wow! certainly we've come very far in systems performance vs. power/noise ratio!
Thanks for the great article Lawrence!

P.S.: did yo have the chance to try OC'ing the CPU?

I wonder how different it would be replacing the stock case fans with 3x Noiseblocker Multiframe.
Also I don't understand why one GPU is so much hotter than the other (I presume GPU1 is the one upper on the motherboard).

Because fresh air from the right case fan is stealed/blocked by the CPU-cooler?

It would've been really interesting to see the results with a) a dual tower CPU cooler (this is enthusiast air after all, $50 isn't breaking the bank at $1700) and b) had you guys tested a card besides the Zotac.

The Zotac is an outstanding card by all measures and reviews, but, besides it's high minimum fan speed, it's heat pipes run up and down the length of the card, making it less than optimal for the 90* orientation of the chassis. It would've been interesting to see if a card whose heat pipe orientation was better suited to the FT05 would perform in SLI (ASUS, MSI).

Anyway, thanks very much for this! I very literally just picked up a MSI 970 Gold Edition for my FT02 and I'll be picking up another when I upgrade my monitor - this is exactly the review I've been looking for in my shopping to convince me I won't have to worry about moving to water.

Edit: Also, a dual-tower CPU cooler would've decreased case temperatures by some margin. This is particularly important in the FT02/FT05 due to the airflow design of the case - non reference cards, although much quieter, dump hot air into the case that, according to user's experiences and a great review by expreview that I can't find any longer, increase overall temperatures in comparison to reference cards by a reasonable margin (especially in SLI). A dual-tower CPU cooler would've helped tame overall heat to be exhausted from the case and might've decreased GPU temps overall.

Indeed.

I'm disappointed that this wasn't investigated further (or at all...).

With this great discrepancy in temps, I would almost suspect a faulty card/cooler. Not only is the card 20c hotter, the fan is also spinning faster.

The "top" GPU - in this case the one on the right - should get worse airflow. It doesn't get any from the right bottom fan and at least 80% of the left bottom fan's airflow is blocked by the other GPU. To make it worse, in my experience, a fan's airflow drops significantly towards the edges.


Nonsense. How would an additional fan in the middle of the case help exhaust air?

Lawrence may as well have attached a second fan to the Scythe Kotetsu. Or installed a dedicated exhaust fan.

But why? All the air that the two 180mm intake fans suck into the fan must be exhausted anyway. Otherwise the case would balloon up and burst.

I've noticed newer nvidia drivers designating the second card as the primary card - maybe that's the case here?

Meh, believe it if you want to or not - many have tested and shown that non-reference GPUs increase CPU temps in the FT/RV series. More CPU headroom might decrease overall temps, another fan on the Kotetsu would've helped, SLI is probably the one situation where GPU TDP plays a role in CPU temperatures, et cetera.

Would have done this build very differently - open air test bench case like a Lian Li T60 and use Raijintek Morpheus' on each GPU with 2 quiet fans blow down the heatsinks. Fans could run at 600rpm which would be well below 20dB.

Non-reference GPUs increase CPU temps in just about ANY closed case.




How?




GPU TDP will always play a small role in CPU temperatures, as hotter GPUs increase case temps.

You are not making any sense.

Could this claim be tested by rotating the case 90 degrees such that it's a 'standard' front2back airflow case?


As the fan speeds are driven bytemperature the faster spinning fans aren't a surprise are they?

Take a serious look at the table on http://www.silentpcreview.com/article1443-page7.html

there are only 4 HS on the list above the kotetsu and all are way more expensive.

NZXT Kraken X61 ($100 and up) (not a dual tower design)

Prolimatech Genesis ($90 and up)

Thermalright Silver Arrow ($100 and up)

Noctua NH-C14 ($65 and up) but not a dual tower design


So of the 4 higher ranked heatsinks only 2 are dual tower and both are over twice the cost of the kotetsu.

So tell us which dual tower heat sink that is $50 do you want them to test? Is it already on that table? If not did it come out within the last few months and is somehow vastly superior to other heatsinks on the list?

