Topic: Speed Difference 680 GTX vs 980 GTX

[Tim M.]

Just wondering if there would be anyway to estimate seconds per frame that would be saved by switching from a 680 GTX to a 980 GTX.

My current benchmark on the 680 GTX is 14220 and I see one of the highest for the 980 GTX is 25020.  Just trying to gage how much of an increase we would see to justify to my boss to spend the money.

Still looks like the  8GB 980 GTX Ti is just a rumor so far from recent post as to when it will be released.

[Morten]

Just wondering if there would be anyway to estimate seconds per frame that would be saved by switching from a 680 GTX to a 980 GTX.

Hi Tim M., someone with both graphics cards would need to render the same scene on the same PC I guess.

[Tim M.]

Thanks Morten.

Just curious and wouldn't hold you to it ofcourse, but if I divide the benchmark of the 980GTX of 25020 by the 680GTX of 14220 I come up with about a 1.75% difference.

Would that be an approximate increase in speed I could use?

[IdeaImmagine]

The difference between 14220 and 25020 is about +75%, not 1.75%!

[Morten]

Would that be an approximate increase in speed I could use?

I don't know. If someone else has some comparison render times where they rendered the same scene on the same PC but with different graphics cards, I'd also be curious to see the results.

[peterm]

Hi Tim M.

The average Benchmarks we have for the GTX 980 is 20025.  That is, it reflects different PC setups.

You would not be able to expect a direct performance change by comparing cards only.  This is because you will be placing the new card in to an existing system, so your CPU clocks, RAM speed and motherboard specifications will have an impact on how well the GTX 980 got used.  So for an older system you might need to arbitrarily reduce expected change by say? 25%.

[IdeaImmagine]

You would not be able to expect a direct performance change by comparing cards only.  This is because you will be placing the new card in to an existing system, so your CPU clocks, RAM speed and motherboard specifications will have an impact on how well the GTX 980 got used.

Isn't Lumion rendering based exclusively on CPU?

[peterm]

Hi IdeaImmagine

Isn't Lumion rendering based exclusively on CPU?

I think you might mean "..on GPU?" 

Not exclusively, but the biggest majority.  The performance of how well the card performs in terms of final render output, is also determined by how well the rest of the system supports the GPU to do it's work.  The same card placed in two different PC setups will not have the exact same results. 

If you were to look at Passmark results, for example the GTX 980, you will see that each of the last 5 Baselines submitted is different, in some cases by large differences.  A small part off that difference could be due to differences in manufacture, quality of chips used, better card design etc.  The remainder is therefore explained by the rest of the system. 

[Morten]

I think you might mean "..on GPU?" 
Not exclusively, but the biggest majority.

In addition, we ought to mention that if the CPU isn't fast enough, it may act as a bottleneck for the GPU which will lower the framerate in Build mode and in Movie mode.

However, this usually only happens if your CPU is quite old/slow.

The optimal PC is one where neither the CPU or the GPU act as a bottleneck for each other. This is a fine art that some very dedicated people waste a lot of time on spend a lot of time on

[IdeaImmagine]

I think you might mean "..on GPU?" 

Sorry for my bad English. But what is the difference?

[Morten]

Sorry for my bad English. But what is the difference?

Click here to learn what the difference between a CPU and a GPU is.

[IdeaImmagine]

It was not my English, but my keyboard: I pressed C instead of G!
Yes: Lumion is GPU-based, of course.

[peterm]

   Good to know you have a clearer understanding IdeaImmagine. 

[IdeaImmagine]

  Good to know you have a clearer understanding IdeaImmagine. 

The difference was just between a C and a G.