exxos's Atari, Amiga & retro forum

rubber_jonnie wrote: 22 Aug 2023 15:13 16Mhz failed at EXACTLY the same point. 2376 folders left and 23510 files left.

I'm going to try to finish it again, but this time at 8Mhz, since last time it finished fine at 16.

Very odd. Maybe it's a clue again :shrug:

What I thought it was some sort of speed issue because it is always the folder names which are corrupted.

I did create a large folder amount only, like 1,000 folders to delete, but I did not see any problems with that. But I did not do a lot of testing with folders only.

So I guess if you have time give that one a try ? maybe even copy them several times for 7,000+ folders to test.


These files may work better.

viewtopic.php?p=104135#p104135

All the files are around 1K. So lots more files and folders, so greater chance of it failing.

And it's a big fat nope:


@8Mhz it died after a short time. I didn't see any folder/file name corruption this time.

rubber_jonnie wrote: 22 Aug 2023 15:28 And it's a big fat nope:

@8Mhz it died after a short time. I didn't see any folder/file name corruption this time.

So was that "folders only" test ? Not sure which fileset your using ?

exxos wrote: 22 Aug 2023 15:29

rubber_jonnie wrote: 22 Aug 2023 15:28 And it's a big fat nope:

@8Mhz it died after a short time. I didn't see any folder/file name corruption this time.

So was that "folders only" test ? Not sure which fileset your using ?

No, that was the original set of test folders for all of my tests.

I have just run a 99 cycle DMA test on my MSTE with my test harness and all cycles passed. I'm running it again now with 1000 cycles, but I doubt the result will change.

rubber_jonnie wrote: 22 Aug 2023 15:35 No, that was the original set of test folders for all of my tests.

I have just run a 99 cycle DMA test on my MSTE with my test harness and all cycles passed. I'm running it again now with 1000 cycles, but I doubt the result will change.

:thumbup:


I guess one course of action which could work, if anyone has a TF536/ST536 who could create a large RAM disk and copy my test files to it and delete . This way it would rule out the entire DMA system then, as it would be all within RAM.

Second test, ironically crashed in exactly the same folder name!

2010 - 984 = 1026
3308 - 1600 = 1708

1708 + 1026 = 2734 total files and folders.

It is mighty coincidental as it exactly matches the folders and files it deleted before when it crashed! (2734) .


I will connect the Pico drive backup so I can copy the folders only again and try a large amount next.

deletion completed - did not bomb-out
STe / 2MB RAM / TOS2.06 / HDDriver 11.13 / Blitter disabled / fold220

will test again with blitter enabled now

dad664npc wrote: 22 Aug 2023 15:45 deletion completed - did not bomb-out
STe / 2MB RAM / TOS2.06 / HDDriver 11.13 / Blitter disabled / fold220

will test again with blitter enabled now

That coincides with my findings so far as well then.

exxos wrote: 22 Aug 2023 15:37

rubber_jonnie wrote: 22 Aug 2023 15:35 No, that was the original set of test folders for all of my tests.

I have just run a 99 cycle DMA test on my MSTE with my test harness and all cycles passed. I'm running it again now with 1000 cycles, but I doubt the result will change.

:thumbup:


I guess one course of action which could work, if anyone has a TF536/ST536 who could create a large RAM disk and copy my test files to it and delete . This way it would rule out the entire DMA system then, as it would be all within RAM.

1000 cycles completed with no error, running 2000 now. I'll test just the folders once thats done.

rubber_jonnie wrote: 22 Aug 2023 16:08 1000 cycles completed with no error, running 2000 now. I'll test just the folders once thats done.

:thumbup:

I'm deleting 8009 folders currently.

EDIT:

They all deleted just fine.

:dizzy: