Bug? Gen killing render fps, please test
Hi folks,
I've had a recurring issue with gen strangely killing the framerate of my GL rendering. It pops up every now and then when I add new gen objects to my patch. Very annoying as I have to delete seemingly random gen objects to get my fps back. I sent this over to C74 but they weren't able to reproduce. Can you guys check it out?
'_m_skeleton 92 bug report.maxpat' is the main patch. Activating the big green switch should render some static white points in the window.
The qmetro is set at 70hz but on my machine rendering runs at 40hz. Deleting any one of the subpatches listed below make the fps jump up to 70hz, as if something gets unlocked. Most of these are just sitting idle, not receiving any input. They all contain jit.gen objects. I've seen the same behavior with gen~ objects.
- in skel_abstractor : d_AE and d_length
- in handler : osc : osc_AE
- in physics : phys_AE and jit.gen amp
My machine is a hackintosh running OSX 10.7.4 & Max 6.0.7, i7 2600k CPU, Radeon HD6870 gfx, 8GB 1600mhz RAM. I verified on my genuine MacbookPro (core2duo, radeon x1600 gfx) and the same behavior is there as well. This was happening in max6.0.5 as well.
Tanx!
Hello, Dieter
I've just had a look at your (awesome) patch and I can confirm your observations. (Win7, Intel Q9450, max 6.0.7)
- Just starting it up gives me ~100hz (metro set to 300hz)
- If I delete two! randomly selected jit.gens it jumps up to ~120hz. (128hz if I delete every jit.gen)
- If I then add one new jit.gen object (just somewhere in the patch. Default state, no connections whatsoever) the framerate drops back to ~100hz. As if there is some threshold regarding the amount of jit.gens used in a patch.
Hope this helps,
Groeten,
Ah! Good to hear I'm not crazy (or jinxed)
The threshold you mention is also the feeling I got. So it appears on both Win and Mac...
Tanx for checking!
PS: awesome patch?! Totally meaningless stripped down fragment of a big project you mean? ;)
C74, is this on the radar?
Hey dtr,
as a side question: I see in your patch that you can click through the different subpatchers by clicking on tabs with their name on top.
How exactly are you doing this? I can't seem to find out directly looking at your patch, but it's really handy!
Tnx,
Vincent
tnx a lot!
C74 still wasn't able to reproduce so I made this stripped-down-to-the-minimum patch to help 'm out. Hope some more people wanna test it!
The behavior I'm seeing:
on load : runs at around 80fps
delete one jit.gen : jump to 115fps
delete 10 more : jump to 200+fps
delete all : same 200+fps, that's apparently what my system can squeeze out
Now when I go back and start duplicating the gen's the fps gradually drops down. With around 200 gen's fps is at 10 fps. NB: the .genjit is completely empty and the objects have nothing connected in or out.
I set qmetro at 240hz because my gfx card's hardware framerate limit is 240hz.
Please state your systems OS and hardware specs when you post your results.
Tanx! D.
So far we had these systems displaying the issue:
dtr's macbookpro2,2, core2duo 2.33ghz, 2gb RAM, ATI X1600 gfx, OSX 10.7.5, Max 6.0.7
dtr's hackintosh desktop, i7 2600k (Z68 chipset), 8GB ram, Nvidia GTX 660Ti, OSX 10.8.2, Max 6.0.7 (this same system had the issue as well when running: Max 6.0.5, OSX 10.7.4, ATI HD6870)
mark2e's Core2Quad Q9450, Win7, Max6.0.7
Same here:
on load: around 90fps
delete one jit.gen: 120fps
delete 10+: around 210fps
delete all: around 210fps
MBpro, 10.7.5, intel i7 2,4GHz , 4gig DDR3 1333MHz, AMD Radeon HD 6770M 1024 MB - Max 6.0.7
I was curious, so i redid your test.
On load: 166fps
No gen: 500fps
Beginning with zero jit.gen objects I started adding them one by one. Just ALT-Drag.
Nine gens: 500fps
Ten gens: 250fps
Twenty: 166fps
Thirty: 125fps
...
Win7, Core2Quad Q9450, RadeonHD 4800 series, 4Gb DDR2, Max 6.0.7
Same scheme but different figures on my:
MBP 6.2, Nvidia GT330M, Max 6.0.7
grt
C74 was now able to reproduce and will fix in the next update. Tanx for testing guys!
This issue 's doing my head in once again... Testing tells me it hasn't been resolved in Max608 nor 611. Is that right?
Hi there. I can definitely reproduce. I've registered it in our bug reporting system and it's marked as high priority. Thanks for the clear patch.
Looking forward to it!
(Did it slip through the mazes though? It was already acknowledged by support 5 months ago...)
Looks it. It was registered, but never fixed.