Mental Ray - Fatal Error 031005 ??

[dave-3d]

Like this :

http://www.dmmultimedia.com/errors/00_error.jpg

D.

[dave-3d]

OK .. this is very strange.

I took this scene home and rendered the exact same scene at 4000 x 3000 and it worked perfectly !! That was on a 1.2 GHz 700 Mb Laptop !!

What can possibly be making my twin processered 2 Gb memory system have this memory problem ???????? Is it a problem with twin processor systems ????????

This cannot be a simple memory problem. Can anybody advise ??? Can anybody suggest any tools to find out WHY mental ray is acting this way ????

I have searched the internet and I am not alone in finding this error. Perhaps it is not isolated to memory after all.

This is VERY annoying.

My Killer system falls down but my wee laptop takes it in its stride !! Mad !!

D.

[BrianHarbauer]

Hey man, same problem here when rendering an extreme amount of polygons. Takes about 4-5 min to load the scene on my 2.5GHz HT w/ 1GB ram, over 60GB free on both hard drives. But, from what i remember, it rendered fine (but slow) on my brother's 2GHz single processor with 512mb ram. Could it be that mental ray has some issues with dual processors? I don't know, but i'd be interested also for an answer too.http://www.cgtalk.com/images/icons/icon9.gif

[Tora_2097]

Same problem over here. I posted a thread related to this topic a while ago but nobody was able to give me answers..
It mostly occurs to me when rendering high res images (3000 pixels and above) during the finalgather prepass step. I tried everything out and the only thing left for me to was reducing the resolution...
I would be REALLY grateful to if anyone knows a solution to that...

[dave-3d]

Thanks for the feedback guys !

Lets see if we can see if there is something common that we all have and keep this thread going. I asked Discreet but they said after the few weeks you have for installation they don't offer support ! W.T.F. :shrug: I have also posted on their discussion forum. We are NOT alone with this problem.

Do you guys have :

Twin Processor system [ ]
Over 1 Gb Ram [ ]
Mixed memory - bought extra Ram to increase the original [ ]

I'm [x] to all the above.

I have been experimenting and can render, so far, to at least 4000x3000 on my laptop (1.2 Ghz 720 Mb Ram), while my DELL Precison 450 Workstation with 2 x Xeon 2.4 GHZ and 2 Gb ram falls over.

Here is latest screen grab.

http://www.dmmultimedia.com/errors/01_error.jpg

And here is the output (shortened - LOTS of "outside of a frame" lines removed) from Mental Ray. Does anyone know what is happening ?????

DISP 0.2 error 211005: timeout waiting for finished rectangles
DISP 0.1 error 211020: received rectangle 1584 432 1631 502 start outside of a frame scope
DISP 0.1 error 211021: received rectangle 1584 432 1631 502 stop outside of a frame scope
DISP 0.1 error 211020: received rectangle 1536 432 1583 502 start outside of a frame scope

RC 0.2 info : rendering statistics

DISP 0.1 error 211020: received rectangle 1471 192 1535 239 start outside of a frame scope

RC 0.2 info : type number per eye ray

RC 0.2 info : eye rays 287642 1.00

DISP 0.1 error 211021: received rectangle 1471 240 1535 287 stop outside of a frame scope

RC 0.2 info : transparent rays 166350 0.58

RC 0.2 info : reflection rays 65545 0.23

DISP 0.1 error 211020: received rectangle 1471 144 1535 191 start outside of a frame scope

RC 0.2 info : refraction rays 128 0.00

RC 0.2 info : shadow rays 746320 2.59

DISP 0.1 error 211021: received rectangle 1471 192 1535 239 stop outside of a frame scope

RC 0.2 info : environment rays 2210 0.01

RC 0.2 info : finalgather rays 119196 0.41

DISP 0.1 error 211020: received rectangle 1471 56 1535 95 start outside of a frame scope

DISP 0.1 error 211021: received rectangle 1471 288 1535 335 stop outside of a frame scope

RCI 0.2 info : main bsp tree statistics:

DISP 0.1 error 211020: received rectangle 1471 96 1535 143 start outside of a frame scope

RCI 0.2 info : max depth : 40

RCI 0.2 info : max leaf size : 195

RCI 0.2 info : average depth : 36

DISP 0.1 error 211021: received rectangle 1680 432 1727 502 stop outside of a frame scope

