The Official Graphics Card and PC gaming Thread

Oh wow - there is tons of room!!!! Thanks for finding that!

Thanks for the info again - good stuff Rubix and Sam. I took note of those model numbers Rubbix - maybe I'll end up buying the d0 on ebay.

PUSH-PULL, THE REPEATER EFFECT
Yeah I know Sam you're not a major fan of push/pull. You believe it offers limited benefit, and maybe you're right.

I think of it the same way I think of intake fans being balanced by exhaust fans - the exhaust fans could be removed, and you would have flow out the passive exhaust vents - similarly you could turn off the intake fans, and you would have airflow into the case from the pull-through of the exhaust fans. But having both sets of fans is like adding repeaters to an electrical circuit, like a phone circuit, or like the booster the cable guy recently added to my cable line - all my tv channels are a lot sharper now!

A repeater in the form of an extra fan tries to make up for the friction drag on the air after it leaves the first fan. In the case of the TRUE hsf, that is particularly valid, because you expect a lot of turbulence through the aluminum heat sinks in order to get good thermal performance from the hsf. You expect a lot of drag on the air blowing into the TRUE. Sure, you're going to get most of that air coming out the other side of the TRUE, toward the back of the case, but the pressure will be dramatically reduced, I believe, by the drag from the aluminum fins. I have to believe that you can speed up that airflow through the TRUE, and increase the heat removal, by placing the vacuum cleaner of another fan right on the back side of the TRUE and sucking that slowed down hot air out!

Intuitively it makes sense, right?

Furthermore, Sam, your great points about the huge amount of heat produced by the 920, gives me more reason than ever to try to achieve a push-pull solution. I don't know if we can find any comparison tests anywhere, but I would love to see push-pull compared to just push (or just pull) to see if anyone has documented the two-fan benefit on the TRUE. Are you aware of any test like that, Rubbix?



PUSH-PULL, PLUS ANOTHER BACK CASE FAN ONE INCH AWAY?
So, Sam, let me ask: When you say interference from the back case fan, I assume you are questioning the fact that the back case fan will end up being about 1" from the pull fan on a TRUE push-pull fan setup.

So, (assuming we're not talking about physical interference - it looks like that won't be a problem) from an airflow perspective, will it interfere? That's a good question. What if we removed the pull fan from the TRUE? What would happen?

Having already made my case that I think push-pull is better than only one fan on the TRUE, then, unless I am wrong in my basic assumption, if I eliminated the pull fan from the TRUE, I don't see how I could expect the back case fan to do the work of the vacuum cleaner suction of the pull fan sitting right on the TRUE sink. The back case fan is at least 2" from the TRUE, and the power of the suction would not be nearly as great - maybe 50% less than applying the pull fan right to the TRUE sink. Does that make sense?

If that makes sense, and if we believe that the second pull fan is a benefit, particularly considering the huge amount of heat produced by a 920, then, let's leave the pull fan on the TRUE, and mentally, let's remove the back case fan.

But now that leaves the back TRUE fan with the job of not only sucking the sluggish hot air out of the TRUE, but also expelling it out the back of the case. But again, that TRUE fan is about two inches away from the case wall - and at least 20-30% of that air is going to be deflected off the vents and bounce back into the case I would think.

I have to conclude that adding a back case fan in direct contact with the vent will push the air out of the back of the case better than trying to let the back TRUE pull fan perform that job, in addition to cooling the TRUE.

Tell me if you see a major hole in my logic, but it seems to me that one would not want to eliminate any of those fans - it seems to me that one would want the push-pull on the TRUE, AND one would want the repeater effect of the back case fan, even if the two fans are only 1" from each other, lol.

Other than negligible increased noise, and negligible increased power usage, I don't see any negatives, but only positives. Where have I messed up (other than not being born a limie or an aussie? hahaha)

Rich

 

Adding a fan like that doesn't necessarily help, as it forces the fan speed of the exhaust fan to rise, which harms its longevity, and it also does not create the pressure required for the extra amount of airflow to be effective.
Realistically, your best bet is to have the CPU HSF blowing into as big or as powerful a fan as you can get. With a big top fan like the HAF (and possibly the spedo, I forget), put it bottom to top vs front to back.

 

IMO positive pressure is the best for cooling, and sam have you tried the push pull config as ALOT of people use it to great effect

 

Hmmmm.

Here's another picture with dfi and back to front push pull.

