with enhancing card's rigidity with backplate.
LOL, i even thought of new product: Universal Backplate . One with lot of prolonged screw holes or simply with big bunch of them for most of gpu families.

[stren]

There simply isn't much of heat dissipated on that side to begin with. Imho it might also enhance cooling by more mount pressure / evened out contact, with enhancing card's rigidity with backplate.
LOL, i even thought of new product: Universal Backplate . One with lot of prolonged screw holes or simply with big bunch of them for most of gpu families.

yeah it could be that it squished the thermal pads down more - here's a shot of the thermal pads, they were thin to begin with but were squidged pretty good:

Well that's pretty cool, the non active backplate is cheaper too assuming you can buy one in the US someday soon. What happens if you put some thermal pads under one of the other backplates for the EK block in the same location? Which location is this anyways?

Yeah i'd like to try this. Here's how I set it up for that run:



After removing the backplate:



Thermal pads had stuck to the board, but if we reconstruct it on to the backplate and compare to the EK (bottom):

Now that is interesting! Glad you tested this, as I just could not understand how a small diameter, single way water pipe could have that much effect. I would expect some water flow in and out of the back plate section with passive circulation but it could surely only be a fraction of a proper two way in out flow design.

Yeah I was surprised too, I was expecting small gains at best.

XSPC is on the testbench now:

[stren]

Well it looks like the card is dead. The weight of the XSPC block slowly pulled it out of the motherboard once (stupidly I was running without the clamp on my tech bench that holds the gpus in place because I've never had a problem before, but I think the extra flex of the impact just meant it had too much freedom). I'll have to RMA or get a new card and redo the testing that I've done already /sigh.

[Justintoxicated]

Oh Man I hate when stuff like this happens. You sure it's not the MB that died?

[stren]

Yeah my 650ti boost is purring along happily in the same board, so this means I've started CPU block testing on 4770K. First on is the supremacy with both orientations, then the others, then the full cover cpu/motherboard impact block from EK and BP.

[Martinm210]

Well it looks like the card is dead. The weight of the XSPC block slowly pulled it out of the motherboard once (stupidly I was running without the clamp on my tech bench that holds the gpus in place because I've never had a problem before, but I think the extra flex of the impact just meant it had too much freedom). I'll have to RMA or get a new card and redo the testing that I've done already /sigh.

First Skinnee and now you. Don't think I'll ever try a FC roundup. Just way too much risk. Hope you get it replaced.

[stren]

First Skinnee and now you. Don't think I'll ever try a FC roundup. Just way too much risk. Hope you get it replaced.

Yeah we'll see, I've had good luck with RMAs recently, but bad luck otherwise (speeding tickets and a bunch of other things breaking). I'm still hoping it will magically fix itself now that the air cooler is on. That titan in the titan block roundup started behaving weirdly until I reflashed the bios. I guess I could try reflashing the bios again in case it's similar, I did save the stock bios just in case.

[Justintoxicated]

Why dosen't my EK backplate drop temps?
Contact seems good.



How's the RMA going Stren?

After removing the sticker around the resistor or caps or whatever my temps while mining on VRM 1 is 30C with a 65 ambient. Not scientific like your testing but there it is

[stren]

Why dosen't my EK backplate drop temps?
Contact seems good.



How's the RMA going Stren?

After removing the sticker around the resistor or caps or whatever my temps while mining on VRM 1 is 30C with a 65 ambient. Not scientific like your testing but there it is

Sapphire are slow slow slow. Last time I buy one of their cards (I hope). First it took over a week of back and forth between their tech support and their outsourced RMA team just to get me a shipping address. I shipped the card early last week but haven't heard anything back yet. Hopefully soon, because I'm moving house in 1.5 weeks...

You mean 30C over ambient or just 30C? I assume the latter - but either way not bad on VRM1 to be honest. My testing was hitting 45C over coolant on the EK backplate, add on 5C for coolant to air delta and an ambient of 25C then the actual temps were about 75C with the EK block.

[Justintoxicated]

I made a mistake VRM1 is 37c (Just 37c) under load, granted my ambients were up to around 75 when I was checking. VRM2 was the lower one it was around 34c at this time.

I'm using Fujiploy extreme pads for everything except the backplate (I applied thermal grease on the graphics card side before applying the pads, but forgot to add grease on the waterblock side (oops). I just re-used the the backplate pads that came on the used block since the backplate isn't critical, just added some non conductive thermal grease to ensure contact as you can see in the picture above.

My card is a Sapphire card as well.

[stren]

I made a mistake VRM1 is 37c (Just 37c) under load, granted my ambients were up to around 75 when I was checking. VRM2 was the lower one it was around 34c at this time.

I'm using Fujiploy extreme pads for everything except the backplate (I applied thermal grease on the graphics card side before applying the pads, but forgot to add grease on the waterblock side (oops). I just re-used the the backplate pads that came on the used block since the backplate isn't critical, just added some non conductive thermal grease to ensure contact as you can see in the picture above.

My card is a Sapphire card as well.

Nice - the RMA place is only up in Pasadena and they confirmed that it shipped today so hopefully it will arrive tomorrow The most annoying thing is just having to retake all the data on the blocks I already ran.

I think I'll start with XSPC with and without backplate, then do EK without backplate as I hadn't done that before, and then back to the others. Hopefully some more blocks will come in soon. Watercool were supposed to ship a while back but are normally slow to do so. I haven't seen anything about BP, Swiftech or Alphacool being ready to ship yet though.

[Ace123]

When you're done with that Aquacomputer and active backplate.... you let me know! lol

[stren]

Been a busy month what with moving house, but while I waited for the new R9-290 to come in I figured I could test some other things including some 4770K testing and a review of the EK Full Cover Maximus VI Impact block review (also includes Supremacy vs CPU-380i as reference points) - full write up here:

http://www.xtremerigs.net/2014/03/03...-block-review/

I now have a 290 again so I can restart testing all over again.

I was thinking to use my shiny new aquaero 6 to log date from flat temperature probes inserted between the VRMs on the front of the board and the block itself rather than purely relying on those internal temperature sensors are not so useful. I thought about this because the AC backplate showed some weird data and I wonder if the temp sensors themselves are not actually that close to the VRMs.

Any thoughts? I feel like worst cause it's just more data to look at and log/analyze.

[stren]

So I added the thermal probe right about here:



zoom---enhance



Then secured it with some liquid tape:



I don't think it will affect any of the performance but should help rule out if there's anything funky going on with the AMD VRM temperature sensor.

Tried to take data yesterday but the clock on the test rig is jumping around and confuses the data logs...

[stren]

More minor setbacks - turned out the clock was jumping because of the temperature probe leaking very very slowly. Luckily the aquaero was logging another coolant probe so I had backup data to use that I can still do apples to apples comparisons with in future.

So then I thought I was done with the XSPC block when it turned out GPU-Z log file got full and it never tells you, just stops logging. So I have to redo the xspc block for the fourth time. No kidding. So that's on the list for tomorrow. In the meantime, I took a look at the CaseLabs S8 case, have a read if you're interested:

http://www.xtremerigs.net/2014/04/04...bs-mercury-s8/