ASUS Direct CU II GTX 670 question? - Page 2

I wouldn't focus on the oc alone but the actual performance. These are not standard OCs we are talking about here and depending on the situation the boost clock rate can alternate from one clock rate to another. That includes non OC and OC cards. So I wouldn't suggest to look at binning or max boost clock rate alone. You also have to consider what will be the average performance you would expect. For example if one card can do 1.2GHz while the other can do slightly higher their wouldn't be much if any perceived difference in performance in games. If you simply want to see the highest clock rate available then that's something different. But again, I'm looking at it from a performance perspective.

A non OC card is only a few percentages slower then a GTX680 and that's before you manually OC it! LOL, anyway I'm not seeing the need for an OC version because the way the cards work now is not the traditional way. You don't have to worry about OC the card right off the bat.
Edited by EastCoast - 5/12/12 at 1:26pm

They're the same price though as the reference model. If his case can handle non-reference cooling, then why not just get one of the non-reference models that are better? As for the Asus TOP, it's $20 more but you get a binned chip, if that comes out to a ~8% performance increase because you can hit a higher overclock (1300mhz vs 1200mhz for example), then that's damn cheap (~$2.5 per percent increase). An 8% increase in OC potential will translate to roughly a 3% increase in FPS. You also get the back-plate on both Asus 670's which is worth $20 by itself. Now i can understand not choosing to get the TOP version for $20 more, but why would you go out of your way to buy a card that's inferior for the same price?
Edited by SeanPoe - 5/12/12 at 1:34pm

If you are implying they are the same then why would he need to by an OC version for the same price? It would be the same or similar performance based on what he gets from the boost.

It's just that I get the impression you are saying get a card to get higher boost. When he really won't now what boost and performance he will get until he get the card . Lets not forget that he won't be able to set a specific clock rate for GPU/Ram using Percision X.

And what good is the backplate for? What person that doesn't have a backplate found the card to be handicap without it?
Edited by EastCoast - 5/12/12 at 1:38pm

They aren't the same though. It's not about the clock differences on the OC models vs the non-OC reference models, the non-reference models (the gigabyte and asus cuii) have improved cooling that's significantly (read: 25%+ quieter for the asus) compared to the reference cooling. Both cards also have improved PCB components that will increase the life of the card and also allow them to handle 24/7 overclocking easier. And the best part is, they're the exact same price.

But cooling alone won't guarantee you higher boost. And who has been complaining about reference design not giving them a decent boost? Sure, you can get a card with a non reference design but if you have a decent enough PC case that can keep your GPU cool I don't see that as much of a problem. The OP is using a CoolerMaster HAF-X Modded. I would find it hard to believe that he would have a cooling problem.

It's not about the cooling man, it's about the noise. If you were to take a reference 670 and a Asus 670 CUii and overclocked both to 1300mhz the asus would be about 25% quieter. The asus also has a backplate (for free), as well as tons of PCB upgrades (for free). I don't know why i'm even debating this. Can you can state one reason to get the reference model over the non-reference model aside from the fact the reference model expels it's hot air outside of the case (which you already said is a non-issue for him)?

Lets not turn this into an argument. You are jumping from one thing to another here. The noise comment isn't really saying anything. Again, the OP has a decent enough case that heat or noise shouldn't be a problem. The card you are talking about boost up to 1137 MHz GPU or so not 1300Mhz. That's manual OC and that's not really guaranteed as you are not able to input 1300MHz as your OC. Furthermore, the results shown so far aren't that impressive enough to me to warrant the $20 difference because he should get something similar or the same with the non OC version. Just use Precision X.

Since this is turning into argument I'm ending it here. However, I found something else worth mentioning. People are currently waiting for a driver update to fix whatever is causing the stuttering. I original thought it was just the 680s but someone with a 670 is saying they are having the same problem. It's something I've notice in this thread. They are working on a fix but just in case the OP notices it that's something they are currently working on.
Edited by EastCoast - 5/12/12 at 2:45pm

It's not turning into an argument .

