Sat 03-Aug-13 08:59 PM | edited Sat 03-Aug-13 09:12 PM by mirek21
>At 100Mhz, the chip transfers 4 bits of data per clock cycle. >For those that do not know, it takes 8 bits to make a byte, so >two clock cycles are needed. Hence the data transfer rate is >half the clock rate with this particular chip.
Correct. However, a DDR50 mode allows to transfer these 4 bits two times per clock cycle (during signal rise and fall). Not sure if this is used in this case though.
More importantly, there is one important part behind the SDIO interface - a flash memory chip. In real wold is is the memory chip that defines the real-world performance of a SD card, not the SDIO signalling speed.
> Also, it is quite possible >that even a 1/8 of a second of error in the measurement is >skewing the results of the file transfer measurements. 1/8th >of a second would equal another 6.25Mbytes of data transferred >and adding that to the observer 39.5 would equal >45Mbytes/second, or, exactly what Sandisk claims. >
No, the results are repeatable. I can consistently take 17 extra RAW frames when having the shutter button pressed for 30s burst with Lexar 90mb/s card compared to 45mb/s. When repeating the error is maximum 1 frame.
17 frames is quite a difference, since D600 can only take 5.5 fps when buffering, so it can't be explained as time measuring error.