Should the difference in performance between the GTX 770 and my new GTX 970 be greater than this...?

[GuruMeditationError]

The improvement I'm getting in frames per second in the benchmarks for the following games by moving from a factory overclocked MSI GTX 770 to a factory overclocked Gigabyte GTX 970...

 

My question: Should I be seeing bigger improvements in frame rate than these?

 

 

Valley:

 

 

(Minimum)... +6 fps

 

 

(Average)... +8 fps

 

 

 

Hitman Absolution:

 

 

(Minimum)... No improvement (over GTX 770 with FX-9370)

 

(Minimum)... +6 fps (over GTX  770 with i3 4130)

 

 

(Average) (compered to GTX 770 with FX-9370)...  -3 fps (minus 3 fps)

 

(Average) (compared to GTX 770 with i3 4130)... +6

 

 

Shadow of Mordor:

 

 

(Minimum)... +10.5 fps

 

 

(Average (compered to GTX 770 with FX-9370))...  -0.5 fps (minus 0.5 fps)

 

(Average (compared to GTX 770 with i3 4130))... +3.5 fps

 

 

 

Tomb Raider:

 

 

(Minimum)... +21 fps

 

 

(Average)... +14 fps

 

 

Thief:

 

 

(Minimum)... +1.3 fps

 

 

(Average)... +6 fps

 

 

So (being generous) that's a total average improvement of about 8.2 frames per second.

 

So...so far, it's cost me £120 and 25% of my Graphics card RAM for roughly an extra 8.2 fps

 

 

Should I be getting a bigger improvement than this?  

 

 

 

 

[Jade]

We could tl;dr this: Are the settings different? Could you give us the actual FPS values rather than the differentials?

[Guest]

On that i3 from your signature? yes of course.

 

edit: I wanted to say yes this situation you are having is expected by me because of that i3.

[GuruMeditationError]

We could tl;dr this: Are the settings different? Could you give us the actual FPS values rather than the differentials?

 

Not really? I posted the information in that form so as to facilitate getting a simpler answer...

[Jade]

Not really? I posted the information in that form so as to facilitate getting a simpler answer...

For all we know you could have gone from 300 fps to 309. So no, that's not particularly helping to get a simpler answer.

 

Before someone calls me out on the '300'; it's an example.

[werto165]

overclock your 970 further, they're beastly overclockers in general. 

[ShadowCaptain]

The 970 is not THAT much more powerful than the 770

 

4+ fps in 4k on Crysis 3

13.1 fps at 1080p on Crysis 3

 

[R_Golden]

That was not a smart upgrade. Did you really expect a 50% increase or something?

[GuruMeditationError]

For all we know you could have gone from 300 fps to 309. So no, that's not particularly helping to get a simpler answer.

 

Before someone calls me out on the '300'; it's an example.

 

But it's not really relevant to the question I'm asking? :0/

 

 

That was not a smart upgrade. Did you really expect a 50% increase or something?

 

Well in a word...no. But I was expecting more than this.

 

[GuruMeditationError]

On that i3 from your signature? yes of course.

 

edit: I wanted to say yes this situation you are having is expected by me because of that i3.

 

Okay, so it could be the i3 that's limiting it? Even the minimums?

 

Edit:

 

Also (something less "CPU bound")...

 

http://www.guru3d.com/index.php?ct=articles&action=file&id=12500&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1

[Guest]

Okay, so it could be the i3 that's limiting it? Even the minimums?

I would say yes.

Not a lot but...

[GuruMeditationError]

I would say yes.

Not a lot but...

 

I guess so, I mean I'm getting minus scores for some of them when switching from the FX to the i3  hmmm.... :0/

 

Edit: ...but to be honest a lot of this is just noise. I'd probably need to run this stuff over and over again to get a genuinely usable result.

[johngo90]

770 to a 970... Yeah I wouldn't have bothered upgrading. 

 

Only option is to overclock that 970. 

[GuruMeditationError]

overclock your 970 further, they're beastly overclockers in general. 

770 to a 970... Yeah I wouldn't have bothered upgrading. 

 

Only option is to overclock that 970. 

 

Do they really overclock that well?

 

I got to 30% over stock with my last GPU overclock but that was pretty extreme. I guess if I could equal that then I can add approximately an extra 16 fps to my average scores in valley (but is that really worth the cost of an Accelero Hybrid II?

 

But then if my CPU is limiting the graphics card...(although I gained 10 fps on one value just downgrading to an i3)...to be honest, this all just seems like noise to me.

 

I guess I'll hang onto the card. I wanted to learn about this stuff and I guess there it is right there...

 

:0/

 

:0)

[D2ultima]

970 is ~25-30% stronger than a 770... but you're using terrible CPUs for it.

 

Listen, let me make this clear OP.

 

Just because and i3 is better than an AMD CPU for lots of games out there, does NOT MEAN THAT AN i3 IS A GOOD GAMING CPU.

 

I don't care if 90% of this forum disagrees. It's not. Quadcores are what you should really hunt for. i5s and i7s are the way to go. There's going to be games that won't launch unless you have a quadcore eventually. 

[GuruMeditationError]

970 is ~25-30% stronger than a 770... but you're using terrible CPUs for it.

 

Listen, let me make this clear OP.