Or put another way, they've already tested the vast majority of heatsinks worth testing. You probably can't find a $50 heatsink that beats the kotetsu*.

*This is SPCR, all heat sinks tested here are ranked at low noise levels. You can slap two fans on a HS to move more air but there is a limit of how much air you can move for 22 [email protected] and fan combinations above that threshold won't make the cut.

This has been done and it showed that it is indeed not the vertical airflow that's the key, but the two (or three in the FT02) giant 180mm fans. And then there's the small gap between the intake fans and the GPUs. In regular ATX cases the intake fans are not only further away from the cards, but there's usually impeding HDD trays in between.

It has been pointed out on SPCR before that once fans come into play, convection doesn't matter any more. (on the scale of computer cases...)


Sure.
But faster spinning fans also provide better cooling. So not only is GPU1 20c hotter, the thermal solution is also working harder to "achieve" that.

Easy their pal, no need to be unfriendly...I know - that's what I said

Think of the the heatsinks as huge radiators emitting heat - if the total heat of the heatsink is lower, the case temperature will be lower...

Yes... again, exactly?

I dunno you seem to understand perfectly and appear to be mean for no reason

Whoaa, settle down buddy! Please re-read my post - spending another $50 on a dual-tower heatsink (as opposed to ~$50 for the Kotetsu), bringing the total to $1700, was what I said.

Indeed, but where the D15 really pulls away is under extremely high TDPs, the kind that spcr avoids and overclockers push for. In those environments the D15 and other dual-tower coolers tend to pull far away from monolith towers.

Before you get all excited, of course having a bigger tower wouldn't decrease the temps of the GPUs directly (or change the TDP of the CPU, or magically absorb GPU heat output, et cetera), but it would decrease the temperature of the heatsink itself, which would decrease overall chassis temperature.

Anyway, sorry to freak you guys out, talk to you soon!

Using different gpu cooling than stock would improve results greatly. Just get cheapest reference 970 and buy proper cooler.
Then you can use quality 120/140mm fans at very low rpm (500-600 for example), and even achieve better gpu temps.
Going for 1 stronger card is another alternative, if it's needed.

Unfortunately the MK-26, Morpheus, and similar wouldn't fit on any motherboard currently available (at least not in SLI). All of these coolers require three slots without fans and a fourth with fans.

The only reason to go SLI with cards of this caliber is if you're a) at 4k, b) at 3440x1440, c) a very serious but non-professional video editor, d) benching/epoints. Taming SLI heat is notoriously difficult on air - even with high-end aircooling (that fits) it's not unusual for the first card to throttle. I know this is a silent enthusiast forum, but SLI is always a compromise, even if you don't care about noise.

Starting to think I'm the only one in here making any sense...

No, you said "non-reference GPUs increase CPU temps in the FT/RV series".


Heatsinks don't emit heat.
CPUs do. The heatsink just passes it along.
If the CPU consumes 100 watts of power, those 100w will be released into the system as heat. 8th grade physics, law of conservation of energy.
How hot the heatsink gets in the process doesn't matter for the thermal energy being released.

But you actually said "More CPU headroom might decrease overall temps".
What headroom?


No, not exactly.

You said "SLI is probably the one situation where GPU TDP plays a role in CPU temperatures, et cetera."


You don't even seem to understand yourself what you're saying.


No. See above.

Yeesh, kids on the internet these days. Fine, you are correct - good job? But wait! The FT/RV series is uniquely affected by this circumstance, as has been demonstrated over and over in reviews and owner's threads across the internet and is specifically documented by Silverstone in their FAQ and component recommendations in their manuals. You can ask expreview about their 30 page article on it if you'd like.


Sure, and if a the heatsink is physically cooler than there's less heat for the fans to cool... I'm starting to wonder if your eighth grade physics degree is legit here... if heatsinks don't emit heat, how do they work? Wouldn't they melt or blow up from all of the CPU heat?


My bad, sorry dude! I meant thermal "headroom" from having a larger heatsink. As many FLIR tests have shown (and your acute patronization ), larger heatsinks emit less heat into a chassis than smaller heatsinks (by nature of having a larger area with which to absorb heat).

