GALAXY GeForce GTX 660 Ti GC 3GB Video Card Review

Man, this card is starting to get on my nerves.

This damn rattling is practically nullifying the benefit of the quiet fans. And from what I can see, it happens on Galaxy's 670 as well, which uses the same PCB and cooler. So I don't even know if RMA'ing will solve the issue.

Further, my particular card does an okay 1199 core in Heaven, but I can't OC the memory one bit. Looks like I'm really meant to play within the handicap of this card.

And even at that max overclock, I have to listen the rattling which gets louder and more sporadic in rhythm thanks to the fan ramping up.

What's going on?
 
Just got my 2 Galaxy 660 Ti cards in the mail and have done some preliminary tests to compare to my Gigabyte GTX 680 SLI setup. After clocking the 660's to around 1250/7000 there is about 6-13 FPS difference in average FPS so far in Sleeping Dogs, Heaven, and Dirt Showdown at 8x MSAA with my 680 SLI setup which was at 1215/6500. My monitor is 2560x1440 BTW. Just wanted to let people know the cards overclock well especially on VRAM which seems to somewhat negate the crippled memory bus.

Here's a little something I've capped so far:

Everything is max and the same in all tests for both sets of cards, 8x MSAA in Dirt Showdown with Ultra settings, 8x MSAA in Batman AC all settings max (edit: Physx high on both, BTW).

680 SLI













660 Ti SLI











It seems the only scores to suffer noticeably at 8x AA are maximums but averages and minimums are pretty good and fairly competitive with the 680 SLI setup at highest settings with current gen games especially considering the $370 price difference.

So far I'm pretty impressed with what the Galaxy 3GB models are able to accomplish.
 
Last edited:
Ajiki-kun said:
Man, this card is starting to get on my nerves.

This damn rattling is practically nullifying the benefit of the quiet fans. And from what I can see, it happens on Galaxy's 670 as well, which uses the same PCB and cooler. So I don't even know if RMA'ing will solve the issue.

Further, my particular card does an okay 1199 core in Heaven, but I can't OC the memory one bit. Looks like I'm really meant to play within the handicap of this card.

And even at that max overclock, I have to listen the rattling which gets louder and more sporadic in rhythm thanks to the fan ramping up.

What's going on?
Click to expand...

I'd exchange it. The two I have are silent except at max fan speed which still isn't as loud as my 680 Windforce models.
 
convexion , you spent nearly 700 bucks for a setup that a single $450 7970 ghz could basically match at 2560.
 
convexion said:
I'd exchange it. The two I have are silent except at max fan speed which still isn't as loud as my 680 Windforce models.
Click to expand...
So it's just me that keeps getting shit on this gen.

Alright, I'll go return this card which I know should do much better. The returns guy at NCIX is starting to hate me...

EDIT:
cannondale06 said:
convexion , you spent nearly 700 bucks for a setup that a single $450 7970 ghz could basically match at 2560.
Click to expand...
What is this, the AMD board? We get it. The 79xx is better for raw performance and higher res. at the same price bracket. You don't have to keep bringing it up.

I myself have just returned a $320 7950 for which AMD cannot seem to release drivers that don't suck ass. It worked extremely well when it worked. Other times, it was a nightmare. Same goes for the 7970 I had to sell. I decided no more AMD from that point on. The 660 ti is the best nVidia deal there is right now, and if it's at least close to a 7950 at $300, I'll be happy with it.
 
Last edited:
Ajiki-kun said:
So it's just me that keeps getting shit on this gen.

Alright, I'll go return this card which I know should do much better. The returns guy at NCIX is starting to hate me...

EDIT:
What is this, the AMD board? We get it. The 79xx is better for raw performance and higher res. at the same price bracket. You don't have to keep bringing it up.