RCI 0.2 info : average leaf size : 10

RCI 0.2 info : leafnodes : 993675

RCI 0.2 info : bsp size (Kb) : 52685

DISP 0.1 error 211020: received rectangle 1536 96 1583 143 start outside of a frame scope

DISP 0.1 error 211021: received rectangle 2400 384 2447 431 stop outside of a frame scope

DISP 0.1 error 211021: received rectangle 2448 384 2498 431 stop outside of a frame scope

DISP 0.1 error 211021: received rectangle 1920 432 1967 502 stop outside of a frame scope

PHEN 0.2 progr: calling output shaders

RC 0.2 progr: rendering finished

RC 0.2 info : wallclock 0:06:57.32 for rendering

RC 0.2 info : allocated 203 MB, max resident 235 MB

GAPM 0.2 info : triangle count excluding retessellation : 235320

GAPM 0.2 info : triangle count including retessellation : 235320

MEM 0.0 info : heap size limit set to 3000 MB

JOB 0.0 info : slave rendering OFF

MEM 0.0 info : heap size limit set to 3000 MB

JOB 0.0 info : slave rendering OFF

MEM 0.0 info : heap size limit set to 3000 MB

JOB 0.0 info : slave rendering OFF



D.

[-Vormav-]

Found this on it:
http://www.highend2d.com/boards/showflat.php?Board=xsirender&Number=159236

Sounds like you just don't have enough memory to handle the rendering job.

[dave-3d]

Hi Vormav,

Thanks, I have seen that link. That is what I first thought.

BUT.

I tried the EXACT same scene on my 1.2 GHz DELL X300 laptop with 720Mb RAM at 4000x3000 and it rendered perfectly (although slowly). It is not a huge scene with many polygons.

I tried it again this morning on my 2x2.4 Ghz Xeon, DELL Workstation Precision 450 with 2 GB Ram at 3000x2250 and it gave me that error. So it CANNOT be a memory problem.

It might be a problem with Mental Ray actually "seeing" my system memory - but there is more than enough memory in the system.

The fact that others have this problem AND that in one other case the same scene renders fine on a lower spec model while crashing on a high spec PC makes me suspicious !

The lack of help from Discreet is NOT GOOD.

D

[JeffPatton]

I just rendered a 6000x4500 test of a diamond scene I've made (Caustics & GI). It completed without any problems. Here's the basics of my PC:

BOXX system:
Dual 3.2 Xeon (1mb cache each)
2048 RAM (2 x 1024 sticks both from same manufacturer)

I wonder if it's because of the different brands of RAM, or as someone else mentioned maybe an odd number of RAM? IE having 3 sticks instead of 4? Either way, very strange and I'm shocked at what Discreet told you. Was that what your reseller said, or an actual Discreet rep?

Jeff

[dave-3d]

Thanks for the reply Jeff. I'll rip out my Ram and have a look - and test without the extra RAM as people have suggested.

I added the extra 1GB RAM as a set from Kingston to take it up to 2GB ordering EXACTLY what was suggested for my DELL model - it adds in to balanced banks.

I just upgraded to Version 7. The reply I got from Discreet was ...

---------------------------------------------------------------------------
Your request for technical assistance could not be processed as your 30 days of installation and configuration support for this product expired on 11/06/2004.
Here are your options to receive technical assistance:

*For assistance and for additional information about Discreet software support contracts, please contact your local Discreet Reseller. To locate a reseller near you, please visit the reseller look-up on the Discreet website at: http://www.discreet.com/resellers

*Our per incident support option with a valid credit card. Please call +44-207-851-8000 for more information.

*Our website contains downloads, valuable FAQs and an updated product knowledgebase at http://www.discreet.com/support

*Our product discussion forum, provided free of charge, is found at http://support.discreet.com (http://support.discreet.com/).

Regards,

Discreet Support
-------------------------------------------------------------------


Looks like they expect me to pay for support. And I feel a bit of a mug having boxes for Version 3,5,6 and 7 on the shelf.

D.

PS. Thanks for the thread on Mental Ray and your website, Jeff :thumbsup: very, VERY helpful.

[PiXeL_MoNKeY]

You may also want to contact your reseller. Any time I have an issue I contact my reseller and he can help me out or will see what he can find out from his Discreet contacts. If your reseller cannot help or is uninterested I suggest looking into another reseller. It doesnt surprise me that they want to sell support contracts as many other companies do the exact same thing with requiring you to purchase a years worth of support to get updates and support from the company. Be happy that Discreet offers different levels of support based on what you need. If Discreet didn't charge for support the cost of the package would go up or things of the package would be effected as they would have to shift company funds to support all 280,000 registered users.

