Best Graphics Cards For The Money: January 2012 (Archive)

[mapesdhs]

dingo07 writes:
> Yes Ian, I have 2 in SLI - right now they OC at 808 core clock, 1955 mem clock, 1v
> EVGA SuperClocked 01G-P3-1372-TR / PNY SuperCLocked VCGGTX4601XPB-OC

Hmm, so when they're running oc'd, they're pretty much the same speed as a normal 460 Sonic
Platinum (that card is 800MHz native). From specs I gather the EVGA is normally 763, and the
PNY normally 765. That puts them right in the middle of the range of default 460 speeds.

I know the Platinum oc's very well without the need for a different cooler, so it may be
you're going to be limited not so much by the cooler as the rest of the board (is it a reference
design?). A better cooler might not give you the extra oc headroom you're expecting.


> I'd appreciate your help in getting the most out of them - I'm currently looking into

Note that if one card has better cooling and the other not, you'll be limited by the least
capable card.

I used a Gelid Icy Vision Rev 2 to test the oc potential of a whole range of 460s back in
March. Results varied widely; seems like some cards are just already near their peak
performance, while others have a lot more headroom (bizarrely, one of the best cards I
have is a boring Palit 460 768MB 675MHz, but it oc's without a sweat to 868MHz).

Sometimes two cards of the same model will behave very differently aswell, eg. I have
an EVGA 460 V2 which happily runs at 1025 core @ 1037mV, whereas another one of
the same model won't go past 975 and needs a lot more voltage to get there (1062mV).
Trying a 3rd card achieved a slightly better balance, but still unable to reach 1025MHz
core with both:

http://www.3dmark.com/3dm11/6616515


Overall though, what's your end goal with this? From all the tests I've done, TBH I'd have
to say just getting a single newer card with more RAM is a better idea if you can, eg. a 670,
or a 7870/7970. Since I've been using my two main EVGA GTX 460 FTW cards for lots of
benchmarking (the ones that run 850MHz native), I decided on a different approach for the
time being, replaced them with two 900MHz EVGA GTX 560Ti 1GB cards (total cost 177 UKP,
uber bargain as they both came fitted with Gelid Icy Vision Rev 2 coolers), which together
run very well at stock speed (faster than a 670):

http://www.3dmark.com/3dm11/6035982

while oc'd they easily beat a standard 680:

http://www.3dmark.com/3dm11/6037434

And it frees up the FTW 460s for benching. The games I'm playing aren't affected by the 1GB
RAM limit, so that's not an issue for the moment (odd thing, Crysis2 with the settings I use will
grab 1370MB if the VRAM is available - tested with a 2GB 460 - but it runs fine with my 1GB
cards, don't know why). What games are you playing? Is VRAM capacity a factor for you? I'm
just thinking in terms of how much you'd end up spending on better coolers and long term
power consumption, perhaps that would be better spent on a single newer/better card (baring
in mind the sale value of your existing cards), whether used or 2nd-hand.

Ian.

 

[cats_Paw]

Someone explain to me why every 6 months the "minimal price bracket" keeps getting higher in price.
About a year ago it was 99 $ and lower, now its 125 and lower.
At this rate we will have soon 200$ and lower as the minimal bracket.

 

[lunyone]

Someone explain to me why every 6 months the "minimal price bracket" keeps getting higher in price.
About a year ago it was 99 $ and lower, now its 125 and lower.
At this rate we will have soon 200$ and lower as the minimal bracket.

Because they want to fit in whatever card fits the budget. I know it can get frustrating, especially if there isn't any "new" GPU's to change things up. I remember 1 year that a particular GPU below $100 almost lasted the whole year as a recommendation for a given price, since there wasn't really much competition. It varies each year, but I always look at the <$200 price range, since most people can only afford in that price bracket.

 

[king smp]

Yes the 5670 was a long long time favorite under a $100 (great card I owned one) and they still mention it LOL

 

[Onus]

I still don't get why they don't mention the GDDR5 version of the HD6670. That's the one to get on a tight budget.

 

[ubercake]

I still don't get why they don't mention the GDDR5 version of the HD6670. That's the one to get on a tight budget.

That's what I'm talking about.

 

[Nevinvv]

Update the chart with intel iris and intel iris pro graphics please

 

[Sakkura]

Update the chart with intel iris and intel iris pro graphics please

Not reviewed yet. They'll get there eventually.

 

[mapesdhs]

dingo07, how did you get on in the end? What happened?

