From: Richard Biener <rguenther@suse.de>
To: juzhe.zhong@rivai.ai
Cc: "richard.sandiford" <richard.sandiford@arm.com>,
gcc-patches <gcc-patches@gcc.gnu.org>,
Pan Li <incarnation.p.lee@outlook.com>,
"pan2.li" <pan2.li@intel.com>,
"kito.cheng" <kito.cheng@sifive.com>
Subject: Re: Re: [PATCH] RISC-V: Bugfix for rvv bool mode precision adjustment
Date: Wed, 1 Mar 2023 14:03:18 +0000 (UTC) [thread overview]
Message-ID: <nycvar.YFH.7.77.849.2303011402110.27913@jbgna.fhfr.qr> (raw)
In-Reply-To: <nycvar.YFH.7.77.849.2303011358180.27913@jbgna.fhfr.qr>
On Wed, 1 Mar 2023, Richard Biener wrote:
> On Wed, 1 Mar 2023, juzhe.zhong@rivai.ai wrote:
>
> > Let's me first introduce RVV load/store basics and stack allocation.
> > For scalable vector memory allocation, we allocate memory according to machine vector-length.
> > To get this CPU vector-length value (runtime invariant but compile time unknown), we have an instruction call csrr vlenb.
> > For example, csrr a5,vlenb (store CPU a single register vector-length value (describe as bytesize) in a5 register).
> > A single register size in bytes (GET_MODE_SIZE) is poly value (8,8) bytes. That means csrr a5,vlenb, a5 has the value of size poly (8,8) bytes.
> >
> > Now, our problem is that VNx1BI, VNx2BI, VNx4BI, VNx8BI has the same bytesize poly (1,1). So their storage consumes the same size.
> > Meaning when we want to allocate a memory storge or stack for register spillings, we should first csrr a5, vlenb, then slli a5,a5,3 (means a5 = a5/8)
> > Then, a5 has the bytesize value of poly (1,1). All VNx1BI, VNx2BI, VNx4BI, VNx8BI are doing the same process as I described above. They all consume
> > the same memory storage size since we can't model them accurately according to precision or you bitsize.
> >
> > They consume the same storage (I am agree it's better to model them more accurately in case of memory storage comsuming).
> >
> > Well, even though they are consuming same size memory storage, I can make their memory accessing behavior (load/store) accurately by
> > emiting the accurate RVV instruction for them according to RVV ISA.
> >
> > VNx1BI,VNx2BI, VNx4BI, VNx8BI are consuming same memory storage with size poly (1,1)
> > The instruction for these modes as follows:
> > VNx1BI: vsevl e8mf8 + vlm, loading 1/8 of poly (1,1) storage.
> > VNx2BI: vsevl e8mf8 + vlm, loading 1/4 of poly (1,1) storage.
> > VNx4BI: vsevl e8mf8 + vlm, loading 1/2 of poly (1,1) storage.
> > VNx8BI: vsevl e8mf8 + vlm, loading 1 of poly (1,1) storage.
> >
> > So base on these, It's fine that we don't model VNx1BI,VNx2BI, VNx4BI, VNx8BI accurately according to precision or bitsize.
> > This implementation is fine even though their memory storage is not accurate.
> >
> > However, the problem is that since they have the same bytesize, GCC will think they are the same and do some incorrect statement elimination:
> >
> > (Note: Load same memory base)
> > load v0 VNx1BI from base0
> > load v1 VNx2BI from base0
> > load v2 VNx4BI from base0
> > load v3 VNx8BI from base0
> >
> > store v0 base1
> > store v1 base2
> > store v2 base3
> > store v3 base4
> >
> > This program sequence, in GCC, it will eliminate the last 3 load instructions.
