Sign in with
Sign up | Sign in
Your question

GTX 460 to GTX 660---Will I see much difference?

Last response: in Graphics & Displays
Share
November 27, 2012 6:51:45 AM

I want to upgrade but will I see much difference between a GTX 460 and GTX 660?
For gaming, especially upcoming games like Crysis 3, Far Cry 3 etc.
1920x1080
TIA

> Mainboard : Asus SABERTOOTH X58
> Chipset : Intel X58
> Processor : Intel Core i7 950 @ 3066MHz
> Physical Memory : 6144MB (3 x 2048 DDR3-SDRAM )
> Video Card : NVIDIA GeForce GTX 460
> Hard Disk : SAMSUNG HD103SJ ATA Device (1000GB)
> Hard Disk : Seagate ST31000528AS ATA Device (1000GB)
> Hard Disk : Western Digital WD5000YS-01MPB1 ATA Device (500GB)
> Hard Disk : SAMSUNG HD204UI (2000GB)
> DVD-Rom Drive : Toshiba-Samsung CDDVDW SH-S223C
> Monitor Type : VE247 - 24 inches
> Network Card : Realtek Semiconductor RTL8169/8110 Gigabit Ethernet Adapter
> Operating System : Windows 7 Home Premium Home Edition Media Center 6.01.7601 Service Pack 1 (64-bit)
> DirectX : Version 11.00
> Windows Performance Index : 5.9 on 7.9
>M-Audio Profire 2626 driver version: 5.10.0.5082
a c 87 U Graphics card
November 27, 2012 7:04:55 AM

There will be a difference and you'll probably see it. Whether or not it's enough of a difference for the cost is not something that I'm sure of.
m
0
l
November 27, 2012 7:08:30 AM

Thank you! I'm assuming that my other hardware is not holding me back....much.
m
0
l
Related resources
a c 87 U Graphics card
November 27, 2012 7:14:18 AM

The rest of your hardware is pretty good. Not great, but decent.
m
0
l
a b U Graphics card
November 27, 2012 7:15:27 AM

Your other hardware is good and 660 would be a big jump framerate wise.

I noticed a massive difference going from a 460 to a 560ti and a 660 is about 20% faster than a 560ti.
m
0
l
November 27, 2012 7:48:30 AM

Ofcourse you will see a big difference :) 
m
0
l
November 27, 2012 8:18:29 AM

Have you thought about getting a ssd in to your system as well ?, it will help with load times and over all performace.
m
0
l
November 27, 2012 9:50:20 AM

You will see a massive difference! gtx 660 is borderline equivilant to GTX 580, very good card, cant go wrong with that =). Playing on 1080p monitor is perfect, you will be at a constant high fps range with minor dips for all current titles (New medal of honor being the most gpu demanding). A worthy upgrade, id probably oc that cpu a little as well, im running the same cpu @4.5ghz, decreases bottlenecking a fair bit.

http://www.videocardbenchmark.net/high_end_gpus.html

That should give you some idea of the difference in terms of how powerful the card truly is.
m
0
l
a c 87 U Graphics card
November 27, 2012 10:01:33 AM

azed3000 said:
You will see a massive difference! gtx 660 is borderline equivilant to GTX 580, very good card, cant go wrong with that =). Playing on 1080p monitor is perfect, you will be at a constant high fps range with minor dips for all current titles (New medal of honor being the most gpu demanding). A worthy upgrade, id probably oc that cpu a little as well, im running the same cpu @4.5ghz, decreases bottlenecking a fair bit.

http://www.videocardbenchmark.net/high_end_gpus.html

That should give you some idea of the difference in terms of how powerful the card truly is.


3dmark and other such synthetics are very inaccurate ways of testing cards and those results are outdated anyway. Testing games in FPS is already too inaccurate, going for an outdated and entirely synthetic test is just asking to get the wrong idea.

For further clarification, I just looked at the link and the placements are almost entirely wrong.
m
0
l
November 27, 2012 10:53:28 AM

blazorthon said:
3dmark and other such synthetics are very inaccurate ways of testing cards and those results are outdated anyway. Testing games in FPS is already too inaccurate, going for an outdated and entirely synthetic test is just asking to get the wrong idea.

For further clarification, I just looked at the link and the placements are almost entirely wrong.