Another option would be to look into the bitmap pager it is a buggy feature in past versions of Max, I haven't tried it in 7, but if used properly it will store pieces of the image to harddrive and save out a final image once it is done. In past versions it would crash Max when the Bitmap pager was enabled anytime you try to preview a bitmap. So disable file input and output preview options and close and don't open the material editor while it is enabled.

My thoughts on the subject,
-Eric

[opus13]

i had a similar thing ahppen not long ago. what resolved it was going from an absolute final gather sample count, to specifying amount and pixel diameter min/max or 35/5 it still looked good, and completed rendering :)

[skinnee]

I've had the same problems rendering 1280x1024 finalgather with 5million polys. It was on a Dual Xeon 2.8 1GB ram. I ended up lowering the fg samples to 250, and lowered the bsp tree depth to 32. My memory usage was still at 1.89 GB physical.

[dave-3d]

Thanks for the tips and feedback guys. I've tried a few already i.e. Bitmap Pager off, BSP lower (No change with my problem).

B U T

My biggest problem is this:

1. PC A is a DELL Precision Workstation, 2x Xeon 2.4 GHz with 2 GB Ram (Thoroughly checked the memory - 2 Banks DELL, 2 Banks Kingston), using dual monitors.

2. PC B is a standard DELL X300 laptop, 1.2 GHz with 720 Mb Ram

With EXACTLY the SAME scene and settings - PC A has a fatal error and falls over, while PC B renders it perfectly.

Therefore, it is not a matter of actual resources on PC A.

This is what is driving me f%^£ing mad !

It is also making me wonder if the errors others see ARE actually related to memory resources OR not. Is it physical memory or graphics memory ??

At the moment I am using my Laptop to render large scenes - and that defies logic. (While being MUCH slower).

I am trying to find out what settings (hardware and software) are wrong with PC A. Otherwise why bother buying a good PC with more memory when you use Mental Ray ?

I will get in touch with my reseller. This is certainly a bug - at least on my machine.

D.

[tyyp]

i have the same prob, i'm using P4 3,4Ghz HT and 2Gigs of ram..
still no workaround?

rgrds,

tanel

[cpan]

As I know since 3.0, MR has an automatic memory management feature that removes unused things from memory when it uses more than you setted in Render Globals and then, if it needs them back, it will calculate again.
So all you have to do is to set your maximum memory setting in RG to a something like 90% of your installed memory.

cheers.

edit :: I don't know if max's implementation of MR has this thing

[dave-3d]

Hi Guys.

Nope, no change here.

I have modified all the correct settings to tell MR to use as much memory as possible. But still it crashes with this error.

I have 2 Gigabytes of RAM (Max Swap file in XP) on a Dual Xeon machine and the only way to render a scene with Final Gathering radius set low and a size of 2000x1500 or greater is to take it home to my Dell X300 laptop with 700kb ram ! How the F££K does it run out of memory on my Dual Xeon system ?? :banghead:

I suspect some bad memory - although I have low-level checked that and it appears fine.

I am building a new machine soon and will experiment with the old machine before I move my 3DS Max 7 across.

Dave

[tyyp]

well, try
bitmap pager on,
manage maps on,
conserve memory
and use placeholder objects.
that helped me.

[zalooa]

I have been trying to render my scene for a couple of days now on 3ds Max 7
It worked fine on 3ds Max 6 and rendered with no problems
But here It is giving me the error you guys are recieving too...
I have a P4 3 GHZ HT and 1GB RAM
As i said before I rendered fine in 3ds Max 6
AS i was desperately looking for ways to render I switched to Default Scanline...As i said i was desperate :)
And the error i recieved for the same scene was not the mental ray error obviously but a RAYTRACER error saying the "Acceleration Control" factors or what not cannot render the scene. So what did i do i looked for raytracer options and found the ACCLERATION Control thing it gave me an error too...So i increased the default settings for the Face Limit to 30 and Voila i was able to render the High polygon model that would not render...Unfortunetely when i added the rest of my scene it did not go smoothly....So i guess there is something that needs to be tweaked on this Acceleration Controller...Im not sure yet but i think it also can be applied to Mental Ray renderer too....Ill let u guys know if i find out anything

[digtlpete]

I have had this error occur as well. I am using a dual zeon system with 2gb of ram. What I did to solve the problem was I cleared out my mat editor of all materials and saved the scene while in a wire frame view port then ro-opened and hit render. I think the error has something to do with video memory or lack of. Give it a try it may work.