I myself have returned a 7950 that AMD cannot seem to get drivers for that don't suck ass. It worked extremely well when it worked. Other times, it was a nightmare. Same goes for the 7970 I had to sell. I decided no more AMD from that point on. The 660 ti is the best nVidia deal there is right now, and if it's at least close to a 7950 at $300, I'll be happy with it.
Click to expand...
I am sorry but if someone is spending 700 bucks for two cards to only keep up with a single $450 card then I will bring it up. I prefer Nvidia but I am simply pointing out the facts.
 
cannondale06 said:
convexion , you spent nearly 700 bucks for a setup that a single $450 7970 ghz could basically match at 2560.
Click to expand...

No big deal, I bought the cards to test for shits and giggles, but not necessarily to keep. My next benches are going to be against one of my 680 Windforce cards which boosts over 1300 Mhz to see if the SLI 660 Ti setup is worth keeping. If the 680 is within about 15 fps I'll probably just stick with that and sell the other three cards because I really can't tell the difference between 2x and 8x AA at 2560x1440.

As far as AMD goes I put a 7850 in my HTPC a couple weeks ago which is really great hardware but the drivers ruin it for me. This is coming from a fan of ATI (not so much of AMD right now). I've had a lot of cards from them over the years, my last one being a 6970 but I'm shocked at how bad the drivers have gotten since then. I'm currently running that card on the 12.3 Cats because everything later than that causes problems of some form or another whether it be in home theater software, games, remote desktop issues, etc. Because of that I also picked up a 660 Ti Zotac to replace it, so you could say I spent nearly a $1000 in cards that could potentially be beaten by a 7970. :p
 
cannondale06 said:
I am sorry but if someone is spending 700 bucks for two cards to only keep up with a single $450 card then I will bring it up. I prefer Nvidia but I am simply pointing out the facts.
Click to expand...
I'm sorry too. I know you're not a troll or anything. I'm just upset right now.
 
cannondale06 said:
I am sorry but if someone is spending 700 bucks for two cards to only keep up with a single $450 card then I will bring it up. I prefer Nvidia but I am simply pointing out the facts.
Click to expand...

I wouldn't call it keeping up, the 660 ti SLI overclocked definitely beats out the 7970 from the few benchmarks I've been able to find. Unfortunately there isn't a lot of info out there @ 25x14 at least with my current comparisons. It's a matter of price vs. performance though, that's for sure. The closest I could find to compare in a quick search for Heaven and Sleeping Dogs was Techpowerup and the neogaf forum.

Unfortunately the guy on Neogaf didn't disable vsync before benchmarking so the score has a ceiling and Techpowerup isn't specific on settings other than 4x AA so I went ahead and kept everything maxed but AA at 4x. I also matched the Neogaf guy's settings with sync on and off for comparison. My scores below theirs.

heaven_2560_1600.gif


Neogaf guy's Sleeping Dogs score.


660 ti SLI



Vsync on and off



 
cannondale06 said:
at 2560 a single 7970 ghz is within 6% of gtx660ti sli with 4x AA and within 3% with 8x AA. it makes WAY more sense to get the much cheaper 7970 ghz card for the same playability and no dual gpu issues. http://www.computerbase.de/artikel/grafikkarten/2012/test-nvidia-geforce-gtx-660-ti/7/

I know you had 3gb cards but all that does is waste another 80 bucks as it adds no increase in performance.
Click to expand...


I'm not arguing price vs. performance that the 660 ti in SLI is a great deal at this point, but those numbers seem... odd. I'm guessing if they're correct there is absolutely no overclock on the 660 SLI setup which is a shame because they're so capable. I didn't test the 660s at stock to know for sure because that really would defeat the purpose and just wouldn't be [H]ard. ;)

Any other benches showing the 7970 at 25x14/16? I'm having a hell of a time finding any for these specific programs at these settings. I need to reinstall Metro 2033 and some other games to get a good comparison as I'm just using stuff I already had on the HD.
 
cannondale06 said:
at 2560 a single 7970 ghz is within 6% of gtx660ti sli with 4x AA and within 3% with 8x AA. it makes WAY more sense to get the much cheaper 7970 ghz card for the same playability and no dual gpu issues. http://www.computerbase.de/artikel/grafikkarten/2012/test-nvidia-geforce-gtx-660-ti/7/

I know you had 3gb cards but all that does is waste another 80 bucks as it adds no increase in performance.
Click to expand...

