MSI GeForce GTX 670 2GB GDDR5 PCI Express Video Card (N670GTX-PM2D2GD5/OC) $322 after $20 rebate + Free Shipping (expired)

Quote from nolf66 : If you don't have to buy today than wait for a better card over this one imo. For the 660ti the only one I would buy is the Galaxy 3gb version. Even if I had to spend more for nvidia I would just for physX but thats just me. I can't see playing BAC or borderlands 2 without it.

I'm planning on having all the parts I need by the end of this coming week (I've ordered everything else but GPU), so waiting is not an option for me.

Quote from osv1 : if the sli build is done right, the gpu temp differences are fairly insignificant, and won't affect anything... we are talking maybe 4-7 degrees difference, under heavy use. for the most part, companies that sell video cards match their top binned chips to aftermarket cooling solutions, because that's where the best cooling is.

Quote from nolf66 : Yes, CPU's OC at 4.8ghz and GPU's boost to 1300ghz. Settings are different for the GPU's but the end results are the same.

I don't want to defend blower over axial. That is silly. Each card, each case will need its own solution. I simply take issues with BestJinjo's claim

Quote from BestJinjo : After-market open air designs are superior to 99% of reference blower designs in temperatures, overclocking, and acoustics

That's broad and inaccurate generalization, and one that's easy to disprove. I have had a different experience than osv. I happen to have that same Galaxy 670 GC. But my card (even with it's non "top binned" chip) manages to rock heaven maxxed [overclock.net] (one of the top five 670s at overclock.net @ 54.7 fps) harder than whole host of people with radial designed coolers, waterblock coolers, etc....

The cooling solution is NOT the answer or indicative of higher clocks, only noise levels. And you're fooling yourself if you think you can stack the "silicon lottery" in your favor by piking a different cooling solution or brand.

Quote from Runamok81 : Do you have any evidence to support that claim?

No one said anything about SLI or CF. However, if you have a modern case and are willing to go SLI or CF, you can pick a motherboard that has sufficient space between 2 cards to allow for good airflow [imgur.com]. In this case, the 2 cards will still run much cooler and quieter. Now if you go for a smaller board like Asus Gene V [newegg.com] or Tri-SLI where the space does not allow to have after-market open air designs, then of course reference blower design is superior by necessity. However, you have brought a claim that reference design is superior since it exhausts the heat out of the case, but this is not true for single-GPUs. In a modern case, whether or not the card exhaust the heat out makes hardly any difference based on scientific testing (2-4*C differences, see links below). What matters the most for system temperatures with a single-GPU is NOT the choice of reference blower vs. open air designs but the type of case [pugetsystems.com] you have and ambient room temperatures. The choice of case can have a 10-20*C difference for your CPU and GPU temperatures. [bit-tech.net] The choice of GPU style open air vs. reference blower hardly changes the internal case temperatures in a well ventilated case. This is why in a modern case you get the full benefits of lower noise levels and temperatures with an open air design. Most system builders still do not understand this and often buy cases with terrible airflow and then try to deal with this by buying a reference blower GPU. This is a backwards approach. The problem is not the GPU, but the case itself.

In a modern case with good airflow, whether you have reference or open air design hardly changes the case temperatures or the temperatures of adjacent components (assuming you do not have a small board like Asus Gene or go Tri-SLI). The reference blower myth and the heavy metal heatsink will damage my motherboard myth continue to persist despite both both been proven wrong a long time ago.

The reason I brought it up is because you earlier said this card's blower fan design is preferable to open air designs since it exhausts the heat out. However, this is a myth as no scientific evidence supports the view that a modern case with good airflow cannot handle the heat being dumped into the case from an open air design. Thus, for single-GPUs in a modern case, the reference blower is inferior - both acoustically and temperature wise - due to inferior heatsink and fan airflow design. The reason reference cards exist is to accommodate cases with terrible airflow, which happen to be the majority of OEM systems. This is why AMD and NV must build reference blower cards.

If you compare a reference HD7950 (Sapphire) to a Sapphire dual-fan (Sapphire Superclock), the internal case temperatures [hardware.fr] are hardly different in a modern case. MSI Lightning card despite dumping > 200W of heat into the case actually results in a cooler case than a reference HD7970 card (see the same link).

Where the major difference is in the card itself. Reference cards run hot and loud and after-market open air design cards do not [hardware.fr].

Using SLI/CF to prove a point that reference blowers can work better is an exception not the rule. If you look at a single-GPU case, in a modern a case with good airflow, an open air after-market card will run cooler, and quieter than a reference blower design, thus lowering the overall noise levels of a system, and lowering the GPU temperature, as a result allowing for higher overclocks as a side bonus. A blower fan design does not drop case temperatures by much (if at all) in a modern case against a good open-air design but has actually results in louder noise levels and GPU temperatures. That's what one would call a terrible trade-off. Unfortunately, most people continue to believe that a modern case cannot deal with 300-400W of heat being dumped into the case which is simply not true.

