Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 275927 times)

Possie

  • New to the forum
  • *
  • Posts: 5
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3250 on: September 15, 2022, 03:47:45 PM »
Holy cow ! Redid the benchmarks.

Samsung hitting 99 write speed - high 80’s read. Insane !

SanDisk 92/85 write/read (most consistent performer)

Lexar does only high read speeds 80’s - low write 50
After low-level format of the card it’s reversed -barely hitting 60 write. It’s a wonky card I use for my M2. 

That play button trick worked a charm

gabriielangel

  • Member
  • ***
  • Posts: 102
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3251 on: September 15, 2022, 03:50:09 PM »
Something is worth taking into account:

You need to restart the camera, load a new preset and try to record a short clip at least twice after copying the new files to the SD Card.
If you don't, you may get the impression that it isn't working, but it maybe will...

Does anyone knows why?

In my case, at first I got a few error messages, saying "frame corruption" on screen, but after restarting, even my Sandisk Extreme 256GB (Red and Gold) works @240MHz!!!!!
bench10-Sandisk-Extreme-256" border="0

The Sandisk Extreme Pro 170 MHz 128GB which did not work a few days ago @240 also works now!
Will report frame corruptions later.


Here are the error messages I got before the 3rd restart, if it can be of any help:

ML ASSERT:
0
at mlv_lite.c:2892 (compress_task), task compress_task
lv:1 mode:3

compress_task stack: 1de948 [1de9d8-1dd9d8]
0x0009EC30 @ bddb24:1de978
0x0009E558 @ 9ec8c:1de948

Magic Lantern version : Nightly.2022Sep15.EOSM202
Mercurial changeset   : e4663790da85 (crop_rec_4k_mlv_snd_raw_only) tip
Built on 2022-09-15 10:08:56 UTC by da.
Free Memory  : 219K + 1891K

ML ASSERT:
0
at mlv_lite.c:2873 (compress_task), task compress_task
lv:1 mode:3

compress_task stack: 1ded50 [1dede0-1ddde0]
0x0009EC30 @ bd6d10:1ded80
0x0009E558 @ 9ec8c:1ded50

Magic Lantern version : Nightly.2022Sep15.EOSM202
Mercurial changeset   : e4663790da85 (crop_rec_4k_mlv_snd_raw_only) tip
Built on 2022-09-15 10:08:56 UTC by da
Free Memory  : 219K + 1890K

ML ASSERT:
0
at mlv_lite.c:2873 (compress_task), task compress_task
lv:1 mode:3

compress_task stack: 1ded50 [1dede0-1ddde0]
0x0009EC30 @ bd69b0:1ded80
0x0009E558 @ 9ec8c:1ded50

Magic Lantern version : Nightly.2022Sep15.EOSM202
Mercurial changeset   : e4663790da85 (crop_rec_4k_mlv_snd_raw_only) tip
Built on 2022-09-15 10:08:56 UTC by da
Free Memory  : 219K + 1895K

ML ASSERT:
0
at mlv_lite.c:2873 (compress_task), task compress_task
lv:1 mode:3

compress_task stack: 1ded50 [1dede0-1ddde0]
0x0009EC30 @ bd6920:1ded80
0x0009E558 @ 9ec8c:1ded50

Magic Lantern version : Nightly.2022Sep15.EOSM202
Mercurial changeset   : e4663790da85 (crop_rec_4k_mlv_snd_raw_only) tip
Built on 2022-09-15 10:08:56 UTC by da
Free Memory  : 220K + 1911K

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3252 on: September 15, 2022, 04:02:38 PM »
Samsung hitting 99 write speed - high 80’s read. Insane !

Lexar does only high read speeds 80’s - low write 50

That play trick worked a charm

Play mode is a method to get card/cam limit without that much overhead by running tasks. You won't get near those numbers during recording.

We need a little more info about your cards.
Samsung prints a string like MB-SD128K on card's contact side. This string is necessary to tell card generations apart and that can make a really big difference!
Lexar is not that easy to read. Manufacturer product string goes like LSDMI32GCBNL1000R. You may find the string on your bill/invoice or somewhere on blister.

