Max 5 metro vs. qmetro FPS differences with 4.6

May 6, 2008 at 7:39am

Max 5 metro vs. qmetro FPS differences with 4.6

This is with Max 5.01 on a PC with XP.
After all sorts of weird behavior with a big video patch and overdrive, I did a little sanity check with metro and qmetro.

In 5.0 a metro of 33.333 settles out to roughly 30.0 fps.
In 4.6 a metro of 33.333 settles out to roughly 30.0 fps.
in 5.0 a qmetro of 33.333 settles out tot roughly 27.5 fps
In 4.6 a qmetro of 33.333 settles out to roughly 29.99 fps
This is with a metro or qmetro attached to a FPS object, with a toggle to start it… nothing else.

Moreso, I normally notice an increased FPS in my patch when I enable Overdrive… but in 5.0 my fps drops from 27.5 fps using a qmetro to like 17 fps. Obviously there is a lot of patch dependency here… but the general issue is previously overdrive sped up my patch.. now it slows it down a LOT.

I didn’t think 4.6 and 5.0 were supposed to be that different under the hood, but what I’m seeing is a bit frightening.

Why would qmetro vary that much from metro?

#37577
May 6, 2008 at 7:49am

Have you looked into the refresh interval scheduler preference? Sounds like you might want to investigate it.

David Z.

#129941
May 6, 2008 at 6:46pm

– Pasted Max Patch, click to expand. –

The fps in my example are with all the scheduler settings at defaults. The only really major difference in those value occurs when I change event/scheduler intervals, or turn overdrive off and on. So refresh settings aren’t changing things at this simple level.

#129942
May 12, 2008 at 7:09pm

Is this expected behavior?
I can use a smaller value for qmetro and get a higher FPS, so I’m not “capping out”.

Id like to know if other people get the same variation, or if something is wrong with my system.

metro 33.33 = 30 fps
qmetro 33.33 = 27.5 fps

And sure, these are arbitrary values, but it makes me feel like I’m using something approximating appropriate video framerates and shuts up gadflys that check out my patch over my shoulder.

#129943
May 12, 2008 at 7:49pm

I agree based on your example there might be something wrong with qmetro in Max 5, although timing accuracy is not really a reason to use it. I’ll investigate further.

David Z.

#129944
May 12, 2008 at 8:33pm

qmetro’s long-term accuracy (better) and short term variability (worse) will be restored to 4.6 levels in the next post-5.0.2 update.

David Z.

#129945

You must be logged in to reply to this topic.