You keep referencing the same computerbase article to illustrate your point about 660 Ti SLI. Maybe we should wait for some more SLI reviews before we go around declaring that 660 Ti SLI is a failure. The Tom's review (about the only other SLI one I could find) showed the 660 Ti about 50% faster than the 7970 in most games.

http://www.tomshardware.com/reviews/geforce-gtx-660-ti-benchmark-review,3279-10.html

The PCPer review didn't test a 7970, but it showed the 660 Ti SLI handling most games pretty well (although not all were tested with MSAA).

http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-660-Ti-2GB-Kepler-Graphics-Card-Review
 
Last edited:
Forceman said:
You keep referencing the same computerbase review to illustrate your point about 660 Ti SLI. Maybe we should wait for some more SLI reviews before we go around declaring that 660 Ti SLI is a failure. The Tom's review (about the only other SLI one I could find) showed the 660 Ti about 50% faster than the 7970 in most games.

http://www.tomshardware.com/reviews/geforce-gtx-660-ti-benchmark-review,3279-10.html
Click to expand...
um that is not the ghz card and that is no where near 50% overall either. BF 3 and Batman do look really good on gtx660 ti sli though.
 
cannondale06 said:
at 2560 a single 7970 ghz is within 6% of gtx660ti sli with 4x AA and within 3% with 8x AA.
Click to expand...

BTW, comparing Techpowerup's Heaven score for the 7870 GHz edition above to mine shows a 34% increase in performance on my end, fairly significant but it costs for sure.
 
convexion said:
BTW, comparing Techpowerup's Heaven score for the 7870 GHz edition above to mine shows a 34% increase in performance on my end, fairly significant but it costs for sure.
Click to expand...
I never pay much attention to synthetic benchmarks though. I typically only use them to make sure my card is performing like it should.
 
cannondale06 said:
um that is not the ghz card and that is no where near 50% overall either. BF 3 and Batman do look really good on gtx660 ti sli though.
Click to expand...

I didn't do the math, but it is closer to 50% than it is to 6%. I'm not arguing either way, I'm just saying we need to wait for more reviews before we go around calling people out for choosing 660 Ti SLI over the 7970.

Max Payne is the one where SLI seems to really suffer.

Edit: I'm not sure how computerbase gets their average, but if you look at the per-game results there are a couple of games where SLI isn't working (and Dirt Showdown has negative scaling) so if they don't throw those games out, it'll really screw up their average numbers.
 
Last edited:
Forceman said:
I didn't do the math, but it is closer to 50% than it is to 6%. I'm not arguing either way, I'm just saying we need to wait for more reviews before we go around calling people out for choosing 660 Ti SLI over the 7970.

Max Payne is the one where SLI seems to really suffer.
Click to expand...
it ranges anywhere from 10% to 65% faster with average of 31% in the games they tested. that would probably be about 17-18% faster if it was the ghz card. so yes its much better than what that site I link to was. Toms used NO AA at all in BF 3 though, only high settings in Crysis 2, only 2x MSAA in Max Payne, only high settings and NO AA in Metro 2033. crank those settings up and you will see the gtx660ti lose a lot of performance and probably come down to about what we see in the link I provided earlier.
 
cannondale06 said:
I never pay much attention to synthetic benchmarks though. I typically only use them to make sure my card is performing like it should.
Click to expand...

