7970 overclocked vs a gtx 670 overlocked? - Page 9  

Did you guys see this?

The OP doesn't need anymore help deciding whether to go AMD or nVidia. Raghu, you've made a valiant attempt, (again) investing many countless hours (that you'll never get back) in your seemingly endless quest to convince the world that the 7970 is the greatest thing since sliced bread. But will you please, GIVE IT A REST, buddy

Now we need to switch our focus to helping the OP choose his 670. My vote is for either FTW, MSI, Asus, or Gigabyte. Personally I have the Gigabyte, I LOVE this card, and I highly recommend it. But the other three are great as well.

No more Tahiti vs. Kepler ... the whole subject has been BEATEN TO FRIGGIN DEATH on this forum, and I am TIRED of cleaning up the messes that result

+1 on the gigabyte 670, Nice and smooooooth in games

If you dont have an 8 pin connector, I found this!! : http://www.newegg.com/Product/Product.aspx?Item=N82E16812200420

Guys, I'm serious ... no more Tahiti vs. Kepler arguing on this thread. Help the guy pick a 670 so we can be done here.



Edit: I've decided to break my rule slightly for the purpose of educational value. The fact of the matter is, EVERY Kepler review I've ever seen do so calculates Kepler scaling WRONG. And the direction of the error is always towards making it look worse than it really is.

The problem is that they take a card that says it's Boost Clock (like in GPU-Z) is, say, 1000, then they overclock it by, say, 150MHz. They then arrive at a % overclock value of 15% ((1150-1000)/1000). But this is wrong, because they're failing to account for the real-time 'Kepler Boost'. That 1000 stock-boost card is actually most likely running at around 1100-1130MHz, not 1000MHz, to begin with. Take a card with a nice KB value of 130, and do the calculation again: (1280-1130/1130) and you get a % OC value of 13.2% (note: see the first link in my sig if you're not sure what I mean here ...)

If we found that the card got +10% fps after the OC, by the first calculation we'd get a scaling % of 66% ... most people would call that sub-par. If we use the CORRECT value, we come up with a much more respectable 75% scaling.

So make sure, if you looking at Kepler 'scaling' in some article somewhere, please keep this in mind. Also keep in mind that the memory on Kepler cards generally has a TON of headroom, and memory clock MATTERS in a lot of cases ...much more so than it did on Fermi. An OC on Kepler of equal % value to the core and memory, when scaling is properly calculated, will get you between 80 and 90% scaling depending on the test. I know it, cause I've done it. It's nothing to sneeze at my friends

The real difference between Tahiti and Kepler afa OC'ing goes lies in the % HEADROOM, not the SCALING % ... Original reference Tahiti cards clearly on average have more headroom, and that much is inarguable in my esteem. Although the GHz editions will change that equation substantially.

Hey, hey, whatdya know ... you were right about the new Beta driver. Here's a new Heaven run, same settings as last time (1360/3496).

304.42 FPS = 57.8
304.79 FSP = 58.4

So we ended up exactly tied now, despite our minor clocking differences




Also, for the purposes of a certain chil, uh, member ... who's been incessantly trolling me via PM about how my card is teh SUCKZORZ cause my power target only goes to 111%, and the FTW 670 is ZOMGBBQTEHGREATEST670EVAH cause it's POWERTARGETZFTW goes to 148% I proffer the following evidence that he ... uh ... doesn't know what the hell he's talking about.

Here's the same test, only instead of running my Power Target at 111%, I ran it at 85%.

And what do you know ... the score is the EXACT SAME. Both as the FTW with it's 148% power target, and as my last run with my target at 111% @ this person ... you are quite the smart guy!

Well personally I love EVGA for myself and the EVGA GTX 670 FTW would be a solid choice. Performance quality and a company that encourages over clocking with exceptional customer support.

The Gigabyte GTX 670 Windforce I hear from those who own it only good things whether it's from AMD or Nvidia owners. As for customer support I cannot comment.

Definetly work your way over to the [Official] NVIDIA GTX 670 Owners Club and you'll get some good feedback on the cards by owners.

Another card that's newer to the market is the MSI GTX 670 Power Edition Twin Frozr IV you might want to look into. There's a thread on it you should read through.

However if it's not the Gigabyte Windforce the EVGA FTW would be a very solid choice IMO.

About that MSI TWIN FROZR 670, I certainly cant find it on Newegg. Must be super new, but I much prefer the look of the gigabyte with three fans

The Gigabyte looks class ... check out my sig rig pic

Hard to see, too much " junk " in the way >.>

I suggest the Zotac GeForce GTX 670 AMP! Edition, it could be quieter but it performs just as well as a Ref. GTX 680

Or the EVGA GeForce GTX 670 FTW or SC,
ASUS Direct CU II Top isn't that bad either xD