From: Richard Henderson <richard.henderson@linaro.org>
To: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>, qemu-devel@nongnu.org
Cc: qemu-riscv@nongnu.org, palmer@dabbelt.com,
alistair.francis@wdc.com, dbarboza@ventanamicro.com,
liwei1518@gmail.com, bmeng.cn@gmail.com,
Swung0x48 <swung0x48@outlook.com>,
TANG Tiancheng <tangtiancheng.ttc@alibaba-inc.com>
Subject: Re: [PATCH v4 02/12] tcg/riscv: Add basic support for vector
Date: Wed, 11 Sep 2024 11:41:13 -0700 [thread overview]
Message-ID: <0d591570-02c6-48c9-9e3f-ef47ac20ce7d@linaro.org> (raw)
In-Reply-To: <20240911132630.461-3-zhiwei_liu@linux.alibaba.com>
On 9/11/24 06:26, LIU Zhiwei wrote:
> From: Swung0x48<swung0x48@outlook.com>
>
> The RISC-V vector instruction set utilizes the LMUL field to group
> multiple registers, enabling variable-length vector registers. This
> implementation uses only the first register number of each group while
> reserving the other register numbers within the group.
>
> In TCG, each VEC_IR can have 3 types (TCG_TYPE_V64/128/256), and the
> host runtime needs to adjust LMUL based on the type to use different
> register groups.
>
> This presents challenges for TCG's register allocation. Currently, we
> avoid modifying the register allocation part of TCG and only expose the
> minimum number of vector registers.
>
> For example, when the host vlen is 64 bits and type is TCG_TYPE_V256, with
> LMUL equal to 4, we use 4 vector registers as one register group. We can
> use a maximum of 8 register groups, but the V0 register number is reserved
> as a mask register, so we can effectively use at most 7 register groups.
> Moreover, when type is smaller than TCG_TYPE_V256, only 7 registers are
> forced to be used. This is because TCG cannot yet dynamically constrain
> registers with type; likewise, when the host vlen is 128 bits and
> TCG_TYPE_V256, we can use at most 15 registers.
>
> There is not much pressure on vector register allocation in TCG now, so
> using 7 registers is feasible and will not have a major impact on code
> generation.
>
> This patch:
> 1. Reserves vector register 0 for use as a mask register.
> 2. When using register groups, reserves the additional registers within
> each group.
>
> Signed-off-by: TANG Tiancheng<tangtiancheng.ttc@alibaba-inc.com>
> Co-authored-by: TANG Tiancheng<tangtiancheng.ttc@alibaba-inc.com>
If there is a co-author, there should be another Signed-off-by.
> Reviewed-by: Liu Zhiwei<zhiwei_liu@linux.alibaba.com>
> ---
> tcg/riscv/tcg-target-con-str.h | 1 +
> tcg/riscv/tcg-target.c.inc | 126 ++++++++++++++++++++++++---------
> tcg/riscv/tcg-target.h | 78 +++++++++++---------
> tcg/riscv/tcg-target.opc.h | 12 ++++
> 4 files changed, 151 insertions(+), 66 deletions(-)
> create mode 100644 tcg/riscv/tcg-target.opc.h
Anyway,
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
r~
next prev parent reply other threads:[~2024-09-11 18:42 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-09-11 13:26 [PATCH v4 00/12] tcg/riscv: Add support for vector LIU Zhiwei
2024-09-11 13:26 ` [PATCH v4 01/12] util: Add RISC-V vector extension probe in cpuinfo LIU Zhiwei
2024-09-11 18:34 ` Richard Henderson
2024-09-18 5:14 ` LIU Zhiwei
2024-09-18 10:14 ` Richard Henderson
2024-09-11 13:26 ` [PATCH v4 02/12] tcg/riscv: Add basic support for vector LIU Zhiwei
2024-09-11 18:41 ` Richard Henderson [this message]
2024-09-18 5:17 ` LIU Zhiwei
2024-09-18 10:11 ` Richard Henderson
2024-09-18 10:43 ` LIU Zhiwei
2024-09-18 14:27 ` Richard Henderson
2024-09-20 4:01 ` 0x48 Swung
2024-09-20 4:27 ` LIU Zhiwei
2024-09-20 14:26 ` LIU Zhiwei
2024-09-21 15:56 ` 0x48 Swung
2024-09-21 17:17 ` Daniel Henrique Barboza
2024-09-20 11:26 ` Daniel Henrique Barboza
2024-09-20 11:37 ` Markus Armbruster
2024-09-11 13:26 ` [PATCH v4 03/12] tcg/riscv: Add vset{i}vli and ld/st vec ops LIU Zhiwei
2024-09-11 22:57 ` Richard Henderson
2024-09-22 4:46 ` Richard Henderson
2024-09-23 4:46 ` LIU Zhiwei
2024-09-23 10:10 ` Richard Henderson
2024-09-11 13:26 ` [PATCH v4 04/12] tcg/riscv: Implement vector mov/dup{m/i} LIU Zhiwei
2024-09-11 23:07 ` Richard Henderson
2024-09-11 13:26 ` [PATCH v4 05/12] tcg/riscv: Add support for basic vector opcodes LIU Zhiwei
2024-09-11 13:26 ` [PATCH v4 06/12] tcg/riscv: Implement vector cmp/cmpsel ops LIU Zhiwei
2024-09-11 23:14 ` Richard Henderson
2024-09-11 13:26 ` [PATCH v4 07/12] tcg/riscv: Implement vector neg ops LIU Zhiwei
2024-09-11 13:26 ` [PATCH v4 08/12] tcg/riscv: Implement vector sat/mul ops LIU Zhiwei
2024-09-11 13:26 ` [PATCH v4 09/12] tcg/riscv: Implement vector min/max ops LIU Zhiwei
2024-09-11 13:26 ` [PATCH v4 10/12] tcg/riscv: Implement vector shi/s/v ops LIU Zhiwei
2024-09-11 23:15 ` Richard Henderson
2024-09-11 13:26 ` [PATCH v4 11/12] tcg/riscv: Implement vector roti/v/x ops LIU Zhiwei
2024-09-11 23:24 ` Richard Henderson
2024-09-11 13:26 ` [PATCH v4 12/12] tcg/riscv: Enable native vector support for TCG host LIU Zhiwei
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=0d591570-02c6-48c9-9e3f-ef47ac20ce7d@linaro.org \
--to=richard.henderson@linaro.org \
--cc=alistair.francis@wdc.com \
--cc=bmeng.cn@gmail.com \
--cc=dbarboza@ventanamicro.com \
--cc=liwei1518@gmail.com \
--cc=palmer@dabbelt.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=swung0x48@outlook.com \
--cc=tangtiancheng.ttc@alibaba-inc.com \
--cc=zhiwei_liu@linux.alibaba.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 NNTP newsgroup(s).