Well hopefully some kind 7970 owner will post their scores using Sleeping Dogs at 25x14/16 with all settings max or thereabouts and hopefully without vsync on. :)

As it stands the closest I could find with vsync disabled is a 670. The scores I posted above at similar settings are nearly double, so that's pretty good as SLI is working as it should with a good overclock on lesser cards:

1we7i.png
 
cannondale06 said:
it ranges anywhere from 10% to 65% faster with average of 31% in the games they tested. that would probably be about 17-18% faster if it was the ghz card. so yes its much better than what that site I link to was. Toms used NO AA at all in BF 3 though, only high settings in Crysis 2, only 2x MSAA in Max Payne, only high settings and NO AA in Metro 2033. crank those settings up and you will see the gtx660ti lose a lot of performance and probably come down to about what we see in the link I provided earlier.
Click to expand...

No doubt it will, but it's not like either setup is doing anything in Metro 2033 with 4xMSAA. They are both sitting at 23 fps. Same for MP3.

Take a minute to dig into the per-game tests on the site - it is a bit more instructive. For example, for the DX11 tests, in BF3 the SLI is 144% of the GHz, Batman 133%, DA2 173%, Crysis 2 no SLI scaling , Dirt Showdown negative SLI scaling, MP3 100% (but unplayable for both), Metro 2033 111% (but unplayable for both), Project Cars no SLI scaling. So in games where SLI seems to work, the SLI looks to be about 30% faster even if you count the two games where the frame rates are in the 20s. Ignore those, and it is 150%.
 
Forceman said:
No doubt it will, but it's not like either setup is doing anything in Metro 2033 with 4xMSAA. They are both sitting at 23 fps. Same for MP3.

Take a minute to dig into the per-game tests on the site - it is a bit more instructive. For example, for the DX11 tests, in BF3 the SLI is 144% of the GHz, Batman 133%, DA2 173%, Crysis 2 no SLI scaling , Dirt Showdown negative SLI scaling, MP3 100% (but unplayable for both), Metro 2033 111% (but unplayable for both), Project Cars no SLI scaling. So in games where SLI seems to work, the SLI looks to be about 30% faster even if you count the two games where the frame rates are in the 20s. Ignore those, and it is 150%.
Click to expand...

I agree on most of your points here, the 660 Ti in SLI is faster than a 7970. I am not going to differentiate between a 7970 and the GHz editions, the GHz edition is just a way for AMD to charge/con a bit more for the same card IMHO. Though your post shows the problems that dual card setups always have somewhere down the line. It doesn't mater if it is SLI or Crossfire, at some point some game will give negative scaling, or poor scaling, or the almost ever-present micro stutter. This leaves a dual GPU user in the pathetic position that he or she has purchased a very expensive paperweight on an alarming number of occasions. Admittedly SLI always fares better here than Crossfire when it comes to finding solutions but these problems still occur a lot.

Anyone who has used dual GPU in the past will have had problems at one point or another. Most will admit that if they could get a card that was within a decent margin of their dual card setup they would have went for it. IMHO what many people are doing here is seeing something new and shiny and purchasing without any real thought.

IMHO purchasing an SLI GTX 660 Ti setup for $600-$650 when a non reference HD 7970 or GTX 670 can be purchased for $400-$420 is not a great move. At least not until some more conclusive reviews are published first. Especially when we consider the inevitable dual GPU issues that will crop up with alarming regularity.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125423

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125413

Even if we are being generous and give the GTX 660 Ti SLI rig a 25% advantage overall, then it just doesn't make financial sense to purchase them given the 50% price increase compared to a single HD 7970 or GTX 670. That's before we even consider the multi GPU headaches that come as standard :)
 
Last edited:
Hopefully Brent is hard at work on a SLI article! A single GTX 670 would be a downgrade from my current CF 6950 setup (2GB, unlocked & OC), if SLI GTX 660s isn't much better then I'll definitely be skipping this generation, just not enough performance per dollar to warrant an upgrade (for me). I think prices of last gen's cards may have spoiled me... Or this gen is still overpriced (relatively) despite recent drops.
 
ICDP said:
IMHO purchasing an SLI GTX 660 Ti setup for $600-$650 when a non reference HD 7970 or GTX 670 can be purchased for $400-$420 is not a great move.
Click to expand...