That test was updated 27th november 2012=p The placements of the dual gpu cards are wrong (since the benchmark test for these results utlise single gpus) but it still gives you a rough estimate on where the cards lie in terms of performance. However i do stand by my statement by saying there will be a large increase in performance from a gtx 460 to 660, and that the benchmarks are indeed upto date. Reading reviews and fps readings also are a good way to determine the difference when upgrading to a better card, it helps make the decision when it comes to purchasing a card, there's never going to be an accurate measure.
m
0
l
a c 87 U Graphics card
November 27, 2012 10:59:20 AM

azed3000 said:
That test was updated 27th november 2012=p The placements of the dual gpu cards are wrong (since the benchmark test for these results utlise single gpus) but it still gives you a rough estimate on where the cards lie in terms of performance. However i do stand by my statement by saying there will be a large increase in performance from a gtx 460 to 660, and that the benchmarks are indeed upto date. Reading reviews and fps readings also are a good way to determine the difference when upgrading to a better card, it helps make the decision when it comes to purchasing a card, there's never going to be an accurate measure.


If I missed anything saying that it was updated just today, then my bad. However, it is still wrong and I'm not just talking about dual-GPU cards. Placements of most of the single GPU cards are all messed up.

FPS reading is only somewhat marginally better than a synthetic benchmark. Frame latency tests are much closer to real-world than FPS could ever be. FPS hides several types of stutter, frame rate variation, and more. Looking at average frame latency, an aggregate frame latency chart, and a few tables with percentage of time spent over 50ms frame latency, 33.3ms frame latency, 16.7ms frame latency, and 8.33ms frame latency is what really gets into the real-world performance metrics. Unfortunately, most reviews do not take this into account (how many reviews have you read where the reviewers are *mystified* by higher FPS not necessarily meaning smoother experience nor even as smooth of an experience compared to setups with less FPS?) and are thus next to useless. One of the few examples of reviewers that measure things properly is Tech Report.

I agree that the 660 is a big leap over the 460, but how much of a leap greatly depends on the model. Also, the 660 is closer to the 570 than it is to the 580. The 660 Ti is much closer to the 580 than the 660 is and even then, their messed-up memory interfaces are not something that I particularly like.
m
0
l
a b U Graphics card
November 27, 2012 11:23:55 AM

460 is a good card even today. A friend sold off his 660ti within a week and went back to 460 SLI @ 1920x1080 he did not see enough gain.
m
0
l
November 27, 2012 11:26:59 AM

blazorthon said:
If I missed anything saying that it was updated just today, then my bad. However, it is still wrong and I'm not just talking about dual-GPU cards. Placements of most of the single GPU cards are all messed up.

FPS reading is only somewhat marginally better than a synthetic benchmark. Frame latency tests are much closer to real-world than FPS could ever be. FPS hides several types of stutter, frame rate variation, and more. Looking at average frame latency, an aggregate frame latency chart, and a few tables with percentage of time spent over 50ms frame latency, 33.3ms frame latency, 16.7ms frame latency, and 8.33ms frame latency is what really gets into the real-world performance metrics. Unfortunately, most reviews do not take this into account (how many reviews have you read where the reviewers are *mystified* by higher FPS not necessarily meaning smoother experience nor even as smooth of an experience compared to setups with less FPS?) and are thus next to useless. One of the few examples of reviewers that measure things properly is Tech Report.

I agree that the 660 is a big leap over the 460, but how much of a leap greatly depends on the model. Also, the 660 is closer to the 570 than it is to the 580. The 660 Ti is much closer to the 580 than the 660 is and even then, their messed-up memory interfaces are not something that I particularly like.


i agree with you 100% I personally experience some chuggyness with my setup even though i get high fps readings. Latency and frame timings are also more important, believe it or not i limited my frames to 40 fps on my 1080p monitor and got much smoother gameplay then when i was running at 65+ (could also be from slight dips). However yes i understand what youre saying, but for the purpose of this persons question i suggested he look at fps readings and benchmarks to help him make his decision. 660 actually surpasses 570 and is inbetween 570 and 580. 660ti would be extremly close if not surpass 580, and would overclock much better. I also agree about their memory interface, even their gtx 680 interface is kinda limited. But yea youre 100% right, most people dont look at latency cause its harder to benchmark and for people to understand.
m
0
l
a b U Graphics card
November 27, 2012 11:37:38 AM

azed3000 said:
i agree with you 100% I personally experience some chuggyness with my setup even though i get high fps readings. Latency and frame timings are also more important, believe it or not i limited my frames to 40 fps on my 1080p monitor and got much smoother gameplay then when i was running at 65+ (could also be from slight dips). However yes i understand what youre saying, but for the purpose of this persons question i suggested he look at fps readings and benchmarks to help him make his decision. 660 actually surpasses 570 and is inbetween 570 and 580. 660ti would be extremly close if not surpass 580, and would overclock much better. I also agree about their memory interface, even their gtx 680 interface is kinda limited. But yea youre 100% right, most people dont look at latency cause its harder to benchmark and for people to understand.