Sam, you're saying that two fans in close proximity will affect each other causing fan revolutions to increase, thereby reducing fan life? I guess I could see that rpms could go up. But when you look at a scythe with 50,000 hours, how much life would you lose - half - I doubt it, but even at that, 25,000 hours if I game 10 hours a week is 250 weeks or 5 years. I can accept that.

Shaff's point is well taken. Hardware Cannucks seems to do a good job on their reviews, and Thermalright added those extra clips for some certain reason. Is push/pull just a marketing gimmick, or have any tests been done to show increased thermal transfer?



Anyway, Sam, yes, the spedo has a ceiling exhaust. You're saying, turn the TRUE around, north/south, (top/bottom) put a large push fan underneath, and blow the air into the ceiling exhaust.

Hmmm. Once the bottom clips are mounted on the motherboard, the TRUE can be mounted either way. I guess maybe some testing would be very interesting to see which way would work better. Blowing the air up would allow both the upper case exhaust as well as the ceiling exhaust to vent that air - but that configuration doesn't allow push pull. Implementing push-pull, you're blowing most of the heat out the lower of the two back case exhausts, with some additional help from the upper and the ceiling fan.

Somewhere on google, somebody has done some push-pull testing.

Edit1: Well, without spending too much time looking, here's one guy claiming 3 degrees further temp drop by adding the second fan on a TRUE cooling a Q6600.

Edit2 - Evidence in the other direction:
Some more google research: "I did this test last winter and found that running two fans will make no difference on the temp. I did find that it cool quicker in dual mode to get back to idle temp much faster. It also made no difference when overclocking either." Different author, same page: "I've seen reviews where they proved that a second fan doesn't really make a big difference with the Ultra-120 Extreme."

Edit 3: Hardware Canucks Thermalright Ultra-120 eXtreme 1366 RT CPU Cooler Review 1/21/09

. . . . . . .

Rich

 

I Apologize for multiple post! Another thread is helping.

 

Another thread - Kevin you traitor!! Hahaha.

Sam helps out on about a dozen threads, I think. Shaff also helps out on about a dozen - well, to be more accurate - he has been banned from a dozen and is now down to a half dozen. LOL

Rubbix - I am coming across some amazing data on the stepping d0 - some wild results - 10% better overclocking at lower voltages - stuff like that. Check out this article: Anandtech (Sam's fav site!) threw it on a DFI H8! Tank guys has the chip for $319.

WHERE DID THIS COOLER COME FROM? YOU GUYS HAVE BEEN HOLDING OUT!!


Thermalright IFX-14 !! Compatible list includes the DFI H8. I've been googling all night - never heard of the thing before and now I've seen some reviews and they're showing 4 degrees lower than TRUE. The IFX takes up to THREE 140mm fans in "push, push, push like a sonavabitch" ... er, I mean, push push pull.

Is it a gimmick? It's by Thermalright, the makers of TRUE. Or is it as Thermalright claims, the best there is in air cooling?

Coolers for Intel Core i7 Overclocking:

Here it is in the picture below, in the xbits lab Antec 1200 case, Asus deluxe x58 mobo, 3 1 gig Dominator modules, with three 140-mm Scythe Kaze Maru fans at 1260rpm mounted in the north south orientation that Sam favors pointing up to that big 21cm ceiling fan.



The DFI H8 board is on the compatability list, but I don't know if I would have room to mount it that way with a graphics card in slot 2 - the hot slots on the DFI being 2 and 6.

Well, here's another view. Maybe I would have room to put the graphics card in slot 2.



Another quote: "Thermalright IFX-14 cools the CPU 3°C better if its heatpipes are positioned horizontally (we are talking about tower cases here)." So they are saying that Sam is right - point the fans up toward the ceiling exhaust fan. That sounds quite convincing for Sam's argument - however northbridge (PWM) cooling has to be taken into account as well - and I would still possibly be interested in real world testing in the spedo case to see how all temps were affected by the two different orientations, with all four gpus running at load.

One more picture - tight fit with three but not six dominator memory modules - same problem as the TRUE.



Okay - the temps: Here is the chart showing the same fan configuration on the IFX and the other 5 coolers - 4 degrees cooler than TRUE.



The 3-fan 140mm configuration is at the bottom - those 140s were running at a lazy and probably inaudible 1260 rpm. Temps not as low as the two 120s running at 2600 - but same stable overclock at 4.0. (Note - this is not the new higher-performaing d0 stepping.)

In their closing remarks, xbit labs says:

So what have you guys heard about this cooler?

