What Is The Maximum Sd Class One Can Benefit From?


quadomatic

DingooWiki Admin
Joined
Mar 26, 2007
Messages
1,233
Location
Chicago, IL - USA
Website
www.opendingoo.com
People have recommended swinging for class 10 SDHC cards, since read/write speeds will be important when booting operating systems and large files are being used.

Could anyone with a Pandora test to see the read and write speeds they're able to achieve with their SDHC cards, so we can determine whether Class 10 cards can actually be utilized to the full extent, or if we're maximizing capability with just Class 6 (or 5, 4 etc.).
 
quadomatic said:
Could anyone with a Pandora test to see the read and write speeds they're able to achieve with their SDHC cards

Hmm, I'm not sure how this would be achieved? I've copied files from one folder to another on the SD card within the pandora but it doesn't display read/write speeds. Unless there is a program for this I haven't found? I have a 16GB class 4 and it seems to be adequate.
 
Last edited by a moderator:
quadomatic said:
People have recommended swinging for class 10 SDHC cards, since read/write speeds will be important when booting operating systems and large files are being used.
Uh... who has recommended that? I know that I've been pretty emphatic about not spending your money on such expensive fast cards. A class 2 should be fast enough to load a SNES ROM in about 1 second, is several times faster than the PSX load speed, able to playback a few dozen MP3s at the same time, and even able to read a DVD ISO in real time. Read operations are not the major bottleneck of the boot: although it may speed things up a little bit, you shouldn't need to reboot all that often that saving a couple seconds is worth twice the cost of storage.
The only benefit I can honestly really see is when writing large files to the SD card, or lots of little files. But the way I see it, if I'm transferring my entire MP3 collection onto a slow SD card, I have no problem just setting it, walking away, and checking back half an hour later.
 
Last edited by a moderator:
It's merely anecdotal, but, using my Pandora I copied a file from my Class 6 card, to another spot on my Class 6 card, and then tried the same with one of my Class 4 ones. I couldn't discern much difference - I really think Class 10 would be overkill, personally.
 
Thanks for the replies. I guess Class 4 is good enough, which is nice since Class 6 and 10 cost a bit more.

I think the manufacturer probably matters more. Some A-Data and RiData cards aren't very reliable (not ALL of them, just some of them)
 
Prometheus said:
It's merely anecdotal, but, using my Pandora I copied a file from my Class 6 card, to another spot on my Class 6 card, and then tried the same with one of my Class 4 ones. I couldn't discern much difference - I really think Class 10 would be overkill, personally.

But does that actually move the data? I thought it just repointed it?
 
Last edited by a moderator:
No, it does create a copy. It took a little while. :p Moving is near-instant regardless of the card's class, in my experience.
 
fearofshorts said:
I think ED stated that the Pandora couldn't read/write at class 10 speeds anyway.
Sure it can. Class 10 just means it has a "minimum" write speed of 10MB/s. The Pandora has a theoretical max write speed of 24MB/s, but has seen up to 16MB/s in practice.
 
Last edited by a moderator:
Prometheus said:
No, it does create a copy. It took a little while. :p Moving is near-instant regardless of the card's class, in my experience.

duh me. Yea copy, no move! Sorry!!!
 
Last edited by a moderator:
Would be nice to see some actual tests. Something like dd will tell you how long an operation took, but you'll only be able to do sequential tests with that, which should be fine for testing bus speed.
 
It's a simple case of the marketing values not telling you what you want to know. The class number is relevant if you're recording with a video camera. On the pandora, how often do you expect to be performing only one long continuous write? No other scenario is accounted for in that classification, and reads (which are undoubtedly more common) are always faster (unless the bus is already saturated). I went for a card with a fast rate of commands instead, but finding documentation on this is not trivial.
If those with Pandoras want to test, try running Bonnie++ (available in Ångström). It's meant for file system benchmarks, but with that being equal, we'll get some performance data for the cards.
 
If it's going to be a dedicated SD card for your Pandora, 4/6 should be enough. If you want to some day (or presently) use it for video, then consider a faster class (and also remember, it will wear down faster with such excessive use).
 
Here's a decent high-performance sd-card test, from february:
http://www.tomshardware.com/reviews/sdhc-memory-card,2143.html

While reading, many cards saturated his card-reader, and are close to SDHC's speed-limit. However, there are significant differences in write-performance and in random I/O, with the best cards basically twice as fast/good as the lesser tested cards...

Almost all of the tested cards reached class 10-speeds, so that classification is hardly worth looking at...

In the end, it depends on what you want to do with a particular card. If you're going to put it in a write-heavy situation (your home-dir for a desktop-environment...), get the best cards. Otherwise, it doesn't really matter at all...
 
Edit: Hrm...I guess I'm going to have to eat my words here

Tom's Hardware did another review comparing class 10 cards. The only ones that did better were the Sandisk Extreme and Lexar Professional cards. For workstation benchmarks, the only ones that were even useful were the Sandisk and Lexar cards.

http://www.tomshardware.com/reviews/compactflash-sdhc-class-10,2574-10.html

I think what I may do, is buy a high quality 8GB Sandisk Extreme Class 10 card for ~$40. If I want to use an operating system, I'll put it on the 8GB card (which is more than enough for simplistic OS). Any applications and large files can go on there as well. Then, for extra storage, I'll use a 8GB microSDHC card I have. I'll replace it with another card later.

But, then I have to wonder. The Sandisk Extreme card is a 30MBps card. The read throughput is ~27MBps, and the write throughput is ~24MBps. It seems to maximize the Pandora's theoretical rate. WizardStan said here that they'd seen the Pandora go to 16MBps write throughput. But, unless they were using a card like this one, it's unlikely the Pandora's potential has actually been tested.

Anyone willing to buy a 4, 8 or 16GB Sandisk Extreme card and test read and write throughput, and I/O performance?

edit: Apparently Tom's Hardware has performance charts for SD cards as well. Sandisk Extreme 16GB trumps

http://www.tomshardware.com/charts/sdhc-memory-card-charts/benchmarks,40.html
 
SDHC is 50Mhz, 4bit at max, giving ~25MB/s max speed. Considering overhead, 20MB/s is apparently the max possible speed supported by SDHC. So, that Sandisk Extreme-card is simply out-of-spec, and supports an un-standardized speed-setting (I guess they took SDXC's speed-modes...)...
 
WizardStan said:
fearofshorts said:
I think ED stated that the Pandora couldn't read/write at class 10 speeds anyway.
Sure it can. Class 10 just means it has a "minimum" write speed of 10MB/s. The Pandora has a theoretical max write speed of 24MB/s, but has seen up to 16MB/s in practice.

Actually that is not true. Class 2, 4 and 6 SD cards do fit that convention, true. But "Class 10" is actually not a standard. It's just a marketing name given to really-high-speed cards by SD card makers.
 
Last edited by a moderator:
fearofshorts said:
Actually that is not true. Class 2, 4 and 6 SD cards do fit that convention, true. But "Class 10" is actually not a standard. It's just a marketing name given to really-high-speed cards by SD card makers.
It is more-or-less true. "Class 10" may not be officially blessed by the standard, but they're still following the conventions of the standard.
 
Last edited by a moderator:
Back
Top