660ti is already a decent but faster than the 580.. BUT the 580 overclocks better yet still is not as fast.

660 oc's really well
m
0
l
a c 87 U Graphics card
November 27, 2012 12:14:13 PM

squirrelonfire said:
Yes but little difference is unlikely to be noticeable

http://www.hwcompare.com/13329/geforce-gtx-460-2gb-vs-g...


HWCompare is usually nearly useless for comparisons. They are as far from real-world as you can get with most of their measurements not even being synthetic benchmarks, rather simply listing rated specifications without taking pretty much everything into account.

The difference between the worst 660 and the best 460 is probably noticeable for most gamers. Whether or not the 660 is a worthwhile upgrade over that is a more debatable thing because some 460 models can get fairly close to the 660. Still a ways off in performance, but close enough to make it fairly similar (some 460s meet or beat even a reference 560 Ti) for an upgrade to be questionable. If it's a reference 460 256 bit or even worse, a 460 192 bit or 460 SE, then the upgrade will undoubtedly be worth the money IMO.

If overclocking is taken into account, then it is worth the money regardless of which 460 is owned, although at that point, Nvidia takes a serious back-seat to AMD.
m
0
l
a b U Graphics card
November 27, 2012 2:58:12 PM

blazorthon said:
some 460s meet or beat even a reference 560 Ti


Can you post a benchmark link demonstrating that?

GTX660 will deliver a huge benefit - I'd estimate over 50% higher framerates on average. Couldn't say about the frame latency stuff, sounds interesting though. Not the first time I've seen frame/second criticised as an accurate indicator of performance - hopefully it's something the tech media will take notice of.
m
0
l
a b U Graphics card
November 27, 2012 3:17:51 PM

sam_p_lay said:
Can you post a benchmark link demonstrating that?

GTX660 will deliver a huge benefit - I'd estimate over 50% higher framerates on average. Couldn't say about the frame latency stuff, sounds interesting though. Not the first time I've seen frame/second criticised as an accurate indicator of performance - hopefully it's something the tech media will take notice of.


EVGA SSC+ or MSI talon hawk
m
0
l
a c 87 U Graphics card
November 27, 2012 3:21:18 PM

sam_p_lay said:
Can you post a benchmark link demonstrating that?

GTX660 will deliver a huge benefit - I'd estimate over 50% higher framerates on average. Couldn't say about the frame latency stuff, sounds interesting though. Not the first time I've seen frame/second criticised as an accurate indicator of performance - hopefully it's something the tech media will take notice of.


The difference between the weakest 460 and the best 460 is probably something like 40-50%. The 460 has huge variance in performance. It's one of the reasons for why I'm against using it as a vantage point for measuring performance improvement when we're not even given more info on it. There are so many versions of it and even among the same version, frequencies can have huge variance, so it's a difficult card to work with without more info about OP's exact model such as the frequencies and memory interface.

No benchmark that I'm likely to find is going to be any newer than several years old, so there's not much point in looking IMO, but if you really want me to, I'l take a look and see what I can find.
m
0
l
November 27, 2012 3:32:47 PM

Yes, you would notice a difference! Not only performance wise, but with lower heat and lower power consumption as well. Not to mention, the Kepler chips overclock quite nicely. A 660 would compliment your system nicely. Some people are saying that you won't notice much of a difference, I respectfully disagree. You might not get a lot of extra FPS, but you will get consistently good FPS (not the same thing). For example, some people want the highest possible FPS, but then they enable vsync which essentially locks the FPS to 60 (or 30, if the GPU can't maintain 60), but, if your GPU is good, it will maintain that 60 FPS, which is what you want. There is a way WAY better chance of the 660 being able to do that compared to a 460. And I have to say it again: less heat and better power efficiency. It matters!
m
0
l
a b U Graphics card
November 27, 2012 3:34:31 PM