You'll notice that I clarified later that a larger heatsink doesn't decrease the TDP of a CPU, and that I had clarified that, as OC reviews have shown, that dual tower heatsinks, like the D15, handle higher TDPs, that SPCR doesn't test, better than single tower heatsinks, like the Kotetsu, given the same conditions (same system, same fans/speed, etc). As demonstrated by those tests, the actual temperature of the *physical heatsink* will be lower (because the CPU temperature is lower thanks to the greater efficiency of dual-towers, which, call me crazy, probably has to do with volume in which to absorb CPU heat..), which, when sandwiched mm's away from a GPU, might decrease the temperature of the GPU or even, and I might be reaching here ( ), chassis temps...



K anal internet guy. In any case, allow me to clarify, and all apologies if I offend you along the way: the effect of a single GPU on CPU temps is demonstrably significantly less than that of two GPUs. Is that better?



Meh, get mad about it? At least I'm not a prick to strangers on the internet! And I can exercise patience in humility when engaging in argument - perks of being a reasonable adult! Audi, 5000

Edit: Readability!

Got any links?

All *I* have are those:

http://www.overclock.net/t/408195/the-o ... wners-club
http://hardforum.com/showthread.php?t=1429251

Go ahead, count the number of users with non-reference cards that work marvellously!


Link?

All I know is the recommendation against a certain kind of heatpipe-orientation. Seeing as one of the two GPUs in this review stayed at 65°c, I'd reckon it's not the problem here...


Never heard of expreview. Doesn't even show up in a Google search.


You're making an absolute fool of yourself.


Larger heatsink are cooler, because they have more surface area to spread the heat. They can be cooled more efficiently, because larger fans are attached to them.

A 100w CPU will still emit 100w of heat energy into the case!

If you cool the CPU down with a more efficient cooling solution, you are NOT lowering the energy output. You are just better at transferring the heat away from the CPU.

The airflow around the CPU area will definitely impact GPU temps. If you have a large, passive heatsink, hot air is not being evacuated swiftly and GPU/CPU temps will suffer. This is, however, a question of fans and airflow design, not of heatsink size.


It's not my fault that you are
a) wrong
b) incapable of understanding why
c) incapable of logic discourse


Better. But that wasn't the point you were wrong about.

You said SLI is the only situation where GPU TDP has an effect and CPU temps. It's not. Depending on the case and cooling setup, even one midrange GPU can significantly drive up CPU temps.


No, you're just obtuse to strangers on the internet.

Hmm, I wonder what results would a pair of 290X + Arctic Xtreme IV custom cards like my single would get. Then again, I wonder what case and motherboard they would fit with a slot between them.

You better re-read what I wrote. Mount the fans perpendicular (parallel to mobo) to the heatsinks. A pair of fans can cool both heatsinks. Fit all current mobos that can hold 2 cards and have the right gap.

I've seen it done and it's the ultimate way to air cool for SLI.

*Yawn* I've been a member of both of those threads for years, you're putting words in my mouth (I never said they didn't work, I said they don't work as well as reference cards...), et cetera. Hell, I've gone through four GPU setups in my FT02... I'm all over those threads. Please feel free to review those threads and count the complaints about the performance of dual non-reference cards. I'll be posting in there in a couple of days when my second 970 and NH-D15 come in, so I suppose I can test all of these hypotheses then anyway.

No, sorry you're lazy - it is literally on the product page.

Google better, or just add ".com"? "expreview.com" My first result was their twitter: https://www.google.com/?gws_rd=ssl#q=expreview

Says childish banter man on internet. You know full well what I meant - greater surface area = greater distribution of heat = less energy/heat per square inch of heatsink

yawn again, l2r.

you are still boring me with your selective reading, poor english, and incapability of *showing* deliberation in thought rather than *insisting* it.

snooze, kids these days...

