Thanks for the hint Ali.

Going over the trace from instruction queue, I am almost positive it is a
FloatAdd.

system.cpu0.fetch: [tid:0]: Instruction PC 0x8986 (0) created [sn:8792].
system.cpu0.fetch: [tid:0]: Instruction is:   ldrneon8_uop.w
system.cpu0.iq: Instruction is ready to issue, putting it onto the ready
list, PC (0x8986=>0x898a).(0=>1) opclass:4 [sn:8792].
system.cpu0 T0 : 0x8986.0  :   ldrneon8_uop.w           : FloatAdd :
D=0x00000000051319b0 A=0xbefffc18
12837500: system.cpu0: Removing committed instruction [tid:0] PC
(0x8986=>0x898a).(0=>1) [sn:8792]

** opclass: 4 is defined as FloatAdd in src/cpu/op_class.hh.

Thanks,
Amin

On Thu, Jul 26, 2012 at 8:36 PM, Ali Saidi <[email protected]> wrote:

> **
>
>
>
>
>
> On 26.07.2012 19:44, Amin Farmahini wrote:
>
> Hi,
>
> I'd like to change the latency of a couple of Neon instructions. I started
> with VLD1 instruction which is called ldrNeon in Gem5. This instruction is
> classified as a "FloatAdd" opClass. So I can modify the latency (opLat) of
> "FloatAdd" class in O3_ARM_v7a.py file. However, this obviously changes the
> latency of all instructions belong to FloatAdd which is not accepted. So I
> think to change the latency of VLD1 instruction, I need to define a new
> opClass for this instruction. Is this approach right? And if so, how can I
> classify VLD1 under the new opClass?
> I took a look at arch/arm/isa/insts/fp.isa and
> arch/arm/isa/insts/macromem.isa to get some idea, but could find a way to
> work around this. BTW, if there are any other ways to work around this, I'd
> appreciate it if you could share it.
>
> Thanks,
> Amin
>
>
>
> It doesn't seem like it's a floatAdd and it probably shouldn't be. Anyway,
> your idea is correct and I think it's defined in
> src/arch/arm/isa/insts/macromem.isa:688.
>
>     iop = InstObjParams("vldsingle", "VldSingle", 'VldSingleOp', "", [])
>
> you should be abl eto change it to something like
>
>     iop = InstObjParams("vldsingle", "VldSingle", 'VldSingleOp', {"op_class" 
> : "MyNewOpClass"}, [])
>
>
>
> Ali
>
>
>
>
>
> _______________________________________________
> gem5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
>
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to