> >
> > Then it will become:
> >
> > load v0 VNx1BI from base0 ===> vsetvl e8mf8 + vlm (only load 1/8 of poly size (1,1) memory data)
> >
> > store v0 base1
> > store v0 base2
> > store v0 base3
> > store v0 base4
> >
> > This is what we want to fix. I think as long as we can have the way to differentiate VNx1BI,VNx2BI, VNx4BI, VNx8BI
> > and GCC will not do th incorrect elimination for RVV.
> >
> > I think it can work fine even though these 4 modes consume inaccurate memory storage size
> > but accurate data memory access load store behavior.
>
> So given the above I think that modeling the size as being the same
> but with accurate precision would work. It's then only the size of the
> padding in bytes we cannot represent with poly-int which should be fine.
>
> Correct?
Btw, is storing a VNx1BI and then loading a VNx2BI from the same
memory address well-defined? That is, how is the padding handled
by the machine load/store instructions?
Richard.
next prev parent reply other threads:[~2023-03-01 14:03 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-16 15:11 incarnation.p.lee
[not found] ` <9800822AA73B1E3D+5F679DFB-633A-446F-BB7F-59ADEEE67E50@rivai.ai>
2023-02-17 7:18 ` Li, Pan2
2023-02-17 7:36 ` Richard Biener
2023-02-17 8:39 ` Li, Pan2
2023-02-21 6:36 ` Li, Pan2
2023-02-21 8:28 ` Kito Cheng
2023-02-24 5:08 ` juzhe.zhong
2023-02-24 7:21 ` Li, Pan2
2023-02-27 3:43 ` Li, Pan2
2023-02-27 14:24 ` Richard Sandiford
2023-02-27 15:13 ` 盼 李
2023-02-28 2:27 ` Li, Pan2
2023-02-28 9:50 ` Richard Sandiford
2023-02-28 9:59 ` 盼 李
2023-02-28 14:07 ` Li, Pan2
2023-03-01 10:11 ` Richard Sandiford
2023-03-01 10:46 ` juzhe.zhong
2023-03-01 10:55 ` 盼 李
2023-03-01 11:11 ` Richard Sandiford
2023-03-01 11:26 ` 盼 李
2023-03-01 11:53 ` 盼 李
2023-03-01 12:03 ` Richard Sandiford
2023-03-01 12:13 ` juzhe.zhong
2023-03-01 12:27 ` 盼 李
2023-03-01 12:33 ` Richard Biener
2023-03-01 12:56 ` Pan Li
2023-03-01 13:11 ` Richard Biener
2023-03-01 13:19 ` Richard Sandiford
2023-03-01 13:26 ` Richard Biener
2023-03-01 13:50 ` juzhe.zhong
2023-03-01 13:59 ` Richard Biener
2023-03-01 14:03 ` Richard Biener [this message]
2023-03-01 14:19 ` juzhe.zhong
2023-03-01 15:42 ` Li, Pan2
2023-03-01 15:46 ` Pan Li
2023-03-01 16:14 ` Richard Sandiford
2023-03-01 22:53 ` juzhe.zhong
2023-03-02 6:07 ` Li, Pan2
2023-03-02 8:25 ` Richard Biener
2023-03-02 8:37 ` juzhe.zhong
2023-03-02 9:39 ` Richard Sandiford
2023-03-02 10:19 ` juzhe.zhong
[not found] ` <2023030121501634323743@rivai.ai>
2023-03-01 13:52 ` juzhe.zhong
2023-03-02 5:55 ` [PATCH v2] " pan2.li
2023-03-02 9:43 ` Richard Sandiford
2023-03-02 14:46 ` Li, Pan2
2023-03-02 17:54 ` Richard Sandiford
2023-03-03 2:34 ` Li, Pan2
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=nycvar.YFH.7.77.849.2303011402110.27913@jbgna.fhfr.qr \
--to=rguenther@suse.de \
--cc=gcc-patches@gcc.gnu.org \
--cc=incarnation.p.lee@outlook.com \
--cc=juzhe.zhong@rivai.ai \
--cc=kito.cheng@sifive.com \
--cc=pan2.li@intel.com \
--cc=richard.sandiford@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).