Slashdot Mirror
A Memory Card Torture Test
An anonymous reader writes "Would you buy a Ferrari and put regular gas into it? I don't think so. So why are most of us buying expensive digital cameras and using cheap memory cards? If you want to find out how much better a high speed memory card is, check out this group test of high capacity compact flash and SD cards."
If you take 10,000 photos between taking a "once-in-a-lifetime" photo and backing it up onto a tougher media, you pretty much deserve to lose all your work. The biggest loss of digital camera images are caused by loss/theft of the camera, and user error (accidental deletion). Media failure doesn't even register on the scale.
Athletic Scholarships to universities make as much sense as academic scholarships to sports teams.
What blows my mind in this issue is not the memory cards, but the cameras themselves. A friend of mine just bought a new Canon camera. Sorry I don't recall the model, but it was the newest 8 megapixel SLR they had. Nice camera, he paid a lot for it too. It takes full motion movies. He took my advice and got a 1gb card for it. So we take a few movies and some pictures and plug the camera into their new iMac. And wait. and wait. and wait some more. My god, why is this going so slow? It's been 10 minutes and it's not even 10% done!
The computer shows the camera is hanging off the USB FS (full speed, 12mbps) bus. Why? Is there a problem with the computer? Get out the manual for the camera. Oh.. my.. god... the camera is USB full speed, not high speed. (this is a difference between 12 mbps and 480 mbps for USB cable download speed!) I had to look in several places to confirm the horror. What were they thinking? This camera takes 200mb movies. That takes HOURS at that speed to download.
So we shuttle back down to the camera store and bought him a nice firewire card reader. Back home, we dump then entire card in 10 minutes, movies and pictures included.
This is inconvenient but gets the job done. There is simply no excuse to pay thousands for a camera that takes movies, and have the manufacturers shave a little off the price of manufacturing by substituting a slow USB chip in the camera. And that's all it is, one teeny little chip they just picked the slow one over the fast one. (they are functionally interchangeable, there is no need to redesign the camera) At the bulk they buy chips that can't have saved them more than a dollar per unit.
I have owned two Canon cameras myself and then there is this one. They have performed very well in all cases as excellent digital cameras. But incidents like this make me seriously consider changing brands. If that would have been my camera purchase, it would have gone right back to the store where it came from. Go to store, go directly to store, do not pass go, do not collect $200.
I work for the Department of Redundancy Department.
I was interested in checking out this review, then i saw i would have to page through 20 pages of, well, pages, to make sure i hit their quoat of ads. No thanks, ill read up on fast memory at somone elses site.
Yes, i am aware that half the pages on the internet are like this, and i am also aware that the website owners need to make money. I dont care.
All Troll + "offtopic" mods are meta moderated as "Unfair", because you abused the system.
I have an expensive ($800) Digital Rebel XT 8MP camera. It doen't matter how slow of a card I put in it, it works great. It has it's own high speed cache to store like 8 pictures or so depending on settings, and write to the card seconds later. I can easily take a few shots a second, but, it's rare I need to shoot that. While some high end cameras have the write to card weakness, it's certainly not universal among the Ferrari Cameras. Those of you driving around in old yugos might need every bit of speed increase that you can get, you're better off getting a better camera though IMHO.
-- these are only opinions and they might not be mine.
Probably not in America. Everything at a filling pump is assumed to be of a certain quality over here in the UK.
Erm.... Just about all petrol companies offer different octanes. Standard unleaded in the UK is 95 octane, which is a lot higher than the US I think. BP offer "ultimate", which is 97, Shell offer a 98, and Tesco offer a 99. Most others offer higher than standard octane too - where do you buy petrol?
OT - As an aside to those in the US, it's horribly expensive, £1 per litre, which is about $7 a US gallon if my maths is correct. To fill up my car costs well over the equivalent of $100.
Of course you fail to mention that your country is also
:-)
a) lot smaller
b) Not trading in USD
c) Full of weird as small cars [not a bad thing though].