As I've pointed out, the issue would seem to be not price / performance of 2x 660 SLI vs single card but price / performance of 3x 660 SLI vs 2x 670 / 680 SLI.
 
ICDP said:
I agree on most of your points here, the 660 Ti in SLI is faster than a 7970. I am not going to differentiate between a 7970 and the GHz editions, the GHz edition is just a way for AMD to charge/con a bit more for the same card IMHO. Though your post shows the problems that dual card setups always have somewhere down the line. It doesn't mater if it is SLI or Crossfire, at some point some game will give negative scaling, or poor scaling, or the almost ever-present micro stutter. This leaves a dual GPU user in the pathetic position that he or she has purchased a very expensive paperweight on an alarming number of occasions. Admittedly SLI always fares better here than Crossfire when it comes to finding solutions but these problems still occur a lot.

Anyone who has used dual GPU in the past will have had problems at one point or another. Most will admit that if they could get a card that was within a decent margin of their dual card setup they would have went for it. IMHO what many people are doing here is seeing something new and shiny and purchasing without any real thought.

IMHO purchasing an SLI GTX 660 Ti setup for $600-$650 when a non reference HD 7970 or GTX 670 can be purchased for $400-$420 is not a great move. At least not until some more conclusive reviews are published first. Especially when we consider the inevitable dual GPU issues that will crop up with alarming regularity.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125423

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125413

Even if we are being generous and give the GTX 660 Ti SLI rig a 25% advantage overall, then it just doesn't make financial sense to purchase them given the 50% price increase compared to a single HD 7970 or GTX 670. That's before we even consider the multi GPU headaches that come as standard :)
Click to expand...

I agree - we need to see some more reviews to find the true story. I don't know whether the SLI problems that computerbase had with Crysis 2, Dirt: Showdown, and Project Cars are normal or not - anyone know if SLI works with those games normally? Might just be a 660 Ti driver quirk.

I think the [H] testing is going to be instructive - it seems like the times the 660 Ti SLI rig falls down against the 7970/680 is in situations where neither configuration will give you playable frame rates. When you look at lower settings the SLI configuration seems to take a pretty commanding lead, but if you can't convert that into higher settings I'm not sure it is worth it. Might be a really good solution for 1080p/120Hz though.
 
Last edited:
ICDP, I would agree with you on Crossfire being problematic at least in my experience. I had nothing but problems using a 4870x2 back in the day, but I did still love that card because when it worked, it worked very well. Though so far using a GTX 680/660 SLI setup (granted, I've only been using the former for a couple months and the latter for an evening) my only real issue has been with heat generating from the bottom card and sending the top card about 10-15 degrees higher, that's about it. Sleeping Dogs was easy to get going in SLI with a simple modification to its profile using NV Inspector before the newest beta driver release even though NV is usually pretty good about getting profiles out for SLI. Not bad at all for a brand new software release, certainly a lot easier than my Crossfire days. Hopefully that pattern will continue, but even though I decided to have some fun overclocking and testing 2 SLI setups and liked what I saw especially on the 660 front, I agree that a single card is the best solution in most situations.

Though if someone did want an SLI setup come hell or high water, I think I've demonstrated pretty clearly over some limited preliminary tests that the 660 Ti SLI with a good overclock in general is only 6-13 fps average behind a 680 SLI with a moderate overclock even at a high resolution with 8x AA. Not too shabby if you're spending $600 vs. $1000 or $800 for 670s, though the deal would be better if the 660 Ti cards were priced at around $250. I don't think anyone would be complaining too much then.

Ultimately I'll probably stick with a single overclocked 680 or maybe even a single 660 with a high o/c if I can get one of the cards to really go far as I've started to realize I just don't really see a difference between FXAA/SMAA/MSAA 4x/8x at 24x14 resolution and just want to be able turn up every other setting to max. At 1080p the story might be different because AA matters more there, so YMMV.

