IR used to be interactive

[maru]

You cannot call me impartial, but I also haven't noticed any slowdown or other issues with IR in 1.7 hf2 or the latest dailies. It seems that some performance issues may be happening on some specific computers only, or with some specific scene setup... We have a few reports about this open (mainly from dual Xeon users), and we are investigating this.

[NicolasC]

We have a few reports about this open (mainly from dual Xeon users), and we are investigating this.

Oh yes, please ;)

[Ondra]

We AFAIK currently do not have any solid performance regression in interactive that we could investigate.
Before you swarm me with answers ;):

• We know there are cases where interactive is slow. For example manipulating override material, manipulating hair&fur, creating new objects in heavy scenes, ... - but all these problems should occur exactly identically in previous versions
• We know there are unnecessary restarts, for example when navigating slate editor - but again, the same problem was in every Corona version with interactive
• There are unnecessary restarts with 3dsmax physical camera. They are solved to switching to Corona camera. The physical camera in general is quite bugged w.r.t. notification system. Going from old max to new max will give you feeling of slower interactive due to switching from standard camera to physical camera
• There is another issue, where moving around viewport is more heavy on main thread when using camera than free viewport. Just moving the camera around could make the main thread 100% busy, so there is no time for IR to update. We have an idea how to fix this, but it involves waaay too heavy changes. On the other hand, it should also fix stuff like materials not expanding in slate or render region not working. It will require a lot of time to implement. But again, this should be broken exactly the same in older corona versions

[Dalton Watts]

Like fraine7 and others have reported I too can't put my finger on why exactly IR feels much slower on my Dual Xeon 2696v3 system because frankly, I haven't had the time yet to isolate this and especially to figure out when exactly did it start to underperform. It is something very noticeable though.

This bothers me because IR was THE main factor that made me switch to Corona.

[davemahi]

I have a dual Xeon and I am noticing IR to be slower too. Sometimes I find it faster to just keep hitting Render over and over than using IR.

I thought it was just me, but I do notice it.

[maru]

I have a dual Xeon and I am noticing IR to be slower too. Sometimes I find it faster to just keep hitting Render over and over than using IR.

I thought it was just me, but I do notice it.

Have you already contacted us about this through our support? Of not, then it would be greatly appreciated: https://coronarenderer.freshdesk.com/support/tickets/new
We would then ask some extra questions, and log your case in our internal tracker.

[Dalton Watts]

I reckon that, now, using Corona 1.7 hotfix 3, IR doesn't feel that much slower than 1.6 on my dual Xeon workstation :)

[romullus]

I don't have problems with IR slowness or unresponsiveness (maybe because my scenes aren't big and complex enough), but i noticed that Corona tkaes much more time to start IR for a very first time then it used to be, even in the most basic scenes. Can't say exactly when i start to notice that, but it is like that for months now. Anyone else has similar impression?

[Christa Noel]

... but i noticed that Corona tkaes much more time to start IR for a very first time then it used to be...

yes, same here

[NicolasC]

but i noticed that Corona tkaes much more time to start IR for a very first time then it used to be, even in the most basic scenes. Can't say exactly when i start to notice that, but it is like that for months now. Anyone else has similar impression?

Same here ....

[PROH]

.... and here ...

[snakebox]

Yup, again I don't have hard data but it would be interesting if anyone have corona 1.3 or 1.2 sitting around to do a comparison with a current scene?

[effilang]

Same, dual xeons also.

[maru]

Guys, once again, it is best to contact us about this issue at support@corona-renderer.com and provide as much info as possible: problematic scene, screenshots, video recordings, anything that would help us understand the issue and reproduce it.
Currently we are not able to confirm any of the slowdown reports. We have already tested the reported scenes on various computers including i7s, dual xeons, threadrippers. Most of the reports rely on a "feeling" that something is slower, without much evidence confirming it. We really need good reports showing us the numbers and the behavior of IR. Only then we will be able to share the report with our QA, developers, and other guys who will be able to look into this.