Gtx 650 2GB vs Gtx 660 (non ti)

[Techie107]

Hi I'm looking to buy a new graphics card.

I used to have the Gigabyte gt240 512mb, but it's really struggling.

I'm looking at getting the gtx 650 2gb or gtx 660 (non ti).

Here is my system:
i5 760
2x2gb aXe Ram DDR3 2000+
P55A UD3
HuntKey PSU (700-800 watts)
gt240 512mb
Samsung SyncMaster 226bw (1680x1050)
Samsung Syncmaster 2232gw(1680x 1050)

I do a lot of video editing on Adobe Premium Pro, Cinema 4D. With res usually at 1920x1080 or 1280x960.

I play Call of Duty 4, MW2 and Black Ops. and Battlefield 3 seldom.

I want to be able the games listed above at 40 fps avg at least, with aa 8 at least.

Let me know,
Thanks guys

 

[unreal9400]

GTX 660

 

[luciferano]

The GTX 660 is by far a more powerful graphics card than the GTX 650. However, Nvidia cards (especially the GTX 660/660 Ti and below) have trouble with MSAA, so if you want 8xMSAA, then you're probably better off with a Radeon 7870. IDK how that would change up your non-gaming workloads. For those that support Direct Compute and/or OpenCL, AMD's Radeon 7xxx cards that have GCN GPUs such as everything above and including the Radeon 7750 generally make Nvidia's cards look like crap in comparison for compute performance.

 

[UAhmed]

If 7870 is cheaper or costs a little more than 660 i'd recommend getting 7870.....
Anyways between 650 and 660.... i'd say go for 660

 

[luciferano]

Definitely the GTX 660 - those games favor Nvidia anyways.

Get this one for $205 after MIR:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127699

They favor Nvidia, but some heavy MSAA settings can easily change that in even the most Nvidia-favoring of games such as Batman:AC according to Tom's tests, so it can almost definitely change things in these games that are mostly far less Nvidia-favoring.

 

[luciferano]

AMD's cards get absolutely smoked in CoD games... a little extra MSAA isn't going to change that.

http://www.guru3d.com/miraserver/images/2012/msi-gtx660/Untitled-9.png

That's with 4xAA maxed out 1080p. Those games are old and don't stress the memory bandwidth enough to hurt the 6xx cards like newer stuff does.

At no point does that show AMD getting smoked, so IDK what you're going on about. It also shows the 660 stuck around a 570, clearly demonstrating a memory bandwidth and/or ROP reliance.

 

[luciferano]

Well, let me spell it out to you:

In CoD, Nvidia's $200 card matches or beats AMD's $300 card and *almost* matches AMD's $400 card. Nvidia's $300 card matches AMD's $450 card. Nvidia's $400 card beats AMD's more expensive $450 card by almost 10%, and Nvidia's $500 card beats it by 17%.

Now, you may not define that as "smoked", but I do.

AMD's ~$200 card also kept up with Nvidia's $200 card and AMD's $300 card has far more overclocking headroom and both have better MSAA scaling, so it can beat Nvidia's $300 card no problem if you overclock. Also, if this test is in FPS, then comparing say 140 to 170 makes no difference even on a 120Hz display, let alone on a 60Hz display. IDK the test settings for this test since they're not stated, so I can't eve ngues at what impact increased AA ould have on this anyway, but it it's not very high MSAA already, then increasing it (as supported) can change these numbers around. Batman:AC has shown us that it can in fact happen even in the most Nvidia favoring of games.

 

[luciferano]

AMD's $200 card is the 7850, which did not keep up with Nvidia's $200 card. AMD's $240 card kept up with it, though. Overclocking is not relevant to this discussion. That chart is with 4xMSAA, so again the point is not relevant to this discussion (http://www.guru3d.com/articles_pages/msi_geforce_gtx_660_twinfrozr_iii_review,12.html). CoD has nothing in common with Batman:AC in terms of how GPUs perform with it, so again that's not relevant to this discussion.

Look, I get the desire to defend AMD's cards. Obviously I think the 7xxx cards are better than the 6xx as well, but the only relevant point you made to this particular discussion is that all these cards are more than enough performance for CoD.

There is no getting around the fact that Nvidia's cards are flat out better in the Call of Duty games.

