Compiled list of card speeds

Started by Audionut, July 13, 2014, 01:58:17 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.


Hi everyone,
first Post here. So i did some Benchmarks with 5d Mark III and some KomputerBay Cards 64 GB 1000x UMDA7. All purchased within the last week,

First Card:

[Edit]Update Post. Now with Benchmark in Movie Mode and GD Off. No Warmup.[Edit]

OLD First Test with GD On

For own reference here a SanDisk Extreme Pro 90MB/s 16GB:

[Edit]I forgot to tell the SanDisk was just a 16 GB Card. As mentioned by Walter Schulz below, Global Draw was off on the SanDisk. The Test with the SanDisk was not really for reference. More a self check for me how the KomputerBay perform to a Brand Manufacturer. Nonetheless i thought GD On makes the card read/write slower, am i right?[Edit]

I use them all formatted as FAT. The KomputerBay Cards came pre-formatted as ExFAT. Reading speeds on the Mac were horrible so i formatted them within the Camera and they performed very good in Black Magic Speed Test and AJA System Test. Used a Transcend USB 3 Card Reader.

Recording seems to be continuous, tried some 1.5 to 2 minute takes. Also when splitting in 4GB chunks looks good.

best regards,

Walter Schulz

It doesn't make that much sense to compare card benchmarks taken with Global Draw ON with another card's benchmark with GD OFF.


Hi Walter,

sorry for that. Played around with settings between the Benchmarks. Updated post with KomputerBay 64GB 1000x Movie Mode and GD Off.


Hi Why not try Renice S8 Series Compact Flash Card.


I have a Canon 5D Mark III and I just purchased a Lexar 128GB 1000x CF Card UDMA 7, formatted it to exFAT and after doing numerous benchmark tests, with GD on and off and in video/picture mode, I get the same low results of 29.9 MB/s write and 13.6 MB/s read

Those readings sound completely wrong.  I can only record video in ML for about 2 seconds before it skips frames and shuts off (shooting at 1080/23.97). 

Do I have a bad card?  Am I missing something?

Any help would be greatly appreciated.  Thank you!

Walter Schulz

I suppose you purchased a faked card. Contact dealer to get a refund or a replacement.


Not sure whether or not I may have gotten a dud version of KB 128gb 1066x in my hands...

Here are the BM comparisons:
1.1.3 (Playback mode)

1.1.3 (24p)

1.1.3 (30p)

1.1.3 (50p)

1.1.3 (60p)


1.2.3 (Playback Mode)

1.2.3 (24p)

1.2.3 (30p)

1.2.3 (50p)

1.2.3 (60p)

Obviously RAW/MLV write speeds are slightly over the edge in 1.1.3... my question is do you guys think I should keep this KB 128gb 1066x card?


I'm starting to like it as each day goes by... the more I learn its 'sweet spot' for each settings... the more I enjoy the practical freedom of 128gb.

I've also noticed the 5D3 can push it to close 103-106mb* during 3x crop 2.5k which can be just absolutely marvelous! (*obviously only for 10-20+ seconds depending on settings)

I wonder if it can do more with a 1200x speed CF card?
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109


Just wanted to share: I have reverted to 1.1.3 and was able to record FullHD @30fps (though, zero LV feedback), but what is more astounding - 2560*1090 @ 24fps, in crop mode. Over a minute! Needless to say, it looks better than anything natively FullHd. Keeping in mind that 1:1 sharpness in raw is better than sharpness of full frame raw... wow.

Recording was done to CF (64GB, 1066 Lexar) and SD (64GB, 95mb/s Sandisk), in parallel.

Walter Schulz

Quote from: KMikhail on November 22, 2014, 09:29:58 PMKeeping in mind that 1:1 sharpness in raw is better than sharpness of full frame raw... wow.

Samples, please!
Why? Because 5D3 uses binning, other cams (except 7D2) use line skipping in uncropped video recording.


AFIK 5D3 doesn't do line skipping (that's why it is superior in low-light, when compared to FF cameras with line skipping, but I suspect it still throws away information), it does binning. Thanks to this resulting FullHD bayer mosaic-ed image isn't as sharp as optimally resized.

