Danne's crop_rec_4k experiments for EOS M

[cannibalferox]

Quick question Danne, I have updated today accordingly (deleted all sd card content, copiew new over and went through original canon update mechanism etc). After that even 1080p could only record for 3 seconds etc. Turning down to 192MHz also didnt help. After that I swapped out the sd_uhs.mo module and now everything works fine but I have a feeling that there is no option anymore to change the sdcard frequency in the menu. Is it now hardcoded to 240MHz? If so all good because after that 1080p worked fine again and now I have run 2.5k preset (without frtp) at 16:9 with 2400px width and 12 bit this one was really stable for me!

[Danne]

Well, a lot of testing around sd_uhs lately.
Do you mean latest download isn´t working when set to 192Mhz? What sd_uhs module are you using that is working then?
You can test todays download again. I changed to new version a couple of times today.

[Danne]

Added bilal latest code to sd_uhs just now. Download in first post
https://www.magiclantern.fm/forum/index.php?topic=26634.msg240328;topicseen#msg240328

To be tested is 240Mhz cards working now that did not work from before?
Is 192Mhz working as before with cards using the preset successfully before.

Happy testing.

[cannibalferox]

ahaaaaa today in the morning I have downloaded the build and it said uploaded 17min ago or so. I assumed Bilals 240MHz was part of it already.

In short:
Downloaed 2022Sep15EOSM202.zip and only after replacing with the original from Bilal ( https://www.magiclantern.fm/forum/index.php?topic=26634.msg240128#msg240128 ) it worked again but I have the feeling that now the option is missing for sdcard frequency

[Danne]

What exactly is working with Bilals original version? Did you enable the 240Mhz preset? You need to do that manually and it´s under debug menu I think. Enable and restart camera.

Could you also test bilals new version from here? Also needs to be enabled from menu or else nothing is patched. Enable and restart camera.
https://www.magiclantern.fm/forum/index.php?topic=26634.msg240328;topicseen#msg240328

And if they both don´t work. Are you saying 192Mhz is working with older eos m build prior to these new test versions?

[cedricp]

Good news here, the card benchmark reports 40MBs but recording go higher with last build... I've been able to record 2.5K at around 78MBs continuously.
I will do more test this afternoon

[Danne]

Nice. Wonder if benchmark is using the orignal patch but recording 240Mhz2? Or maybe I am asstalking  .
Bilal. Do you know what's going on? Why two different 240Mhz in same code? Maybe just replace wirh the later one fully?

[Walter Schulz]

I suppose you did something wrong, cedricp. Something like setting SD overclocking frequency and running benchmark without restart.
I think it is impossible to have a 40 MByte/s benchmark run (which is standard for non-overclocking benchmarks) and having 78 MByte/s recording without reboot.
Card used, btw?

[cedricp]

I used the 128gb in the picture I attached earlier.
I did reboot after the setting change...

[cedricp]

Ok, I made a write benchmark and I accidentally pushed the set button,what deactivated the live view... Indeed I get 69MBs now 
Thanks for that hack

[Walter Schulz]

Have you tested your other cards yet?

[cedricp]

Not yet, was too excited to test it at such a speed 
I wil for sure do that

[Danne]

Ok, I made a write benchmark and I accidentally pushed the set button,what deactivated the live view... Indeed I get 69MBs now 
Thanks for that hack

Ok, might be confusing with all builds here but here is one which replaced older 240Mhz string with bilals latest. If you could test it that would be good. At 240Mhz and 192Mhz:
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/downloads/sd_uhs.mo
Replaced by eosm full build. Code seems stable.

When benchmarking you could switch to photo mode and just before benchmarks start push play button for maximum readings.

Now only this is present in patch when in 240Mhz:

Code: [Select]

static uint32_t sdr_240MHz[]     = {        0x3,        0x3,                             0x1, 0x1D000001,        0x0,      0x100,      0x100,      0x100,        0x1 };

Code: [Select]

static void WriteClock(uint32_t* regs, uint32_t* stack, uint32_t pc)
{
    memcpy(uhs_vals, sdr_240MHz, sizeof(uhs_vals));
}

[gabriielangel]

Thanks for sharing tests.
What if you disable histogram? Do you enable both hacks as well from bilal? lvface + aewb?

I only use lvface, because I want to use shutter fine-tuning (Shutter speed still drifts when hitting record with +aewb enabled).

If you use lvface+aewb, 2.8k 14bit will stop less often before reaching the 1 minute mark, and it tolerates overexposure more. By that, I mean that if the scene becomes brighter, the recording won't stop immediately.
I did not notice a significant difference at 3k 23.976 fps 14bits when also using +aewb.

Regarding disabling the histogram, I'll record a few clips on my way home later and let you know.

