Performance diff. GTX 1060 vs GTX 1070 in photoscan...?

[JohnyJoe]

Does anybody have an idea or actual numbers, whats the speed improvement in photoscan when using GTX 1060 6 GB vs GTX 1070 8 GB?

Thank you

[Thibaud Capra]

A GTX 1070 8 Gb will obviously outperform a 1060 6 Gb.

If you want a performance differential between them, check their respective benchmarks.

As for time processing in PhotoScan, it'll all depend on your project. The larger the project, the bigger the difference will be.

[JohnyJoe]

Thank but im not an idiot :-). There many game benchmarks, but sadly, gtx 1070 performs worse in profi applications, where there are NOT many (or any?) tests at all...

I didnt find practicaly any, thats why im askign here, if someone didnt test them already?

[Alexey Pasumansky]

Hello JohnyJoe,

According to the card specifications, I can assume that the performance of GTX 1070 would be about 40%-50% higher compared to GTX 1060.

[SvaneSDU]

By googling the topic (CPU vs GPU strain in Photoscan) I found that for 3D work a dual gpu setup works the best. The 1060 can not be used in a SLI setup, thus the 1070 is better in that aspect.

For general alignment and processing (non-3d) it seems cpu-frequency (and multi cores) has more effect.

Its nice that the high RAM need has been lowered in the recent updates.


I use a gtx 1070 myself btw, and it performs very solid. In general (other applications) what has been mentioned (40% gain compared to 1060) is also what I have read.

[Dave Martin]

...... The 1060 can not be used in a SLI setup, thus the 1070 is better in that aspect.......

SLI doesn't help, and has been alleged to damage, PhotoScan performance.

[Alexey Pasumansky]

Hello SvaneSDU,

Actually, the experiments with SLI, although performed in version 1.2, have shown that SLI connection affect the processing speed, so even for the identical cards it is not recommended to use SLI/CrossFire connection.

The GPU-based image matching has been introduced in the version 1.3.0, so the threads related to the older version matching do not cover GPU at all, since Align Photos stage has been performed on CPU only in the older versions.

[Thibaud Capra]

For general alignment and processing (non-3d) it seems cpu-frequency (and multi cores) has more effect.

I'm using a dual-Xeon (2x24 cores) and a Quadro M4000, and I feel the opposite...
Would love to learn a bit more about it!

[feiko.lai]

FYI, we just got a 1070, some benchmarks for dense cloud generation, high quality, building sample

1.2.6

2017-03-01 04:44:06 finished depth reconstruction in 691.51 seconds
2017-03-01 04:44:06 Device 1 performance: 1174.18 million samples/sec (GeForce GTX 1070)
2017-03-01 04:44:06 Total performance: 1174.18 million samples/sec

1.3.0

2017-03-01 05:00:56 finished depth reconstruction in 208.065 seconds
2017-03-01 05:00:56 Device 1: 100% work done with performance: 487.714 million samples/sec (GeForce GTX 1070), device used for 343.541 seconds
2017-03-01 05:00:56 Total performance: 487.714 million samples/sec

[Dave Martin]

With reference to comparing performance, I would just add that Agisoft have mentioned on here that because of different processing in 1.3 compared to 1.2 that the only valid comparison between these versions is the total time taken. Can you advise the total time elapsed?

[Alexey Pasumansky]

With reference to comparing performance, I would just add that Agisoft have mentioned on here that because of different processing in 1.3 compared to 1.2 that the only valid comparison between these versions is the total time taken. Can you advise the total time elapsed?

As I see the time spent on the depth maps reconstruction is also provided: 208 seconds in 1.3 vs. 691 seconds in 1.2.6.