Sorry, I am too lazy to repeat my charts shots :) It becomes immediately obvious, once you pull out full frame dng to ACR - sharpness is lower, but what is worse - the binning artifacts are raising their head if you try to sharpen the image - location of RGGB elements is different in each binned pixel. So you can't really make it SUPER nice at 1:1. Obviously, 1:1 has none of this - it is as pristine as normal photo raw. Though, micro-contrast of full frame is still superior, when compared to crop mode.


In fact, if you think about it - due to 3x3 binning over 2x2 bayer: 1) the best case is 5 green elements over 3x3 2) the worst case is 1 R/B over 3x3. So, if it is straight binning over 3x3, we still lose plenty of incoming light.


Naturally, the issue with crop mode is framing and focusing - crop window isn't aligned with the center of the center of sensor. Though, I think custom crop marks can be added via bmp (there are 5 of them available, but for different purposes).


There are currently three Kombuterbay 64GB cards available: 1000x, 1050x, 1066x. Which one should i get based on your experience?


5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109


So is the 1066x the fastest? I thought the 1000x is the one to go..?


does your KB 1066x 128gb allows for the 1080p 24/25p continously recording?   5D Mark III, 6D, 550D


Quote from: kgv5 on November 28, 2014, 08:08:56 AM
does your KB 1066x 128gb allows for the 1080p 24/25p continously recording?
Confirmed for both 24/25p in 1080p.

Both RAW & MLV.


So far!

Quote from: swinxx on November 28, 2014, 06:58:37 AM
So is the 1066x the fastest? I thought the 1000x is the one to go..?

Sometimes in life it's worth to bite the bullet when timing persist.

Also due to mixed reviews, I'm afraid I've waited too long on ordering the KB 128gb 1066x since now it has given me so much flexibility in shooting RAW/MLV for long period of times in between takes without hesitating on swapping cards as often.

Saving up to get a bundle package one of these days!

I also use few Lexar Pro 32gb 1000x that runs great as well as few Sandisks Extreme Pro 32gb but w limited spaces obviously.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109


I'm a bit confused with the card speed hardware bottlenecks ::). I have a 700D and I am planning on installing ML and I'm about to order a card. If I get it straight, despite CBr, RAW, etc. the camera won't be able to go upwards of about 40 MB/s?

And btw, which card do you recommend? Right now i'm between a Sandisk Extreme 60 MB/s and a Sandisk Extreme Plus 80 MB/s :-\? Those are read speeds, advertised write speeds are 20MB/s less on each.


The 700D is hardware limited to 40MB/s.  Doesn't matter how fast the card is, that is the maximum limit on that camera.

You probably want to ensure your card is faster then this limit, as this ensures you always reach that hardware limit. 
I'm a little impatient, so I get the fastest card I can afford, for reduced transfer times from card to computer.


Hello all, I'm planning to buy a 50D for raw. The first that came up to my mind after seeing te prices of CF cards, was why no one is using SD->CF converters? SD cards are way cheaper and get easy to 90 write and read. example : Samsung MicroSDXC Pro 64GB + SD-adapter. Another pro of it, is it's flexibility to use in other devices like small camera's and even smarthphone. Is there someone who wants to test this out?

Walter Schulz

Please do some research on the web.
You will have a hard time to find hard data for most of those adapters.
You will need at least UDMA-6 covered by CF interface and UHS-I for the SD slot.

And I was not able to find a test saying "Eureka! It works at full speed!". What I found looks like this:
Most tests I found are outdated.


QuoteNo, photo mode.
Movie, 24fps, GD OFF: 90W/120R.
Movie, 30fps, GD OFF: 85.5W/115R.
All on 1.2.3.
Could you share your settings?
I just got 2 of those Lexar 1066 64GB cards and they are both almost identically peaking at 63mb/s when running in-camera benchmark.
In Playback mode they get a bit above 80mb/s - similar to the results I get from them in "AS SSD" and Chrystalmark.
Seems that they are too fast to be fake but also too slow to be real - any thoughts
Actually, performance wise, it seems to be closer to my Lexar 32GB 800X than I'd like :(
Daybreak broke me loose and brought me back...

Walter Schulz

80 MByte/s is quite too slow for this type. Any chance your reader is sub standard? Sometimes virus scanner may slow down benchmark tests. Disconnect from web, deactive antivirus software and re-run benchmark tests.