[martgreg]

Someone do a benchwell with V2 and lets see how fast it is

[Mihai]

When doing render comparisons with V2, the most important thing to remember is don't compare running to the same SL. A smaller SL in V2 will mean in most cases a less noisy render. So the only comparison should be time, not SL.

The benchwell site is ment to compare different computers, it's not so well suited to compare 1.7 vs V2.

[martgreg]

true i guess thanks

but if one can figure out what is comparable to the the sl of 15 which was the goal and see how fast we can get to that it would be a nice comparison...

we will see ...

[tom]

The benchwell scene is rendering ~1.618 (589 / 364 sec) times faster. Also, pay attention that some parts are even clearer.

Maxwell Render 1.7.1 --- SL15 // 9m49s // B1093.881


Maxwell Render 2.0 --- SL13.94 // 6m34s // B1062.317

[brodie_geers]

Interesting how the color and shadow on the table is a tad different between the 2 versions.

-Brodie

[tom]

Interesting how the color and shadow on the table is a tad different between the 2 versions.

They have to be different because in v2;
- BSDF is vastly improved about physical accuracy.
- Glossy portion (0 < Roughness < 100) now has a seamless accurate transition.
- IBL is completely renewed and optimized.

[martgreg]

Interesting how the color and shadow on the table is a tad different between the 2 versions.

They have to be different because in v2;
- BSDF is vastly improved about physical accuracy.
- Glossy portion (0 < Roughness < 100) now has a seamless accurate transition.
- IBL is completely renewed and optimized.

yep what Tom said.... I was gonna say the exact same thing...

thanks Tom for the test.

now i need to decide what comp to buy at work:)

[camel]

v2 SL15
21m47
(have to mention that my system really needs a reboot.. but im too busy :p)

[tom]

v2 SL15 -- 21m47

Thanks for running v2 up to SL15 but this image is not comparable.

[camel]

yes i understand now
but people asked for v2 pictures.. :p

[tom]

Keep'em coming

[Jozvex]

1.618 times faster? That's the golden ratio!

OoooOooOOooo!

[motopiku]

even in the old release was difficult to compare.
if new engines have less noise, which variable would be better to use?
banchwell make a new time-based? SL reaches those ?.... or based on benchmarks?
(maybe even have a benchmark study for the preview of maxwell have a guide on materials optimized.)

I think the benchwell is fine for testing the computer (only power hardware).
To assess the performance of versions of Maxwell (patches and updates) is better to create a new test.