Rich

 

LOL! I couldn't decide whether to post it here, or the OC thread. It has been figured out regardless. See...I wasnt sure whether my 260GTX was 65nm, or 55nm. 2 softwares had different ideas LOL! It has been discovered though, that its indeed 55nm(Thankfully)! If it were 65...my PSU probably would have POPPED by now LOL! And in fact, it may be close to going now Though it only blows cool air!!! Something tells me, I made a stupid purchase buying an OCZ. Its ok, you guys can bash me. I can take it LOL! Regardless, in under a month, I will be buying a Behemoth PSU LOL! I dont like my equipment being limited in options!!!

Kevin

 

The question you have to ask yourself is - is it really worth all that extra effort of making sure it will absolutely fit for the 4c or so benefit? The TRUE120 and Noctua NH-U12P are still IMO the best.

 

I did try it a while back, and couldn't really see any difference. Then again, perhaps it only works best with noisy high speed fans. What is more important however, as I proved, is room setup. At home my PC is on my desk to the right of my monitor, in a small room that gets very hot (due to the heat my PC puts out). Now I'm back in York, the room is much bigger and doesn't heat up very much, and the PC is on the floor. Where it was on the desk, the side fans suck the hot air in from my monitor straight into the CPU cooler. The upshot of this is, moving back to York has dropped my idle temps by 25ºC
What's worth mentioning is that the TRUE has a relatively low fan throughput plateau, i.e. where making the fan faster makes almost no difference to temps. Generally this seems around 1600-1800rpm for a 120mm fan, beyond that you're talking 1-2ºC off for an extra 50% increase in fan speed. Presumably this is because the heatsink is so large, the limiting factor is now the contact with the block and the heatspreader of the CPU.

I was wondering when the IFX-14 would show up. In essence, I think it is a bit gimmicky. I've never seen any tests at normal configs where it offers any more than a couple of degrees over a TRUE. It's marginally better, but more expensive, heavier, more difficult to install and so on and so forth.

For the record, a Kaze Maru is definitely not inaudible at 1200rpm, they're fast and loud fans for the size, but even so, a few more hundred rpms would knock the temps down a little bit.
Did you spot the Zalman PSU in that picture?

 

Hehe. Trust you to pick that up dude

So... I was doing some thinking... how much do you think my idle cpu temp will rise by when I finally get my GTX275s installed later this week? ATM, it's 44c hottest core when room temp is around 30c. I don't know if my modded side fan is going to cut it :S Might have to put another one on..

 

the silverstone fortesss is one of the top air cooled cases around and it uses 2 intake 180mm fans (top and bottom) and 1 120mm exhaust, usign positive prssure to force air though the rest fo the case and then though the vents, and thought the PCIE slots, (which are ventilated).

 

But doesn't it lack a side fan IIRC?

 

yes your right

yeah it was the best for the CPU IICR but for the GPU it was one of the worst lol

 

Very good point - after the excitement wore off last night, I started asking myself the same thing.

Ok Sam, I see your point completely - 25 degrees cooler!! How many more houses do I have to sell to be able to move to York so my system will work as well as yours?

Hahaha - I threw that "inaudible" out there and I figured you'd pick up on it if I were wrong.

I got myself temporarily confused - at the same rpm the bigger fans will push more air, but of course, generally make more noise. I guess there are other factors at work besides size, for example shroud design, pitch of blade, bearing type. The scythe s-flex 120mm fluid dynamic bearing fan is rated at 150,000 hours and optimized for noise reduction. The Kaze Maru is a sleeve bearing design rated at 30,000 hours. Here are the specs:

Kaze Maru 140mm at 1200rpm is 26.58 db, delivering 61 CFM.
Scythe S-flex 120mm at 1200rpm is only 20.1 db and delivers 49 CFM.
Scythe S-flex 120mm at 1600rpm is 28 db and delivers 63.7 CFM

It looks to me like the choice between the 120mm scythe s-flex at 1600 rpm is just about a tossup, compared to the kaze maru 140mm at 1200 rpm. The scythe s-flex will last 5 times longer - basically forever - and for an extra 1.42 db of noise, you pick up 2.7 CFM more flow.

Applying Sam's analysis to the ifx-14, sounds like Sam's analysis is accurate. The xbit labs chart showed 5 degrees lower temps, with two 120s at a screaming 2600 rpm, than with 3 140s kaze maru at 1200 rpm. And as we see by the fan specs, 3 scythe s-flex 120s at 1600 rpm wouldn't make much of a difference from the kaze maru 140s at 1200.