Upgrading from a 460 to 660 will surely give you a noticeable performance boost.A 460 performs similar to a hd 5850.So considering that a 660 can make an unplayable game playable for you.Here's the bench --

m
0
l
a c 87 U Graphics card
November 27, 2012 3:44:39 PM

Anik8 said:
Upgrading from a 460 to 660 will surely give you a noticeable performance boost.A 460 performs similar to a hd 5850.So considering that a 660 can make an unplayable game playable for you.Here's the bench --
http://www.guru3d.com/index.php?ct=articles&action=file&id=1280


Some 460s are right around the 5850, some are a little weaker. Some perform more like the 6950 (bringing the average performance improvement from around 50% as claimed earlier to a much more modest less than 20%) and there are others everywhere between the extremes. Why are several of you people generalizing the 460? There are five versions of the 460 (at least, there might be a sixth or seventh that I'm missing), each with non-trivial performance differences, and they can each have widely varying frequencies for the GPU and memory, further mucking it all up.

I'm not at all against the upgrade, just putting the 460's performance variation into perspective.
m
0
l
a c 225 U Graphics card
November 27, 2012 3:59:05 PM

http://www.tomshardware.com/reviews/gaming-graphics-car...

As you can see, the 460 is three tiers below the 660.

Quote:
I don’t recommend upgrading your graphics card unless the replacement card is at least three tiers higher. Otherwise, the upgrade is somewhat parallel and you may not notice a worthwhile difference in performance.


m
0
l
a b U Graphics card
November 27, 2012 4:04:10 PM

Anik8 said:
Upgrading from a 460 to 660 will surely give you a noticeable performance boost.A 460 performs similar to a hd 5850.So considering that a 660 can make an unplayable game playable for you.Here's the bench --
http://www.guru3d.com/index.php?ct=articles&action=file&id=1280



You missed the point blazorthon was making 460 was base clock of 675MHz yes comparable with 5850
but the really highly clocked ones coming to from the factory at 850MHz + overclock gets up there just the same as the 560 does when overclocked.
m
0
l
a b U Graphics card
November 27, 2012 4:24:35 PM

@blazorthon and @spentshells
So... do we know how much overclocked a 460 is being used by the OP?
And the same thing can be said about the 660,a highly clocked 660 like the 660 hawk performs faster than the 7870 in a majority of titles even if you take latest catalyst drivers to consideration.
m
0
l
a c 87 U Graphics card
November 27, 2012 4:32:08 PM

We're been talking about stock cards. Some 460s run at over 900MHz GPU frequencies at stock and that's huge enough for them to get very close to the 660, making it less of an upgrade without overclocking. I've already specified that with overclocking, the 660 can be brought high enough to alleviate this. Telling me what I've already said changes nothing.

Also, the 660 Hawk doesn't generally best similarly priced 7870s. The 7870 Hawk puts the 660 Hawk to shame in value with a small performance win and a significant freebie win and there's a cheaper yet similarly performing Gigabyte 7870 and another even faster 7870, I forget the brand.

OP has yet to specify his/her 460's memory interface, GPU frequency, and memory frequency, so IDK how it stacks up and won't make direct claims of how much of an improvement the 660 will be until OP gives such info. I've simply been trying to explain why such info is important because of how much variance in performance there is with the 460.

Like I said earlier, I'm not against the upgrade, just unable to make honest and accurate estimations of the performance improvement.
m
0
l
a c 87 U Graphics card
November 27, 2012 4:42:14 PM

dingo07 said:
how about factory OC 460's in SLI like the ones I have - how do they fare in this debate y'all r havin...??? 1920x1200


That'll mostly depend on their memory interface's width, memory frequency, and GPU frequency.
m
0
l
November 27, 2012 4:48:59 PM

You will see a difference, do not know if its worth the money though. I kind adoubt that you will see a bottleneck though with ur i7. I have a 3770k and the 660 bottlenecks the cpu instead of otherway around.
m
0
l
a b U Graphics card
November 27, 2012 4:49:04 PM

The sc versions are 763 or 761 MHz faster than a 660 anyway as they are faster than a 580
m
0
l
a b U Graphics card
November 27, 2012 4:53:38 PM

Alright short answer: hell yes.