Note I've obtained four GTX 580s, have included some benchmark results already,
but lots more to do. I've also obtained two GTX 280s. Results much as expected.
Also had some 3DMark fun last week; on 3dmark.com, search for GTX 580 3-way
SLI for Firestrike Extreme, my system is in 2nd place atm. Currently in 7th
for GTX 280 2-way SLI 3DMark06.

Ian.

 

[king smp]

wonder what a triple 580 is equal to in new cards LOL

 

[slomo4sho]

wonder what a triple 580 is equal to in new cards LOL

About 2x power consumption

 

[mapesdhs]

wonder what a triple 580 is equal to in new cards LOL

What do you need to know? I've been running tests these past few days. Here's some
Futuremark results:

Firestrike Extreme: http://www.3dmark.com/fs/518524
3DMark11 Extreme: http://www.3dmark.com/3dm11/6685280
3DMark11 Performance: http://www.3dmark.com/3dm11/6685226
3DMark Vantage Extreme: http://www.3dmark.com/3dmv/4717898

I'll be testing Unigine, Stalker, CoJ, etc. later.

Ian.

 

[saint19]

wonder what a triple 580 is equal to in new cards LOL

What do you need to know? I've been running tests these past few days. Here's some
Futuremark results:

Firestrike Extreme: http://www.3dmark.com/fs/518524
3DMark11 Extreme: http://www.3dmark.com/3dm11/6685280
3DMark11 Performance: http://www.3dmark.com/3dm11/6685226
3DMark Vantage Extreme: http://www.3dmark.com/3dmv/4717898

I'll be testing Unigine, Stalker, CoJ, etc. later.

Ian.

Not the results that I was waiting for a 3-Way SLI...

 

[mapesdhs]

Not the results that I was waiting for a 3-Way SLI...

That sounds like two sentences blended together. 😀 A 3-way SLI of what, testing what?

Ian.

 

[anthony8989]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

 

[king smp]

It is just a rough guideline as the authors state.
within the tier there is a spread of performance.
I wouldnt hold to it as an exact bible of GPU performance.
I am a very happy 660Ti owner myself and I am not offended

if you notice some older cards still match up in the higher tiers with newer cards since they are just using pure FPS benchmarks for basis.
but in the real world you wouldnt want to have the older card over the newer card due to feature sets and performance vs power usage

I would love to see more SLI/CF setups in the charts

 

[Sakkura]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

The GTX 660 Ti is hit harder by the nerf, because the GTX 670 already has less memory bandwidth than the HD 7950. The 7870 XT loses 20% of 240 GB/s, the GTX 660 Ti loses 25% of 192.3 GB/s. That actually means the 7870 XT has the same memory bandwidth as the GTX 670!

Also, the 7870 XT runs at higher clocks than the 7950, but the GTX 660 Ti runs at the same clocks as the 670.

 

[copichael7]

Thanks for helping to buyers.

 

[anthony8989]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

The GTX 660 Ti is hit harder by the nerf, because the GTX 670 already has less memory bandwidth than the HD 7950. The 7870 XT loses 20% of 240 GB/s, the GTX 660 Ti loses 25% of 192.3 GB/s. That actually means the 7870 XT has the same memory bandwidth as the GTX 670!

Also, the 7870 XT runs at higher clocks than the 7950, but the GTX 660 Ti runs at the same clocks as the 670.

I respectfully disagree. The GTX 660 TI rivals the HD 7950 in performance and now in price with the recent reduction. Every benchmark I've seen involving the two cards can back that statement up. Radeon cards have always been about high numbers, but higher numbers doesn't necessarily mean a better card. It's pretty much a marketing gimmick to make their components more appealing to the customer. For example an Intel Quad core i7 CPU can out perform an AMD FX 8 core CPU.

I agree with what was said about how cards were previously ranked based solely on FPS benchmarks and hence some smeared tiers. But now with programs like FCAT out, we should be judging GPU's on their ability to create frames evenly and smoothly at the visual level - not just how many frames are created at the game engine as measured by FPS apps.

All that aside, I still stand behind the statement that in no logical sense is a 7950 / 7870 XT > GTX 660 TI by a full tier. 😛

 

[Sakkura]

Here's the performance summary from the GTX 660 Ti review. AMDs drivers have improved relative to Nvidia since then. Bear in mind the GTX 660 Ti will perform worse at higher resolutions or with more AA, due to its limited memory bandwidth.

 

[anthony8989]

Here's the performance summary from the GTX 660 Ti review. AMDs drivers have improved relative to Nvidia since then. Bear in mind the GTX 660 Ti will perform worse at higher resolutions or with more AA, due to its limited memory bandwidth.

