LiveView hacks (write speed improvement)

Started by theBilalFakhouri, April 07, 2022, 06:20:22 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

ilia3101

With these new hacks + overclocking + card spanning, 5D3.113 gives 3840x1536 14 bit semi-continously :o

Previously I could only get 12 bit for no more than 10 seconds.

I can't thank you enough!

theBilalFakhouri

Quote from: Walter Schulz on May 25, 2022, 08:18:42 AM
Adata Premier ONE 64GB microSDXC. UHS-II. Rated 275MB/s, V90.

Okay, thanks, do you know if it switch to 48 MHz (drop to 21 MB/s) with 240 MHz overclocking after a while?
It is stable at 240 MHz?

theBilalFakhouri

Quote from: mlrocks on May 28, 2022, 04:06:16 AM
SanDisk 170 mb/s sd card can only write 90 mb/s. maybe 300 mb/s card is needed for 99 mb/s write speed?

I am not sure, but there are more than one SanDisk 170 MB/s version, like:

1-SDSDXXY-128G-ANCIN --> Cameramemoryspeed.com claims it can reach up to 98.643 MB/s write speed.
2-SDSDXXY-128G-GN4IN --> No idea

theBilalFakhouri

Quote from: ilia3101 on May 28, 2022, 08:55:34 PM
With these new hacks + overclocking + card spanning, 5D3.113 gives 3840x1536 14 bit semi-continously :o

Previously I could only get 12 bit for no more than 10 seconds.

I can't thank you enough!

Cool, have fun :)

Walter Schulz

Quote from: theBilalFakhouri on May 29, 2022, 12:13:20 AM
Okay, thanks, do you know if it switch to 48 MHz (drop to 21 MB/s) with 240 MHz overclocking after a while?
It is stable at 240 MHz?

I have to resolve problems with card access and loosing cam's bootflag first.

Walter Schulz

Overclocking in video mode doesn't work well for 650D.
Photomode gives 98 MByte/s max and no fallback to 20 MByte/s. Modules bench and sds_uhs loaded.
In movie mode numbers are much lower and 240 MHz does fall back to 20 MByte/s.
Additionally loading mlv_lite will give "malloc error: buffer=16777216" during read runs.

theBilalFakhouri

Quote from: juno60 on May 12, 2022, 07:26:29 PM
Hi there,

I am awfully sorry to sound like a total bozo... but I cannot get ML to load the modules. I get only 2, namely adtg and bench.
Tried multiple times, no luck whatsoever. I am not new to ML and all crop rec releases have been working fine, including the experimental ones.
This time around I can't figure out what's wrong. 

Running firmware 1.2.3

Thanks a lot, guys!

Hello,

Danne made some of modules (like crop_rec.mo/mlv_lite/mlv_snd.mo probably sd_uhs/dual_iso too) in his custom builds loaded by default and you can't unload/load them at your own, it's always loaded.

Go to "Movie" tab from ML settings and you will see RAW video and Crop mode are there (search for other modules, they will probably be in other ML tabs).

theBilalFakhouri

Quote from: Billsifr013 on May 26, 2022, 10:56:35 PM
I think this time has come, could you make a build for 60d, I would try it myself, but I'm not a programmer)

60D card slot is limited to 21 MB/s write speed, and you can already get 21 MB/s write speed in LiveView. So the new hacks will not make any difference, unless someone get SD overclocking working on 60D.

ShittyWebsite

Quote from: ilia3101 on May 28, 2022, 08:55:34 PM
With these new hacks + overclocking + card spanning, 5D3.113 gives 3840x1536 14 bit semi-continously :o

Previously I could only get 12 bit for no more than 10 seconds.

I can't thank you enough!

I'm sorry being a bit off topic but i have a question

i'm trying with overclocking, these new hacks, card spanning, killing global draw,and i cannot get 14 bit for more than 3 secs (3.5K preset, 24fps)
The scene is exposed to the right, Raw module says 168mb/s required (lossless)

Is it better underexposing by 1 or 2 stops or stick with 10bit properly exposed to the right?

I'll 3 scenes from yesterday:







dpjpandone

I did some testing of my own, adding the lines from the following commits to Danne's 5d3 repo:


8a57a24:  https://bitbucket.org/Dannephoto/magic-lantern_dannephoto_git/commits/8a57a244b767337904f18f3c1bcf1199c3a3ac24

61c2a5a: https://bitbucket.org/Dannephoto/magic-lantern_dannephoto_git/commits/61c2a5a38fedc273156882850ebe5ce302cc858b


LvFace+AEWB nets a realworld gain of 7MB/s

Cartridge cancel has No effect on real world write speed.

This of course only applies to 5D3 1.23


ML700D

I tested on my lexar 1667x using 240mhz setting same result w/wo two more hack
write 96.7 mb/s
read 102.3 mb/s

preset 4K 1x3 12bit continues recording

btw, bilal.. what is the best setting for sandisk extreme pro 170mb/s?

thanks for the hack speed.
EOS 700D

Danne

I have more hacks and 240Mhz included in my build already? What's the difference here?

ML700D

lexar has more write speed than sandisk extreme but less read speed.

I mean using 1+2 hack on lexar has no different result..

EOS 700D

Danne

I was more curious about why dpjpandone added code already in place :).

dpjpandone

Quote from: Danne on January 25, 2023, 01:02:52 PM
I was more curious about why dpjpandone added code already in place :).