[dave-3d]

Thanks for all the replies.

tyyp: Tried all those already with no change for myself.

zalooa: If you find anything let us know.

digtlpete: I'll give that a try, it sounds plausible. What Graphics card are you using ?

Just before my crash I can see the little box squares appear - then it seems to stumble. It can't possibly be system memory (2 Gigs + Max XP swap file on 2x Xeon machine) because it renders fine, although slowly on my laptop (700kb ram).

I'll post my results from clearing the Material Editor.

D.

[digtlpete]

I am using a quadro fx 1000.

[digtlpete]

One more thing I have heard of is that there is a way to modify one of windows system files to allow windows to allocate up to 3gb of memory to applications. Standard windows will only allocate up to 2gb to applications. 1gb to windows and 2gb to applications. By modifying this file windows will grab more memory from your hard drive (virtual memory). I may have this info wrong but it is definetely something to look into. I know we have machines here that are set up in this manner but I am unsure what needs to be done.

[Crew]

The /3GB Switch can be found on discreet's forum. Post your Tips and Tricks / Improve Memory Handling. Don't get your hopes up, thinking it's the fix. I just wanted to include the trick here to close a trail.

[dissidently]

Nearly two years ago I had a similar problem with MR, on one computer it'd work fine, on another not. I traced it to my swap file being on a striped RAID setup and figured that the swap file was corrupting or otherwise stuffing up during the render. Swap files are inherently messy little things. Try moving the swap file to another hard drive in your system, and forcing it to a size by setting min and max size to exactly the same. There is a way to completely disable the swap file that involves setting it to min and max of 0 plus then changing something in the registry, but I can't remember what it was and I've only got 2 gig in my machines so I've never bothered chasing it. Windoze always uses swap files if they're available, even if it doesn't need to.

[zalooa]

Guys I have some VERY GOOD NEWS
While fiddling with the mental ray settings as i was doing for a long time. The answer to our problems...well mine at least was right in fron of me. WHen You hit F10 and the rendering dialog comes up, THere are tabs at the top. I clicked Processing and saw the CONSERVE MEMORY BUTTON...this did not help but right above it and to the righ is the MEMORY LIMIT. IT WAS set to 1500 MB my scene needed more than that so i made it 3000 MB and it RENDERED. THE SAME SCENE that would not render FINALLY renderred. I think this is the problem because I am happy with the results. It would explain why the memory runs out if you tell the program to use only a fraction of it. Let me know if this helps guys
PEACE

[dave-3d]

This is still haunting me !

I have tried EVERY trick mentioned but I don't feel it is actually a lack of memory on my 2Gig RAm 2x2GHZ Xeon Workstation (The EXACT same scenes renders, albeit slowly, on my low spec laptop).

I upgraded to the newest MAx 7.5 - after installing SP1.

I tried to push my image resolution up to 3000x2500 and all it did was draw small white boxes !?

So, I tried again at 800x600 and CRASH. AGAIN. I still get this message :

mem 0.x1 fatal 031008 : cannot allocate x2 bytes

With x1 being either a 0, 4 or 6 and
x2 being anything from 16 to 50 Mbs.

I was watching the memory use as it was rendering right up to the point it crashed and there was over 700 Mbs physical RAM available ! And NO use of any swap file. EXACTLY what 20mbs can MR not allocate ??

http://www.dmmultimedia.com/errors/memory_mr.jpg

I'll try stripping out memory modules now ..

It SHOULD be possible to render a scene at 4000x3000. How the HELL do ILM render out their MR scenes ??

Dave

[cpan]

How the HELL do ILM render out their MR scenes ??

they don't use max's mr :D

[dave-3d]

I can see clearly why they wouldn't want to use Max's MR !

Just start to render Yoda - CRASH.

We would have to wait a lot longer than May 9th for a movie !
Dave

[cpan]

well, at this point, max's mr isn't very well implemented, it doesn't support many features of mray and thus is slower than the standalone version.

.those at ilm use mray standalone for the final render.

