drawstress benchmark limited by vsync about bgfx HOT 6 CLOSED

On my test on Linux Chrome 35 with option --disable-gpu-vsync the score went from 2744 to 6859~8000.

from bgfx.

Thanks! I'll use --disable-gpu-vsync and update results later tonight.

from bgfx.

Probably I'll add it as notice (in the same way those extra FF27 options improve performance), since my preference for test is how end users will see it. If it is possible to change vsync behavior programmatically then I would put results as is.

from bgfx.

Hi Branimir,

In end user browser environment it is unlikely to have this flag. Also the
behavior is not optimal.

It would be good to change the benchmark for example target most cubes at
30fps to get consistent results in all browsers.

On Mon, Mar 17, 2014 at 3:48 PM, Branimir Karadžić <[email protected]

wrote:

Probably I'll add it as notice (in the same way those extra FF27 options
improve performance), since my preference for test is how end users will
see it. If it is possible to change vsync behavior programmatically then I
would put results as is.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/77#issuecomment-37880816
.

from bgfx.

Yeah that's good idea, I'll add separate table just to show results at 30Hz for browsers (with reference results for native, and native+ANGLE at 30Hz), and I'll measure it also at 60Hz with * notice about settings changes.

from bgfx.

I added 30Hz section and updated data: https://github.com/bkaradzic/bgfx#30hz-test-for-browsers

from bgfx.