Nvidia GeForce GTX 650 And 660 Review: Kepler At $110 And $230

[demonhorde665]

[citation][nom]cleeve[/nom]It's true everyone has unique sensitivities. I can also detect 60 Hz vs 120 Hz pretty instantly, but having said that 60 Hz doesn't bother me. I'm quite happy with 30 FPS minimum and 40 FPS average.Then again, I find 120 Hz motion compensation on televisions irritating as hell when its applied to 24 FPS movies...[/citation]

all so very very true ,

i can't tell the differnce between 20 fps and 30 , but I can see a difference between 30 and 60

usually 24 fps min and 35-40 avaerage is good enough for me to game , on .. now sound i've been a musician for 20 years (play bass) but i have yet to have any hearing damage i can still hear a distant car a mile a while and people talking two rooms away from me.

 

[blazorthon]

[citation][nom]demonhorde665[/nom]also some people choose to be fans of a company based off ethics as well , sure brand X may not bea s good as brand A right now , but at least company X did not go around and pay people to not carry brand A. or brand X didn't try to make a monoploy like brand A did for example. now im not really like this but i know many people that refuse to buy intel because intel has done multiple unethical buisness deals trying to control the cpu market. in fact if intel had not done these dirty deals it's likely AMD would be on a much more level playing field today, fincially, as it stands Amd just doesn't have the finances to compete in reasearch and development.[/citation]

AMD has plenty of finances to compete. If you simply take their FX-8120 as an example, you can raise the CPU/NB frequency from a mere stock 2.2GHz to over 3GHz and use PSCheck to cut-down the P states of the second core of each module and prioritize for the first core of each module, that FX-8120 can now compete with even the i7s in quad and eight threaded performance, no problem. It'll even keep up with the K edition (LGA 1155) i7s in overclocking performance and with more manageable power consumption too. AMD simply continually makes stupid mistakes such as not setting up their CPUs like this by default and that's why they can't compete in the CPU market. Heck, with Llano, they had far more APUs made than chip sets, so they had a huge stock of APUs that they couldn't even sell because there weren't enough motherboards for them. Like I said, stupid mistakes.

Sure, AMD is a far smaller and poorer company than Intel, but they are still a multi-billion dollar company and they have enough money to compete in the x86 CPU markets properly. They simply need to stop screwing everything that they can up at every turn.

 

[crisan_tiberiu]

The question is: is it worth upgrading from a GTX 460 1GB/256bit to a GTX 560 non ti?

 

[blazorthon]

[citation][nom]crisan_tiberiu[/nom]The question is: is it worth upgrading from a GTX 460 1GB/256bit to a GTX 560 non ti?[/citation]

It could be if you sell the old card or play PhysX-heavy settings in supporting games that can use the older 460 as a PhysX processing card so that maxed-out PhysX doesn't affect the 560's gaming performance. So long as Nvidia and AMD primarily ignore the ~150-190 market, the previous gen cards are all that there is unless you count highly factor overclocked and/or 2GB models of the 7770.

 

[crisan_tiberiu]

[citation][nom]blazorthon[/nom]It could be if you sell the old card or play PhysX-heavy settings in supporting games that can use the older 460 as a PhysX processing card so that maxed-out PhysX doesn't affect the 560's gaming performance.[/citation]


Typo: "gtx 660 non ti' not 560 😀, my bad 😛

 

[blazorthon]

[citation][nom]crisan_tiberiu[/nom]Typo: "gtx 660 non ti' not 560 , my bad[/citation]

Well, my answer still applies, just switch 660 and 560. At that price point, I'd rather get a discounted 7870, but as others have said, the 660 is priced properly and is considerable. It would definitely be a worth-while upgrade to go to the 660 or the 7870, especially if the 460 is sold or used as a PhysX processing card (it works even if an AMD card is the master card if you use the right drivers).

