What is actually the role of HADDR[31:0] in AMBA?

[novicevlsi]

am new to AMBA, and this would be a very simple question to u all,

what actually is the rold of HADDR[31:0]? and

it given on page 3-5(AMBA spec) that ,
"incrementing bursts, which do not wrap at address boundaries,
wrapping bursts,which wrap at particular address boundaries"


thanks

novice

 

[uditkumar1983]

AMBA question

Hi
role of HADDR is to provide the address of slave to/from which place data transfer will occur..

about ur second question :
if your starting location of ur transfer is 0038 then
If wraped burst of four beat is used then ur address will be 0038,003C,0030,0034.... means after address 003C it will go to 0030 but in case of increment burst it will go to 0038,003C,0040,0044 ....

I think u can get it...
Regards

 

[novicevlsi]

Re: AMBA question

hi uditkumar!

thanks for ur kind reply.

so is it that address lines HADDR[31:10] are used to select slaves and HADDR[9:0]
are used to address internal registers of the selected slave?

thanks & regards
novice

 

[asicengineer1]

AMBA question

hi udit,
i've another question based on the bursts, in which cases do we prefer the wrap around burst and where do we prefer incremental burst ?

i guess there is also fixed burst, where the entire data is fetched from an address only, my guess is this might be used for ROM based applications. for the others, i'm not sure. could you please clarify ?

 

[amitgvlsijune06]

Re: AMBA question

u will find answers here

 

[amitgvlsijune06]

Re: AMBA question

A conventional interleaved memory provides an "even" memory bank and an "odd" memory bank. Data having even memory addresses are stored in the even memory bank. Data having odd memory addresses are stored in the odd memory bank. Hence, any two sequential memory locations are stored in separate memory banks. If the two sequential memory addresses are to be accessed, a first memory location is read from the first memory bank and, while the first memory bank is read, the second memory bank is pre-charged. Then, the second memory location is read from the second memory bank, while the first memory bank is pre-charged. Hence, if two sequential memory locations are accessed, the data stored at the locations may be fetched quickly without any intervening pre-charge cycles required for typical random access memory.

Added after 1 minutes:


Burst mode sequences are typically either "linear sequential" (e.g., AMD, and Motorola burst sequences) or "aligned sequential" (e.g., Intel burst sequences), and may have a burst address space of any size, although four-word and 256-word burst address space sizes are typical. For memory chips to be used for burst mode access, burst mode processing circuitry is provided directly on the memory chip. More specifically, a means is provided directly on the chip for loading the initial address, generating sequential addresses within the burst address spaces, and accessing the data at the initial and sequentially-generated addresses, based on a predefined burst mode sequence and burst addressing space size

 

[uditkumar1983]

Re: AMBA question

hi uditkumar!

thanks for ur kind reply.

so is it that address lines HADDR[31] are used to select slaves and HADDR[9]
are used to address internal registers of the selected slave?

thanks & regards
novice

Hi
This is the minimum requirment ( due to 1KB Address boundary , Every slave connected to AMBA AHB bus must have address space of 1KB) ..

Regards

Added after 4 minutes:

hi udit,
i've another question based on the bursts, in which cases do we prefer the wrap around burst and where do we prefer incremental burst ?

i guess there is also fixed burst, where the entire data is fetched from an address only, my guess is this might be used for ROM based applications. for the others, i'm not sure. could you please clarify ?

hi
As i am also thinking same like you , but there may be any othor more advantages ...
So its good if any other member suggest for it..

Regards

 

[kushagrak]

Re: AMBA question

We can use wrapping when we are dealing with cache, and we can go for incremental when we are dealing with buffers....

 

[cuteyes14]

AMBA question

hi
i'm new to this forum.. am working on the AMBA AHB.. i really found the posts useful..

thanx

 

[pascal465]

To come back to the original question, why HBURST, HSIZE and HTRANS is sent to slave ? it just need HADDR.