|
Message
From: "Damjan Lampret" <lampret@o...>
Date: Thu, 26 Jun 2003 19:00:45 -0700
Subject: Re: [openrisc] WB:RISC Clock Ratios
----- Original Message -----
From: "Brian Adams" <brian.adams@a...>
To: <openrisc@o...>
Sent: Wednesday, June 25, 2003 8:51 AM
Subject: [openrisc] WB:RISC Clock Ratios
> Hello All,
>
> I have a (hopefully) quick sanity check to ask. For the SOC we are
> designing we want to have the ability to run WB:RISC clock ratios of 1:1
> or 1:2. To do this I would have to (?)
>
> 1) Define OR1200_CLKDIV_2_SUPPORTED
> 2) Drive clmode_i on or1200_top appropriately (we plan to use an
> external switch or jumper)
> 3) Drive iwb_clk_i and dwb_clk_i with either the full or half rate WB
> clock (depending on how we set #2)
Exactly right.
>
> Also, are there restrictions on the 1:2 WB/RISC clock relationships?
> ie. Must the WB clock must be EXACTLY 1/2 the RISC clock, and phase
> aligned. We plan to have both, but I was wondering if either was
> absolutely necessary.
It needs to be phase aligned because there is no logic to combat
metastability if clocks are not in sync. There can be skew between main
clock and 1/2 clock, but this simply means reduction in max clock frequency
(you STA will tell you that). But yes the clocks need to be in sync.
>
> I it appears that OR1200_CLKDIV_2_SUPPORTED, OR1200_CLKDIV_4_SUPPORTED
> and clmode were added to make it easy to do just what I have proposed,
> but I have been wrong before. =)
You are right this time. =)
regards,
Damjan
>
> Thanks,
> Brian
>
>
>
>
>
>
|
 |
Browse