[mienda]

I have a dell precision 650 dual 3.06 with 2 GB and i had a big problem with memory in after effect to render 2K pictures ( "not enough memory to create.....".) After many many research in forum i found a solution with the /3GB switch ini ( the crew's post).Perhaps you have already installed it , if not try it because i think it could be the solution for the 2Gb system (i'm almost thinking that you should not had any problem with only 1 GB on your dual xeon).
This is the link of the adobe forum: http://www.adobeforums.com/cgi-bin/webx?13@235.H0SCeVSehwn.0@.3bb50d64

And a link for the 3GB : http://www.eyeonline.com/Web/EyeonWeb/Support/TechFAQ.aspx#15

I hope it could help you

[Jasont028]

Hey Dave_3d,

I have recently come up with this problem also and am wondering if you have fixed this problem on your machine? I am running a Amd 3300 with dual 512mb Corsairx Ram.

[dave-3d]

Meinda : Thanks for those links. I have tried that on a borrowed laptop and I can create 4000x3000 images at high-res.

But in my DELL when I enable the /GB switch it crashes or locks-up ???!

So, no Jasont028, I have not yet found out what is the problem. I still get error messages on my DELL and MR crashes and I have to restart.

I think it is something to do with the memory handling by the DELL.

Either way it is F$*&ing annoying. I can use my Dual processor Workstation for testing and have to take it to my small laptop to make a final render.

Dave

[mienda]

hello Dave_3d,
it is strange because on my dell 650 i have no problem with /3GB switch.
Have you try the Diagnostic Dell's CD then you could test your ram ?
Perhaps try with only 1 GB.
Is your system ok( no conflicts, no other applications bugs..?)? if you have windows service pack 1 the /3gb switch needs a patch to work ( see the microsoft site ) but if you are on SP2 there is no problem.

i hope it could help you.

[phlesch]

I've got a similar problem with max 7 and mental ray, so I'm just throwing this out to see if there's a common thread.

I've got a new system, asus m-board, dual 3.2 ghz xeons with 2 GB ram (matched), nvidea fx500. I have a striped RAID, but it's not selected for my pagefile.

The offending geometry seems to be a large-ish scatter object. When I reduce the number of objects being scattered I can get it to render. However, as I edge up the number, it crashes before rendering. Thing is, the scene is only using just over half of my RAM at render time - I've done much larger scatters in the past with smaller systems and the scanline renderer (indeed, max's scanline has no problem rendering this same scene).

I'm only trying to render at 1280x1024 and I don't have any large maps in the scene. The scene is under 5 million polygons at render time (I say "at render time" because I know that scatter objects are created at render time and can therefore "balloon" dramatically. I'm referring to RAM usage as given by the renderer while it's rendering, not in . I'm not using any caustics, indirect lighting, raytracing, or sss materials (ie. it's a "straight" old max scene). In My "memory limit" is set to 2000 MB.

[phlesch]

Mienda!

Thanks so much for finding that alteration to the boot.ini file (adding the /3GB switch) in XP. It seems to have fixed my memory problems while rendering with mental ray on a dual xeon with 2GB of ram.

If anybody else is having this specific problem, here is the link again, as posted by mienda:

http://www.eyeonline.com/Web/EyeonW...TechFAQ.aspx#15 (http://www.eyeonline.com/Web/EyeonWeb/Support/TechFAQ.aspx#15)


Lars

[john_the_eagle]

I`ve got "mental ray has encountered a fatal error and the application will now shutdown. The error is:

MEM 0.2 fatal 031008: can`t allocate 300161424 bytes. "

I try with "/3GB" and other.... but always with no efect :/


Can anyone help?

[yamyam]

Same problem here on a brand new DELL Precison 670 Workstation with 2 x Xeon 3 GHZ and 2 Gb ram, trying to render on Mental Ray under Max7 and Windows XP.

In the same way, the scene is rendered on old notebook single old P4 1Gb RAM, same softwares and Win XP, taking days; so I guess it's not a software or file problem. I push down the resolution, until 600x480 and the Dell still not able to render it!!

I have already tried all tips on this topic, but nothing works here. I apreciate any sugestion.

"The error is: MEM 0.2 fatal 031005: cannot free memory"

[cpan]

Same problem here on a brand new DELL Precison 670 Workstation with 2 x Xeon 3 GHZ and 2 Gb ram, trying to render on Mental Ray under Max7 and Windows XP.