I mean for me to drive home to my folks place is equivalent to [roughly] driving entirely from one end of England to the other. And I don't even leave the province I'm in to do my trip!!! Talk to me when you live in a country that is 3000Km wide about the price of gas.
That and yeah, if oil wasn't traded in USD you'd probably have an easier time buying gas. Of course the next logical choice is the euro, not the pound. So you're still fucked.
Tom
Someday, I'll have a real sig.
Just to expand a bit...
Actually relating to the ferrari question, if Ferrari tells you to put regular gas and a cup of water into the gast tank, you put regular gas and a cup of water in the gas tank! THEY designed the engine. Back in the 50's engines were so inefficent that it didn't make much of a difference, but with all the sensors, and gizmos on modern engines to get a better burn you need to put the correct octane in your car.
Some engines produce less power if you put a higher octane than the car is rated for because certain calibrations make assumptions about how the gas will burn. Best rule of thumb is to read directions and put whatever the car owners manual tells you to =)
So anyway the correct answer is to do whatever Ferrari tells you to.
Some engines require higher octane fuel to prevent detonation though. If I put 87 octane in my car and start up my datalogger, I see knock and it kicks the timing back a few degrees to prevent it. The car runs at reduced efficiency. If the timing wasn't backed off it would detonate the fuel, which means it explodes instead of burns. This generates increased temps which can burn holes in the pistons. If it cannot back off the timing anymore, like on a hot dry day, it will start backing off the boost (turbo pressure) from 23psi to something around 14psi.
Putting high octane fuel in your chevy cavalier is a waste of money, but putting it in certain types of cars is a requirement to prevent engine damage.
Need Free Juniper/NetScreen Support? JuniperForum
Far be it from me to point out flaws in other peoples tests but these guys are using card readers which are simply slow. Getting 2.2 MB/sec write speed out of a Lexar Pro 133x CF card is pathetic, see more realistic results here (as well as tests on real cameras):
3 xcards.asp
http://www.dpreview.com/news/0601/06011701lexar13
As a Brit living in the US I can attest to the fact that (a) petrol is a lot cheaper here and (b) lots of people drive absurd cars with terrible fuel consumption. However, one thing which might alleviate your concerns a little is to know that the US Gallon is smaller than the UK Gallon (1.2:1). So those numbers probably aren't directly comparable.
---- Den ene knappen er powerknapp, den andre er Bender voice knapp "Bite My Shiny Metal Ass"
That's 36 shots in that package, and they cost at least $0.50 each when everything's included. When I was recently on Safari I took over 5000 shots in a week - with your materials that would have cost $2500 and filled up a pretty decent sized case. As it was it cost nothing (I already had the cards) and everything fitted on my laptop. Sure it's partly a matter of convenience but if I were using film I simply couldn't have taken the shots I did - film was expensive out there and in short supply, and luggage space severely limited. So in that particular case, shotting digital enabled me to take shots I wouldn't otherwise have got.
---- Den ene knappen er powerknapp, den andre er Bender voice knapp "Bite My Shiny Metal Ass"
There are a lot of different aspects to testing nand flash performance. The burst speed of a 25MHz 4-bit bus (used in original SD) would be about 12.5Mbytes/sec. But data is not written immediately to flash, but stored in a buffer. An often quoted read/write speed of 9Mbyte/sec likely involves writes to consecutively addressed blocks and the SD memory block management system has a ready supply of erased blocks. Put a filesystem on top of the NAND memory block management system, and things get more complex. Fragmentation is going to be a problem here eventually as well. Did this test do any long term testing? Another factor (for PC testing) is the SD interface. Is this over USB or and SD slot such as those found in a laptop. The peak rate may be 60Mbyte/sec, but add protocol overhead, and again, random access times can be heavily affected. I went through this a little while ago and wrote a test program which measures peak USB flash memory performance 'under' the filesystem to as to try to attain the quoted peak speeds. I have write and read results for plain blue Sandisk (5 and 8 MByte/sec) and Lexar (5 and 4 MByte/sec) at http://s3u.sf.net/
Precisely. I was vaguely interested in the article - will a more expensive card improve the shooting speed on my camera? I wondered, or more precisely - would it reduce the delay between being able to take pictures?
Page 2 of the article: "many of our digital cameras have limited write speeds too, so the full potential of these so-called high-speed cards will be restricted.".
So nothing to see here, move along.
I have heard stories of some of the highest speed cards breaking in older readers, perhaps the autonegotiation was designed with the assumption that cards would always be slower than a reader's capabilities, or those readers aren't fully meeting the specification and no one noticed.
Most early USB 2.0 (not USB 1.1) readers just plain don't work. Even using the words "high speed" with those readers makes me laugh. They fail at USB/1.1 speeds.
The most common early USB 2.0 flash reader chipset (commonly used in 5-way and 6-way readers or thereabouts) had critical timing-sensitive bugs that were fixed in later revs that cause massive data corruption for transfer sizes over about 512 bytes at a time. At periods proportional to the transfer size (I think), one block is substituted with a copy of block zero, usually right in the middle of your files. This only occurs on OSes that actually try to transfer data in large chunks ant buffer it like Mac OS X and Linux, resulting in bizarre banding on photos, and since the data is cached, there's no way to re-read the data with that reader and get back the right data. You simply have to pull the card and stick it in a different reader.
On Windows, the same readers usually "just work" because Windows does I/O in smaller chunks and AFAIK never buffers anything from removable devices. The problem you describe may very well be the same bug, just appearing at smaller transfer sizes due to the data becoming available more quickly or something.
I once did a hacked driver that capped transfers at 512 bytes to work around the problem. Disgusting, and performance sucks, but it should theoretically make some really seriously broken USB card readers "work".... I don't have the reader anymore, so I don't really care.... I figured out that workaround after realizing that even with the broken reader, I could use 'dd' to clone the card one block at a time, resulting in a valid .iso image with no corruption....
Anyway, after that experience, it's FireWire readers or bust....
Check out my sci-fi/humor trilogy at PatriotsBooks.
I can easily exceed the rate of whatever memory I put in my Nikon D70s when I shoot pictures back to back. Then again, I often shoot in raw mode to allow me to perform better post processing of the images. Each image is 5-6MB in size, so at 3 pictures per second the flash will not keep up. The faster flash definitely makes a difference if I am shooting a lot of pictures, since the raw buffer is only 4 pictures.
Granted, the higher end cameras have larger buffers which helps mitigate the problem, but faster flash is definitely an advantage in these cases. Now with JPEG, it would be a different story, especially with normal compression.
This post is encrypted twice with ROT-13. Documenting or attempting to crack this encryption is illegal.
USB 1.0 has two modes:
Full Speed - 12Mbits/s
Low Speed - 1.5Mbits/s
1.5Mb ~ 187 KB
100MB/600sec ~ 167 KB/s
If he was using Windows XP, the various service packs &/or patches fscked up his USB settings. I'm pretty sure it's a known problem with WinXP's SP/patches, but I can't be bothered to look it up.
Alternatively, the camera only supports Low Speed Mode.
[Fuck Beta]
o0t!
Depends. If you take your pics in single JPEGs it hardly matters.
On my Panasonic FZ-10, I can take a burst of 4 pictures in about 2 seconds. On an cheap slow card, the camera writes almost 10 seconds before letting me take more pictures. With a 133x card, the recycle time is only about 2-3 seconds.
Same cards again in my friends FZ-30, taking pictures in RAW format. At about 15 megs each, it took almost 10 seconds to write a single picture, but on the 133x card perhaps two.
So that camera can write at least 15 megs/ 2 seconds, I don't know what would happen with faster cards, but cards writing slower that 7.5 megs/sec would slow the camera down.