回信： [oc] Re: about OR16

[jimmy87@sunplus.com.tw]

In fact, I'll try my best.
Either w/ '32' or w/o '32' is ok to me.

If you change it, just notify me and I could adapt myself. After all,
VHDL/Verilog RTL could only distinguish binary codes.

In fact, I REALLY need the three docs you wrote.
Lampret, could you send me those docs once when you finished.
Best regards,
Jimmy





"Damjan Lampret" <lampret@opencores.org> 於 2000/03/16 10:59:46 PM

請回應 給 cores@opencores.org

收件人：  jimmy87/Sunplus@Sunplus
副本抄送："Johan Rydberg" <johan.rydberg@netinsight.se>, cores@opencores.org

主旨：    [oc] Re: about OR16



> I want to pose 4 new instructions to extend the DSP computing powers.
> What do you think about it?
>

Sure.

> h.ma32s rA,rB => r14 = r14 + rA * rB (multiply add signed)
> h.ms32s rA,rB => r14 = r14 - rA * rB (multipy subtract signed)
> h.mod32s rA, rB => rA = rA % rB (signed)
> h.mod32u rA, rB => rA = rA % rB (unsigned)
>

BTW I have another version of OR32 without these '32' in every mnemonic. I
think it looks better w/o. What do you think?

> binary codes are
> 1101_0011_A[4]_B[4] -> h.ma32s
> 1101_0100_A[4]_B[4] -> h.ms32s
> 1101_0101_A[4]_B[4] -> h.mod32s
> 1101_0110_A[4]_B[4] -> h.mod32u
>
I am working on three docs to be published. Once they are published and in
CVS everyone can modify. I hope Jimmy will change architecture so it will
also support OR16 properly. Until then I am trying to modify things so Jimmy
won't have problems later.

Names of the manuals are:

OpenRISC 1000: System Architecture Manual
OpenRISC 1000: External Bus Interfaces
OpenRISC 1003: Implementation Manual

regards, Damjan