In the same way, the scene is rendered on old notebook single old P4 1Gb RAM, same softwares and Win XP, taking days; so I guess it's not a software or file problem. I push down the resolution, until 600x480 and the Dell still not able to render it!!

I have already tried all tips on this topic, but nothing works here. I apreciate any sugestion.

"The error is: MEM 0.2 fatal 031005: cannot free memory"

i guess it's max 7.0's fault then, not mray's. Try updating to a newer/better mentalray implementation

[dave-3d]

Nope - I have the latest version of 3D max and MR and it still does it.

Interestingly, when I push my Dell with a large Stitcher scene it again falls over complaining of a memory problem - and again the scene works well on my laptop.

It must be something to do with the DELL.

Dave

[yamyam]

The local Dell support said: --We don't know nothing about software!

I guess any child is able to give us the same .... support!?

Strange: the same software runs well on other machines, including older dell, at any resolution, while the Precision is able to do a maximum of 400x300; larger than that, it crashes!

Shame on you, Japan Dell!

[dave-3d]

I've decided to build a new PC - will not be going with DELL for my own Workstation again (they're fine for corporate machines).

Just sourcing parts now. One priority is a quality case with good quiet cooling. My dual DELL Precision sits like a vacuum cleaner when it gets hot. Ridiculously loud and annoying.

Dave

[LoneRobot]

Hi Dave,

I had to get up to speed on this thread since i hadn't read any of it before. It sounds to me that it's an issue with your Dell Workstation, not the Implementation of MR in Max, simply due to the fact that your laptop, with a fraction of the power can render it (albeit slowly)
Try removing your ram and testing each stick one at a time in bank 0. (it should be marked on your MOBO what that is) If you still get the issue id see if you could borrow some ram if possible to try to ascertain if it is the problem. If it's a dodgy stick of ram then it should, like the laptop, just render it more slowly with the one good stick of ram if that's the problem.

My other question is how many buckets are you seeing? With a dual processor you should get two buckets rendering at a time. My Dell utilises Hyperthreading, so in MR i get 4 buckets rendering a frame. If you are using hyperthreading on the Dell then id suggest turning it off if while you work out if its the RAM or not. Im no expert on these things but disabling any form of acceleration is advisable until you can locate a problem. The laptop which renders fine would not be using any thing like that any way.

[piu]

Try to use backburner ( net render) with Split Scan Lines check on and divide per 2 or 4.

First you can create a .pmap and .fgm files with a low resolution (e.g. 320 X 240) and then re-use them.

[thearchitect]

Hey looks like this thread has been going on for awhile. I think I might know what your problem is. Its your video card. What kind of video card do you have? I had a problem very similar to this about a year ago and it was pissing me off. I too had 2gigs of RAM and it was not even close to utilizing it all when it would crash. Anyway after trying everything else I decided to swap out my video card thinking that it would do nothing after all the video card shouldnt have anything to do with rendering. However it fixed the problem. It was a radeon 9800xt 256mb, i swapped it out with an old nvidia 64mb ti and it rendered just fine even though it was a far worse card. I was not using mental ray just scanline. I would consider trying to swap out the video card just to test. It is reletively easy to ask a friend to borrow one for an hour or something. Good luck and what kind of vid card do you have?

[dave-3d]

Thanks for all the comments.

I have just built a new workstation for myself and had no problems with any scenes :) It flys. Packed it with 4 Gb Memory.

So the DELL can be relegated to menial tasks.

I might have a crack at changing the graphics card - but I think I will reformat the HD and install EVERYTHING from new.

Dave

[o-h]

Argh, I had this problem too. Strangely, it disappeared when I deleted my light source
(mr omni) and replaced it with a spot.
I really don't know if the mr omni was the true source of the problem, and I might experiment to find out (it will take a bit of time). I was just pleased that the crashing stopped.

[banzato]

I have the exactly same problem and my machine is a p4 2.5 with 1.5 ram. The problem is not with memory and not with textures becouse i had a procedural textured scene and in a weak machine the render was ok. I'm trying a lot of things and I hope to find what is this fuc* message. I tryed 3 versions of Max and 2 versions of MR. Come back soon for results.

[tanel]

oh!, any results?

