Hi Francisco,

Based on my understanding, TLBs are modeled in SE mode for x86. With what
accuracy and preciseness TLB timing is modeled is debatable though.
However, it does not seem like that is the case for RISC-V. Specifically,
if you take a look at the src/arch/riscv/tlb.cc (TLB::translate()), you
will notice that the address translation request is sent to the process
page table (rather than looking into the tlb) if FullSystem mode is not
used. I guess that is the reason you do not see any TLB related debug
printings.

-Ayaz

On Fri, Dec 11, 2020 at 12:55 PM Francisco Carlos via gem5-users <
[email protected]> wrote:

> Hello everyone,
>
> I am using gem5 SE mode and investigating memory operation latencies and
> how virtual memory can affect them in a superscalar processor (DerivO3CPU).
>
> Does the SE mode consider TLB delays, TLB hit and miss, for instance, or
> this is implemented only in FullSystem mode?
>
> I saw in the stats file that there are stats related to itb and dtb, but
> in my simulations, they are always zero. Is that an expected behavior?
>
> Additionally, I was trying to have a better understanding of how TLB works
> in gem5, so I enabled TLB debug flag and nothing was printed.
>
> I am currently using the gem5 20.1.0.2 Risc-V ISA and se.py as the
> configuration file, so there is nothing special in my configuration file.
> Any help would be appreciated.
>
>
>
>
> ----------------------------------------------------------------------------------------------------------------------------------------------
> Francisco Carlos Silva Junior
> Phd Student at University of Brasilia
> _______________________________________________
> gem5-users mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
_______________________________________________
gem5-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s

Reply via email to