Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Re: initial run

1 view
Skip to first unread message

[Instr] Albert Chae

unread,
Aug 10, 2008, 4:37:45 PM8/10/08
to
On Sun, 10 Aug 2008, EECS Instructional Account wrote:

> Hey ijust realized that i cant initialize all the lru and and valid bits in
> accessMemory to zero every time the accessMemory function is called, because
> you only want all this initialization to happen in the first call to a cache.
> I was trying to figure out a way to know when its the first call, by possibly
> assuming that the addr given in to accessMemory is 0x00400000 (the first PC
> value, and use that to know, but I am not sure if I can assume that this will
> be what gaurantees the first call? Am I doing it right or is there another
> way I am just not thinking of. Ive been stuck on this for a while

These should be initialized for you. You don't have to worry about
initializing it in accessMemory.

>
> also, we dont have to make sure an instruction fetch given to accessMemory is
> separate from a data fetch given to accessMemory, almost like keeping two
> caches?
>

No you don't.

Albert

EECS Instructional Account

unread,
Aug 10, 2008, 3:17:10 PM8/10/08
to
Hey ijust realized that i cant initialize all the lru and and valid bits
in accessMemory to zero every time the accessMemory function is called,
because you only want all this initialization to happen in the first call
to a cache. I was trying to figure out a way to know when its the first
call, by possibly assuming that the addr given in to accessMemory is
0x00400000 (the first PC value, and use that to know, but I am not sure if I can assume
that this will be what gaurantees the first call? Am I doing it right or
is there another way I am just not thinking of. Ive been stuck on this for
a while

also, we dont have to make sure an instruction fetch given to accessMemory

0 new messages