CompuServe Thread

#Again: still frame error

11 messages in this thread
#178846From: Volker UffelmannJul 7, 1995 8:18 AM
Hi Gary, I 'warned' you that I will return with a report concerning odd errors in animations rendered to GIF or TIF stills. Well, here I am <g>: Despite very excessive and extensive testing I haven't been able to put my finger on the responsible part of hardware/software. I tried various 3DS-Projects on various computers using various setups and I _always_ got errors, not only with GIFs and TIFs but also with Flics. One obvious reason why I didn't see errors in uncompressed file formats is that compression algorithms get very confused (to the point of crashing the system) if a bad byte occurs, whereas uncompressed pictures only show a miscolored pixel (which is quite hard to find in megabytes of information…). Used computers systems: – Pentium 90 EISA, Adaptec 2940W SCSI, various combinations of RAM modules from 16 MB up to 96 MB, Seagate 12450W 2HD, Hercules Dynamite Pro PCI – Intel 486 DX2/66 VLB, Adaptec Total Advantage VL SCSI/EIDE combo, various combinations of RAM modules from 16 MB up to 96 MB, Quantum LPS240AT, Quantum Lightning 730S, no name Tseng ET4000 W32 VLB video adapter – Intel 486 DX2/66 VLB, no name VLB IDE controller, 64 MB RAM, Quantum LPS240AT, Quantum LPS 270 AT, Elsa Winner 1000 VLB video adapter – Novell 3.12 Server Intel 486 DX2/66 VLB, Adaptec 1542 ISA, 32MB RAM, Fujitsu 1GB HD SCSI, DEC 1GB HD SCSI, no name ISA video adapter. Used setups: – Memory managers: QEMM 7.0 – 7.04, EMM386 from WfW 3.11, none – Network: connected to Ethernet (Novell 3.12 drivers, either master or slave mode), unconnected – various TSRs and driver programs activated or removed – 3DStudio Releases 3 and 4 All these tests resulted in errors. <SIGH!> One significant occurence is that frames located more to the beginning of an animation are less likely to be corrupted than frames further into the animation. One particular project which was rendered at least ten times never produced an error before frame 110 and then every fifth frame or so was bad. What could it possibly be that I'm overlooking? I can't believe that there are four completely different computers with exactly the same hardware fault. But I also can't believe that there's a bug in 3D Studio which seemingly only strikes at me and nobody else. After hundreds, maybe thousands (computer) hours of testing I'm beginning to get a bit weary and uninspired in inventing new tests. Maybe someone else is willing to try one of these projects (my favorite) on his own computer? Could the fault be project related? Is it my kismet to _not_ get flawless pictures? <g> Any help, suggestions etc are, as usual, appreciated. Volker
#178854From: David J. MarksJul 7, 1995 8:54 AM
>> One particular project which was rendered at least ten times never >> produced an error before frame 110 and then every fifth frame or so >> was bad. Let me throw some suggestions at you. I fought with a similar situation and spent weeks trying to nail it down. In my case, every third or fourth frame was trashed. There would be a bad byte that would scramble the entire image. What – I – was doing was Net rendering on 3 machines and saving TGA files to a common server. Nearly EVERY file generated on one machine was corrupt. So, I rendered to a local drive on that machine to see if it made a difference. The files looked good, so I copied them to the server — and EVERY one from that machine was now corrupted on the server, but GOOD on the local drive. After doing file comparisons, and copying files by floppy disk, and so on and so on and so on… We discovered that for some bizarre reason, network copy operations were subtracting 02h from every byte!!??!! To make a long story short, we changed our Ethernet card, and everything ran flawlessly again. To see if yours is a similar problem to mine, check your net render log and see if bad frames are coming from one particular machine. Are they always saved to a common server? Do you always VIEW them from a particular machine? Work completely LOCAL on each machine, and see if the problem still occurs. Hope this helps. – Dave
#178887From: Volker UffelmannJul 7, 1995 11:28 AM
David, thanks for your suggestions, but: yes, indeed we've had the problem with a corrupt ethernet adapter, in our case in the common file server. This adapter has been shredded to dust <g> and the new one doesn't show any byte errors when copying large amounts of bytes. As for the exclusively local rendering, yes I've already tried that one to no avail – the output is damaged. I'm sure that I haven't tried _all_ the possible combinations of the listed setups but quite many and the most usual anyway. And that's why I'm beginning to despair slowly but very surely… I guess that makes me a 'desperado', doesn't it? <g> Well, I'm afraid I have to seek new ways in order to get a grip on this #*@! (sorry!) bugger… Volker PS: Yes, the faulty ethernet card did cost me weeks, too.
#178861From: Yost GroupJul 7, 1995 9:30 AM
As David says, it certainly _can_ be a hardware problem. (And in your case, I'm sure it is, because we've had tens of thousands of hours testing that code, and we're positive that it's rock solid.) – G
#178888From: Volker UffelmannJul 7, 1995 11:28 AM
Gary, I believe that you did test the code thoroughly – I didn't intend to sound doubtful on your quality assurance – it's just that I can't see a probable explanation for this phenomenon. One thing, however, do the used computers have in common: all of them are located in the rooms of my bureau – maybe there exists an evil influence, earth rays or something… <g>. I think I'm gonna install 3DS on a computer elsewhere and try again for some weeks… SIGH! Volker
#178984From: Yost GroupJul 8, 1995 12:06 AM
You might check to see if you're underneath an ozone hole. Those have been known to cause massive corruption of data (genetic and otherwise). – G
#179020From: Volker UffelmannJul 8, 1995 10:03 AM
>> ozone hole … cause massive corruption … << heck, NOW I understand the close relationship of the reason of the unprecedented buying-out of sun blockers and the naming of texture space coordinates – rumor has it they cause a lot of trouble, too… And one of these micro-gamma-rays seems to have short-circuited a certain region in my brain… I can't control my typing anymore… now I'm able of conceiving a devilishly ingenious plan… as I'm into testing for the next six months or so anyway… why don't ya send me the R5 beta CD… I'm unbeatable in seeking the unforeseeable… HARHARHARharharhar…. […mad laughter trailing off into infinity…] Volker PS: So much for the corruption of genetic data <G>. Evidence is hot that I live indeed underneath an ozone hole.
#179027From: Yost GroupJul 8, 1995 11:20 AM
(that makes two of us) – G
#179614From: Volker UffelmannJul 12, 1995 4:42 AM
>> (that makes two of us) << gen… gen… genetic fai… fai.. failures? Volker
#179653From: Yost GroupJul 12, 1995 10:24 AM
mutations is more like it – G
#179705From: Volker UffelmannJul 12, 1995 3:24 PM
ah, hence the tentacles on my head… Volker