Lexar 128GB SDXC - Who's going to be first on this?

[dreamer]

By then I'd expect to have the >2TB version and wonder how I ever put up with those tiny 32GB cards.

Except those huge cards won't work on the Pandora

 

[Grench]

By then I'd expect to have the >2TB version and wonder how I ever put up with those tiny 32GB cards.

Except those huge cards won't work on the Pandora

And you expect to still be using your Pandora daily in 20 years instead of something that can emulate today's hottest workstation technology?

 

[Nardella]

And you expect to still be using your Pandora daily in 20 years instead of something that can emulate today's hottest workstation technology?

Yes

 

[Prometheus]

Yeah. In twenty years, retro will be the new retro, so we'll put a retro in your retro so you can retro while you retro.


Or something.

 

[Blue Protoman]

You suck at memes. BAN ME!

 

[Neelix]

By then I'd expect to have the >2TB version and wonder how I ever put up with those tiny 32GB cards.

Except those huge cards won't work on the Pandora

And you expect to still be using your Pandora daily in 20 years instead of something that can emulate today's hottest workstation technology?

Of course not... in 20 years you'd be using your Pandora version 4 with a Pandora emulator running on it so you could still use all the current software


-Neelix

 

[Grench]

Back on topic a bit...


We really really need a basic benchmarking program to show the performance statistics of SD/SDHC/SDXC cards (and USB sticks, USB HDD, etc...) ON the Pandora.


I like the one that is bundled into Ubuntu:


http://ubuntuforums.org/showthread.php?t=1468995


No, I have no idea of how to get it out and re-compiled to ARM and put into a PND.


I do have a fairly good assortment of cards that I would like to benchmark though - and I'd really like to know what the fastest card is on the Pandora.


Before the "Class 2 should be fast enough for anyone" crowd chimes in - I'm booting from SDHC and I -multitask- on my Pandora. Faster cards DO make a difference.


So - how do we get from where we're at (no benchmarking software) to something nice packaged up in a PND?

 

[Grench]

Here's a good picture of what the Ubuntu benchmark tool looks like when it's run:


http://www.techpowerup.com/forums/showthread.php?t=121432


It should even scale nice to our screen.... maybe. I have no idea really.

 

[Prometheus]

Serious question, now the goofing off is over with: Would that benchmarking application help in spotting problematic cards that don't work properly with the Pandora, though? That is, those cards that use oddball controller hardware that won't play nice with certain filesystems, or just plain won't play nice with the Pandora at all (note that I'm not talking about fake-capacity cards here, but known-genuine ones).


To my knowledge, this is still a bit of an issue, and it mostly affects Class 6 and Class 10 cards, so it's something that people buying fast cards really need to watch out for (I actually ended up having to replace a Class 6 Verbatim card because its controller kept losing writes from the Pandora no matter what the filesystem was, and I almost lost my entire novel because of it once - it was a good thing I'd just backed it up! ).


If the benchmarking app can mark those out, that would make it a genuinely useful tool to have on the Pandora - after all, it's no good having a fast card if it's going to be losing writes all the time due to poor controller hardware, or suchlike. I certainly wouldn't want to risk booting from such a card (it's bad enough using one when booting from NAND).

 

[Grench]

Serious question, now the goofing off is over with: Would that benchmarking application help in spotting problematic cards that don't work properly with the Pandora, though? That is, those cards that use oddball controller hardware that won't play nice with certain filesystems, or just plain won't play nice with the Pandora at all (note that I'm not talking about fake-capacity cards here, but known-genuine ones).

The short answer: I don't know.


What is the easiest way to spot the incompatibility? Do they fail to read/write? Do you need to run a long format on them to figure it out?

To my knowledge, this is still a bit of an issue, and it mostly affects Class 6 and Class 10 cards, so it's something that people buying fast cards really need to watch out for (I actually ended up having to replace a Class 6 Verbatim card because its controller kept losing writes from the Pandora no matter what the filesystem was, and I almost lost my entire novel because of it once - it was a good thing I'd just backed it up!

).

I no longer consider Class 6 cards to be 'fast'. Class 10 is the fastest 'standard' I've seen, but once you cross the 10MB/sec minimum write there is a -lot- of variation in how fast the cards are. Since I/we are very interested in the read rates as well as the writes, the 'class' standard isn't overly helpful by itself.