Once again Sakkura, we don't measure a card's quality solely on the average FPS generated at the game engine as measured by FPS programs any more. Those charts are nearly useless now. We don't even see those Average FPS on the monitor that your chart shows. If you would humor me by looking at the 3 reviews on Tech Report done on the GTX 660 TI vs the HD 7950, you'll see what I mean about fluidity in the game play. Sure the 7950 generates some higher FPS on average in a lot of games, but higher fps at the game engine does not mean a smoother game play. Like I just said, we don't even visually see those FPS. I trust Tech Report because they analyse in between the seconds of game play . And when you look in between the seconds, the 660 ti > HD 7950, and according to Tech Report, even in Average FPS the 660 TI wins.

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/11

 

[Novuake]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

I would disagree on that, the 650 boost is as good as a 7850 and in no way does it blow it out of the water.
http://www.tomshardware.com/reviews/geforce-gtx-650-ti-boost-gk106-benchmark,3463.html

Here's the performance summary from the GTX 660 Ti review. AMDs drivers have improved relative to Nvidia since then. Bear in mind the GTX 660 Ti will perform worse at higher resolutions or with more AA, due to its limited memory bandwidth.

http://media.bestofmicro.com/9/V/348979/original/Average%20Perf.png

Once again Sakkura, we don't measure a card's quality solely on the average FPS generated at the game engine as measured by FPS programs any more. Those charts are nearly useless now. We don't even see those Average FPS on the monitor that your chart shows. If you would humor me by looking at the 3 reviews on Tech Report done on the GTX 660 TI vs the HD 7950, you'll see what I mean about fluidity in the game play. Sure the 7950 generates some higher FPS on average in a lot of games, but higher fps at the game engine does not mean a smoother game play. Like I just said, we don't even visually see those FPS. I trust Tech Report because they analyse in between the seconds of game play . And when you look in between the seconds, the 660 ti > HD 7950, and according to Tech Report, even in Average FPS the 660 TI wins.

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/11
http://blogs.nvidia.com/wp-content/uploads/2013/03/FCATA.jpg

Did you bother reading all the follow-ups on that exact same article? The latency has been RESOLVED in al but Borderlands 2, FFS! Geez. So thus the HD7950 is once again superior to a GTX660ti in my books.

No point in bringing up issues long dead and gone!

EDIT : I can not believe how many times I have had to say exactly that...

 

[anthony8989]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

I would disagree on that, the 650 boost is as good as a 7850 and in no way does it blow it out of the water.
http://www.tomshardware.com/reviews/geforce-gtx-650-ti-boost-gk106-benchmark,3463.html

Here's the performance summary from the GTX 660 Ti review. AMDs drivers have improved relative to Nvidia since then. Bear in mind the GTX 660 Ti will perform worse at higher resolutions or with more AA, due to its limited memory bandwidth.

http://media.bestofmicro.com/9/V/348979/original/Average%20Perf.png

Once again Sakkura, we don't measure a card's quality solely on the average FPS generated at the game engine as measured by FPS programs any more. Those charts are nearly useless now. We don't even see those Average FPS on the monitor that your chart shows. If you would humor me by looking at the 3 reviews on Tech Report done on the GTX 660 TI vs the HD 7950, you'll see what I mean about fluidity in the game play. Sure the 7950 generates some higher FPS on average in a lot of games, but higher fps at the game engine does not mean a smoother game play. Like I just said, we don't even visually see those FPS. I trust Tech Report because they analyse in between the seconds of game play . And when you look in between the seconds, the 660 ti > HD 7950, and according to Tech Report, even in Average FPS the 660 TI wins.

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/11
http://blogs.nvidia.com/wp-content/uploads/2013/03/FCATA.jpg

Did you bother reading all the follow-ups on that exact same article? The latency has been RESOLVED in al but Borderlands 2, FFS! Geez. So thus the HD7950 is once again superior to a GTX660ti in my books.

No point in bringing up issues long dead and gone!

EDIT : I can not believe how many times I have had to say exactly that...

