* [OpenRISC] OpenRISC isa changes proposal
@ 2016-10-11 19:23 Alexey Baturo
2016-10-15 14:57 ` Jeremy Bennett
0 siblings, 1 reply; 3+ messages in thread
From: Alexey Baturo @ 2016-10-11 19:23 UTC (permalink / raw)
To: openrisc
Hi, everyone!
I've been looking at openrisc isa recently, especially at orvdx part and
have some ideas to improve it.
First of all, there is no ORVDX32 extention and I don't see why shouldn't
we have it? It would not be very efficient, but why not?
Next one is related to vector instructions encoding. Let's say there are 2
types of vector instructions: the ones who operate on bytes and another
operate on vector. To distinguish them it would be useful to assume the
byte ones have LSB of 0 and 1 for halves. To do that we have to slightly
tune the encoding: i.e. make lv.avg.b and lv.avg.h be encoded with 0x38 and
0x39 respectively, instead of 0x39 and 0x3a. I could provide suggested list
of changes.
Thanks for attention.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.librecores.org/pipermail/openrisc/attachments/20161011/d9af975e/attachment.html>
^ permalink raw reply [flat|nested] 3+ messages in thread
* [OpenRISC] OpenRISC isa changes proposal
2016-10-11 19:23 [OpenRISC] OpenRISC isa changes proposal Alexey Baturo
@ 2016-10-15 14:57 ` Jeremy Bennett
2016-10-22 7:56 ` Stafford Horne
0 siblings, 1 reply; 3+ messages in thread
From: Jeremy Bennett @ 2016-10-15 14:57 UTC (permalink / raw)
To: openrisc
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 11/10/16 20:23, Alexey Baturo wrote:
> Hi, everyone!
>
> I've been looking at openrisc isa recently, especially at orvdx
> part and have some ideas to improve it.
>
> First of all, there is no ORVDX32 extention and I don't see why
> shouldn't we have it? It would not be very efficient, but why not?
>
> Next one is related to vector instructions encoding. Let's say
> there are 2 types of vector instructions: the ones who operate on
> bytes and another operate on vector. To distinguish them it would
> be useful to assume the byte ones have LSB of 0 and 1 for halves.
> To do that we have to slightly tune the encoding: i.e. make
> lv.avg.b and lv.avg.h be encoded with 0x38 and 0x39 respectively,
> instead of 0x39 and 0x3a. I could provide suggested list of
> changes.
Hi Alexey,
I wish you well with this project. Although the original spec included
a set of vector instructions, I don't remember anyone ever
implementing them. IIRC they were never added to Or1ksim either.
So you should have plenty of scope to refine the details of the spec.
Best wishes,
Jeremy
>
> Thanks for attention.
>
>
> _______________________________________________ OpenRISC mailing
> list OpenRISC at lists.librecores.org
> https://lists.librecores.org/listinfo/openrisc
>
- --
Tel: +44 (1590) 610184
Cell: +44 (7970) 676050
SkypeID: jeremybennett
Twitter: @jeremypbennett
Email: jeremy.bennett at embecosm.com
Web: www.embecosm.com
PGP key: 1024D/BEF58172FB4754E1 2009-03-20
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iGMEARECACMFAlgCQ8AcHGplcmVteS5iZW5uZXR0QGVtYmVjb3NtLmNvbQAKCRC+
9YFy+0dU4ZGNAJ4upMBOQjcVz1fCAjEmeWUJLHZl/QCdENfZzxvYn6awIWkiog0i
A6C0o1U=
=l6UW
-----END PGP SIGNATURE-----
^ permalink raw reply [flat|nested] 3+ messages in thread
* [OpenRISC] OpenRISC isa changes proposal
2016-10-15 14:57 ` Jeremy Bennett
@ 2016-10-22 7:56 ` Stafford Horne
0 siblings, 0 replies; 3+ messages in thread
From: Stafford Horne @ 2016-10-22 7:56 UTC (permalink / raw)
To: openrisc
On Sat, Oct 15, 2016 at 11:57 PM, Jeremy Bennett
<jeremy.bennett@embecosm.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 11/10/16 20:23, Alexey Baturo wrote:
>> Hi, everyone!
>>
>> I've been looking at openrisc isa recently, especially at orvdx
>> part and have some ideas to improve it.
>>
>> First of all, there is no ORVDX32 extention and I don't see why
>> shouldn't we have it? It would not be very efficient, but why not?
>>
>> Next one is related to vector instructions encoding. Let's say
>> there are 2 types of vector instructions: the ones who operate on
>> bytes and another operate on vector. To distinguish them it would
>> be useful to assume the byte ones have LSB of 0 and 1 for halves.
>> To do that we have to slightly tune the encoding: i.e. make
>> lv.avg.b and lv.avg.h be encoded with 0x38 and 0x39 respectively,
>> instead of 0x39 and 0x3a. I could provide suggested list of
>> changes.
>
> Hi Alexey,
>
> I wish you well with this project. Although the original spec included
> a set of vector instructions, I don't remember anyone ever
> implementing them. IIRC they were never added to Or1ksim either.
>
> So you should have plenty of scope to refine the details of the spec.
>
Hello, just FYI, I sent a previous mail about moving the architecture
proposal process over to openrisc.io. I just sent a pull request [1]
to add the new process on openrisc.io/architecture.
Once/If the pull request is accepted. This might be a good candidate
for using and giving any feedback on the new process.
[1] https://github.com/openrisc/openrisc.github.io/pulls
-Stafford
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2016-10-22 7:56 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-10-11 19:23 [OpenRISC] OpenRISC isa changes proposal Alexey Baturo
2016-10-15 14:57 ` Jeremy Bennett
2016-10-22 7:56 ` Stafford Horne
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.