The 660 is a $230 card, yet you called it the $200 card, so I'd appreciate it if you didn't argue that with me over the 7870s when you started that logic process unless you want to correct it yourself. The 7870 can also be found cheaper than the 660, so that's not an argument. Higher MSRP doesn't necessarily mean higher retail prices.

Also, from that review:

AMD Catalyst (12.1 ~ 12.8)

Tell me, where does it tell you what driver version each tested card is using? Clear evidence of Guru3D's habit of not retesting old cards when they throw them in their comparisons. I question at least most of the 7xxx placements in that chart as a result of this. Naturally, all Nvidia cards are said to run the most recent Nvidia driver at the time, 306.23, so that review is really not seeming very unbiased.

 

[luciferano]

The 660 is a $200 card at this moment in time:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127699

The 7850 is technically a $180 card or so. The 7870 is a $230 card.

That's one card with a specific promo and is $205 anyway. I can get a 7870 for $200 with a promo if I want to.

 

[luciferano]

Link please?

Meh, promotion ended yesterday, but w/e. It was an HIS Radeon 7870 at $219.99 that had a $20 promo to bring it down to $199.99 and its around periodically, but people buy it up and it goes out of stock every other week or so.

 

[luciferano]

That was yesterday. This is today.

Point accepted. I still can't use that review as unbiased evidence given the driver BS that went on with it. Do you have any links that didn't try pulling that crap? I'll try Google too.

 

[luciferano]

Again, just like that guy eric was doing in the thread the other day, dismissing information in reviews as "biased" for no other reason than because you don't like it is not a valid form of argument.

Guru3d doesn't retroactively update results with new drivers. In fact, the only site that I know to do that is TechPowerUp, and they stopped including CoD games in their benchmarking suite. That means that the results of older cards (both AMD and Nvidia) on Guru3d's suite, are from whatever the most current driver was when the cards were reviewed. It's not ideal, but I don't know of any more up to date info on the game (I believe CCC 12.3 was the last driver TPU has results for CoD from).

It's not a conspiracy. When review sites disprove or disagree with your general preconceptions of things, it's not always because there's some anti-(your favorite company) conspiracy going on.

Dismissing results because they used irrelevant drivers is entirely unbiased, especially when they had the gall to use fully updated drivers for Nvidia and left AMD with mostly old drivers which can be nothing other than trying to exemplify Nvidia's cards as Guru3D is known to do. The review proved nothing other than AMD's older drivers didn't work well for them, as should be expected, so it's irrelevant in the face of newer drivers. That most certainly can be an accurate conspiracy, not that I care if it is or isn't.

So, no, AMD's cards weren't reviewed with the most current drivers, only Nvidias and maybe a few AMD models were (wasn't specified which ones had proper drivers AFAIK, so it doesn't matter if a few di because you don't know what ones did).

 

[luciferano]

http://skepticalteacher.files.wordpress.com/2011/04/conspiracy.jpg

AMD was tested with far outdated drivers for the time while Nvidia got tested with the most current drivers. How can you explain that, huh?

 

[luciferano]

The same way I explain how the government faked the moon landing and covered up the UFO evidence in Area 51

You're the one dismissing what you shouldn't here. The review is invalid because it did not use proper test system configuration. It is no more useful than looking up the May reviews of the GTX 680 to compare them against the Radeon 7970 today and that's completely useless.

 

[luciferano]

No. You're arguing like a fanboy conspiracy theorist who is trying to figure out how to dismiss info you don't like.

Guru3d does NOT retroactively update it's driver results for EITHER company. Get rid of the tinfoil hat.

PROOF:

@680 Launch:
http://www.guru3d.com/miraserver/images/2012/gtx680/Untitled-10.png

@670 Launch:
http://www.guru3d.com/miraserver/images/2012/gtx670/Untitled-9.png

@660ti Launch:
http://www.guru3d.com/miraserver/images/2012/gtx660-palit/Untitled-9.png

@660 Launch:
http://www.guru3d.com/miraserver/images/2012/gtx-660-evga-sc/Untitled-9.png

Notice anything about those results? Like maybe how they don't change? Which indicates that either Nvidia's drivers are updated with no affect on performance, or more likely - they weren't updated either. In either case, get rid of the stupid tin foil please.

Your review link clearly states that 306.23, the newest drivers for the time (and even today IIRC), were used for all Nvidia cards right below where it says that AMD driver versions varied between 12.1 and 12.8. It'd be a messed up conspriacy theory if you own link didn't specify this, so stop with the BS and mockery when you're the one who gave a link that states this right in the test system page of the review. Just because Nvidia's performance is similar doesn't mean that they should cripple AMD with older drivers while Nvidia gets new ones.