I will update my previous post with those details.

[gabriielangel]

ahaaaaa today in the morning I have downloaded the build and it said uploaded 17min ago or so. I assumed Bilals 240MHz was part of it already.

In short:
Downloaed 2022Sep15EOSM202.zip and only after replacing with the original from Bilal ( https://www.magiclantern.fm/forum/index.php?topic=26634.msg240128#msg240128 ) it worked again but I have the feeling that now the option is missing for sdcard frequency

When using Bilal's codes, you have to go in the DEBUG tab, at the bottom. You will see the Overclock frequency there.

[cedricp]

Quick test with the 32Gb/90MBs and it looks like perfs are better. I have been able to record 2520x1070@12bits continously. (Benchmark says 62MB/s, got 48MB/s before the 240MHz hack)

[roterabe]

Ok, might be confusing with all builds here but here is one which replaced older 240Mhz string with bilals latest. If you could test it that would be good. At 240Mhz and 192Mhz:
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/downloads/sd_uhs.mo

When benchmarking you could switch to photo mode and just before benchmarks start push play button for maximum readings.

Now only this is present in patch when in 240Mhz:

Code: [Select]

static uint32_t sdr_240MHz[]     = {        0x3,        0x3,                             0x1, 0x1D000001,        0x0,      0x100,      0x100,      0x100,        0x1 };

Code: [Select]

static void WriteClock(uint32_t* regs, uint32_t* stack, uint32_t pc)
{
    memcpy(uhs_vals, sdr_240MHz, sizeof(uhs_vals));
}

Color me impressed. The new sd_uhs module managed to squeeze 55.2MB/s out of a Samsung Evo Plus MicroSD. It's quite impressive. We should also note that 240Hz works at all.

[cedricp]

Same with the microSD 64Gb/90MBs, 62MB/s now (that was the worst case before the hack with 42MBs)
Super happy

[cannibalferox]

Okay, sorry for all the confusion let me explain a few things in correct order because I have now run a lot of benchmark tests.
1. I reverted to the magic lantern version I used before with 192MHz. Results in 59,15MB/s with Apr27 build


2. I updated back to Sep15 the newest build @240MHz with 19.95MB/s and my fault in the morning was when changing sdcard frequency I havent rebooted but now after the reboot 192MHz works as before with 59.x-ish MB/s so sorry for the confusion


3. I then switch the sd_uhs.mo danne suggested with the 104 patch and voila now I get 67.3MB/s @240MHz


4. Lastly I tried out at least to my knowledge the 16:9 mode with highest pixels out 2.5k without frtp at 16.9 maximum resolution which is I think 2512px wide and 1400-is height. The setting was on purpose a touch overexposed (f-stop down no nd filter on) and with 14bit it worked only for a couple of seconds but with 12bit it was already pretty stable (!)


If a mod can help out please resize the images or tell me what I need to add so I can make them smaller

All in all thanks a lot to all of you guys! Impressive to get almost another 10MB/s

[Walter Schulz]

The new sd_uhs module managed to squeeze 55.2MB/s out of a Samsung Evo Plus MicroSD.

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!

We should also note that 240Hz works at all.

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

[theBilalFakhouri]

It's better to run benchmarks in Play mode (outside LiveView):

-Load bench.mo from "Modules" tab in ML menu, restart camera
-Turn off RAW video
-From ML menu, "Debug" tab, Click on "Benchmarks" --> "Card Benchmarks" -->   Run "Quick R/W benchmark (1 min)"
-Press Play button quickly

Share the result

[cedricp]

In play mode the 128Gb Extreme pro 170MB/s reports a fabulous 95Mb/s

[cannibalferox]

By the way I wanted to mention while I did a couple of minutes test recordinings as in the fourth picture I started to smell something Which was the hot sdcard so I can definetly confirm higher temperatures.

Towards the Liveview/Play Mode I have now only repeated it with the last version plus patch I am running now:
Writespeed1 95.3MB/s
Readspeed1 85.6MB/s
Writespeed2 94.7MB/s
Readspeed2 85.6MB/s

[Possie]

SanDisk Extreme Pro 128GB  170 U3
high 60’s with spikes up to 77 read and 86 write

Samsung Pro Plus 128 GB
73 write only

Lexar 64GB 1000x
almost 70, dropping to 49 write only
After 1 run dropping to low/mid 50’s

On EOS M sept 15 build + patch

[Walter Schulz]

SanDisk Extreme Pro 128GB  170 U3
high 60’s with spikes up to 77 read and 86 write

Play mode or MV-xxx?
To enter play mode press play button after starting benchmark run.

If benchmark is not running like a clock in play mode you should consider running a low-level format in cam and redo tests.