• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Which graphics card for my system - 770 v 7970

Because it has a lesser GPU? :confused:

However, if you observed the falloff in performance from increasing AA, the 3GB 580 would lose less FPS (as a percentage) from increasing AA/res than a 670 would.

Additionally, a 3GB 580 isn't the standard model, the standard model is 1.5GB on a 384 bit bus.
I highlighted that bit just to make it stand out more. I'm actually quite surprised that being a graphic card forum such as ours, there are SO many people don't even realise having a narrower memory bus mean frame rate would dip at a faster rate (in terms of percentage) than wider memory bus.

It is also the same reason why the GK104 cards would have frame rate drop to below AMD card of the same tier level on the higher res. Far too many people going round in circle talking about the amount of vram, whne it is the bus size that's holding back the cards the most.
 
Last edited:
However, if you observed the falloff in performance from increasing AA, the 3GB 580 would lose less FPS (as a percentage) from increasing AA/res than a 670 would.

Yeah that was my point, that doesn't happen.


Additionally, a 3GB 580 isn't the standard model, the standard model is 1.5GB on a 384 bit bus.

Yes I know it comes in 1.5GB and 3GB versions, hence me specifying the 3GB version for the example.

My point was that the 3GB 580 with its 384bit bus should (if you were right) begin to reel in the 2GB 670 with its 256bit bus as things get heavy, but it doesn't.

The reason the GTX670 doesn't drop off faster than the 580 despite having 2/3 the bus is because the bus is only one part of the equation, both the 580 and 670 have 192GB/s memory throughput because despite the 670 having a smaller bus its countered by faster ram (the old bandwidth*speed=throughput equation.
 
Last edited:
Yeah that was my point, that doesn't happen.
Is it "I believe that it doesn't happen on my GTX670", or have you actually "looked up some benchmarks and comparing the frame rate of the 384-bit bus card vs 256-bit card at both 0xAA and 4xAA and comparing the % of frame rate drop going from 0xAA to 4xAA"?

I'm not trying to have a go at you here, but I just want to point out...there's no harm actually try reading things up on some subjects. It's quite dangerous to just make blind assumption and stick by it.
 
Last edited:
Yeah that was my point, that doesn't happen.

No it wasn't as per your last comment below:



Yes I know it comes in 1.5GB and 3GB versions, hence me specifying the 3GB version for the example.

It's not the RAM quantity that's the issue.

My point was that the 3GB 580 with its 384bit bus should (if you were right) begin to reel in the 2GB 670 with its 256bit bus as things get heavy, but it doesn't.

No, the RAM quantity isn't the issue, it's memory bandwidth.

That aside, the 670's GPU is a lore more powerful than the 580's.

Put it this way:

If you took 2 GTX670s and put them in two computers with identical specs, but one of the 670s had a memory clock of 6Ghz, giving a memory bandwidth of 192GB/s, and the other had a memory clock of 8Ghz, giving a memory bandwidth of 256GB/s.

Keeping both cores at the same speed, the card with the 8Ghz RAM would see less of a drop in frame rate from increasing resolution and levels of AA than the card with 6Ghz RAM.

This is evidenced by RAM overclocks on GTX6 series cards improving the falloff in frame rates, this means that the memory bus is being saturated at 192GB/s.

This is why the GTX770 doesn't suffer as much as the GTX670 with increased resolution and AA amounts, as it has 224GB/s of bandwidth due to its 7Ghz VRAM.
 
Is it "I believe that it doesn't happen on my GTX670", or have you actually "looked up some benchmarks and comparing the frame rate of the 384-bit bus card vs 256-bit card at both 0xAA and 4xAA and comparing the % of frame rate drop going from 0xAA to 4xAA"?

I'm not trying to have a go at you here, but I just want to point out...there's no harm actually try reading things up on some subjects. It's quite dangerous to just make blind assumption and stick by it.

Well you can see this in 670/680 vs 7970 comparisons. The 7970 (and even the 7950) tend to do relatively better in 1440p and 1600p where post processing such as AA is applied i.e. any prior advantage is increased and any deficit is narrowed.
 
Well you can see this in 670/680 vs 7970 comparisons. The 7970 (and even the 7950) tend to do relatively better in 1440p and 1600p where post processing such as AA is applied i.e. any prior advantage is increased and any deficit is narrowed.

And the mail reason is available memory bandwidth.

I think some people in this thread are trying to make the discussion something it isn't because for some bizarre reason they feel the need to continually recommend cards with lower memory bandwidth and lesser memory.

When you're buying now, there's just no reason why you should do that due to the fact that a card with more RAM and memory bandwidth will have more longevity in them.
 
