What does Burst Mode......Burst Transfer...refers to?

hi everyone.........


What does Burst Mode......Burst Transfer...refers to?



Thanks

SDRAM, when issue one command read/write and able to read/write 2/4/8/ full page of data

if u want to write in continnous memory locations it is useful

sry i could n't get you ..............plz

can you be more specific in the way you describe it.........

i mean what is that 2/4/8 signify............


thanks

hi Guru

Letz start from the very basic of memory accesses. When u want to access a single memory location at a time, what u do is tht u just specify the memory location (and the data to be written in case itz a write operation) and that it will be done. Again if u want to access another location u do the same. U know when u give address this way (one at a time), then each time the access will take same time (including accessing that location + reading/writing).

But since we know memory accesses are very slow compared to CPU speed, so a lot of time is wasted in case u want to read say contiguous memory locations and u provide those contiguous locations one at a time.

We can see it this way....say a memory read access takes say N memory cycles ...then if u want to read say M locations n u provide one address at a time...then it will take N*M memory cycles (ideally) for the entire access. Right. But if u really c its a whole lot of time....n we r not gaining anything even if the locations being contiguous.

By this wat i mean to say is that normally in SDRAMs, the access to a memory location is by first opening a particular row, followed by accessing the intended column in that row...n then that row is closed after access. So if u want to read say 4 contiguous locations a time (meaning 4 adjacent columns say N to N+4 of the same row say row no. R) then if u provide all those locations one at a time, then for every access, it will first take some time to open row R , then will access the column N, then will take time to close row R. Again for accessing column N+1 , it will again take time to open row R, then some time to access column N+1, then again will close row R.....

SO u can see unnecessarily its opening n closing the same row again n again n wasting time in that ( when we know that the locations are adjacent )....So a better option is to go for so called BURST MODE...where u just need to give input for once only in the form ofrow address, 1st column address, n the no. of columns to access. Wat SDRAM will do is tht it will open the row for the first time, n then will access columns one after the other without closing the row....n when finally all columns r accessed it will close that row.....

Here watever i said may be not exactly correct in terms of say u cant access any no. of colums (normally its 2 or 4 or 8 or the entire row)....but overall i suppose u'll wat i mean to say....

in case any doubts ...feel free to ask.....
u can also go thru this pdf...plz go thru the timing diagrams for Burst mode access (read/write)

**broken link removed**
or
**broken link removed**

though it may seem like a lot to read....but i suppose its quite easy to understand...so in case u hav time....just go thru it once...at least u'll get the very basic idea of Burst Mode....

the post is really good my doubt is why are the address's given in power's of 2 and not odd numbers.