I worded that poorly. What I meant was I added the changes

FROM: (Danne's) committs
TO: MY (based on mainline crop_rec) mlv_lite

theBilalFakhouri

Quote from: ML700D on January 25, 2023, 05:11:07 AM
btw, bilal.. what is the best setting for sandisk extreme pro 170mb/s?

What do you mean by "best setting"? What are you looking for?

Quote from: ML700D on January 25, 2023, 05:11:07 AM
thanks for the hack speed.

No problem, have fun :)

Danne

Quote from: dpjpandone on January 25, 2023, 01:35:16 PM
I worded that poorly. What I meant was I added the changes

FROM: (Danne's) committs
TO: MY (based on mainline crop_rec) mlv_lite
I see. Do you see any difference in speed/performance here between the two?

ML700D

Quote from: theBilalFakhouri on January 25, 2023, 02:00:00 PM
What do you mean by "best setting"? What are you looking for?

I mean is it just use 1+2 hack and 240mhz only or anything else need for speed..?
somehow the speed is drop sometimes when I benchmark again..
EOS 700D

dpjpandone

Quote from: ML700D on January 25, 2023, 03:20:24 PM
I mean is it just use 1+2 hack and 240mhz only or anything else need for speed..?
somehow the speed is drop sometimes when I benchmark again..

easy to test. forget about card benchmark it's meaningless. set your resolution to something higher than you can record continuously. record a short clip. check raw menu for sustained bitrate. add a hack. test again. add another hack, test again, keep going back to raw menu to check the write speed that was achieved during previous recording. use the settings that produced the highest write speed during recording

dpjpandone

Quote from: Danne on January 25, 2023, 02:45:18 PM
I see. Do you see any difference in speed/performance here between the two?

LV+AEWB adds 7MB/s
Carteridge makes no difference

I should note that this is 5D3 tested with Sandisk Extreme Pro 170MB/s SD card @240 MHz. Best record speed was around 88MB/s.  closer to 81MB/s without LVFace+AEWB. These are real world recording speeds.

  Speeds with  my CF card are about 10MB/s lower across the board, the CF I own is the bottleneck in that case so no reason to test CF. No desire to purchase faster CF since SD has enough bandwith for my use.

theBilalFakhouri

Quote from: ML700D on January 25, 2023, 03:20:24 PM
I mean is it just use 1+2 hack and 240mhz only or anything else need for speed..?

For me, I get the best write speed when all hacks are enabled.
Also, check this article, also mentinoed in first post in my 650D/700D thread:

Quote from: theBilalFakhouri on September 18, 2020, 07:51:30 PM
-Looking for more recording times?

-Check Image quality and its effect on memory (more RAM 4 free).


Quote from: ML700D on January 25, 2023, 03:20:24 PM
somehow the speed is drop sometimes when I benchmark again..

How do you do your benchmarks? like in Play mode or Video mode?
Do you change Camera or ML settings among benchmark passes?

First step to do if you have speed drops is to perform a "low level format" in camera from Canon menu, then run benchmarks as many time as you want in the same mode/settings . . benchmarks should give very similar results everytime.

theBilalFakhouri

Quote from: dpjpandone on January 25, 2023, 05:53:20 PM
Carteridge makes no difference

Nope, it does make a difference . . at least my benchmarks prove that.

Quote from: dpjpandone on January 25, 2023, 05:53:20 PM
Best record speed was around 88MB/s.  closer to 81MB/s without LVFace+AEWB. These are real world recording speeds.

How did you make your calculations?

ML700D

Quote from: theBilalFakhouri on January 26, 2023, 05:32:52 AM
For me, I get the best write speed when all hacks are enabled.
Also, check this article, also mentinoed in first post in my 650D/700D thread:

noted..

Quote from: theBilalFakhouri on January 26, 2023, 05:32:52 AM
How do you do your benchmarks? like in Play mode or Video mode?
Do you change Camera or ML settings among benchmark passes?
I benchmark using video mode..
I benchmark first then taking video 10s each with changing ML setting like dualiso, crop preset and iso when I feel the speed drop I check benchmark again.

Quote from: theBilalFakhouri on January 26, 2023, 05:32:52 AM
First step to do if you have speed drops is to perform a "low level format" in camera from Canon menu, then run benchmarks as many time as you want in the same mode/settings . . benchmarks should give very similar results everytime.

oh..ok thanks.
I'll try to low level format and enable all hack then.
EOS 700D

ML700D

Quote from: dpjpandone on January 25, 2023, 05:52:15 PM
easy to test. forget about card benchmark it's meaningless. set your resolution to something higher than you can record continuously. record a short clip. check raw menu for sustained bitrate. add a hack. test again. add another hack, test again, keep going back to raw menu to check the write speed that was achieved during previous recording. use the settings that produced the highest write speed during recording

ok thanks.
EOS 700D

theBilalFakhouri

Quote from: ML700D on January 26, 2023, 07:12:25 AM
I benchmark first then taking video 10s each with changing ML setting like dualiso, crop preset and iso when I feel the speed drop I check benchmark again.

Write speed is affected by many factors like resolution and FPS, it varies from video mode to another, also from crop preset to another.
Higher resolution and FPS means reducded write speed compared to less resolution preset or lower FPS. Anyway the hacks help in all kind of settings.

You shouldn't compare write speed in a preset to another, but compare write speed in one preset like this: once with hacks disabled and once with hacks enabled.
Also, all hacks got disabled if you are not recording, benchmarks in video mode will always show write speed without the hacks.