[tom]

Hi Maxer, we couldn't yet reproduce this but trying to figure what could be the problem you're experiencing.

[Maxer]

What do you mean you can't reproduce this, what kind of tests are you doing because I can do it every time I send anything for cooperative rendering.

[b-kandor]

Just to be clear, there are 2 completely different issues in this thread as far as I understand:

Maxer - the mxi saved to the location specified in render options is a higher sl than the sl of any individual node - but not as high as you think it should be.

Me - the mxi saved to the location specified in render options is a lower sl than the sl of my highest node - yet I'm able to find an appropriately high sl mxi file by searching the temp folders of all my systems.

Maxer, in your previous test, can you provide the sl reached by each node and the final sl of the merged mxi?

[Maxer]

All of that info is on the first page where I posted my findings.

[b-kandor]

ok, I see it now

Something is definately not adding up. It's strange because in my case the sl of the merged file seems about right for the time spent rendering, it's just that the file is in the wrong place. Your situation is a comparison thing where 8 cores are beating 22 cores - which suggests a hell of a lot of overlap occuring.....

[Maxer]

I don't know if my setup may be affecting cooperative rendering but this is how I've set it up. I have a dual processor AMD system running Windows XP and Maxwell 32 as my manager machine; all the render nodes I used for this test are dual core Intel's running Windows XP 64 and running Maxwell 64.

I don't know if the 32 bit machine could be affecting the cooperative merger process in some weird way but I was under the impression that you could mix different operating systems and different versions of Maxwell with no problems.

[b-kandor]

Maxer, there is a thread somewhere talking about how win64 is much slower rendering than win 32.... can't remember where.

found it here:

http://www.maxwellrender.com/forum/view ... c&start=30

[Maxer]

Lovely, this keeps getting better and better.

[b-kandor]

I'm using 2 win32 machines (single core each) and 2 dual core machines win64 each so it's really hard for me to tell. My one dual core is dual boot as well - so I'll run the benchmark test on it in both os's and see....

[tom]

Could be a stupid question, so forgive me but when you're adding a job, do you define MXI output path and where?

[Maxer]

MXI output is saved to a network drive that all nodes have access to, in the test I posted the final MXI was successfully written to that drive and in the network manager it showed up as a successful merger. If you have a test scene you would like for me to test please send it to me.

Let me add that this is the same system that I have been running V1.1 on for over a year and have never experienced this problem or anything like it.

[b-kandor]

Could be a stupid question, so forgive me but when you're adding a job, do you define MXI output path and where?

Hi Tom,

I also specify a network path to save the mxi file. I did another test today, the outputed mxi is a low sl, and the proper high sl mxi file is found named 'cooperative.mxi' in the temp folder of the system that is running manager.

Twice in a row it's been this way?

[Maxer]

Please someone give me some idea what is going one with this, I have work that need to be done I can't wait forever.

[Q2]

Man. This is BAD!

Can't get this thing worked out here either. The Windows version doesn't work properly, writing the file somewhere so we have to look for ourselves and the OS X version doesn't do it's job either. Dissapointing that is!

Please NL, let us know when we can hope for an answer or a fix for this situation.

I got tons of work to do and the pressure rises.

Thank you!

Q! Berlin

[Mihai]

MXI output is saved to a network drive that all nodes have access to, in the test I posted the final MXI was successfully written to that drive and in the network manager it showed up as a successful merger. If you have a test scene you would like for me to test please send it to me.

Let me add that this is the same system that I have been running V1.1 on for over a year and have never experienced this problem or anything like it.

Maxer, from what I understood, the problem you are having is that the final SL is lower than expected? This could be due to Win64. We're looking into what could be the problem rendering on this OS.

Can't get this thing worked out here either. The Windows version doesn't work properly, writing the file somewhere so we have to look for ourselves

Could you post a screenshot of the Add Job window? Where does the final mxi get saved? What is the name of the final mxi? Is it the name you gave in the add job window, or something else?