Possie

  • New to the forum
  • *
  • Posts: 5
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3253 on: September 15, 2022, 04:12:06 PM »
Thanks for the help Walter.

Samsung invoice reads MB-SD128K/EU

The 4 min test in 5K mode passed - even with the whole screen filled with blown highlights it stays orange with occasional red.

Lexar card is from the GH2 hack days so pretty ancient. No blister package or invoice. I’d say it’s irrelevant to test my specific old card further.


Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7504
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3254 on: September 15, 2022, 04:26:52 PM »
Something is worth taking into account:
Could be several things causing these issues. What preset used, fps, bitrate, using histogram etc will affect what is going on.
Do all tests without any liveview extras and if you then decide to run all bells and whistles it's up to you but tests with extras on will not be a good idea.
Thanks for reports

theBilalFakhouri

  • Developer
  • Hero Member
  • *****
  • Posts: 1067
  • UHS-I
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3255 on: September 15, 2022, 04:34:30 PM »
Yeah, thanks all for your reports guys!
If you face another issue with 240 MHz, you know where to find me :P

... I started to smell something :D Which was the hot sdcard so I can definetly confirm higher temperatures.

I remember my first day too with 700D and 240 MHz :P
It seems my 100D get hotter than 700D, due to compact size and close distant for internals hardware.
Waiting some new DIGIC 5 models . . .

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3256 on: September 15, 2022, 06:25:27 PM »
And another one stable @240 MHz:
Samsung PRO Plus 256 GB, R160/W120 MB-SD256KA

Seems to have same performance as 128 GB variety. Lower capacities should perfom not as well according to Samsung specs.

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3257 on: September 15, 2022, 07:38:12 PM »
Adata Premiere One V90 (UHS-II): Sadly not reliable at write for 240 MHz. Once stellar 98.0 MByte/s then in the eighties. Or other way round.
Insanely bad price/performance ratio for UHS-I anyway.

Papa-Yeti

  • Just arrived
  • *
  • Posts: 1
Re: Danne's crop_rec_4periments for EOS M
« Reply #3258 on: September 16, 2022, 01:07:35 AM »
Adding an SD card test result to the discussion.

SD Card: Lexar Professional 256GB 1066x SDXC UHS-I (Silver Series)
Read:Write (OEM claimed): 160MB/s:120MB/s
Preset/bit rate: 5K anamorphic frtp / 14bits
Resolution/Aspect ratio: 1264x2134 / 1:2
Recording Time: User stops @ 2 minutes



Will keep testing this card with other resolutions and report back but as of now, I'm happy with the current hack. Thank you, developers! :D

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7504
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3259 on: September 16, 2022, 05:45:03 AM »
Thanks for testing Walter.
I uploaded a build with the latest code which seems stable enough coming from reports. Please keep on testing and also report if for instance 192Mhz isn´t working, especially if it was working before and so on.
Expected cards now fully working are sandisk extreme pro cards 170Mb/s and probably also 95Mb/s. Seems a few other cards do work as well but most common card seems sandisk atm.

roterabe

  • New to the forum
  • *
  • Posts: 4
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3260 on: September 16, 2022, 07:15:52 AM »
Samsung prints a string on card's contact side. Something like MB-MC128G. We need this string to tell which generation is in use because it can make a big difference!

Ehm ... do you mean "I tested this card with 240 MHz overclocking and it works"?

Right then, sorry. The Samsung Evo Plus MicroSD manufacturer code is MB-MC128H.

And yes, you are correct. I did test 240MHz with that result.
It seems stable for the moment. For a rather cheap card I had lying around, it’s doing well at 14bit 1080p and slightly below 2.3K 12bit. Anything above that is pretty much impossible, but still. Impressive nonetheless.

gabriielangel

  • Member
  • ***
  • Posts: 102
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3261 on: September 16, 2022, 08:04:32 AM »
Some tests to check for corrupted frames.

Sandisk Extreme 256 Micro SD (Red and Gold) SDSQXA1-256G-GN6MA

crop_rec_4k_mlv_snd_raw_only_2022Sep15.EOSM202.zip
Didn't touch anything, except for the 2.39:1 Aspect Ratio.