And yes, please post those benchmarks. I don't care if they're old - we're still comparing different instances of the same model. If you can show GTX460s outperforming other GTX460s by 50% then I'll be absolutely blown away.
m
0
l
a b U Graphics card
November 27, 2012 5:01:25 PM

I say wait til next year... Then you might be able to snag a 256-bit Kepler card in the $200 range at about the GTX 580/660Ti level. Personally, I'm waiting for the GTX 760 - not impressed with these 192-bit cards this time around...

Your current card is still good. Good enough to play Crysis 3 on High (judging from the Alpha). And by next year, all the patches for those games will be out, too - you can get a new card and bump the game up to Ultra, and that's always fun.
m
0
l
a b U Graphics card
November 27, 2012 5:05:54 PM

jessterman21 said:
I say wait til next year... Then you might be able to snag a 256-bit Kepler card in the $200 range at about the GTX 580/660Ti level. Personally, I'm waiting for the GTX 760 - not impressed with these 192-bit cards this time around...

Your current card is still good. Good enough to play Crysis 3 on High (judging from the Alpha). And by next year, all the patches for those games will be out, too - you can get a new card and bump the game up to Ultra, and that's always fun.


+1 to this (except maybe the current obsession with memory bus width which is highly debatable...) - I'm on a crappy old backup card since my main card died, but I'm gonna hold out until something like GTX770. PC gaming has such an awesome catalogue of games going right back over 20 years that you're not short of excellent games you can play in the meantime. I'm personally gonna fire up DarkStar One (2006) after I finish with Borderlands 2 and I know I'm gonna love it :-D Plus DOSBox! I know that's maybe an extreme, but you can honestly still have a lot of fun with games like Heretic, Lemmings, Flashback, Syndicate Wars...
m
0
l
a c 87 U Graphics card
November 27, 2012 5:18:20 PM

sam_p_lay said:
Alright short answer: hell yes.

And yes, please post those benchmarks. I don't care if they're old - we're still comparing different instances of the same model. If you can show GTX460s outperforming other GTX460s by 50% then I'll be absolutely blown away.


EVGA GTX 460 FTW has an ~29% memory bandwidth advantage over the reference GTX 460 192 bit and an ~26% GPU frequency advantage over reference. I'll see if I can find any benchmarks, but the math is easy to do without them anyway at around 40-60% higher performance. There are a few other models that are almost as fast as it such as the Gainward GTX 460 GS-GLH and KFA2 GeForce GTX 460 LTD OC.

At least we're not counting GTX 460 SE and 465 in this, they'd widen the gap in both directions even more.
m
0
l
a b U Graphics card
November 27, 2012 5:31:32 PM

blazorthon said:
EVGA GTX 460 FTW has an ~29% memory bandwidth advantage over the reference GTX 460 192 bit and an ~26% GPU frequency advantage over reference. I'll see if I can find any benchmarks, but the math is easy to do without them anyway at around 40-60% higher performance. There are a few other models that are almost as fast as it such as the Gainward GTX 460 GS-GLH and KFA2 GeForce GTX 460 LTD OC.

At least we're not counting GTX 460 SE and 465 in this, they'd widen the gap in both directions even more.


You can't just add percentages together like that. And you know it's pretty much never the case that a 10% clock speed gain equals a 10% framerate gain? Typically you get a lot less. This kind of speculation is totally pointless - it's exactly the reason we have benchmarks. No need for speculation on hypothetical performance gains when you can just look at real results.
m
0
l
a c 87 U Graphics card
November 27, 2012 5:53:57 PM

sam_p_lay said:
You can't just add percentages together like that. And you know it's pretty much never the case that a 10% clock speed gain equals a 10% framerate gain? Typically you get a lot less. This kind of speculation is totally pointless - it's exactly the reason we have benchmarks. No need for speculation on hypothetical performance gains when you can just look at real results.


I didn't add percentages and you can in fact do math with them like that. It's easier for me with the newer cards because I have more practice with my math on them (which has very accurately predicted the performance of every Kepler card launched since the 680), but I can work with older cards. I guarantee that ~40% is a very realistic number as approximately the most common minimum and up to around 60% in some situations.

Also, if you read my numbers, you'll notice that I didn't treat a 10% GPU nor memory frequency gain as a 10% performance gain.

http://www.tomshardware.com/reviews/evga-geforce-gtx-46...