 

[luciferano]

OK, I'll grant your conspiracy theory. Guru3d nefariously updates its drivers for Nvidia only, in an attempt to make AMD look bad. Unfortunately, they were very sad to discover that this didn't change performance at all, not even by a meager 1fps, with any card, and the results were the same as if they had not.

Go stare at those graphs for a while. There is no preferential treatment given to Nvidia. Their performance numbers don't gradually increase over time while AMD's stay the same. No - what they get at launch is where they stay on the chart, for good or ill.

Now, you might hypothesize that AMD's driver updates have since granted greater performance improvements in this game than Nvidia's, but you have posted precisely zero proof for that claim. It's possible, but there's just no evidence of it. Just because AMD has received larger relative performance boosts in other games via driver updates than Nvidia does not in any way mean that would be the case in this game.

Haven't you read the Tom's reviews and reviews from other sites?

Also, more important than performance enhancements were fixes for poor performance. For example, before Catalyst 12.7 and its CAPs, not even MSAA could save AMD from being greatly outperformed in Batman:AC, yet with 8xMSAA, AMD now wipes the floor with Nvidia in that game to such a point that Radeon 7750 CF has been shown to beat the 660 Ti when used in such a way.

I don't know how much this game was affected, but AMD made leaps and bounds in a lot of games and I'm just saying that they should do a lot better in this game too, not that performance isn't already so high that it doesn't matter. Heck, look at the improvements that AMD made in other games that OP is interested in such as in BF3. You can't just make a decision based on a single game that OP plays, especially if OP's non-gaming software supports OpenCL or DirectC acceleration well (very likely at least for the Adobe software given Adobe's adamant and increasingly good support of OpenCL).

 

[luciferano]

I think you need to reflect on what this means for your argument. BF3 will perform similarly on a 660 as on a 7870, and neither will be powerful enough for (multiplayer) ultra 1080p unless he turns down MSAA anyways.

You're grasping at straws. If you know of a way that an AMD card will be superior in Adobe Premium or Cinema 4D, then by all means post it - I don't do video editing and am not familiar with which programs can use which GPUs.

However, hypothesizing about how maybe perhaps possibly somehow if the stars align AMD would be a better option for his stated objectives isn't helpful.

You're grasping at straws no less than I am. Not only did AMD lose to Nvidia prior to their more recent drivers such as Catalyst 12.7 and up, but they lost in a lot of games that they're on-par or winning in right now. You have no proof to disprove anything either.

AMD has a significant advantage in OpenCL and Direct Compute performance. The 7870 beats the GTX 680 in this and the 79xx cards are in a league of their own.

 

[Techie107]

Hi Guys,

Thanks for all the replies!!

I'm going with the gtx 660. Probably a gigabyte as I know the distributor so can get them for cheaper.

Thanks for the advice and benchmarks really helped a lot.

 

[Techie107]

Only thing I'm worried about is the PCI-e 3.

I know you can put the card backwards, but what will I lose in performance.

 

[esrever]

The argument about mw2 is really pointless as the game is pathetically low requirements means any card over $100 will max it. Battlefield 3 I can see used in an argument but come on? 150 fps vs 160 isn't the end of the world for 1 game...

 

[Techie107]

Best answer selected by Techie107.

 

[cravin]

OK, I'll grant your conspiracy theory. Guru3d nefariously updates its drivers for Nvidia only, in an attempt to make AMD look bad. Unfortunately, they were very sad to discover that this didn't change performance at all, not even by a meager 1fps, with any card, and the results were the same as if they had not.

Go stare at those graphs for a while. There is no preferential treatment given to Nvidia. Their performance numbers don't gradually increase over time while AMD's stay the same. No - what they get at launch is where they stay on the chart, for good or ill.

Now, you might hypothesize that AMD's driver updates have since granted greater performance improvements in this game than Nvidia's, but you have posted precisely zero proof for that claim. It's possible, but there's just no evidence of it that I've seen. Just because AMD has received larger relative performance boosts in other games via driver updates than Nvidia does not in any way mean that would be the case in this game.

Reading this thread to get information for my own purchase it is PAINFUL listening to you. Do you have to try to insult people in every comment you make? You linked to a shitty benchmark GET OVER IT. jesus chrit.