Results: (Click to enlarge)
Sandisk-Extreme-SDSQXA1-256-G-GN6-MA-Report-2" border="0

Camera Temperature peaked at 54 Degrees C.
I did not notice anything obvious as far as image quality goes, which could have happened because since the camera gets hot to the touch, the sensor could run hotter too.
The card stopped recording when I was close to full capacity (About 10GB free space left). At first I thought it was because of the heat, but only erasing all the files on card restored the speed.

There is a drop of speed when you get in the vincinity of 54 Degrees C. Happened after about 30 minutes recording the 30 seconds clips, sooner when recording 3 minutes clips (Those are still copying to disk as I am writing this).
Taking the card out and blowing air on it fixed it.
I only got corrupted clips when I got close to the full capacity of the card. Everything else is corruption free.

Method:

I shot the following chart at 3k, 2.8k, 2.5k; at every bitrate. (If someone has a more appropriate chart, please let me know)
Since the moiré happening because I am recording an LCD screen has some movement into it, I guess it should put enough strain on the compressor to be realistic.
ML-Test-Recording-Chart" border="0

Shot at 4 Exposure levels, to check how dark we have to go...
63percent" border="0
70percent" border="0
80percent" border="0
92percent" border="0

Files:

MLVFiles1" border="0
MLVFiles2" border="0
MLVFiles3" border="0
Finder-Files1" border="0
Finder-Files2" border="0

Benchmarks

@240Mhz
bench-Sandisk-Extreme-256-240-MHz" border="0

@192Mhz
bench-Sandisk-Extreme-256-192-MHz" border="0

@160Mhz
bench-Sandisk-Extreme-256-160-MHz" border="0

Overclock Off
bench-Sandisk-Extreme-256-Oc-Off" border="0

I can also upload the file size and MLV app screenshots for the 3 minutes test, if someone needs to see those.

Also, my Sandisk Extreme Pro 170 128GB which didn't work @240MHz now works flawlessly.
Will add max recording times to the mix when I get a chance.

Thank you to all those involved for pushing it a step further!

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3262 on: September 16, 2022, 10:19:08 AM »
Thanks for testing Walter.
I uploaded a build with the latest code which seems stable enough coming from reports. Please keep on testing and also report if for instance 192Mhz isn´t working, especially if it was working before and so on.
Expected cards now fully working are sandisk extreme pro cards 170Mb/s and probably also 95Mb/s. Seems a few other cards do work as well but most common card seems sandisk atm.

I think you got it wrong. Using latest sd_uhs modules "Sandisk Extreme Pro 170MB/s" is no longer the only card able to have highest transfer speeds and there are more than just "a few". It may be too soon to have a final word for recording but for benchmarks we got a clear picture: Any UHS-I card labeled W90 or better works above 90 MByte/s writing in EOS M. Including Samsung PRO Plus and EVO Select. I expect others (Kingston Canvas, for example) to do as well.
In Germany you will find special offers for Samsung cards quite frequently. I bought my EVO Select 256 GB for just 21 Euro.
There are two things differing from Sandisk:
You will likely get a deal frequently.
You can get the card size you need for a decent price. For Sandisk you often have to pick a big one (like 400 GB for 60 Euro) or other sizes with much worse Euro/GB ratio. Disclaimer: Samsung cards smaller than 64 GB do perform substantially worse.

Just ordered a EVO Plus 256 GB for 20 Euro and look forward to test it. I'm quite optimistic.*

Euro/GB ratio = 0.078 for Samsung
Euro/GB ratio = 0.15 for Sandisk 400 GB

TL;DR: No more "Gold Standard"? Waiting for more reports.

*EDIT: Results are in and my optimism was on spot this time. https://www.magiclantern.fm/forum/index.php?topic=25781.msg240456#msg240456

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7504
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3263 on: September 16, 2022, 10:24:18 AM »
Sounds very nice  8)
EVO Plus 256 GB. They are micro-sd cards right?

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3264 on: September 16, 2022, 12:14:08 PM »
Right then, sorry. The Samsung Evo Plus MicroSD manufacturer code is MB-MC128H.

2020 gen. Specified by Samsung for R100 and W60.
For this generation 256 GB variety is specified R100/W90.
Samsung has different performance levels for smaller and bigger cards (=storage size). For this product line and generation splitt happens between 128 GB and 256 GB. Other types may have gap between 64 and 128.