Even compared to another slightly factory overclocked GTX 460 256 bit, this model managed an average lead of about 14-28%. It only has a 20% GPU frequency advantage and an ~11% memory bandwidth advantage over that model. That's with very old drivers, new ones increase the *overclocking* efficiency a little compared to way back then.
m
0
l
November 27, 2012 6:05:08 PM

Thanks to all, I'm learning allot.
"OP has yet to specify his/her 460's memory interface, GPU frequency, and memory frequency"
Sorry, I just barely know what those thing are.
I already have a solid state drive, I'm just now adding it to my list.

> Mainboard : Asus SABERTOOTH X58
> Chipset : Intel X58
> Processor : Intel Core i7 950 @ 3066MHz
> Physical Memory : 6144MB (3 x 2048 DDR3-SDRAM )
> Video Card : NVIDIA GeForce GTX 460
> Hard Disk : SAMSUNG HD103SJ ATA Device (1000GB)
> Corsair 120 gig SSD
> Hard Disk : Seagate ST31000528AS ATA Device (1000GB)
> Hard Disk : Western Digital WD5000YS-01MPB1 ATA Device (500GB)
> Hard Disk : SAMSUNG HD204UI (2000GB)
> DVD-Rom Drive : Toshiba-Samsung CDDVDW SH-S223C
> Monitor Type : VE247 - 24 inches
> Network Card : Realtek Semiconductor RTL8169/8110 Gigabit Ethernet Adapter
> Operating System : Windows 7 Home Premium Home Edition Media Center 6.01.7601 Service Pack 1 (64-bit)
> DirectX : Version 11.00
> Windows Performance Index : 5.9 on 7.9
>M-Audio Profire 2626 driver version: 5.10.0.5082
m
0
l
a c 87 U Graphics card
November 27, 2012 6:09:01 PM

Well, if you have a model number for your 460, I can find the needed info myself. You can also find it if you download the tool called GPU-Z (shouldn't need to be installed, but it might ask if you want it to). It has memory capacity, interface width, frequency, GPU frequency, and much more all listed in plain text and has captions to describe things a little further (although I think that some of the explanations leave a little to be desired).

However, it's really only important if you want an accurate presentation of what sort of performance jump to expect. If you're happy knowing that it should at least be a noticeable improvement, then don't worry about this.
m
0
l
a b U Graphics card
November 27, 2012 6:23:53 PM

sam_p_lay said:
Alright short answer: hell yes.

And yes, please post those benchmarks. I don't care if they're old - we're still comparing different instances of the same model. If you can show GTX460s outperforming other GTX460s by 50% then I'll be absolutely blown away.


2 years ago that would not have been an issue if you are allowing for overclocking
m
0
l
a b U Graphics card
November 27, 2012 6:28:45 PM

blazorthon said:
I didn't add percentages and you can in fact do math with them like that. It's easier for me with the newer cards because I have more practice with my math on them (which has very accurately predicted the performance of every Kepler card launched since the 680), but I can work with older cards. I guarantee that ~40% is a very realistic number as approximately the most common minimum and up to around 60% in some situations.

Also, if you read my numbers, you'll notice that I didn't treat a 10% GPU nor memory frequency gain as a 10% performance gain.

http://www.tomshardware.com/reviews/evga-geforce-gtx-46...

Even compared to another slightly factory overclocked GTX 460 256 bit, this model managed an average lead of about 14-28%. It only has a 20% GPU frequency advantage and an ~11% memory bandwidth advantage over that model. That's with very old drivers, new ones increase the *overclocking* efficiency a little compared to way back then.


Some decent performance gains there but I'm still not seeing 40-60%. Do you remember what exactly this GTX660-grade GTX460 was? Who made it?
m
0
l
a c 87 U Graphics card
November 27, 2012 6:41:19 PM

Here's a review that compared the EVGA 460 FTW to the 460 768MB (a 192 bit model):
http://www.fudzilla.com/reviews/item/20517-evga-gtx-460...

It averaged out closer to 35% than 40% ahead of the 460 768MB, but like I said above, this was with old drivers, doesn't include the slowest of 460s and maybe not the fastest, so it's potentially a considerably conservative number. Considering that some of the tests in the earlier review already neared 30%, it goes to reason that some of the same tests would reach for about 50% above the 460 768 MB tested in this review.
m
0
l
a c 87 U Graphics card
November 27, 2012 6:42:45 PM

sam_p_lay said:
Some decent performance gains there but I'm still not seeing 40-60%. Do you remember what exactly this GTX660-grade GTX460 was? Who made it?


