The asm {} in D is quite handy, but it seems
that the FISTTP instruction is missing. This
one shouln't be too tricky to implement.

; )

"Bob W" <nospam@aol.com> wrote in message news:d2k694$1c8c$1@digitaldaemon.com...
>
> The asm {} in D is quite handy, but it seems
> that the FISTTP instruction is missing. This
> one shouln't be too tricky to implement.

The FISTTP instruction doesn't appear in the Intel data sheets. Perhaps you mean FISTP?

"Walter" <newshound@digitalmars.com> wrote in message news:d2koue$1tm0$1@digitaldaemon.com...
>
> "Bob W" <nospam@aol.com> wrote in message news:d2k694$1c8c$1@digitaldaemon.com...
>>
>> The asm {} in D is quite handy, but it seems
>> that the FISTTP instruction is missing. This
>> one shouln't be too tricky to implement.
>
> The FISTTP instruction doesn't appear in the Intel data sheets. Perhaps
> you
> mean FISTP?
>
>


No, as opposed to FISTP, FISTTP bypasses rounding
when storing integers, so rounding flags can be left
untouched when a truncated result is desired.


Can be found in the following PDF file:

IA-32 Intel® Architecture
Software Developer’s Manual
Volume 2A:
Instruction Set Reference, A-M
2004


FISTTP: Store Integer with Truncation

Description:
FISTTP converts the value in ST into a signed
integer using truncation (chop) as rounding mode,
transfers the result to the destination, and pops ST.
FISTTP accepts word, short integer, and long
integer destinations.

"Bob W" <nospam@aol.com> wrote in message news:d2l0d4$23il$1@digitaldaemon.com...
> No, as opposed to FISTP, FISTTP bypasses rounding
> when storing integers, so rounding flags can be left
> untouched when a truncated result is desired.

Hmm, must be a new one.

>
> Can be found in the following PDF file:
>
> IA-32 Intel® Architecture
> Software Developer's Manual
> Volume 2A:
> Instruction Set Reference, A-M
> 2004
>
>
> FISTTP: Store Integer with Truncation
>
> Description:
> FISTTP converts the value in ST into a signed
> integer using truncation (chop) as rounding mode,
> transfers the result to the destination, and pops ST.
> FISTTP accepts word, short integer, and long
> integer destinations.
>
>
>

I believe FISTTP is an SSE3 instruction and not (yet) supported by DMC

In article <d2l64r$283h$1@digitaldaemon.com>, Walter says...
>
>
>"Bob W" <nospam@aol.com> wrote in message news:d2l0d4$23il$1@digitaldaemon.com...
>> No, as opposed to FISTP, FISTTP bypasses rounding
>> when storing integers, so rounding flags can be left
>> untouched when a truncated result is desired.
>
>Hmm, must be a new one.
>
>>
>> Can be found in the following PDF file:
>>
>> IA-32 Intel® Architecture
>> Software Developer's Manual
>> Volume 2A:
>> Instruction Set Reference, A-M
>> 2004
>>
>>
>> FISTTP: Store Integer with Truncation
>>
>> Description:
>> FISTTP converts the value in ST into a signed
>> integer using truncation (chop) as rounding mode,
>> transfers the result to the destination, and pops ST.
>> FISTTP accepts word, short integer, and long
>> integer destinations.
>>
>>
>>
>
>

I found it in the Pentium 4 manuals (it was new for that processor). It's added now to the inline assembler.

"pvmoore" <pvmoore_member@pathlink.com> wrote in message news:d2mtp5$tpk$1@digitaldaemon.com...
> I believe FISTTP is an SSE3 instruction and not (yet) supported by DMC

Excellent.

Does this mean that the other SSE3 instructions have been added too?

ADDSUBPD
ADDSUBPS
HADDPD
HSUBPD
HADDPS
HSUBPS
LDDQU
MOVDDUP
MOVSHDUP
MOVSLDUP
MONITOR
MWAIT

It would be nice to have the full set :)

And will they be available to use in DMD 0.120?


In article <d2n7fg$168h$2@digitaldaemon.com>, Walter says...
>
>I found it in the Pentium 4 manuals (it was new for that processor). It's added now to the inline assembler.
>
>"pvmoore" <pvmoore_member@pathlink.com> wrote in message news:d2mtp5$tpk$1@digitaldaemon.com...
>> I believe FISTTP is an SSE3 instruction and not (yet) supported by DMC
>
>

"pvmoore" <pvmoore_member@pathlink.com> wrote in message news:d2onbq$2dqi$1@digitaldaemon.com...
> Does this mean that the other SSE3 instructions have been added too?
>
> ADDSUBPD
> ADDSUBPS
> HADDPD
> HSUBPD
> HADDPS
> HSUBPS
> LDDQU
> MOVDDUP
> MOVSHDUP
> MOVSLDUP
> MONITOR
> MWAIT

Yes.

> It would be nice to have the full set :)

I agree.

> And will they be available to use in DMD 0.120?

Yes. The next DMC++ too.