From: "Amonson, Paul D" <paul.d.amonson@intel.com>
To: "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
CC: Nathan Bossart <nathandbossart@gmail.com>, "Shankaran, Akash"
	<akash.shankaran@intel.com>
Subject: RE: Proposal for Updating CRC32C with AVX-512 Algorithm.
Thread-Topic: Proposal for Updating CRC32C with AVX-512 Algorithm.
Thread-Index: Adqb27jiuXmC5sH1QzmzETmUIqng2AMmmwIw
Date: Fri, 17 May 2024 16:21:19 +0000
Message-ID: <BN0SPR01MB00084DB3E6F61E09F59533FFDCEE2@BN0SPR01MB0008.namprd11.prod.outlook.com>
References: <BL1PR11MB530401FA7E9B1CA432CF9DC3DC192@BL1PR11MB5304.namprd11.prod.outlook.com>
In-Reply-To: <BL1PR11MB530401FA7E9B1CA432CF9DC3DC192@BL1PR11MB5304.namprd11.prod.outlook.com>
Accept-Language: en-US
Content-Language: en-US
x-ms-exchange-antispam-messagedata-chunkcount: 1
x-ms-exchange-antispam-messagedata-0: =?us-ascii?Q?/d2LRm+5LKl8xdrKX4PYKTy2FFzY0YeRk4sPjNdGvbDpVO7hDW1BmmpLlAkE?=
 =?us-ascii?Q?B+S1wuwPR72idN64wLntaEDXEbLVEoPRRMViYuThslB09oiYpg2eNzIMslYE?=
 =?us-ascii?Q?KMtbbRfxSwnLGcc+4pP6wsE1rcFH84D3FEqGDQWtw5zwq0ey45Z0mDaRV8CW?=
 =?us-ascii?Q?s6ArDbzkwXXdSwSgI9kTRSh7JRedkSdOjrPvUKVcmqfQPKr4NdD0f8W02TOb?=
 =?us-ascii?Q?QcOAxY2HQiGCWSb7gU7zN11uoQv787ayRMDKJ9Ycrt62hKIzNzLzZoRlFxGd?=
 =?us-ascii?Q?K2UIiJVBrClWw1/BKEDznx7SjGk1dq0az7LTKRugg1MUgoFdOAtm5ZOHHkqj?=
 =?us-ascii?Q?DGCfrc8WrF4Ltu7nCsYkFiRSA/EYtv5okNBICqb/l9dnceIwMTygtcNY9K03?=
 =?us-ascii?Q?CIwe01TyhRTay6RCB2zivauZBZomSl2BwD6idUFZtMkMAYgnZaYF61LUQyTn?=
 =?us-ascii?Q?xncWVLtUipTpIn9UzXvvJxKufX7mviLMV0lGPb92F1faH5bXd+U6oHych75N?=
 =?us-ascii?Q?EOjnZQMz+jgsC2p8c8HFbO1iOfXk+hdfLRS+XNAho+ufjvUwKxYItQ3P91bP?=
 =?us-ascii?Q?J2jDIs/FJ1IZv9oMpbGUpCl9YtKx2VtQdJgwMh5MldpCAfpictnqZQVQ84cs?=
 =?us-ascii?Q?yRwuSNHunh6j77EBVtYkQkBGT2p7U0H2TM6FxAcDJ+/Jq0DL9fG4tWjiBGt8?=
 =?us-ascii?Q?RYw9ye8zLWORogjbQRtWiphBzBw0WDYzXlyjhj7rPSG/0ZgYAqTzPBvnl49K?=
 =?us-ascii?Q?OSMVC2r+bORfvTklS772RNwlO1Y74fCE0Ca0FlruB9LZAyLY3DTnCtGHq/78?=
 =?us-ascii?Q?EfSDfG3BzqymnZJwJ/gq3i6hs/D9QK8v7/FbwTHR0QBtCUg0zrZJEbfmjdTN?=
 =?us-ascii?Q?iR0TM9O8dgMpaocgUe6dQfDRI7QxfFdpwEzJjVCLQIGugVWWjrv/2yH3qvGp?=
 =?us-ascii?Q?L1H6N1TziVVrQLo1J1XNUSf0wY4GVWwc0S1jpxO/fkNpANArQQwQGeH3YbU4?=
 =?us-ascii?Q?MSJB5o/JF/RLnzy8EbdSUIr97MIFYee+l/jJDZHxxH7pXFM9w/vsq7UGq+WZ?=
 =?us-ascii?Q?7Mkd4Wlmuyndh+fnnNx6KRZaIlbJIL7Twzt4f6cVurtsCd3omp8ve/TyfVm0?=
 =?us-ascii?Q?TsQVfaKXP9qs0A8Ry2gB8pXQrO+uf3yFk8Yn9ZdT917g4HYPcyJe1vf0nDFR?=
 =?us-ascii?Q?RIkAZ54PH+kLcq+AwBr1fwUTlZ+gnm6RDt3LJ2joUE+RMjQ1zKVFjXI53Xas?=
 =?us-ascii?Q?9aAlfnYm+o0c3//MaDRNMaJ+0Es7QoHJv6AtVKRCvHl7CTQt+VkRinsszJXN?=
 =?us-ascii?Q?iae289gs3J3vYGy/eH2oaIyYu611ov25m3n9ogH1xqqgVn5MN6VvV+89fJ+W?=
 =?us-ascii?Q?BGX7ITi0fnEVXtUMiMCuviVGZGUDqRjYC3QCIOcrh95fFuPcsTWz+WwhvD2R?=
 =?us-ascii?Q?zexFJP9THV92wmzL5XM+HBB/4QmZ51pgQ4UwBe+sFlarS5YUx5XbMtI3qoNC?=
 =?us-ascii?Q?6PT/3fXHKXsGiwZuzH+PzybCmyxPzzjgGEUQdGeZnjodkTo14VstWuTyyZZU?=
 =?us-ascii?Q?YzQpolkGCDxcDUvlPqIR4HmXiF8VkZatBCUYKdph?=
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-AuthSource: BN0SPR01MB0008.namprd11.prod.outlook.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 85ad29b3-7ff8-42c2-53fa-08dc768d6293
X-MS-Exchange-CrossTenant-originalarrivaltime: 17 May 2024 16:21:19.6002
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 46c98d88-e344-4ed4-8496-4ed7712e255d
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-CrossTenant-userprincipalname: 7w9vG6Hx8FEh7FQxiNShegsGuBeITe54v92/8TO5Q/KZcS0DtCIeijRokEyS+e/w/O3c4fzCXpobskmur6R/YiypLpucM+llLdWwpVsn4OE=
X-MS-Exchange-Transport-CrossTenantHeadersStamped: PH7PR11MB6883
X-OriginatorOrg: intel.com
List-Id: <pgsql-hackers.lists.postgresql.org>
List-Help: <https://lists.postgresql.org/manage/>
List-Subscribe: <https://lists.postgresql.org/manage/>
List-Post: <mailto:pgsql-hackers@lists.postgresql.org>
List-Owner: <mailto:pgsql-hackers-owner@lists.postgresql.org>
List-Archive: <https://www.postgresql.org/list/pgsql-hackers>
Archived-At: <https://www.postgresql.org/message-id/BN0SPR01MB00084DB3E6F61E09F59533FFDCEE2%40BN0SPR01MB0008.namprd11.prod.outlook.com>
Precedence: bulk