 

Just because and i3 is better than an AMD CPU for lots of games out there, does NOT MEAN THAT AN i3 IS A GOOD GAMING CPU.

 

I don't care if 90% of this forum disagrees. It's not. Quadcores are what you should really hunt for. i5s and i7s are the way to go. There's going to be games that won't launch unless you have a quadcore eventually. 

 

Thanks :0)

 

That's what I'm actually looking at right now, but to be completely honest with you I'm not running enough repeated benchmarks (i.e. the hundreds I'd probably need to) to get a valid statistical result. I've had fps values rise as well as fall going from the FX 9370 to the i3-4130...

 

...but my intention was to try and cap the top end with the CPU but to see if I still get higher minimums (but then I guess cards are as fast as they can be and if that could work we'd probably already have the option to cap maximums to improve minimums and average...

 

...it's kind of an experiment...

 

http://linustechtips.com/main/topic/306820-my-latest-crazy-pc-experiment-low-baller-how-slow-can-you-go-my-new-build/

[Mush Brain]

Dat AMD bottleneck tho... -3 FPS lol.

[GuruMeditationError]

Dat AMD bottleneck tho... -3 FPS lol.

 

No, it's the other way around, that's -3 with the i3 and the GTX 970. With the AMD it was +3 (over the i3 and 970) with the 770 (but to be honest I don't really think it's statistically valid?)

 

But anyway, I've got to get to the post office. :0/

[Mush Brain]

No, it's the other way around, that's -3 with the i3 (but to be honest I don't really think it's statistically valid?)

But anyway, I've got to get to the post office. :0/

"(Average) (compered to GTX 770 with FX-9370)... -3 fps (minus 3 fps)

(Average) (compared to GTX 770 with i3 4130)... +6"

[GuruMeditationError]

 

Dat AMD bottleneck tho... -3 FPS lol.

 

No, it's the other way around, that's -3 with the i3 and the GTX 970. With the AMD it was +3 (over the i3 and 970) with the 770 (but to be honest I don't really think it's statistically valid?)

 

But anyway, I've got to get to the post office. :0/

 

"(Average) (compered to GTX 770 with FX-9370)... -3 fps (minus 3 fps)

(Average) (compared to GTX 770 with i3 4130)... +6"

 

Yup...the average I got for that one value, with one benchmark run each, was 3 fps fewer with the i3 and the 970 than with the FX and the 770...

 

...but, as  I say, I really don't think it's statistically valid. The results are all over the place, it's just noise.

 

[sgloux3470]

The performance difference between the two is just small.

It'd be like going from a 280x to a 290x.

Generally if you have a 7xx card upgrading to 9xx doesn't make sense unless you jump into a higher bracket.

[R_Golden]

You shouldn't have upgraded to a 970, but an i5.

[GuruMeditationError]

You shouldn't have upgraded to a 970, but an i5.

 

I've been getting stuttering in one part of thief and when I checked the CPU usage it was at 100% but it could have been 100% the entire time I was playing (I only looked when it started stuttering) :0/

 

An explanation of why I went for the i3 here (in case you're interested)...http://linustechtips.com/main/topic/326178-should-the-difference-in-performance-between-the-gtx-770-and-my-new-gtx-970-be-greater-than-this/#entry4430652

[Briggsy]

Nvidia release new cards more often, with smaller increments in performance. You should expect a huge leap if upgrading from a 570, but realistically the 680/770 is only a jump of one generation, and usually its recommended that people wait for 2 generations with Nvidia.

 

 

*edit* I should explain what I mean: starting with Kepler, Nvidia are staggering the release of architectures, by releasing a smaller version first, and then later releasing a bigger version. Instead of releasing the GK110 as the 680, they staggered it to make it look like 2 generations, by releasing it later as the 780.

Edited March 11, 2015 by Briggsy

[GuruMeditationError]

Nvidia release new cards more often, with smaller increments in performance. You should expect a huge leap if upgrading from a 570, but realistically the 680/770 is only a jump of one generation, and usually its recommended that people wait for 2 generations with Nvidia.

 

 

*edit* I should explain what I mean: starting with Kepler, Nvidia are staggering the release of architectures, by releasing a smaller version first, and then later releasing a bigger version. Instead of releasing the GK110 as the 680, they staggered it to make it look like 2 generations, by releasing it later as the 780.

 

Sounds like good advice. At this point it seems there's not really much value even in selling a card to upgrade (where I'm sitting at least), which might actually be a an argument in favour of SLI? ...but that brings its own issues :0/

 

For me, really, the bottom line: if I returned this card (which I could still do) I'd be back to my GTX 470 (because I've already sold and shipped my 770). Bu that would mean going from a smooth frame rate to a stuttery frame rate...and, basically, that's the tipping point for me in keeping this card. I want to play Thief etc., without stutter. :0/

 

The next step is probably to play-test it in the areas I know there to be slowdown and to try and ascertain if that slowdown is attributable to the CPU, then upgrade it to an overclockable i5 (probably) and see what the improvements are to the overall experience with the i5 at stock and when overclocked (to be honest I think the figures are just noise (or at least they are the way I'm testing, not sure about how binning works?!  :huh: ((b4 etc....for everyone out there about to jump on that: I mean how binning actually works)) but at this point I don't have much faith in it).