 

[tomfreak]

exactly, except GTX670, it seems the entire line up of Kepler are not beating the price/performance of Radeon HD7000. I am kinda disappointed.

 

[mohit9206]

and i had thought that the gtx650 would compete with 7770 but it cant even beat a 7750. nvidia will never win the price to performance battle in entry level segment. but atleast the gtx660 fares good and slots in right where the price point would suggest.

 

[blazorthon]

[citation][nom]Tomfreak[/nom]exactly, except GTX670, it seems the entire line up of Kepler are not beating the price/performance of Radeon HD7000. I am kinda disappointed.[/citation]

Honestly, I like the 660's price/performance so long as the MSAA isn't overloaded on it. Instead of increasing AA, there is TXAA (although I think that it still isn't well-supported) and TXAA on Nvidia versus MSAA on AMD is probably pretty similar in AA efficiency. If TXAA is/gets better supported than I think that it is now, then Nvidia might be able to compete in performance at a given AA quality much better than they do now.

My problem with the 670 is that a Radeon 7950 with a good after-market cooler can take the best 670s on in an OC versus OC comparison quite well, so I'dd rather get a discounted 7950 for $280-320 or even just get two discounted 7870s in CF and really the 670s and even the 7970s and 7970 GHz Editions quite significantly for about the same price.

Honestly, I too am overall disappointed in Nvidia's showing this generation and that's beyond their MSAA woes. They took Fermi, a more or less great architecture, and pretty much chopped out much of what was good about it. Had Nvidia built on it like AMD built on thier previous archs, maybe they would have pulled ahead of AMD.

 

[blazorthon]

[citation][nom]mohit9206[/nom]and i had thought that the gtx650 would compete with 7770 but it cant even beat a 7750. nvidia will never win the price to performance battle in entry level segment. but atleast the gtx660 fares good and slots in right where the price point would suggest.[/citation]

If Nvidia drops the price on the 650 down to the $90-100 then it could be considerable, depending on the games that you play. It lost to the 7750 in three tests, it beat the 7750 similarly significantly to its losses in two tests (in fact, good enough to keep up with the 7770), and it kept up with the 7750 in the remaining three tests. That's not a win over the 7750 overall, but it's not much of a loss. Heck, things might change up a bit with a wider game selection, for better or for worse. I think that I'd like to see Anand's comparison because they often have large game selections, although I'm honestly not expecting much of a difference, just curious to find out for sure.

 

[jaquith]

Video - http://www.youtube.com/watch?v=ol3O9oAPsG0
I recommend you play @ HD and pause occasionally especially where there's motion. The video was running an AMD GPU.

So in games while you are running 8xMSAA with chop and killing a building, your opponents with FXAA High and no chop are shooting you in the head.

The way I dial-in a game is to have my average FPS with quality settings running an average of 70~80FPS+ and resulting in very few if any frames below 40FPS, set Adaptive vSync so if I'm hitting 59FPS or less I'm not being dragged down to 30FPS (Conventional vSync & Stutter) and in the process eliminating tear. This can only be done on an nVidia GPU.

Yes, a Hatchet Job!

 

[blazorthon]

I could just as easily pull up an image showing even 4x MSAA beating FXAA in quality.

 

[jaquith]

Unless you tell us what AMD graphics card was used in that test and what driver version was used , your example to prove Tom's did a "hatchet job" is a "hatchet job" of an example. I could just as easily pull up an image showing even 4x MSAA beating FXAA in quality without any context, such as this, and it wouldn't prove nor disprove any of us because of its lack of context.
http://techgage.com/reviews/gaming/max_payne_3/4xaa_vs_fxaa_thumb.jpg

Did you even look at the video link and description? Sapphire Radeon HD 7970 @ 1.1GHz The image was a simple screen capture, and try it for yourself. Stills and low motion it's very difficult to tell the difference, so unless while gaming you're standing still as a target -- you're going to be moving around. The 8xMSAA created ghosting and also increases the blur.