[bchan3rdi]

Guys,....I just ordered a Dell670 Dual 3g Zeon 2Gigs o ramo today!!! Im having the same issue on my franken-pc Int-3g 1gig Ram. I went is search of answers and found this! I think im going to cancell my order ASAP! Anyone have this similar Dell setup that DOESNT get this memory issue? Im best putting my 4gs elsewhere I suppose. :( Sorry to see all your issues guys,..I know feel your pain.

Brian

[dave-3d]

I have new machines now.

But I am going to completely re-configure and install my DELL from scratch and I'll let you know what I find.

I also have a problem loading in Stitcher files that load fine on my laptop (~700MB RAM), but crash with memory errors on the DELL (2GB RAM). It just doesn't make sense. Even with the 3GB switch.

I'm pretty sure it is a memory configuration problem with the DELL.

I'll let you know my findings when I have re-configured/re-installed my DELL in a few weeks time.

Dave

[Clanger]

I have this memory error when I try and render anything over around 3500 pixels (varies a little from file to file) On two machines one with 1 meg of ram one with 2, nether are Dells.

I can only render this high with 'conserve memory' switch on otherwise the maximum res is a fair bit lower.

I had this problem ever since Max included Mental ray, can't remember which version now, so long ago.

Why isn't it fixed!!! They had so much time to do something about it, and no suggested fix has worked for me.

[RickStefani]

You guys may not believe this but I found a fix for the problems we have been having. My school would render the scene only with 1 gig ram but mine with 2 gigs would not. Your laptop would but your workstation would not. What is the deference? Let me guess, you laptop does not have a graphics card and has to run software texturing? No open GL or direct X9. My computers at school default at software and I forgot to change the setting. I got hope, Set the graphics mode to software, the scene that crashed every time last night now renders fine and fast. Might just be dumb luck for me and not the problem but who knows.

[bchan3rdi]

Rick, thanks for that info. I will try this later in the day. I hope it is the solution. I ended up going with my dual Xeon and yes,..it crashed the scene before. Grrr (still a nice machine though) Lets hope this works,..thanks for sharing. I'll post back later with my findings!

[logotomie]

Ok, had the problem in our art department myself. Imho we are talking about at LEAST 2 problems here:
1st people who simply dont have enough memory for their scenes or too large scenes
2nd people who have plenty of memory and still cant render especially high res images


we are of the second category here. Mental ray usually fails with an error message about not beeing able to create a temp file caled "fb000.something" because it does not have enough memory. This alone is strange because creating a file handle needs very little memory and mental ray tells us that it used only about halve of our system memory. Our guess here is that this is a so called "memory mapped file", a file that is mirrored in the main memory. So what actually happens is that mental ray tries to allocate a big chunk of memory in one go and fails. The bigger your screen resolution is the bigger is the chunk of memory mental ray needs, in our case about 100MB and thats the problem. Every application in 32Bit Windows gets max 2 or 3 GB of memory. mental ray is not a separate application but runs within max and uses the memory Max got assigned. Max itself is a memory hog and uses some 100 MB itself. Before you render max has to load a scene, create a internal representation and display it on screen. So it does already do a lot of memory operations which fragment the memory. Mental ray has to live with that fragmented memory and although everything works well as long as it allocates small chunks of memory at a time, it might fail when it tries to allocate a 100MB big chunk simply because the memory is already too fragmented.


Solution? The best solution might be to render it standalone so mental ray gets a fresh and clean memory space.
If you insist rendering from within max, start rendering directly after you start max, switch to software Z Buffer (Opengl seems to fragment memory more, thanks RickStefani, worked for me), try the the 3GB swith so the fragmentation is slower, and if tall this does fail, think about how else you can reduce the memory usage of max itself. One of our artists had to disable the splitting of his viewport and display everything as box. After a Max restart his scene rendered. Sounds like voodoo, doesnt it?

hope it helps
Jan Jordan

[Werre]

I tried using the grid method of raytrace accelration and it took care of my extremely large file (15 xref'ed buildings into the major retail center site) problem. I too had the "out of memeory" issue. This grid method worked the best, hands down.

I hope this helps someone else.

[Rockin]

Damn, I wish I had seen this thread yesterday :(.

We were using a fg map for our 30 sec animation (fg samples set to 200), and after about 100 frames max crashed with that error. We have tried it on two P4 dualcore machines (one @ 2800mhz and one @3000mhz) with 2GB of RAM - the same make; GeForce 6600GT videocard. There were still about 500MB free when it crashed, and setting 'conserve memory' and all the other options on that pannel only helped rendering a few more frames.

