+1jamesh wrote:I think a 8GB card is a sensible size vs price compromise.
Tell you what, I'll agree with you when you are right, and you can agree with me when I am right.pluggy wrote:+1jamesh wrote:I think a 8GB card is a sensible size vs price compromise.
Even if James and I are somewhat notorious for not agreeing
You can't really argue with the swag store offering at 4 quid.
Except that one is showing as "out of stock" at the moment.pluggy wrote:You can't really argue with the swag store offering at 4 quid.
Why? Storage performance seems very important for having a system that feels snappy in use. My new B+ has a class 4 card whereas my old B has a class 10 card. The new B+ feels very sluggish. I would say that the difference in responsiveness is greater than any CPU overclocking could achieve.BMS Doug wrote:Class 4 or class 6 cards are better for use with a pi than class 10 cards.
Night and day between my Class 10 Samsung SD card and Class 4 Kingston MicroSD card, in favour of the Samsung. So it was not a waste of money at all. Although I didn't choose it because of the class rating but because of the reported read speed. So, instead of buying class 4 or 6 cards, I would recommend to check what is the the card's read speed in megabytes per second as reported by the manufacturer. I suspect that big companies as Samsung, Sandisk , Kingston and Transcend do report their speeds accurately. If you can not find the speed reported anywhere, do not buy the card - because most probably it is a slow card.pluggy wrote:The class has nothing whatever to do with the read performance of a card. The read speed which is what matters is usually much faster than the write speed which is what class is about. Class 4 and class 6 cards are often (but not always) faster than a class 10 when it comes to reading. At best, a Class 10 is just a waste of money but theres a fair chance it could be slower. The Sandisk class 4 I use read at around 20-22 MB/s although it writes at about 4 (hence class 4) a Class 10 might only read at 12 MB/s although it writes at 10. The extra performance in writing is completely lost in everything else happening (slowly) on the Pi.
Isn't the problem with all of those benchmarks the same as with every benchmark. Benchmarks are only good for showing how a system performs while running a benchmark rather than running any realistic workload. We all have to admit there's a field full of male bovine excrement written about the whys and wherefores of SDCards and the RPi.BMS Doug wrote:a card optomised for the best write speed is not needed for the pi, see the sd card benchmarks thread here