"BUT WUTT ABOUT THOSE WORDS YOU USED ABOUT HOW HEATSINKS DON'T EMIT HEAT? OMG YOU ARE SO WRONG ABOUT THAT GO BACK TO THE EIGHTH GRADE AND RELEARN YOUR PHYSICS LOL UR SO DUMB"

Ahhhh, I see! That would be interesting to see, although now I wonder if there's enough space for fans mounted there in the FT02/FT05 (165mm clearance on the FT02, 162mm on the FT05), but definitely an interesting option...

Thought it might be unfair to refer to a review that's been archived from a Chinese website, so here you go:

[url]https://web.archive.org/web/20120618202117/http://en.expreview.com/2010/11/15/90°-rotationhow-to-install-heatpipe-cooler/11843.html[/url]
http://www.computerbase.de/2014-08/silv ... 05-test/4/ - this one is more about reference vs non-reference in these cases

*Prepares to be called an idiot because neither article is anywhere near 30 pages*

Less snark from all, please. Provide information, not personal attacks.

According to the German review you provided in the next post, they do work as well as reference cards.

When they rotated the case by 90° - turning it into a conventional one - the reference card's temps rose by 4c, the DirectCU II card's temps rose by 3c.


In dual GPU configurations, EVERY CASE benefits from reference (i.e. exhaust) cooling solutions. That's not "unique" to the FT02/RV02.


So it is the thing where they say a specific heatpipe orientation is suboptimal.

That is not what you claimed, i.e. that Silverstone recommends the use of reference cards.


Ooh, their twitter. Awesome.

I googled "expreview ft02" and "expreview rv02" and it didn't find anything.

And that Archive.org link that you brokenly posted gives me a "Wayback Machine doesn't have that page archived."
Yeah, but I'm lazy...


That is not what you said and not what you meant.

That is what *I* said.

You said these things:


...repeatedly demonstrating a misunderstanding.


Yes, I am insisting on it.

Heatsink size and/or efficiency has no effect on the amount of heat a CPU generates!


Glass house ... stones ...


Are you seriously trying to call me out for using your own words against you?!

Also, I'm likely older than you are, it's no use calling me a kid every time you run out of arguments.

And stop summoning straw men for me to fight, it won't work. I'll not forget what we are argueing about and I will come back to poke you with it every time.

Are we reading the same review?

I think you are looking at CPU temperatures. Try the pull-down menu and change it to GPU - the numbers should be a little more drastic for the reference card then what you are seeing, which speaks to the RV/FT series unique… privileging (sorry, that’s a poor word choice) of reference GPUs.



Sure, yes, but if you put two of the exact same cards in a different case you get better temperatures than you do in the FT/RV series.

You are correct, my fault! I am thinking of posts from the OCN SS owner’s club - whoops!

That is unfortunate - it is working on this end, right now, and users on OCN confirmed that it worked two months ago, so sorry for your luck? Perhaps try again later? Anyway, here are their reference card thoughts:

And their non-reference card thoughts:

It is worth noting that these results are old, and that users with newer revisions of the Accelero and T-Rad coolers like the Windforce (basically an accelero) have had better experiences - sintered heat pipes maybe?

Sure, yes, but if you put two of the exact same cards in a different case you get better temperatures than you do in the FT/RV series.

You are correct, my fault! I am thinking of posts from the OCN SS owner’s club - whoops!

Yet, somehow, the words are right there...

You keep insisting that you are having this argument in good faith and yet refuse to give me the benefit of the doubt. I appreciate that I may have been unclear/you may have misread, but at this point it’s ludicrous. You have consistently misread every post I’ve made in this thread from the beginning, where you somehow read my talk about CPU heatsink size as need for more exhaust…

Your primary beef with me (I struggle to call this an argument) appears to be this:

Which I totally agree with.
Let’s take a look back in context.
A disclaimer from my third post, before this particular argument started, that you appear to have missed:

To which you responded:
Again, I’ll insist that I know that and, in fact, said that a heatsink has no effect on the TDP/thermal output of a CPU when I very literally said “[having a bigger tower wouldn’t] change the TDP of the CPU.”
I then responded like so:

I’m not sure how I can be any clearer or why you insist that I am saying the opposite of the words that are posted here, which all leads me to believe that this is in poor faith and that maybe you’re just trolling me.