If the benchmarking app can mark those out, that would make it a genuinely useful tool to have on the Pandora - after all, it's no good having a fast card if it's going to be losing writes all the time due to poor controller hardware, or suchlike. I certainly wouldn't want to risk booting from such a card (it's bad enough using one when booting from NAND).

What test can be executed that will show, 'this card is bad'?

 

[Grench]

Wow - just noticed that there are code snipets on that SD_compatibility_list that are decent benchmarks. They don't get into seek or plotting the block size curve - but they're something.


Any chance we can get those run for the 128GB Lexar cards? I see that there are stats for the 64GB version, but I'm not sure if they're using the same tech.

 

[Prometheus]

I'm sure that there was a test written up by cosurgi (it was documented on the forums somewhere), but I can't seem to find it right now. It involved having to format the card first, though.


With some (such as the 32GB Verbatim Class 6 and the 32GB Kingston Class 4), it's so bad that you don't even need to do the test to see it - you'll just keep losing writes in an obvious and noticeable fashion (i.e., text files being corrupted and blanked when saved, settings seemingly being lost at random, and so on).


With all of the things people might want to test their cards for (speed, genuine-capacity, whether writes get lost in genuine cards due to bad controller hardware) I really do wish there was a more comprehensive SD testing suite out there. :lol:


EDIT;

Any chance we can get those run for the 128GB Lexar cards? I see that there are stats for the 64GB version, but I'm not sure if they're using the same tech.

I'm not quite sure I'm understanding you fully here, but for the cards to work in the Pandora at all, they'd have to be using the same technology, no?


Paging anyone who has one of these cards, I guess.

 

[Grench]

Looking into it more...


I tried that read test on both of my 32GB Patriot LX cards in my Pandora.


Left slot EXT2 first partition (also running OS) 15.3MB/s


Right slot FAT32 whole device 15.9MB/s


The chart is wrong about EXT2 not working - not sure why the other person had issues. I've been running my OS off of an EXT2 on one of these cards for months.


The note by the Lexar 64GB cards seems to be very true. It appears that there is a read-capability limit on the Pandora right around 16MB/s.


So - it looks like I may have topped out for speed on the cards in my Pandora. Now I'm wondering if this is a hardware limitation or if an improved driver/clock/other could boost it more...


Still learning.

 

[Prometheus]

I can't answer the rest, but;

The chart is wrong about EXT2 not working - not sure why the other person had issues. I've been running my OS off of an EXT2 on one of these cards for months.

The reason for this is because some manufacturers of cards use whatever controllers, etc., that they can get their hands on at any given time, rather than manufacturing their own. The result is what you've pointed out here - some people will have problems, and others won't, because what is ostensibly the same card actually has different hardware in different individual units. (This has been observed with Kingston cards here, for example. You won't run into it with, say, SanDisk, because they make all of their own stuff.)

 

[ImmaculateChimp]

Speaking of SanDisk, I'm still waiting for them to bring out a 128GB SD card of their own...(and not demand an obscene price for it.. ).

 

[juliano]

Finally got mine. It is wonderful, and the speed is very good. I can confirm that Daphne runs much smoother with this card in comparison with my old class 4 8GB card.

 

[Linux-SWAT]

Actually the best :


http://www.pretec.com/products/item/sd/sdxc-class16


3.0

 

[Grench]

Actually the best :


http://www.pretec.co...sd/sdxc-class16


3.0

Next time - instead of off-topic necroposting all of the other card threads - can you create a new one specific to the product you're looking at?


You've posted that same thing in two threads - or more, claiming it as, "the best," without anything to back that claim up.


If you could please read THIS thread you will find that transfer rates past 16MB/sec are capped on the Pandora and that THIS thread is talking about a 128GB Lexar card.


So, by all means, start a new thread talking about that pretec card and benchmark it or find out if anyone has one to benchmark and show that it's "the best". But please lay off of digging up old threads to cap with your claim?

 

[Linux-SWAT]

Just added some info for thread followers.

 

[Zuk]

When I did mine I had to make one small change from the directions. I had to use:


sudo fdisk /dev/mmcblk0


instead of the full:


sudo fdisk /dev/mmcblk0p1


Then delete the existing partition and then follow the rest of the instructions. It might of had something to do with me skipping the update to dosfstools at first. When the first go without the update failed, I updated and started over.


Now to start loading these bad boys up...