Messages

@Grognard

Not really , the showing writespeed is incorrect in other modes. See this test:

https://www.magiclantern.fm/forum/index.php?topic=12862.msg228848#msg228848

I see that there's a lot going on here with the writespeed calculations.

You also asked what happens when I try to use SD overclocking. I am using a 95mbps Sandisk Extreme Pro 32gb SDHC card. I have a 128gb 170 coming in the mail on Monday.

When I use the overclocking, I turn it on in the settings, turn off the camera, turn back on, wait for the prompt to say it's been patched. Then when I try to record, I have gotten both "card full," and "recorded has stopped" errors. No frames recorded. I have tested it spanned with my CF card (pixelflash 1107x 64gb) and on it's own. No luck. I have also tried both 3123 and 3113 builds to see if that made a difference. Nothing. I also tried formatting multiple different ways (exfat, fat32), but I don't actually know anything about formatting cards so that was just at random and it didn't do anything.

Hi there! I wanted to say thank you for all the work on this build. It's been such a huge upgrade.

I found a weird quirk using the card spanning module. I'm using the July 15 3113 build. Not using sd overclocking as it's not working for me.

The 5x crop preset has an extra 5mbps without global draw using card spanning vs UHD or anamorphic. In those modes it tops out at 111mbps (91 cf - 20 sd) but in 5x 3.5k, the CF card climbs to 94-95, for a total of 115mbps! I think that mode is using fewer hardware resources or something and that translates into more recording bandwidth. I don't know if the way these modes behave is already well known, but it seemed worth mentioning. I am noticing differing (but consistent) speeds using different presets, with or without global draw.

I'm using a Pixelflash 1107x 64gb card and a Sandisk Extreme Pro 95mbps 32gb.

Kill Global Draw set to ON.

Edit: also it looks to me like the recording rate estimation is using 14bit raw recording in it's calculation, even though the files are in 12bit, giving a larger estimate than actual. Is this true? It seems like my recording times reflect what I would expect to get with 12bit.

Okay I downgraded my firmware to 1.1.3 and tried it again.

I am still getting slower rates. However, they are slightly faster. With global draw on I'm getting about 68mb/s, off is 73mb/s. 48p mode jumps to 76mb/s with global draw off.

Still slower than I wanted but a bit more usable. Unfortunately I think there is something inherent to running the camera at a higher frame rate that lowers write speeds. I don't know if this is a bug or a hardware bottleneck, but I am surprised I haven't seen anyone mention it before?

Thank you for your response.

I had considered that the compression may be eating up CPU resources which is slowing down the card. I just tested it at 10bit and unfortunately, the card is still slowed down for some reason, this time to around 73, which is an improvement, but still significantly slowed.

It seems like it's related to setting the camera to 60fps from the Canon menu. Once I turn that on, my card speed drops during recording. Turn it off and the card speed jumps back up to 95ish.

I'm not sure what's going on but that's where I am so far. Because the issue seems to happen when changing settings from the Canon menu, and not from magic lantern's menu, I am going to attempt to downgrade my firmware, and see if that changes anything. I will update if it does.

It's possible that changing the frame rate is also using CPU resources or something like that, but I'm going to try anyway.

Using a Canon 5dmk3

Experimental 4k raw recording build with lossless compression

firmware 3.123

PixelFlash 1107x 64gb Supersport card

Loading modules crop_rec, mlv_lite, mlv_play, mlv_sound.

My card typical records between 90-100mb/s, but when I switch to the 50p/60p crop recording mode, the write speed drops to 63mb/s ish and stays there. If I try to record larger video it predictably stops recording within a few frames.

Is this a known issue? Am I doing something wrong?