Actually, no, you said some nonsensical thing about how heatsinks don’t emit heat - see above. I know that you know that this is not true, that heatsinks in fact absorb heat from the CPU and then fans push that heat through the chassis, and I do apologize for taking advantage of this rhetorical slip-up later in the thread (I’ll explain below what I mean by "taking advantage" if you do not know what I am referring to - re: sarcasm used to poke fun).
Please explain how this ideas are incongruous. Bigger heatsink = greater surface area = greater distribution of heat = less energy/heat per sqaure inch…

Please re-read - I see very well that you are insisting but you are failing to demonstrate why…

No, I was sarcastically calling you out for suggesting that heatsinks don’t emit heat:

And then make fun of your rigorous demand for semantic consistency to hold you accountable for saying something that you know is untrue…

because, if it were, heatsinks would melt and or explode, so, therefore, have a sense of humor:

You seem to be operating on the false assumption that age and maturity are correlated.
It’s pretty clear that you have some bone to pick, but I’m really not sure how to make it any more plain or to remind you exactly what it is: when I am talking about using a larger heatsink rather than a smaller one because of the benefit of it’s increased size I am pretty plainly making the “bigger heatsink = greater surface area = greater distribution of heat = less energy/heat per square inch” argument. You can continue to insist that I am not, and you’re welcome to, but if you think that makes you a good faith ensurer of truth, or that fighting about semantics isn’t embarrassing for everyone in this thread, well, I guess that’s too bad.

Edit: Relavant aside, a user just posted a 970 sli reference setup in a FT05 with a D15 over at OCN... AP182s at ~570rpm constant, single CPU fan (middle) at ~500rpm on load, and they report that the GPUs stay below 70 while playing BF4 (significantly more demanding than RE6)... but, you know, conjecture isn't proof and they offer no GPU fan speed info

I think you are reading the graph wrong.
Pay attentention to the 5V and 12V.
For 5V the DirectCU II gets 80c when rotated 90° and 83c normally.
At 12V it's the same 3c difference.

I would not call that dramatic.


You are debating dishonestly.
You have asserted TWO things, that I have challenged:

1) Non-reference GPUs are worse in the FT02.

The only actual proof you have for this is the review/image you posted above and as I said, I think you are mistaken on this one.

2) A dual-tower heatsink would help GPU-temps, because it keeps the CPU cooler, which keeps case temps down.

You have claimed that multiple times, I have quoted you on this and I don't think the quotes are contentious.
It is wrong, because a better cooling solution is just more efficient at transferring heat energy from the CPU to the case. It doesn't affect the overall heat energy building up inside the case.


It isn't relevant.
The argument was never whether two GPUs with reference coolers would be better for a SLI setup. Of course they would be.

It's a straw man, you are being dishonest again.

-

At this point, I'm out. We have all but killed this thread, for which I'm sorry.

I have fulfilled an xkcd comic, but I've reached the point where I don't care if you're wrong. I'll swallow my pride and not return here, CA_Steve, no worries.

I have this case and am currently running a 4790k on a DH 15 and sli reference 970s. I upgraded the fans to 182s and run them at a constant and very quiet 570ish rpms. While gaming, GPUs get up to around 70 and the CPU stays in the 30s. This case is a pain to build in but it's the best performing case that I have ever owned.

I also own the case, also a SLI of 970s, also upgraded to AP182 and its a great combo (you can check my signature for the build), i usually run the ap182 @ 360rpms and have the bios ramp them up according to CPU temps, to me its inaudible in 3ft away from me, even gaming, but most of my games atm are not that hard to drive games like WOW and LOL. Benchmarking with Unheave does make the first gpu reach temps where the first gpu goes into above 2k rpms where its noisy, but on usual gaming loads its fine, either way, im pretty happy with the FT05, the only thing i wished it had more room for CM, and to come with AP182 instead of AP181s, but the rest is a great gaming case, that can be built quiet and can take a lot of hardware.

Post Deleted