CF-Cards fast enough for 1920x1080p24/25 5D3 MLraw video

Started by grooveminister, June 23, 2013, 05:35:58 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

grooveminister

Hi all,

the Transcend cards are very close to achieving the required 83 MB/s sustained write speed, and I´ve heard rare success reports who said it would work after low-level formatting the card with the "autoformat" tool provided by Transcend.

I´m having a hard time on deciding if I should spend money on Lexar or Komputerbay cards - so it would be nice to know if it´s worth trying the Transcend cards...

Some even wrote that Transcends 600x cards had faster write speeds, can anyone confirm that?

Best wishes,
Andreas

Stedda

I have no problem with a Transcend 1000x 32GB with full HD 1920x1080 continuous. This is with several cards.

With the recent builds I can even go one resolution higher continuous.

I have a 600x also... it can't handle it.
5D Mark III -- 7D   SOLD -- EOS M 22mm 18-55mm STM -- Fuji X-T1 18-55 F2.8-F4 & 35 F1.4
Canon Glass   100L F2.8 IS -- 70-200L F4 -- 135L F2 -- 85 F1.8 -- 17-40L --  40 F2.8 -- 35 F2 IS  Sigma Glass  120-300 F2.8 OS -- 50 F1.4 -- 85 F1.4  Tamron Glass   24-70 2.8 VC   600EX-RT X3

grooveminister

Hey Stedda, good news!
You can modify your voting, I´ve added the 32 GB version!
Did you have to use the low-level formatting tool?

Stedda

I do nothing special with the cards.
I format them in camera every few times but I did that before RAW video.

There is a card benchmark thread where Alex was looking for some info.... search it out it has everything you're looking for and more.
5D Mark III -- 7D   SOLD -- EOS M 22mm 18-55mm STM -- Fuji X-T1 18-55 F2.8-F4 & 35 F1.4
Canon Glass   100L F2.8 IS -- 70-200L F4 -- 135L F2 -- 85 F1.8 -- 17-40L --  40 F2.8 -- 35 F2 IS  Sigma Glass  120-300 F2.8 OS -- 50 F1.4 -- 85 F1.4  Tamron Glass   24-70 2.8 VC   600EX-RT X3


Stedda

You got it. Just about every card you could imagine with graphs posted by Alex.
5D Mark III -- 7D   SOLD -- EOS M 22mm 18-55mm STM -- Fuji X-T1 18-55 F2.8-F4 & 35 F1.4
Canon Glass   100L F2.8 IS -- 70-200L F4 -- 135L F2 -- 85 F1.8 -- 17-40L --  40 F2.8 -- 35 F2 IS  Sigma Glass  120-300 F2.8 OS -- 50 F1.4 -- 85 F1.4  Tamron Glass   24-70 2.8 VC   600EX-RT X3

mkrjf

The chart posted by Alex is not that useful:
It shows one value - is that the max for all cards. Average?
With what buffer size / config? Special card format?

I just got two new transcend 64GB 1000x cards - with sequential serial numbers - and one writes less than 82MBps while the other goes up to 96MBps. And transcend claims 120MBps write capability (for who knows what buffer size).

I would like to see write speed distribution by brand / card vs vendor claimed spec.
My data points are 120 claimed, 1: < 80 avg 82 peak; 2: < 90 avg 96 peak
So of example if the polluter at cards are really Lexmark rejects there should be a wide deviation in card write rates while maybe for lexar there is very little deviation.
I asked elsewhere but do not seem to get answers in the forums I post to - now do you set the buffer size? So if 5 minute benchmark test shows best results for 16384 buffer - can I force that buffer size so it will actually work for raw video?
Thx
Mike

mkrjf

Sorry I did not preview last post for spellcheck errors;)
"If komputerbay cards are lexar rejects"

grooveminister

Quote from: mkrjf on June 23, 2013, 08:10:26 PM
Sorry I did not preview last post for spellcheck errors;)
"If komputerbay cards are lexar rejects"
Yeah, but I liked the first version even more!  ;)

Danne


Danne

I have a transcend 1000x 64gb. Works fine for 1920x1080 25fps

jpjp

Transcend 1000x 32Gb, no problems running 1920x1080 25fps, no need for low-level format with autoformat here.

grooveminister

Is anyone using the Transcend 128 GB version successfully @ 1080p25?
I fear entering the same desaster as with my Komputerbay 128 GB version which is slower than it´s 64 GB counterparts...
The one in the poll is an error, that was before I added the 32 GB version to choose...

Mido

First ordered Komputerbay 128GB 1000x and it peaks at 76MB/s, so not enough. Then ordered two Komputerbay 64GB 1000x one is perfect (around 97MB/s) but second one is simply corrupted. Can't format it either on os x or camera (sending it back for replacement). Today I ordered Transcend 64GB 1000x and I hope that it will work (at least 86MB/s)

Danne

Quote from: jpjp on June 24, 2013, 12:59:16 AM
Transcend 1000x 32Gb, no problems running 1920x1080 25fps, no need for low-level format with autoformat here.
how does it even work? Do I add the .exe file to the card?  I,m on a mac. My card works fine but maybe I could squeeze a few more mb in ;)

haemma


grooveminister

Quote from: Danne on June 24, 2013, 06:48:44 AM
how does it even work? Do I add the .exe file to the card?  I,m on a mac. My card works fine but maybe I could squeeze a few more mb in ;)
Visit someone with a PC and a decent card reader.
Guess you need to have the .dll and the .exe in the same folder there and start the .exe which will low-level format the card.
After that you´ll need to format in camera again.

