[MattNo5ss]

It sucks that you can't run what clocks you want for clock-for-clock tests because the boost clock varies throughout, and how high the boost goes and when it changes isn't the same during multiple runs of the same test either. I can do "same OC" tests if you want, max OC tests wouldn't be a good choice for comparison if one can run higher clocks than the other in some tests.

[Archer0915]

I just think they are screwed up. The card is not much slower than the top of the line and priced like a top of the line card. To me $400 is overpriced. Not that the performance is not there but marketing this is a nightmare. nVidia should down clock it rebadge it a 665 and be done with it. I think they are hurting themselves with this.

[Frakk]

It is overpriced, there all overpriced from both colours.

But its still $100 cheaper than its slightly bigger brother.

The whole lot needs to come down in price, its just ridiculous especially in this climate.

[Brolloks]

Originally Posted by MattNo5ss

It sucks that you can't run what clocks you want for clock-for-clock tests because the boost clock varies throughout, and how high the boost goes and when it changes isn't the same during multiple runs of the same test either. I can do "same OC" tests if you want, max OC tests wouldn't be a good choice for comparison if one can run higher clocks than the other in some tests.

Then people should not say the 670 is the same as the 680, it is not, OC both, you cant OC one and let the other run at stock, same applies to any component.
If you game yes, I'd say get the 670 but for benching the 680 will stomp the 670

[Frakk]

Playing games is what its for, most people are more interested in what sort of actual performance there getting for there money than they are throwing around useless numbers in some game of one upmanship

[xander89]

Originally Posted by Frakk

Playing games is what its for, most people are more interested in what sort of actual performance there getting for there money than they are throwing around useless numbers in some game of one upmanship

++++++++

[clubber_lang]

WOW!!.....That is one hell of a card for $400.00!! Thank you for the test!

[Janus67]

Originally Posted by Frakk

Playing games is what its for, most people are more interested in what sort of actual performance there getting for there money than they are throwing around useless numbers in some game of one upmanship

I love gaming as much as the next guy, but don't try to say that to a competitive benchmarker as it is as much of a game as anything.

[hokiealumnus]

Originally Posted by Janus67

I love gaming as much as the next guy, but don't try to say that to a competitive benchmarker as it is as much of a game as anything.

To quote xander89, "++++++++".

Different strokes for different folks. There is no need to try to imply one is more worthy than the other Frakk. Many people (myself included) find benchmark scores to be far from useless. I don't game. Can't, don't have time, have no real desire any more...but I sure do benchmark! That's the game I choose. That it's not the same game as your preference doesn't invalidate it.

[MattNo5ss]

Originally Posted by Brolloks

Then people should not say the 670 is the same as the 680, it is not, OC both, you cant OC one and let the other run at stock, same applies to any component.

That's why I didn't compare an OC'd card to a stock card. I ran both at stock, and both boost to whatever they boost to.

What I'm saying is that it is IMPOSSIBLE to get a clock-for-clock comparison between 600 series cards because of the boost clock, so any comparison we do will not be ideal. Even if I would OC both cards so their max boost is the same clock, that would still not be a clock for clock comparison since boost clock isn't consistent throughout testing.

[Bobnova]

Can't you set static clocks if you short the current sense resistors?

[EarthDog]

Originally Posted by MattNo5ss

That's why I didn't compare an OC'd card to a stock card. I ran both at stock, and both boost to whatever they boost to.

What I'm saying is that it is IMPOSSIBLE to get a clock-for-clock comparison between 600 series cards because of the boost clock, so any comparison we do will not be ideal. Even if I would OC both cards so their max boost is the same clock, that would still not be a clock for clock comparison since boost clock isn't consistent throughout testing.

You could have it throttle via temp or power threshold, but thats certainly not worth it..

I agree you tested at stock, we cant really control boost easily so it is what it is.

[Moussa93]

There we go

[super2007]

Originally Posted by MattNo5ss

Nah, the GTX670 uses 2x 6-pin PCIe power connectors unless you are getting a non-reference version, then they might be using different power connectors.

Ya i am thinking about getting the gigabyte version which uses 8 and 6 power connectors. Would it be ok for me to use those power connections adapters that come with my card? since my psu has 2x 6 pin

[MattNo5ss]

Originally Posted by EarthDog

You could have it throttle via temp or power threshold, but thats certainly not worth it..

I agree you tested at stock, we cant really control boost easily so it is what it is.

Even when they throttle due to temps, the clocks aren't consistent

Originally Posted by Moussa93

There we go

Nice! Don't forget to take the film off the light gray brushed aluminum and the fan hub

Originally Posted by super2007

Ya i am thinking about getting the gigabyte version which uses 8 and 6 power connectors. Would it be ok for me to use those power connections adapters that come with my card? since my psu has 2x 6 pin

Nothing is wrong with using adapters, you'll be fine.

[Moussa93]

Originally Posted by MattNo5ss

Even when they throttle due to temps, the clocks aren't consistent



Nice! Don't forget to take the film off the light gray brushed aluminum and the fan hub



Nothing is wrong with using adapters, you'll be fine.

Oops...

*Opens the computer*

[lordkosc]

Originally Posted by super2007

Ya i am thinking about getting the gigabyte version which uses 8 and 6 power connectors. Would it be ok for me to use those power connections adapters that come with my card? since my psu has 2x 6 pin

Mine is coming on 5/16 from Amazon, damn backorder , but the use of adapters for power should be no issue. The windforce setup sold me on this card, from reviews I've read its between 10c to 20c temperature difference from the OEMs.

I'll be going from a GTX460 to a GTX670 , anyone know what performance increase I can expect? 3x better?

None of the reviews even mention the GTX460 anymore, which I guess makes sense as its 2 years old.

[ivanlabrie]

Consider 2 460s scale like a single 580...and pit that against a 670/80.

[lordkosc]

Originally Posted by ivanlabrie

Consider 2 460s scale like a single 580...and pit that against a 670/80.

Thats what I wanted to hear!

[xander89]

Originally Posted by hokiealumnus

To quote xander89, "++++++++".

Different strokes for different folks. There is no need to try to imply one is more worthy than the other Frakk. Many people (myself included) find benchmark scores to be far from useless. I don't game. Can't, don't have time, have no real desire any more...but I sure do benchmark! That's the game I choose. That it's not the same game as your preference doesn't invalidate it.

I do get where you are coming from hokie.I used to bench but not so much anymore. But i do agree for the majority of people its real game performance that matters.

But also with benching i tend to find it wont be about how high the numbers are for your overall rig, but how high they are for that specific setup. Sure a 680 will stomp a 670 in benches, but a 690 will stomp a 680 ( all obvious i know)

Personally im not too bothered about it getting stomped in benches, because if i was to bench i wouldn't bench a 670 against a 680. I would bench a 670 against other 670 so i suppose its all relative! It only matters really what you are benching when it gets to the top end where people are going for the highest numbers possible.