I don't have any idea which follow-ups you're talking about but according to this one, the problem was NOT fixed:

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited

 

[Novuake]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

I would disagree on that, the 650 boost is as good as a 7850 and in no way does it blow it out of the water.
http://www.tomshardware.com/reviews/geforce-gtx-650-ti-boost-gk106-benchmark,3463.html

Here's the performance summary from the GTX 660 Ti review. AMDs drivers have improved relative to Nvidia since then. Bear in mind the GTX 660 Ti will perform worse at higher resolutions or with more AA, due to its limited memory bandwidth.

http://media.bestofmicro.com/9/V/348979/original/Average%20Perf.png

Once again Sakkura, we don't measure a card's quality solely on the average FPS generated at the game engine as measured by FPS programs any more. Those charts are nearly useless now. We don't even see those Average FPS on the monitor that your chart shows. If you would humor me by looking at the 3 reviews on Tech Report done on the GTX 660 TI vs the HD 7950, you'll see what I mean about fluidity in the game play. Sure the 7950 generates some higher FPS on average in a lot of games, but higher fps at the game engine does not mean a smoother game play. Like I just said, we don't even visually see those FPS. I trust Tech Report because they analyse in between the seconds of game play . And when you look in between the seconds, the 660 ti > HD 7950, and according to Tech Report, even in Average FPS the 660 TI wins.

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/11
http://blogs.nvidia.com/wp-content/uploads/2013/03/FCATA.jpg

Did you bother reading all the follow-ups on that exact same article? The latency has been RESOLVED in al but Borderlands 2, FFS! Geez. So thus the HD7950 is once again superior to a GTX660ti in my books.

No point in bringing up issues long dead and gone!

EDIT : I can not believe how many times I have had to say exactly that...

I don't have any idea which follow-ups you're talking about but according to this one, the problem was NOT fixed:

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited

That is an even older article. So what you are saying is that you post things without being sure of resolution? Careful with that. Read : http://www.tomshardware.com/forum/386945-33-testing-method-intro-techreport-hd7950-gtx660ti

EDIT : One more thing. This was affecting ALL HD7xxx series cards till Cat13.2. So stop saying its isolated to the HD7950...

 

[anthony8989]

It seems strange to me that the GTX 660 TI is not in the same tier as the GTX 670, and yet the 7870 XT LE is in the same tier as the HD 7950, which is in the same tier as the GTX 670... The two groups of cards could not possibly be more identical. The 660 and 7870 XT are just nerfed versions of their big brothers and they all compete with each other so closely you might as well flip a coin between them. Not to mention the 660 TI blows the 7870 non-Tahiti out of the water.

I would disagree on that, the 650 boost is as good as a 7850 and in no way does it blow it out of the water.
http://www.tomshardware.com/reviews/geforce-gtx-650-ti-boost-gk106-benchmark,3463.html

Here's the performance summary from the GTX 660 Ti review. AMDs drivers have improved relative to Nvidia since then. Bear in mind the GTX 660 Ti will perform worse at higher resolutions or with more AA, due to its limited memory bandwidth.

http://media.bestofmicro.com/9/V/348979/original/Average%20Perf.png

Once again Sakkura, we don't measure a card's quality solely on the average FPS generated at the game engine as measured by FPS programs any more. Those charts are nearly useless now. We don't even see those Average FPS on the monitor that your chart shows. If you would humor me by looking at the 3 reviews on Tech Report done on the GTX 660 TI vs the HD 7950, you'll see what I mean about fluidity in the game play. Sure the 7950 generates some higher FPS on average in a lot of games, but higher fps at the game engine does not mean a smoother game play. Like I just said, we don't even visually see those FPS. I trust Tech Report because they analyse in between the seconds of game play . And when you look in between the seconds, the 660 ti > HD 7950, and according to Tech Report, even in Average FPS the 660 TI wins.

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/11
http://blogs.nvidia.com/wp-content/uploads/2013/03/FCATA.jpg

Did you bother reading all the follow-ups on that exact same article? The latency has been RESOLVED in al but Borderlands 2, FFS! Geez. So thus the HD7950 is once again superior to a GTX660ti in my books.

No point in bringing up issues long dead and gone!

EDIT : I can not believe how many times I have had to say exactly that...

I don't have any idea which follow-ups you're talking about but according to this one, the problem was NOT fixed:

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited

That is an even older article. So what you are saying is that you post things without being sure of resolution? Careful with that. Read : http://www.tomshardware.com/forum/386945-33-testing-method-intro-techreport-hd7950-gtx660ti

EDIT : One more thing. This was affecting ALL HD7xxx series cards till Cat13.2. So stop saying its isolated to the HD7950...

I guess you didn't even read your own article. All I see from that report is the GTX 660 TI beating the HD 7950 in 2 out of the 3 games in both FPS and in frame latency. And the only game the 7950 managed to pull a 5 fps win was in Guild Wars 2, where it also tied in frame latency. Yep that's a full tier higher in performance for sure (sarcasm).