ok, I will try that.
Thanks.
Nathanaël
Le 03/06/2012 16:00, Ali Saidi a écrit :
If you want to account for these (bearing in mind that even a miss won't delay
execution) you can add the stat increments to translateSe.cc in arch/arm/tlb.cc
Ali
Sent from my ARM powered mobile device
On Jun 3,
If you want to account for these (bearing in mind that even a miss won't delay
execution) you can add the stat increments to translateSe.cc in arch/arm/tlb.cc
Ali
Sent from my ARM powered mobile device
On Jun 3, 2012, at 5:18 AM, Nathanaël Prémillieu wrote:
> Then why the stats of the TLBs a
Then why the stats of the TLBs are all zero in my runs ?
Nathanaël
Le 02/06/2012 19:58, Gabe Black a écrit :
There *is* a TLB in SE mode, and address translation. The primary
difference is that in SE mode on misses it's filled from a magical page
table structure which the simulator maintains in
In SE mode I think so. Page faults are generally fatal in SE mode
because they represent a seg fault, except if they're just past the end
of the stack which indicates that the stack was just not big enough.
Gabe
On 06/02/12 11:15, Amin Farmahini wrote:
> So are TLB misses handled instantaneously
So are TLB misses handled instantaneously in same tick?
What about page faults? I would guess no page fault is raised in SE.
Thanks,
Amin
On Sat, Jun 2, 2012 at 11:58 AM, Gabe Black wrote:
> There *is* a TLB in SE mode, and address translation. The primary
> difference is that in SE mode on mis
There *is* a TLB in SE mode, and address translation. The primary
difference is that in SE mode on misses it's filled from a magical page
table structure which the simulator maintains internally, and in FS mode
it uses the normal mechanism defined by the architecture.
Gabe
On 06/02/12 09:27, Ali
Ok, thanks for these details.
Nathanaël
Le 02/06/2012 18:27, Ali Saidi a écrit :
There is no way to have one, "out-of-the-box." You'd need to do a fair amount
of work to build a page table and allocate memory in a non-sequential way. Running in FS
mode is probably the more expedient method.
There is no way to have one, "out-of-the-box." You'd need to do a fair amount
of work to build a page table and allocate memory in a non-sequential way.
Running in FS mode is probably the more expedient method.
Thanks,
Ali
On Jun 2, 2012, at 4:58 AM, Nathanaël Prémillieu wrote:
> Ok,
> Thank
Ok,
Thanks for this answer.
Is there a way to way to had a tlb in SE mode, to have a more accurate
model ?
Regards.
Nathanaël
Le 31/05/2012 18:00, Ali Saidi a écrit :
Hi Nathanaël,
In SE mode VA == PA so no translation is required.
Thanks,
Ali
On 5/31/12 11:54 AM, "Nathanaël Prémillieu"
Hi Nathanaël,
In SE mode VA == PA so no translation is required.
Thanks,
Ali
On 5/31/12 11:54 AM, "Nathanaël Prémillieu" wrote:
>Hi,
>
>I'm running gem5 in SE mode for the ARM architecture. I'm using the
>'arm_detailed' cpu type with --caches and --l2cache options activated.
>However, all the
Hi,
I'm running gem5 in SE mode for the ARM architecture. I'm using the
'arm_detailed' cpu type with --caches and --l2cache options activated.
However, all the statistics about the TLB (dtb and itb sections of the
statistics) are null:
system.cpu.dtb.inst_hits 0 # IT
11 matches
Mail list logo