If the net result in image quality difference is minimal, or indistinguishable, but the frame rates are cut in half then all it proves is MSAA is very inefficient and other forms of AA are more prudent. However, MSAA in motion produces more negative artifacts and kills your frame rate to boot.

BTW - regarding your image add some motion and then decide.

 

[jabliese]

Love the 650. I do not want to use my computer as an extra heater, so GTX 4XX is out of the picture. Of the low power cards, it competes directly with 7750. Now that nVidia has finally come out with a competitor for the 7750, maybe AMD will lower prices. Or maybe I will risk the fanboi label and buy a 650. Either way, things look good for those with an eye on their electric bill.

 

[spentshells]

660 in the wild for 229 great deal

 

[blazorthon]

Did you even look at the video link and description? Sapphire Radeon HD 7970 @ 1.1GHz The image was a simple screen capture, and try it for yourself. Stills and low motion it's very difficult to tell the difference, so unless while gaming you're standing still as a target -- you're going to be moving around. The 8xMSAA created ghosting and also increases the blur.

If the net result in image quality is minimal, or indistinguishable, but the frame rates are cut in half then all it proves is MSAA is very inefficient and other forms of AA are more prudent. However, MSAA in motion produces more negative artifacts and kills your frame rate to boot.

BTW - regarding your image add some motion and then decide.

Ahh sorry, I didn't see the link, my bad.

The FXAA seems a little brighter, but it also looks more out of focus, blurred. I didn't think that it was too difficult at all to tell the difference.

 

[jaquith]

Ahh sorry, I didn't see the link, my bad.

The FXAA seems a little brighter, but it also looks more out of focus, blurred. I didn't think that it was too difficult at all to tell the difference.

If I said the top was FXAA and the bottom was MSAA you're perception would be flip-flopped.

So are you trying to tell me that a 43% frame rate drop is a wise choice quality to quality??

 

[jaquith]

My underlying point is the test parameters are bad, and only prove the nVidia has a problem with anything above 4xMSAA while ignoring it's strengths. In benchmark terms it's an exploit unless other options are compared; there's no comparisons here, and only unsuitable settings for any cards. No gamer in their right mind is going to use them playing.

However and most importantly, if I had an AMD I'd use the best possible settings to produce a stabilized & good quality render. Likewise I'd do the same on an nVidia. Clearly, each has card it's strong suits and the settings would indeed be different (AMD vs nVidia) but the quality the same.

Therefore, I'd look for the best of each and then run a bench and compare as a comprehensive review. Don decided for whatever reason to exploit the worst vs exploit the strengths. Now I can buy that from a noob but not from someone publishing an impartial article.

 

[blazorthon]

If I said the top was FXAA and the bottom was MSAA you're perception would be flip-flopped.

So are you trying to tell me that a 43% frame rate drop is a wise choice quality to quality??

I disagree. I've already shown before that I can distinguish between x4 MSAA and x8MSAA even with your tiny pictures (although x8 MSAA and x32CSAA were too close with those pictures) any you claim that not only is there no discernible difference between x4 and x8 MSAA, but that there is also no difference between FXAA (High) and x4 MSAA? Just because you don't see it doesn't mean that other people don't see it. Also, as Tom's recent benchmarks on MSAA have shown, x8 MSAA doesn't always take nearly that much performance. The Batman AC results showed x8 MSAA being about 85% efficient with the Radeon 7950 and 7870 at 1080p and about 80% efficient at 2560x1440.

 

[spentshells]

My underlying point is the test parameters are bad, and only prove the nVidia has a problem with anything above 4xMSAA.

However and most importantly, if I had an AMD I'd use the best possible settings to produce a stabilized & good quality render. Likewise I'd do the same on an nVidia. Clearly, each has card it's strong suits and the settings would indeed be different but the quality the same.

