Does a higher class card really matter? This guy says "no"!

F

Filmkid2003

Well-known member
Hey everyone. I was checking out the Sandisk Extreme Class 10 16GB SDHC card and came across this guy's review of the card. Surprisingly, he says that for HD video recording, if your SDHC card is TOO fast, it can actually cause problems for the camera. Anyone have any thoughts on this? Below is the review and the link to the amazon page where I found it.


99 of 108 people found the following review helpful:
4.0 out of 5 stars Bits and Bytes, December 6, 2009
By Rob - See all my reviews
This review is from: SanDisk 16GB Extreme - SDHC Class 10 High Performance memory card (SDSDX3-016G-P31, Retail Packaging) - NEW 30MB/s version
Regarding complaints about performance with HD camcorders:

SDHC cards are rated in MegaBytes per second (MB/sec). HD camcorders are rated in Megabits per second (Mb/sec). There are 8 bits per Byte.

Therefore, an HD camcorder with 24Mb/sec video capture requires (24Mb/sec) / (8Mb/MB) = 3MB/sec SDHC card. A Class 4 SDHC card with at least 4MB/sec data transfer is more than enough to capture full HD. Those camcorders are not designed to work with Class 6 and Class 10 cards.

Remember that Full HD, 1920 x 1080 pixels = 2,073,600 pixels, or 2 Megapixels. Digital SLRs shoot at 12 Megapixels with minimal lag time between shots requiring much higher data transfer rates than HD camcorders.

Consequently, people who use Extreme Class 10 SDHC cards with high speed digital SLRs seem happy, but people using them with HD camcorders are not. It can be confusing, but you must use the appropriate card for your application. Faster is not always better.


Here is the page link:

http://www.amazon.com/SanDisk-16GB-...1?ie=UTF8&s=electronics&qid=1277764179&sr=8-1
 
Nonsense. There is no downside whatsoever to using a faster card, other than that it's overkill.

The stock GH1 asks for a class 6 card, and its default HD mode requires at least Class 4. We have been tripling the stock bitrates, and more in the case of the MJPG mode. You definitely, unquestionably, need the fastest card you can get when working with the GH13.
 
I call BS on the whole "too fast" thing. That's just not true. I design electronics and can say that without a doubt, all the years I've worked with memory, there was never a failure due to memory being too fast!

Are some of the faster cards kinda a waste of money if you don't need that speed? Yeah. Are they going to somehow give you problems? no.
 
Right. If the guy wants to say "don't waste your money on a faster card, your camera can't take advantage of it anyway" then that's one thing. Implying that there will be "problems" from using a "too fast" card? Nonsense.
 
Bollocks

Bollocks

Lets first understand what speed is needed. If you are writing 24 mbps it DOES NOT indicate the minimum speed.
The camera will expose for a certain shutter, then it will read the RAW signal, then it will process the signal into a compressed stream AND THEN it will write at the rate of 24 mbps.

The time that the camera will have to actually write out the data is only about 1/2 to 1/3rd the advertised speed.
Also, the RTOS (the Operating Systems) that is used in these devices are usually non-memory protected, single-tasking. So the processor only does 1 task at a time.
The Algorithm goes
Expose -> Read RAW -> Process -> Compress -> Write

The Noise Reduction, Contrast etc is the first Process and the 24P/25P conversion is most probably the last Process. It seems that the firmware first compressed to 50i/60i and then it is converted back to 24P/25P (this is the only way we can account for the problems with 24P/25P).

When you are using the firmware hack so you are increasing the time it will take for the last 3 in this chain.

Hence the speed needed is substantially more.
The main advantage of the faster cards is the higher burst speeds.

We cannot go to the next frame until the previous frame is written out, as this will overwrite the buffer and cause the famous error.
Hence we need faster speeds.

Its really annoying how people with half a knowledge of electronics and computing suddenly become "experts".
 
From the test results I've seen posted here, and the tests I've done myself there appears to be no limit to the advantage to be gained on the GH13 with faster cards. I have consistent repeatable tests that show the difference between class 6 and class 10 cards from Transcend - class 6 can't do the "c" settings, class 10 can. Sandisk class 10 cards (they are faster than Transcend) can do 60,62,70 settings, the Transcend class 10 cards can't. Assuming the Sandisk cards are the fastest available (or close to it); there is clearly something to be gained by using even them over other class 10 cards.

We have plenty of empirical data that flatly contradicts what this guy is saying. That trumps theory...

Chris
 
Last edited:
Again, specifically as related to the GH13, I am totally agreeing. In this particular case there isn't even a question.

In general, with fixed-bitrate codecs like normal AVCHD, there is no advantage to be gained by using a faster-than-specified card. There will be absolutely no difference between AVCCAM recorded on a Class 4 card, and AVCCAM recorded on a Class 6 card, and AVCCAM recorded on a Class 10 card -- so you're perfectly ok to use the less-expensive Class 4 cards with AVCCAM.

But this thread was posted in the GH13 area, and there are obvious and definite benefits to be gained by using the very fastest cards possible in our particular scenario. The GH13 is open-ended in its bitrate and takes advantage of faster speeds.
 
And that, my friends, is why I just ordered a Sandisk Extreme 16GB Class 10 30MB/sec SDHC card to replace my Patriot Class 10 SDHC card.
 
Just did some tests and posted on this thread. Hope this helps.

http://www.dvxuser.com/V6/newreply.php?do=newreply&p=2034127

mojo43 said:
Ok, I just picked up the card and shot the Pappas star chart. I did it with three different cards:

1) Core Micros 16GB Class 6 (cheapo)
2) Panasonic 8GB Class 10 (gold)
3) Sandisk Extreme 8GB Class 10


I shot 1080p with C settings with the 24p native checked with the stock lens.

The 1) lasted on average 20 sec, the 2) lasted on average 25 sec and the 3) lasted on average 40 sec before error.

Hope this helps the next person!
Click to expand...
 
Give your 5 minutes to test real speed of your cards, for example in CrystalDiskMark, and there take a look on 2 values
- "sequential" write speed = maximal throughput for that card (for example mine 4 class card give 16mb/sec ... and exactly the same values writen for many peoples, that like big numbers /marketing/)
- "512k" write speed = in fact must not be less than class rate, otherwise take that value in account as your card class speed ... and by multiplying to 0.8 you will get safe "Limiting Bitrate" for your card in Mbytes/sec

smaller block values may be ignored.
 
Last edited:
Mostly-but not entirely true. The cluster size the GH1 uses is 32k - which corresponds to the minimum block size that can be written at a time. The test program you are using tests 512k and then goes down to 4k blocks. Performance tanks at 4k - but that is 8x lower than the 32k clusters the camera uses. But on the other hand 512k is 16x higher than 32k. I've found that if your card is fragmented performance can be severely degraded. Admittedly, this only happens if you have taken a bunch of stills and little clips and have erased only some of them. So it is important to make sure your card is not fragmented (the best way to do this is to format it in camera before you use it). Also, don't format it in your computer - that can result in a cluster size of 8k - which can also significantly degrade performance.

I guess this is why Panasonic uses settings much lower than the ones we are using with good success. They have to always work, even when a card is fragmented, also contains stills, etc...

Chris
 
You must log in or register to reply here.
Back
Top