Magic Lantern for 5D Mark III - Alpha 3 (1.1.3)

[a1ex]

Yes, with all cards, but especially with that card.

[Zerokenshi]

Is it a safe test? (5d3, Transcend 32gb 400x)

[a1ex]

Source code is included, I think it's 100 times safer than any regular ML build. No other ML code is running besides that code snippet.

[Zerokenshi]

Edit : Just noticed the 1.1.3 firmware requirement, off to update first.

[RenatoPhoto]

Here are the results with 5Dm3

Name  ---  card mode ----------- size bytes ----------- --------------

SanDisk   sd-32GB 45 MB/sec   1025000   1048576*
Lexar Pro   cf-32GB 1000x                  x        x           test did not comple, lockup, one time it showed err 02 in live view
SanDisk   cf-32gb-60MB/s   102500   1048576*
SanDisk   cf-4gb-45MB/s               x        x           test did not comple, one time it showed err 02 (not in live view)
SanDisk   cf-16GB-30MB/s   102500   1048576*
Kingston   cf-4gb 133x (20 MB/sec?)   x          x          test did not comple, lockup, one time it showed err 80 (not in live view)
Kingston   cf-512K-  6MB/s   102500   1048576*

[a1ex]

Hm, that's odd.

Preliminary advice: if you care about your data, try not to use the cards that failed the test in 5D3 (with or without ML, doesn't matter). You can use them in 5D2 though.

Could it be any connection between this and the reports of corrupted videos? (I remember seeing some on canon rumors).

Edit: here are my results:
- Kingston 32 GB 266x CF -> OK (1025000 bytes)
- Kingston 16 GB 266x CF -> fails (tried to format it a few times). The red LED stays on for about 1 minute, then takes the picture, then err20.
- Kingston 2 GB class 2 SD -> OK (kinda slow, but works)

[Lcrusher]

Transcend 400x 32gb ----- size 1,025,000 bytes ------ no error

[Zerokenshi]

Transcend 32gb 400x CF , 0.97 MB (1,025,000 bytes) , No Error.

Edit : Also tested on a Kingston SD 32GB Class 4. Same results as above.

[RenatoPhoto]

One of the card that fails: Lexar Pro 1000x is the one I used the most and never had any data corrupted yet!

[wintix]

edit: got it to work, see later post.

[a1ex]

Here are some reports:
http://www.canonrumors.com/forum/index.php?topic=8732.0
http://www.canonrumors.com/forum/index.php?topic=6191.msg119814#msg119814
http://www.canonrumors.com/forum/index.php?topic=5360.0

It's probably something that doesn't happen every day, but only with certain settings or usage patterns.

I'm looking forward for comments from those users who had problems with ML on 5D3 (there were a few reports).

[a1ex]

Please keep posting the results, even if other users have already posted the results from the same card as the one you use.

I need as many test results as possible in order to draw a meaningful conclusion.

On my card I was able to make an workaround, by slowing down the CF write routine. It can be fixed if you help me.

[CaptainHook]

Here's my results:

Lexar 1000x 32gig CF (#1) > OK (1,025,000 bytes)
Lexar 1000x 32gig CF (#2) > OK (1,025,000 bytes)
Sandisk Extreme 60MB/s 16gig CF > OK (1,025,000 bytes)
Transcend 133x 16gig (#1) CF > OK (1,025,000 bytes)
Transcend 133x 16gig (#2) CF > OK (1,025,000 bytes)
A-Data 'Speedy' 8gig CF > OK (1,025,000 bytes) - (Not sure on speed of this card - prob crap )
Panasonic SDHC (4) 16gig SD > OK (1,025,000 bytes) - (Not sure details of this either - i normally use this SD card for ML)

I ran all the above tests by putting the actual test firmware on the individual cards without having the SD (from bottom of list) in the camera running ML like i have been with the alphas - just to be sure it was only testing those cards.