Well you can see this in 670/680 vs 7970 comparisons. The 7970 (and even the 7950) tend to do relatively better in 1440p and 1600p where post processing such as AA is applied i.e. any prior advantage is increased and any deficit is narrowed.
Yea but far too many people believe that the 79xx cards coping better at higher res doesn't have anything to do with 1920 res, and not bothered to understand the reason behind that, and why is that relevant to 1920 res as well. They insisting on looking at only 1920 results, so the only way to show them is to have them comparing the frame rate difference between 0xAA and 4xAA at 1920 res and comparing the drop rate in percentage on the frame rate between 256-bit and 384-bit memory bus cards.
 
Last edited:
That aside, the 670's GPU is a lore more powerful than the 580's.

Exactly this is the point I'm trying to make, I'm not saying that with unlimited GPU performance available memory throughput wouldn't be the bottleneck, I'm saying that the GPU will become the bottleneck first in real world usage, hence why the 580 cannot match the 670. Yes in theory their bus does give the 79xx a theoretical advantage, but in the real world by the time that advantage would be coming into play it would already be irrelevant as neither the 6xx or the 79xx would be capable f running new games maxed at 1920 any more.

If somebody asked me which card to get I would recommend the HD's over the GTX every day due to the games package, its a no brainier, but not because of a delusion that the HD's will remain viable for longer.
 
Exactly this is the point I'm trying to make, I'm not saying that with unlimited GPU performance available memory throughput wouldn't be the bottleneck, I'm saying that the GPU will become the bottleneck first in real world usage, hence why the 580 cannot match the 670. Yes in theory their bus does give the 79xx a theoretical advantage, but in the real world by the time that advantage would be coming into play it would already be irrelevant as neither the 6xx or the 79xx would be capable f running new games maxed at 1920 any more.
Not entirely true. With the basis of both GPUs being equal, what's gonna happen is the the 770 would need to drop the AA level sooner than the 7970 in order to sustain the frame rate.

Many people actually holding off upgrading for long due the the small progression on on the performance each gen, so it's not longer "upgrading once every 2 gens", thus anything that contribute toward the card being able to last longer (without extra cost) would be a merit.
 
My 670 memory goes to 7.5 ghz, does that mean with lots of AA and a higher res it will keep a higher framerate percentage and slow down the % falloff of FPS?


I will try this if i can be bothered sometime if thats the case.
 
Exactly this is the point I'm trying to make, I'm not saying that with unlimited GPU performance available memory throughput wouldn't be the bottleneck, I'm saying that the GPU will become the bottleneck first in real world usage, hence why the 580 cannot match the 670. Yes in theory their bus does give the 79xx a theoretical advantage, but in the real world by the time that advantage would be coming into play it would already be irrelevant as neither the 6xx or the 79xx would be capable f running new games maxed at 1920 any more.

If somebody asked me which card to get I would recommend the HD's over the GTX every day due to the games package, its a no brainier, but not because of a delusion that the HD's will remain viable for longer.

If you think it's a delusion then you clearly don't understand how graphics cards work.

You also don't seem to know what delusion means either.

You are responding to things I never said. The bottom line is that in the future, you will have to turn down more settings for a 6XX card than you would on a 79XX to get a playable frame rate, this is how the situation is now, so on what planet is it a delusion?
 
My 670 memory goes to 7.5 ghz, does that mean with lots of AA and a higher res it will keep a higher framerate percentage and slow down the % falloff of FPS?


I will try this if i can be bothered sometime if thats the case.

Yes it does. At 7.5ghz, your 670 has a memory bandwidth of 240GB/s, the same as a 7950 assuming it's a stable memory overclock and error correction isn't kicking in.
 
If the 256bit and the 384 bit have their memory running at the same speed, so that their total bandwidth both = 250gb for example, would the wider bus of the 384bit card still give it a performance advantage or would that not mean anything now as the actual bandwidth is the same?
 
If the 256bit and the 384 bit have their memory running at the same speed, so that their total bandwidth both = 250gb for example, would the wider bus of the 384bit card still give it a performance advantage or would that not mean anything now as the actual bandwidth is the same?

I take you mean if you got the memory running so it gave you the same bandwidth on both?

Excellent question.

On my present understanding it would even things out

This is the sort of stuff I'm trying to gen up on.

It would be nice if some kind soul would explain(in layman terms) :confused:

This question may have been better Matt in its own thread :)
 
I take you mean if you got the memory running so it gave you the same bandwidth on both?

Excellent question.

On my present understanding it would even things out

This is the sort of stuff I'm trying to gen up on.

It would be nice if some kind soul would explain(in layman terms) :confused:

This question may have been better Matt in its own thread :)

Yep thats what i meant. No answer so far i see. :D
 
Back
Top Bottom