Will test MB-MC256K soon.


gabriielangel

  • Member
  • ***
  • Posts: 102
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3265 on: September 17, 2022, 06:32:33 AM »
Benchmark for Sandisk Ultra 128GB SDSQUA4-128GB (Red and Grey) Micro SD

This card is a No-Go.

@240 Mhz
SDSQUA4-128-G-240-MHz" border="0

@192 MHz
SDSQUA4-128-G-192-MHz" border="0

@160 Mhz
SDSQUA4-128-G-160-MHz" border="0

OC Off
SDSQUA4-128-G-Oc-Off" border="0


This is a U1 rated card. So I guess I aimed a little Too Low!

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3266 on: September 17, 2022, 06:43:13 AM »
This is a U1 rated card. So I guess I aimed a little Too Low!

Thanks for testing, though. ATM we have found no card doing 90 MByte/s (or more) in benchmark without W90 (or higher) label. AFAIK no W80 tested yet.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7504
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3267 on: September 17, 2022, 10:29:03 AM »
Thanks for reports @gabriielangel. Nice to see the progress.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7504
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3268 on: September 17, 2022, 12:10:18 PM »
Posted a new version in first post:

Now possible to switch to sdr 50 "Access". This is useful for some fast cards becoming slow while patching sd_uhs. All thanks to bilal wizardry:
https://www.magiclantern.fm/forum/index.php?topic=26634.msg240421#msg240421



Sidenote: Acces to sd_uhs is in my eosm version under the Movie menu tab. I moved it from Debug menu tab.

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3269 on: September 17, 2022, 03:17:58 PM »
Samsung EVO Plus 256 GB MB-MC256K performs well in benchmark.


This card is on sale for 20 Euro right now. Euro/GB = 0.078 which is around half of Sandisk Extreme Pro R170/W90 best offer.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7504
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3270 on: September 17, 2022, 03:27:10 PM »
On sale, where?

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3271 on: September 17, 2022, 03:39:23 PM »
Germany, Saturn. They don't ship outside Germany.
https://www.saturn.de/de/product/_samsung-evo-plus-micro-sd-speicherkarte-256-gb-130-mbs-2769067.html
No shipping fees for me.

Other distrubutors may export to other countries but shipping costs will most likely kill the deal. Look for local offers.

EDIT: And I haven't done any real recording sessions with this card. Will take a few hours ... household issues first.

gabriielangel

  • Member
  • ***
  • Posts: 102
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3272 on: September 17, 2022, 05:21:11 PM »
This Samsung EVO Plus 256 GB MB-MC256K you benchmarked sits at a very comfortable price point here in Canada. If those cards prove to be as reliable as the Sandisk  Extreme Pro,
it will give the extra space required to record 3K 1:1 to everyone.

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3273 on: September 17, 2022, 09:47:09 PM »
Samsung EVO Plus performed pretty well in benchmarks and now in recording, too.
Used an older build (02. Aug. 2022) with 14 bit uncompressed, 23.976 fps (Exakt fps) and not much more (Simple hacks = More). And Bilal's "Universal" sd_uhs.mo.

Recorded 4 clips with increasing resolution until recording broke. First 3 clips were stopped manually after 5, 6, 6 minutes.
1824x1026; 5:04; 23,971,658,752 Bytes -> 75.2 MByte/s
1856x1044; 6:03; 29,628,040,704 Bytes -> 77.8 MByte/s
1872x1052; 6:01; 29,973,235,200 Bytes -> 79.18 MByte/s
1888x1062; 0:34/816 frames; 2,874,075,648 Bytes -> 80.6 MByte/s


Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3274 on: September 18, 2022, 11:34:40 AM »
Just about 2 deals in Germany:

1.) You can get a 10 percent discount on Samsung cards via their chat. Always!
Don't know about other countries but you can get a Samsung EVO plus 256 GB MB-MC256K for 17.91 Euro until September, 25th.

2) Amazon and Saturn in their on-going "Anything you can sell cheap I can sell for the same price" duett:
"Gold Standard" Sandisk Extreme Pro 128 GB R170/W90 for 19 Euro. Via their ebay shops.