I never said a GTX 460 that was on-par with the 660. I said that there are some where the performance difference can drop to around 20% on average.
m
0
l
a b U Graphics card
November 27, 2012 6:48:44 PM

blazorthon said:
We're been talking about stock cards. Some 460s run at over 900MHz GPU frequencies at stock and that's huge enough for them to get very close to the 660, making it less of an upgrade without overclocking. I've already specified that with overclocking, the 660 can be brought high enough to alleviate this. Telling me what I've already said changes nothing.

Also, the 660 Hawk doesn't generally best similarly priced 7870s. The 7870 Hawk puts the 660 Hawk to shame in value with a small performance win and a significant freebie win and there's a cheaper yet similarly performing Gigabyte 7870 and another even faster 7870, I forget the brand.

OP has yet to specify his/her 460's memory interface, GPU frequency, and memory frequency, so IDK how it stacks up and won't make direct claims of how much of an improvement the 660 will be until OP gives such info. I've simply been trying to explain why such info is important because of how much variance in performance there is with the 460.

Like I said earlier, I'm not against the upgrade, just unable to make honest and accurate estimations of the performance improvement.

I wasn't going into a 7870 vs 660 debate but was only highlighting its oc advantage by drawing a stock 7870 into comparison.
Moreover before moving into clock speed,bus width and all that we should take into consideration,that the architectures are vastly different and they react differently to frequency,memory,etc.With all honesty do actually think a highly clocked 460 will break the 30 fps barrier in that benchmark.Hitman Absolution is a decent benchmark that includes the best of dx11 with no physx at all.The 560 ti was the replacement for 470 and somehow managed 31 fps.Even playing at 31 fps is not a good idea as the fps won't be pegged at 31 throughout the game.Here the 660 is more safe a solution.
m
0
l
a b U Graphics card
November 30, 2012 4:46:20 PM

Nice! $165 after rebate! We don't get rebates in the UK... and best price you'll find here is £165 - around $250 by exchange rate. That is not fair.
m
0
l
a c 87 U Graphics card
November 30, 2012 4:50:34 PM

Anik8 said:
I wasn't going into a 7870 vs 660 debate but was only highlighting its oc advantage by drawing a stock 7870 into comparison.
Moreover before moving into clock speed,bus width and all that we should take into consideration,that the architectures are vastly different and they react differently to frequency,memory,etc.With all honesty do actually think a highly clocked 460 will break the 30 fps barrier in that benchmark.Hitman Absolution is a decent benchmark that includes the best of dx11 with no physx at all.The 560 ti was the replacement for 470 and somehow managed 31 fps.Even playing at 31 fps is not a good idea as the fps won't be pegged at 31 throughout the game.Here the 660 is more safe a solution.


Kepler and Fermi are not very different architectures. They are extremely similar architecturally.

What's the point of citing all of that ? Yes, a 460 can break 30FPS in that benchmark, no problem. However, that doesn't matter at all. The 660 isn't too far ahead and that's what matters. Besides, we're all relying on these vastly outdated benchmarks. They aren't much good at all for comparing modern performance even with the cards that we're comparing.
m
0
l
a b U Graphics card
December 2, 2012 11:26:54 AM

blazorthon said:
Kepler and Fermi are not very different architectures. They are extremely similar architecturally.

What's the point of citing all of that ? Yes, a 460 can break 30FPS in that benchmark, no problem. However, that doesn't matter at all. The 660 isn't too far ahead and that's what matters. Besides, we're all relying on these vastly outdated benchmarks. They aren't much good at all for comparing modern performance even with the cards that we're comparing.

Did you say outdated? :o  The benchmark is of Hitman Absolution,the latest sequel to the series.The benchmark has been done by Guru3D and the drivers used here are 310.61 and catalyst 12.11 build 8.Please convince me where you find this outdated.Performance wise there is a difference is 11 fps from 25 to 36.If hd 5850 is considered 100%,the 660 is performing at 144% .Furthermore overclocking doesn't help much in below 30s.OCing a 460 to its max may fetch you only 3 fps more and it will still hover in the 20s and make your gameplay unplayable.While a stock 660 averages 6fps higher than the safe limit of 30.That is enough to provide playable gameplay in most demanding titles like this one.
m
0
l
!