If you have a case with poor airflow, no matter what card you get, you'll still have 10-20*C higher CPU and GPU temperatures than you would have in a case with great airflow and an open-air GPU. The problem is, even if you have a modern case, the reference blower card will be louder and hotter than an open air design most of the time (excluding Tri-SLI, smaller boards with no space cases).

Quote from BestJinjo : No one said anything about SLI or CF... .... .... .... ...

Let me see if I understand. It took 750+ words to say that If you have the right modern case, and if you have the right-sized motherboard, and if you're not running SLI, and if you skipped the poorly designed axial coolers, then axial coolers are better. Like 99% of the time?

Quote from Runamok81 : But my card (even with it's non "top binned" chip) manages to rock heaven maxxed [overclock.net] (one of the top five 670s at overclock.net @ 54.7 fps) harder than whole host of people with radial designed coolers, waterblock coolers, etc.

so finally, after multiple posts, you've "proved" your entire argument with the results of only one card.

the fact that you claim that one card may have decent overclockability doesn't disprove the claim that some companies do bin sorting on the chips.

and it doesn't back up your claim that one cooler design is better than the other, because, among other things, you haven't performed any kind of controlled testing against other cooler designs.

Quote from osv1 : so finally, after multiple posts, you've "proved" your entire argument with the results of only one card. the fact that you claim that one card may have decent overclockability doesn't disprove the claim that some companies do bin sorting on the chips. and it doesn't back up your claim that one cooler design is better than the other, because, among other things, you haven't performed any kind of controlled testing against other cooler designs.

If you understand how silly a single point of evidence is, then why were you first out-of-the-gate to offer your system as evidence? I was simply countering that weak-sauce with my own proof. Now, the sour grapes?

Let's clear the air. axials as a group will outperform the blower design. But NOT 99% of the time better. That's bogus and I wanted to set the record straight. While the blower design CAN be improved upon, it is not exactly the cripple that others have made it out to be. Give the nVidia thermal engineers a little more credit.

here this thread needs puppies now.

The only good blower designs are the ones used by HIS's aftermarket IceQ series.

Quote from expansionist : The only good blower designs are the ones used by HIS's aftermarket IceQ series.

Did they ever put that on top of a 670?

Quote from Runamok81 : here this thread needs puppies now.

Rep for you for puppies pic, cause it's a Monday!

Quote from Runamok81 : If you understand how silly a single point of evidence is, then why were you first out-of-the-gate to offer your system as evidence?

let's see, you have totally failed to back up the argument that you started... and you even postulated a straw man about sli, which i responded to.

BestJinjo provided ample evidence, that proves you are wrong.

perhaps the high school that you are in has a class in logic that you could take?

Deal seems to be dead. $350 AR now.

Quote from Traveler2530 : Because it's an expensive premium video card mostly for gamers. Even if last year it was $2000, and now a jaw dropping $200... I still would not buy it! I don't need that level of video performance. Almost 100% of the time I'm on the computer, I use FireFox and MSOffice. The few times I play games it's chess, Wheel of Fortune, crosswords... No games that require high-performance video. My c. 2007-2008 home-brew assembled from markdowns/returns (Q6600, P-35 Gigabyte board, $15 video card AR, Patriot Torx-SATA-II SSD...) has MUCH more power than I'll need for years to come.

Someone like you shouldn't be trolling in a thread about a 3D gaming graphics card.

Quote from osv1 : let's see, you have totally failed to back up the argument that you started... and you even postulated a straw man about sli, which i responded to. BestJinjo provided ample evidence, that proves you are wrong. perhaps the high school that you are in has a class in logic that you could take?

I don't have to supply evidence, because I never made any silly claims. I'm simply debunking his nonsense. But if you need evidence, you can ask BestJinjo. He shot himself in the foot when he supplied the article [tomshardware.com] in this post. You'd be silly to corroborate with BestJinjo and put that article as evidence. How can you read that article and especially this line

Quote : To our surprise, Nvidia's reference card can keep up with the pack. It even pulls in front of two three-slot competitors.

And take-away that axials are always better than blowers? Read it! That article does nothing but show the difference between good and poor axial designs. And while as a pack they outperform the blower, there are some really poor-performing axial dogs in there. This has been beaten to death. Can we please move on to talking about something else now?

Guys...96.4856% of statistics are made up on the spot.

Oh and the cooling solution used to be the answer and indicative of higher clocks. The cooler the card, the more stable and the easier it was to overclock. It's not the case with the 600 series though