Therefore, I'd look for the best of each and then run a bench. Don decided for whatever reason to exploit the worst vs exploit the strengths. Now I can buy that from a noob but not from someone publishing an impartial article.

The article comes of as fair.

The 650 is a joke and the 660 is a champ.

Frankly the 660 looks like the best price performance card of all time

 

[jaquith]

I disagree. I've already shown before that I can distinguish between x4 MSAA and x8MSAA even with your tiny pictures (although x8 MSAA and x32CSAA were too close with those pictures) any you claim that not only is there no discernible difference between x4 and x8 MSAA, but that there is also no difference between FXAA (High) and x4 MSAA? Just because you don't see it doesn't mean that other people don't see it. Also, as Tom's recent benchmarks on MSAA have shown, x8 MSAA doesn't always take nearly that much performance. The Batman AC results showed x8 MSAA being about 85% efficient with the Radeon 7950 and 7870 at 1080p and about 80% efficient at 2560x1440.

Answer this then I'll argue your points.

So are you trying to tell me that a 43% frame rate drop is a wise choice quality to quality??

 

[blazorthon]

My underlying point is the test parameters are bad, and only prove the nVidia has a problem with anything above 4xMSAA while ignoring it's strengths. In benchmark terms it's an exploit unless other options are compared; there's no comparisons here, and only unsuitable settings for any cards. No gamer in their right mind is going to use them playing.

However and most importantly, if I had an AMD I'd use the best possible settings to produce a stabilized & good quality render. Likewise I'd do the same on an nVidia. Clearly, each has card it's strong suits and the settings would indeed be different (AMD vs nVidia) but the quality the same.

Therefore, I'd look for the best of each and then run a bench. Don decided for whatever reason to exploit the worst vs exploit the strengths. Now I can buy that from a noob but not from someone publishing an impartial article.

Well, I can't argue with the facts that having more test parameters to compare as well as comparing video quality itself is better than looking at only a few benches set up with a more singular focus on settings. However, I can argue with your demonizing of Tom's and their tests simply because they aren't what you wanted to see.

 

[blazorthon]

Answer this then I'll argue your points.

So are you trying to tell me that a 43% frame rate drop is a wise choice quality to quality??

If the drop was always 43%, then it probably wouldn't be worth it. However, Tom's has clearly shown that at least with current drivers, the drop is far less than a huge 43%, at least in some tests.

http://www.tomshardware.com/reviews/geforce-gtx-660-ti-memory-bandwidth-anti-aliasing,3283-13.html

 

[blazorthon]

[citation][nom]spentshells[/nom]The article comes of as fair.The 650 is a joke and the 660 is a champ. Frankly the 660 looks like the best price performance card of all time[/citation]

Are you just going to ignore the 7870 at the same price point? MSRP might be higher, but many of the actual prices that you can buy them from are lower than the 660s. The 7850 is also far superior to the 660 when overclocking is considered and the 7870 has a considerable advantage over the 7850 in that too. The 660 in't bad at it's price at all, but it's quite bad at overclocking and that is where the true value lies for any overclocking enthusiast. Heck, it's not like the 7870 doesn't hold it's own even at stock and in the same price range anyway.

 

[cleeve]

My underlying point is the test parameters are bad, and only prove the nVidia has a problem with anything above 4xMSAA while ignoring it's strengths.

A test parameter that proves something isn't 'bad'.
It's ideal, actually, unless you're trying to bury the truth because it doesn't sync with your brand preference.

The point of testing is to expose strengths and weaknesses.

Don decided for whatever reason to exploit the worst vs exploit the strengths. Now I can buy that from a noob but not from someone publishing an impartial article.

You're entitled to your opinion, you're welcome to disagree with me all you like. Everyone is welcome to their own perspective.

But if you can't accept a difference of opinion without insults and baseless accusations I'll wipe you off the board for good.
I've been very patient but enough is enough.

Play nice or you're out of here, chief, along with all the posts you've ever written on this forum. 😉