None of my cards seemed to do anything odd - the led blinked.. then went pretty solid as it wrote the YUV422 liveview buffer file (or whatever), then took a photo in live view mode and exited back into live view. Did i do something wrong or am i just lucky with all my cards? I DID format every card before adding the firmware and testing, maybe that helped?

Thanks Alex!!

[wintix]

Aaaah! Got it.

My problem was that I had the cam in photomode. I just redid the tests with videomode, only the fir file on the cards and it works as it should. Swiftly boots, blinks, takes a pic and doesn't freeze. Also the testfile is on all cards being its 1,025,000 bytes in size.

That is again:

Transcend CF 64 GB 400x and 600x
INX 1GB SD (sloooow and old)

[RenatoPhoto]

Here are the results with 5Dm3

Name  ---  card mode ----------- size bytes ----------- --------------

SanDisk   sd-32GB 45 MB/sec   1025000   1048576*
Lexar Pro   cf-32GB 1000x                  x        x           test did not comple, lockup, one time it showed err 02 in live view
SanDisk   cf-32gb-60MB/s   102500   1048576*
SanDisk   cf-4gb-45MB/s               x        x           test did not comple, one time it showed err 02 (not in live view)
SanDisk   cf-16GB-30MB/s   102500   1048576*
Kingston   cf-4gb 133x (20 MB/sec?)   x          x          test did not comple, lockup, one time it showed err 80 (not in live view)
Kingston   cf-512K-  6MB/s   102500   1048576*

OK, ALL WORKING NOW! 

UPDATE: I put the camera in VIDEO MODE and test all again.  All cards tested a lot faster and passed, no hangups.

[CaptainHook]

^-- What did you do differently?

[a1ex]

On my camera it doesn't matter in what mode I run the test - the 16GB one always fails with ERR02, the other two are OK.

If you run the test with display off, the camera will start in powersave mode, so it will appear locked. Press INFO to turn the screen on, and should be OK.

So... make sure you run the test with display on.

[pascalc]

SDHC Sandisk Ultra - Class 4 - 15 MB/s - 16 GB / No error - TEST.DAT = 1 025 000 octets  2 cards tested
SDHC Lexar FULL-HD VIDEO - Class 8 - 8 GB / No error - TEST.DAT = 1 025 000 octets 1 card tested
CF Sandisk Extreme - 60 MB/s - 32 GB / No error - TEST.DAT = 1 025 000 octets  4 cards tested

[a1ex]

So... only my card is buggy?!

[a1ex]

I'll add this card test (without picture taking) in the next alpha. It will delay the startup by 1 or 2 seconds, but I think the integrity of your data is a lot more important than a small delay.

If the card will be diagnosed as "not compatible", ML will refuse to run.

Code, for reviewing: https://bitbucket.org/hudson/magic-lantern/changeset/5fcc91e8d936

Maybe it's just my card, maybe not, I don't have enough info to tell, just incomplete theories. We'll see.

[wintix]

will it start up cleanly without needing liveview?

[nanomad]

Yes, it should start up without livewview

[platu]

I have been testing Alpha 2 and find it to be really stable.  Just 2 things that don't seem to be working for me...

1.  When I use 1280 60p mode (ALL-I or IPB), I get no overlays at all (no zebra, focus peaking, etc).  When I bring up the magic lantern menu, it says "GlobalDraw is disabled, check you settings".  The overlays work fine in 1080 24p/30p modes.

2.  When no cropmarks are selected, and when returning to live view after previewing recorded footage, the top and bottom info bars lose their black background color and become transparent.  The info is still displayed... it's just displayed over the live footage.

As a side note, I know you're still early in testing, but auto boot would be my wish list addition as soon you feel it makes sense. 

[a1ex]

1. go to GlobalDraw to see why.

2. can you tell me the exact sequence of actions to reproduce it? (I've never noticed that)

[platu]

I left out a critical bit of information...  this only happens when using my SmallHD DP4 monitor via hdmi.  Works fine on the camera's built in lcd.