The 'noise comment' is saying something. Heat and noise are always a factor when you're using fans to cool. Fans inside computers only move air, they do not use evaporative cooling to actually reduce the temperature inside the case. It doesn't matter if his case has 20 fans, all that does is reduce the temperature inside the case to the ambient temperature outside the case. The fans on his GPU are still responsible for cooling the card independent of the case fans (case fans DO NOT cool the card, they only reduce surrounding ambient temperatures which will reduce some of the work on the GPU fans). Reducing temperatures inside the case is a constant for both cards (ie, it benefits them equally). By that i mean, if a reference 670 needs to run its fans at 80% at room ambient temperatures to keep the GPU chip at a safe temperatures and a non-reference 670 only needs to run its fans at 50% at room ambient temperatures, then changing the temperature inside the case will effect them both equally. Lets assume he has crap cooling so ambient temperatures inside the case are bad, now the reference 670 needs to run the fans at an increased rate (lets say 90%, 10% increase over baseline) to keep the chip at a safe temperature, the reference card also will need to boost it's fan speed to compensate (lets say it's now at 60%). That's a 30% difference in fan RPMs. Now lets say he has PERFECT case cooling so case temperatures are ALWAYS right at ambient temperatures, the fans on the reference card will still need to run at 80% to keep the chip at a safe overclock temperature, and the non-reference card would still need to run at 50%. That's a 30% difference. That's the exact same as with poor case cooling. This is because you can NEVER reduce case temperatures below ambient room temperature with fans. Period.

Now what is the correlation to sound/noise you ask? Fan RPM (along with some other factors, like quality of design, fan size, etc) is what determines how loud the fans are going to be. Since the non-reference has a more efficient design and higher quality fans, it doesn't need to run the fans at the same speed to get the same amount of cooling. This translates to a reduction in sound production no mater what the ambient temperature inside his case is. However, since these fans run at 30% less load, this means you could increase their speed 30% more than the reference models if you wanted, resulting in more cooling and the same sound output as the reference coolers.


Also, just for the record, the reference models with stock cooling cost the EXACT same as the non-reference models with improved cooling and other improvements.

OP,

If there is any concern regarding noise you can take a look at this review:
Noise
The difference is about 4dBa difference between the non ref Asus and ref design. Something that you shouldn't notice. And it may be closer then that for the type of case you have. As for performance, I'm not seeing much that an OC couldn't handle on a non OC card. Take BF3 for example. I believe that the difference will be made up using Precision X (even for the other benchmark results).
Edited by EastCoast - 5/12/12 at 3:21pm

Here, look, Reference vs Asus CUi when both are overclocked:

Asus OC:
Max core clock with boost 1293 MHz,
Max memory clock 1890 MHz
80.6 FPS in their BF3 bench
25dbA at load at 1058 MHz
Has tons of PCB improvements that will increase the life of the card, free backplate, free tuning software,

Reference OC:
Max core clock with boost ~1240
Max memory clock 1760 MHz
75FPS in their BF3 Bench
40dbA at load at 915 MHz = normalized to 1058 giving linear scaling (which this cheap fan won't have) = 46dbA at 1058 MHz
Has no extras

To summarize:
+53Mhz (4% increase)
+130Mhz (7% increase)
+5.6 FPS (7% increase)
-21dbA (62% decrease in dbA) [This is a two-fold (~210%) audible fan noise decrease because of how humans perceive sound]
Has a longer lifespam from PCB improvements, free backplate, free tuning software.
Cost EXACTLY the same as a reference card

Please, keep telling us how a reference card for the same price is better. It's entertaining

Also, Guru3D was comparing a STOCK reference 670 to an OVERCLOCKED Asus 670 and the Asus was 4dbA quieter. Their testing methodology for noise is a joke when compared to TPU too, read the methodology for fan noise here, TPU has impeccable, documented and repeatable methodology. Who are you going to believe to be more accurate?
Edited by SeanPoe - 5/12/12 at 5:28pm