On a side note, the Zotac 660 Ti is just about the most precious gaming card I've ever seen! :D Loud little SOB, but adorable.
 
Forceman said:
I agree - we need to see some more reviews to find the true story. I don't know whether the SLI problems that computerbase had with Crysis 2, Dirt: Showdown, and Project Cars are normal or not - anyone know if SLI works with those games normally? Might just be a 660 Ti driver quirk.

I think the [H] testing is going to be instructive - it seems like the times the 660 Ti SLI rig falls down against the 7970/680 is in situations where neither configuration will give you playable frame rates. When you look at lower settings the SLI configuration seems to take a pretty commanding lead, but if you can't convert that into higher settings I'm not sure it is worth it. Might be a really good solution for 1080p/120Hz though.
Click to expand...

Not sure what's up with Dirt Showdown as each card seems to be utilized yet the score is kind of low on both the 680/660 SLI setups but still playable. Probably needs a profile tweak.
 
convexion said:
ICDP, I would agree with you on Crossfire being problematic at least in my experience. I had nothing but problems using a 4870x2 back in the day, but I did still love that card because when it worked, it worked very well.
Click to expand...

I know what you mean re Crossfire, I had 2x 6970s in CF and when they worked they were great, but far too often I had to disable CF to get games working. My last SLI rig was 2x GTX 460s and they fared much better than my AMD CF setup for compatibility. Having said that I got shot of my CF setup before the latest AMD driver with customisable CF profiles, so CF compatibility may have improved..

convexion said:
Though so far using a GTX 680/660 SLI setup (granted, I've only been using the former for a couple months and the latter for an evening) my only real issue has been with heat generating from the bottom card and sending the top card about 10-15 degrees higher.
Click to expand...

I have to admit I benefited greatly from this very issue. A former colleague who I still keep in touch with built himself a GTX 680 SLI rig and he was unhappy after a month or so of use because the top card overheated causing his substantial OC to take a dump. His top card was hitting 85c which meant the card wouldn't overclock and since cards in SLI will only run at the slowest cards speed it meant the lower card would not overclock either. He sold them both to purchase a GTX 690 and I nabbed one of his stock EVGA 680s for £300. :)

convexion said:
Though if someone did want an SLI setup come hell or high water, I think I've demonstrated pretty clearly over some limited preliminary tests that the 660 Ti SLI with a good overclock in general is only 6-13 fps average behind a 680 SLI with a moderate overclock even at a high resolution with 8x AA. Not too shabby if you're spending $600 vs. $1000 or $800 for 670s, though the deal would be better if the 660 Ti cards were priced at around $250. I don't think anyone would be complaining too much then.
Click to expand...

This has always been the case for GPU prices. The top end performance has rarely been worth the premium. The GTX 570 or HD 6970 were no more than 15% slower than a GTX 580 but they were 33%-40% cheaper.

convexion said:
Ultimately I'll probably stick with a single overclocked 680 or maybe even a single 660 with a high o/c if I can get one of the cards to really go far as I've started to realize I just don't really see a difference between FXAA/SMAA/MSAA 4x/8x at 24x14 resolution and just want to be able turn up every other setting to max. At 1080p the story might be different because AA matters more there, so YMMV.

On a side note, the Zotac 660 Ti is just about the most precious gaming card I've ever seen! :D Loud little SOB, but adorable.
Click to expand...

I see you have the same problem I have, I always have this urge to test almost every card that is released. It's not a cheap addiction to have, but at least it's not unhealthy like drink or drugs :)
 
Last edited:
ICDP said:
I see you have the same problem I have, I always have this urge to test almost every card that is released. It's not a cheap addiction to have, but at least it's not unhealthy like drink or drugs :)
Click to expand...