A german amazon user reported success with his card:
First too slow, after low level format the performance had increased.

mkrjf

Can the people who have transcend 1000x 64GB running at 24fps run the 5 minute benchmark and share the results (saves .bmp)? For 16384k buffer is all i care - that is what raw capture will use, right? One of my 2 transcend 1000x 64GB worked yesterday but not today!! Apparently it is just under required speed now! I have complaint in to transcend since they claim 120GB write and I am getting like 82MB/s for both cards now. Reformatting in camera yields same or slightly less than out of the box.
Thread posted by Alex shows max lucky write speeds and is actually a 'false negative' test - there is no guarantee transcend will work.
Can you also share all your changes from config default? I tried changing global draw, etc but no impact on write speed for me. No way in hell I am paying Lexmark prices. ;)
Mike

bluewater

Isn't there any CF card can record 1920*1080 30fps ???

It must be as fast as 107MB/s write speed. :D

Ammar

I have tow cards both work flawless on RAW 1920x1080 24/25P no drop-frames at all, its all on 5DMark3

the 1st one is SanDisk Extreem Pro UDMA 32GB
the 2nd Lexar 1000X 64GB UDMA7

i cant see any difference in these tow cards on RAW 1920x108025 continuous, but any other resolution above 1920 it do drop frame in just max of 2sec.

when i bought the Lexar i thought i can do bigger frame size like 2.5k but its not happening.


any one knows how to format the Lexar 1000X 64GB at Low level on mac? as i understand that it will make it faster a bit.



Ammar


Markus

Bought two komputerbay 64GB cards both records 1920x1152 continous at 25fps =)

gmango

I am getting a write speed of 69.8 mb/s on the Komputer Bay 128 gb 1000 x UDMA 7 (exFat) and the Canon 5D Mark III requires 83 m/s
On a 16 GB 1000X Hoodman steel CF i am getting 69.2 mb/s formated on computer as fat32.
I reformated with exFat and write speed increased to 78.2 mb/s

v1rt

Looking at the poll, looks like the card I need to buy for my Canon 50D is Komputerbay 1000x 064 GB. What do you think folks?

grooveminister

Now I´ve got my own 128 GB Transcend card.

No single 1080p24 recording was stopped by the camera yet!
The idle values start small (1-2 ms) but grow the longer I am recording. Seems the card needs to warm up.

Komputerbays 128 GB cards doesnt record more than a second @ Academy Format 1.85:1!

I didn´t have to use AutoFormat - but I´d like to give it a try.
Transcend Support recommended to Low Level format the card. But what is that?

The Tool offers:
-"Optimised Format" I guess this one is like Quick Format in Windows.
-"No Selection" is this already Low Level formatting?
-"Complete Format" - ok everything is filles with zeros

Now do I have to select nothing or "Complete Format" to Low Level format the CF-card?
At least I know that Low Level formatting on the camera itself (SD-card slot) doesn´t take more than a few seconds, so it can´t fill everything with zeros...

mkrjf

I think there are other variables besides cf card itself. Maybe build / chipset batch for 5dmk3 itself?
I just received 2 Lexmark 64GB 1000x and they are performing almost identically to the two transcend 64GB 1000x.
Highly unlikely. When I complained to reseller about transcend (which I was exchanging for lexar plus $) one of their canon reps said 5dmk3 cn only write a 76GBps max (but with Alex DMA code obviously number is different). But maybe cards are not actually bottleneck but rather build of 5dmk3.

So my 5dmk3 is like 9 months old. Are the people who have same transcend and lexar cards as me working using brand new or 6-9 month old version?
And komputerbay seems to have highest luck but I really don't feel like rolling the dice a third time!

I asked for people to share actual read write rates by card and now also by camera age / serial
Wonder if Alex can try benchmark with higher buffer size replacing the small (and why so many small tests?) sizes with larger buffers - or is 16384K largest possible?

And does crop mode write faster or same? What row length / formats could I get that require less write speed? 2k by some funky length at 24fps is better than no continuous recording!
And is 25fps faster or slower than 24 (you never know). For transcend formatting to exfat had no benefit.

Otherwise it is useless even though it has great potential.
                    W/R MBps for buffer:       2k         3k         4k          16k        32k(other test)
Transcend 64GB 1000x CF #1         56/83    65/93     65/93    81/119      80/
                                       #2         Similar I deleted the .bmp                 78/
Lexmark   64GB 1000x CF #1          35/44    41/44    42/56     79/104
                                       #2.         36/45    42/43    43/57     81/107

Anyway the values for 16k writes are too similar to be limited just by the cards across vendors - it looks like something on the camera / code side.
Please provide any relevant data you have.
And I have asked 5 ways in 5 forums with no answers:
What are ideal settings after reset to defaults for 5dmk3 raw performance.
I have tried most related settings / tweaks (no draw, slow write flag, etc) but don't see any change in write at 16k.
And I have also asked for confirmation that raw recording uses 16k buffer and that other sizes (larger)  not improve chance of working.
Sigh