Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

question about FIFO depth

Status
Not open for further replies.

bradyue

Member level 2
Joined
Jan 18, 2008
Messages
47
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
1,603
how to determine the depth of this FIFO?
write clock is 40Mhz, read clcok is 50Mhz
data is written into FIFO constantly and repeatly, like 0 1 2 3 0 1 2 3
data is read from FIFO repeatly, lie 0 1 2 3 x 0 1 2 3 x
x means no read.
that means every 4 write clock, data in and out of FIFO are the same.
so, how to determine the depth of this FIFO, and how many cycles should data write be ahead of data read?
 

the two clock has the same source
 

I don't think they are the same case.
The write clock is faster than read clock in your referred article.
But in my design, write clock is slower than read clock.



shanmugaveld said:
Watch the below thread may help you.
 

How frequently you need to access the FIFO for read operation.
Whether Burst Read and write are supported?
 

as I stated, in the read side, 4 data out of every 5 read clocks , like 0 1 2 3 x 0 1 2 3 x
in the write side, every write clock per data, like 0 1 2 3 0 1 2 3
accoring to the freq ratio, the duration of 4 data write is equal to 5 data read.
 

bradyue said:
as I stated, in the read side, 4 data out of every 5 read clocks , like 0 1 2 3 x 0 1 2 3 x
in the write side, every write clock per data, like 0 1 2 3 0 1 2 3
accoring to the freq ratio, the duration of 4 data write is equal to 5 data read.

assuming write data width equal to read data width.
assuming no latency when read operating.

write burst length =8, as you said that duration of 8 data write equal to 10 data read. so only 1-level fifo needed to consume one idle cycle during read operating.
To make it stronger, one more level will be better.
 

yes, rhythmrain is right. If you only care about fifo work fine, 2-level fifo will be enough.

but your fifo looks like design for HDMI. so it also depend on your application.
 

yes, it it HDMI
you are so brilliant
so, 2-level is enough?
write pointer is always ahead of read pointer?
it that safe enough?
 

OK, It's Lucky
So how do you want to handle overrun case?
Tell upstream for waiting moment?

maybe we can talk detail by email.
buqingjun(at)gmail
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top