We had no choice but to edit the flickering this caused in combustion and premiere, and live with that since the deadline was tight.

I will try as soon as possible the latest tips, and post again to tell you which helped.

[rasamaya]

For error 211005, I took your advice and deleted my ambient light, but instead of creating a different one, I just created a new ambient light and made it the same color and intensity as before. I be very happy - Thank you

[unknowngod430]

somethings telling me it's with dells ram, maybe something to do with the voltages on it... it could very well just need a voltage boost, because the ram might just be shitting out. however it's weird that i've never had the problem occur till yesterday

[ronok]

I've been having this very same problem........... I tried the idea of changing back to software graphics, but, that just got me past frame two and into 98 (of the 100/750 I set it to render) and died on me anyway... So I just ended up wasting 15 hours. Has anyone come close to finding out whats wrong for sure? :'(

[jacquess]

error here :(
i wanted to speedup rendertimes so i'm using mr distributed render on my neighbour pc.
my comp:
amd sempron 2500+
2x512 MB RAM Kingston
nVIdia 6200 (some lowend)

neighbour:
intel D 2,5 GHz (or something like that)
2x1024 MR RAM
some ati graphic card

now when i'm rendering only on my comp, everything goes ok, when i'm using distributed rendering, it crashes (resolution doesn't matter).

well i think there is some bug in final gather radius settings when rendering on more proc's.

i tried this:
tried to render with distributed rendering in varions resolutions with final gather on 150 samples and radius set on 2mm, min radius set on 1mm
always crashes

tried the same scene only on my comp (single core proc.) - seems to render it

tried to uncheck radius and rendered it with distributed rendering - seems to render it

seems like mr has problem rendering final gather with radius on more than one core?

edit: damned, i tried 4000x2500 and it crashes :( so still none solution :(

[c3drundi]

I find that on my dual 2G Xeon with 3G's of RAM... that the buffered image file gets too fraged I also found that most defragers don't really work... the only one that seems to do something is Diskkeeper 9 or 10. in the boottime defrag. my work around is to run Diskkeeper 2 or more time in normal mode then (step 2) I create a simple box in Max turn on MR and render a 7600 by 5200 image of the box (13 by 19 paper at 400 dpi my normal print/rendering size) this creates a clean new buffer image. then I go back to diskkeeper and setup a bootime defrak and also go back to windows and set a checkdisk... reboot and diskkeeper does it's magic...
I still have to watch my menory (many of my sceans have huge texture maps and the poly count goes up to 6 million faces some times more... and then I have to reduce my image size.) sometimes I even save the Max file in box mode and restart the system clean just to get the max. res. out of my system.

Why that file is such a problem to defrag... I just don't know... but this type of problem goes back to 3ds 4 (no not max 4 but 3D Studio Ver. 4) it's how windows handles memory... the more mem. you have the worse this problem gets I even sometimes run diskkeeper in the background while it is rendering a really big file.

[depleteD]

I have this problem as well, I have dual 3.0 with 2 gigs of ram. My scene is heavily raytraced with fg. I think MR has trouble dumping memory. I have a scene tha thas heavy particle flow geometry and raytraceing. So mr has to translate this into something that it can read. When I crank the number of particles, i get 2 frames. If i halve them, I get about 20 frames before a crash. I know for raytrace you need to play with the accelerator to work so I will try a couple things. Right now I'm gonna reboot to the 3gb switch I read about.

[mateley0]

My machine is basically the same spec as Dave_3d's although i'm running MR for Maya 8.5. Had exactly the same problem and have been pulling my hair out all weekend. The alterations, mentioned previously in this thread, to 'boot.ini' in windows xp pro worked a treat! Many thanks,
Matt

[chrisnobles]

Ive just had this problem, seems anything over 3000 and mr doesnt like it.

i think the solution is to run backburner (making sure you have no render time outs!) and split the scans...

[Snuurtje]

My machine is basically the same spec as Dave_3d's although i'm running MR for Maya 8.5. Had exactly the same problem and have been pulling my hair out all weekend. The alterations, mentioned previously in this thread, to 'boot.ini' in windows xp pro worked a treat! Many thanks,
Matt
Can you tell me were this post is? cause i cant find anything about boot.ini that seams to work previously mentioned in this thread.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.