Too true (but I don't mind a drink every once in a while ;)), and nice price on the 680.
 
convexion said:
Just got my 2 Galaxy 660 Ti cards in the mail and have done some preliminary tests to compare to my Gigabyte GTX 680 SLI setup. After clocking the 660's to around 1250/7000 there is about 6-13 FPS difference in average FPS so far in Sleeping Dogs, Heaven, and Dirt Showdown at 8x MSAA with my 680 SLI setup which was at 1215/6500. My monitor is 2560x1440 BTW. Just wanted to let people know the cards overclock well especially on VRAM which seems to somewhat negate the crippled memory bus.

Here's a little something I've capped so far:

Everything is max and the same in all tests for both sets of cards, 8x MSAA in Dirt Showdown with Ultra settings, 8x MSAA in Batman AC all settings max (edit: Physx high on both, BTW).

680 SLI













660 Ti SLI











It seems the only scores to suffer noticeably at 8x AA are maximums but averages and minimums are pretty good and fairly competitive with the 680 SLI setup at highest settings with current gen games especially considering the $370 price difference.

So far I'm pretty impressed with what the Galaxy 3GB models are able to accomplish.
Click to expand...

Convex,

You can test a single GTX 680 vs 660ti sli at 1920x1080?

The same programs and games that you tested
 
Here's a quick one for Heaven & Metro, both o/c.

660 Ti SLI







Single 680





 
Last edited:
I was lucky enough to get one Galaxy 660 Ti that boosts very well to a stable 1327. Comparing that to my 680 that maxed at 1300 is interesting, the performance is virtually the same in Sleeping Dogs. Big enough for you, hdgamer? ;)

700f55e0.png
 
Last edited:
convexion said:
I was lucky enough to get one Galaxy 660 Ti that boosts very well to a stable 1327. Comparing that to my 680 that maxed at 1300 is interesting, the performance is virtually the same in Sleeping Dogs. Big enough for you, hdgamer? ;)

700f55e0.png
Click to expand...

Wowzas I hope my 660 OCes that well! Did you touch the mem clock at all with yours? I want my mem clock to get to at least 6600, I've heard they can get to 6800 and with that kind of OC, it pretty much negates the 192 bit bus speed. Even able to beat 670's at that point.
 
Velazkid said:
Wowzas I hope my 660 OCes that well! Did you touch the mem clock at all with yours? I want my mem clock to get to at least 6600, I've heard they can get to 6800 and with that kind of OC, it pretty much negates the 192 bit bus speed. Even able to beat 670's at that point.
Click to expand...
6800 does not even come close to negating the 192 bit bus. it would take 8000 with 192 bit bus to match 6000 with 256 bit bus.

and there must be something wrong if the 680 and 660 get the same performance. also that is better than his earlier 660 sli results.
 
Thanks for posting those results convexion

I assume they are both SLI results? If so could you try in single card mode please. Apparently Sleeping Dogs is not scaling well in SLI and in fact as a gaming evolved game it is not good a good tool to compare Nvidia hardware. I suggest trying a more common and more stable game with a confirmed working SLI profile to get a more accurate comparison between your 660Ti SLI and GTX 680 SLI setup.

Even if you don't post the result here at least you will have a better idea for your own reference :)
 
Last edited:
Velazkid said:
Wowzas I hope my 660 OCes that well! Did you touch the mem clock at all with yours? I want my mem clock to get to at least 6600, I've heard they can get to 6800 and with that kind of OC, it pretty much negates the 192 bit bus speed. Even able to beat 670's at that point.
Click to expand...

Yes, both cards I got easily clocked to 7000 without issue.
 
cannondale06 said:
6800 does not even come close to negating the 192 bit bus. it would take 8000 with 192 bit bus to match 6000 with 256 bit bus.

and there must be something wrong if the 680 and 660 get the same performance. also that is better than his earlier 660 sli results.
Click to expand...

It's not at all better than the SLI results, not sure which test you're looking at unless you're thinking of the extreme vs. high benches posted earlier. Here's the SLI result at the same settings again:

4aa30617.jpg
 
Last edited:
You must log in or register to reply here.
Back
Top