I was thinking the same thing - it would have been nice if the xbit labs reviewer had had the 1900rpm kaze maru models, with a fan controller knocking them down, say to 1500-1600 rpm, so we could see the real deal on temps. The 1900 rpm kaze maru at a screaming 41db puts out over 104 CFM - that's a lot of air! - so fan-controller-down-revved to 1500 or 1600 rpm, it should still be moving lots of air, maybe at a 33-35db noise level.(It looks like the xbit labs guy just scrounged around and grabbed the only set of three 140s that he had on hand.)

Ya see, you were holding out!! Hahaha.

Well, around the net, most everybody seems to agree - maybe the ifx-14 is a lot of extra trouble for little gain. But what's everybody's take on pure copper TRUE (with a counter-balancing spring hanging from top of case.) I know, you're all gonna say: "Rich are you really Rube Goldberg in disguise?" haha

A Zalman PSU? I ditto rubbix - good eyes, Sam. Oh yeah, there it is. I remember now. Last night, about the time I was telling myself, "that looks like my antec 1200 that I just traded in for that spedo" before I verified by checking the setup, I noticed the bottom PSU in the third shot. But I didn't pick up on the brand. How's your Zalman working out for you these days?

Shaff, I agree 100% on what you said about positive case pressure. You know, Sam, you once said that you opened the front of your lexa and you were dismayed to find hot air trying to exit the front of the case. As I recall you thought air should have been trying to be sucked in. But positive case pressure means that any opening at all will expell air - and in my mind that's what you want - if you absolutely want to control dust and not have all your heat sinks lose effectiveness over time. So that means more intake pressure than exhaust, and filters on every intake fan. That's going to be my modding philosophy for the spedo.

So guys, do I spring load the all-copper TRUE, or FUGGEDABOUDIT! LOL

Rich

 

remember to have ur top exhaust as an intake , or test to see which one works better

 

Lol. Dunno, depends what you buy. Rent here is cheaper than a lot of places, so perhaps house prices are too?
The area has nothing to do with it though, it's all about the room layout. My room here is much bigger, and is laid out in such a manner that my PC can sit in a much more useful place.

To be honest, don't look at fan specs too closely, they're often inconsistent and not especially accurate. Scythe's are usually pretty good, and from my own experience those figures seem close. However, a good 1200rpm 140mm fan should not be quite 26.6dB. The Marus are rather rough, simple designs compared to the refined S-Flex. For more air per rpm, and better CFM per dB, the 120mm Slipstreams (Kaze Jyuni) are excellent buys, I'm considering picking some up for my HAF. They make them in 5 speeds - 500, 800, 1200, 1600 and 1900 rpm. Pitch and airflow wise, due to their 9 blade design, this is more like 650, 1050, 1600, 2100 and 2450rpm though. (But the noise produce will be a little less than this due to the smaller blades)
With the big 230mm+120mm setup at the top, 140 at the back, and more vents than you can shake a stick at, positive pressure doesn't really bug me with the HAF. The Lexa was a bit cramped, and simplistic as far as airflow goes to cool two X2s properly. The next system going in it should have no issue though, and this time it's silence over performance.
The TRUE Copper is a big waste. It's lethal to motherboards unless the PC lies flat on its side all day, and does't really gain you any more than a loud fan does for a regular TRUE. The limiting factor nowadays is the heat transfer through the heatpipe.

The Zalman, so far, is still proving amazing, still inaudible no matter what I do.

 

its a shame i cant say that my 850W northQ isnt so quite

its quiet but not as silent as my 650W dark power pro

 

It's quiet because you're only drawing 150W and can't hear it over the GPU fan...

 

which PSU?

i cant hear the northQ over the silverstone 80mm fan, the GPU is quiet at 20% fan speed. whic is what i have it at

 

Here's a good question. Once upon a time a few months ago, I saw an alternative means for power for a graphics card. That way, one could upgrade their Video card, without having to upgrade their PSU for a lot of money. Any thoughts on that technology?
http://www.newegg.com/Product/Produ...scription=Independent/Supplementary&x=22&y=31

That way, In removing the GPU, it SHOULD effect the 12V rail rather substantially. Then I can bump my CPU up a notch or two I REALLY believe my CPU is becoming underpowered when I attempt more overclocking than im at now. Infact, to prolong the life of the PSU, I may drop it back 250Mhz. Ive heard its possible also, to destroy one rail, while maintaining others? Should I just underclock the CPU, and wait til I can afford a SUPER PSU? You know...a 1200W or more HIGHLY recommended PSU.