Hi Abdlerhman,

I think your understanding of different requestor IDs for a single core is
correct i.e., these specify instruction and data requests.
"dataRequestorId()" and "instRequestorId()" functions in src/cpu/base.hh
verify this. I think prefetchers should also have their own requestor ids
as any object that can issue memory requests is assigned an id at the start
of the simulation.

-Ayaz

On Wed, Feb 15, 2023 at 8:51 PM Abdlerhman Abotaleb via gem5-users <
gem5-users@gem5.org> wrote:

> I want to identify the source CPU# of a packet.
> I found a field called "pkt->requestorId()"
> """"""""""""
> This field originally can have the following options: "*Source:
> gem5/src/mem/request.hh*"
> *wbRequestorId *= 0, /* writeback requests by the caches */
> *funcRequestorId *= 1, /*  functional requests that  don't come from a
> particular device*/
> *intRequestorId *= 2,  /* message signaled interrupts */
> *invldRequestorId *= std::numeric_limits<RequestorID>::max() /* Invalid
> requestor id for assertion checking only*/
> """""""""""
> Then inside the processor, when create a request this ID is assigned
> either _instRequestorId or _dataRequestorId.
>
> Now when run experiments with 4 cores , single thread :
> I have the following requestor ID values:
> Core 0 (IDs = 0x5,0x6)
> Core 1 (IDs = 0x9,0xA)
> Core 2 (IDs = 0xE,0xD)
> Core 3 (IDs = 0x11,0x12)
> *When have hardware prefetcher enabled*, the IDs get increased.
> """"""
> Now my questions:
> How to associate the IDs found with the source core? How they IDs are
> being assigned?
> Why the same core can generate different requests with different IDs (May
> be one is data and other is instruction) ?
>  In case of HWP enabled , is it considered as an additional source ?
>
> Thank you so much for the help.
>
> _______________________________________________
> gem5-users mailing list -- gem5-users@gem5.org
> To unsubscribe send an email to gem5-users-le...@gem5.org
>
_______________________________________________
gem5-users mailing list -- gem5-users@gem5.org
To unsubscribe send an email to gem5-users-le...@gem5.org

Reply via email to