Hi, forgive the top-post but I have not seen any response to this post?

Thanks,
Paul

> -----Original Message-----
> From: Amonson, Paul D
> Sent: Wednesday, May 1, 2024 8:56 AM
> To: pgsql-hackers@lists.postgresql.org
> Cc: Nathan Bossart <nathandbossart@gmail.com>; Shankaran, Akash
> <akash.shankaran@intel.com>
> Subject: Proposal for Updating CRC32C with AVX-512 Algorithm.
>=20
> Hi,
>=20
> Comparing the current SSE4.2 implementation of the CRC32C algorithm in
> Postgres, to an optimized AVX-512 algorithm [0] we observed significant
> gains. The result was a ~6.6X average multiplier of increased performance
> measured on 3 different Intel products. Details below. The AVX-512 algori=
thm
> in C is a port of the ISA-L library [1] assembler code.
>=20
> Workload call size distribution details (write heavy):
>    * Average was approximately around 1,010 bytes per call
>    * ~80% of the calls were under 256 bytes
>    * ~20% of the calls were greater than or equal to 256 bytes up to the =
max
> buffer size of 8192
>=20
> The 256 bytes is important because if the buffer is smaller, it makes sen=
se
> fallback to the existing implementation. This is because the AVX-512 algo=
rithm
> needs a minimum of 256 bytes to operate.
>=20
> Using the above workload data distribution,
> at 0%    calls < 256 bytes, a 841% improvement on average for crc32c
> functionality was observed.
> at 50%   calls < 256 bytes, a 758% improvement on average for crc32c
> functionality was observed.
> at 90%   calls < 256 bytes, a 44% improvement on average for crc32c
> functionality was observed.
> at 97.6% calls < 256 bytes, the workload's crc32c performance breaks-even=
.
> at 100%  calls < 256 bytes, a 14% regression is seen when using AVX-512
> implementation.
>=20
> The results above are averages over 3 machines, and were measured on: Int=
el
> Saphire Rapids bare metal, and using EC2 on AWS cloud: Intel Saphire Rapi=
ds
> (m7i.2xlarge) and Intel Ice Lake (m6i.2xlarge).
>=20
> Summary Data (Saphire Rapids bare metal, AWS m7i-2xl, and AWS m6i-2xl):
> +---------------------+-------------------+-------------------+----------=
---------+---------
> -----------+
> | Rates in Bytes/us   |     Bare Metal    |    AWS m6i-2xl    |   AWS m7i=
-2xl     |
> |
> | (Larger is Better)  +---------+---------+---------+---------+---------+=
---------+
> Overall Multiplier |
> |                     | SSE 4.2 | AVX-512 | SSE 4.2 | AVX-512 | SSE 4.2 |=
 AVX-512 |
> |
> +---------------------+---------+---------+---------+---------+---------+=
---------+-------
> -------------+
> | Numbers 256-8192    |  12,046 |  83,196 |   7,471 |  39,965 |  11,867 |
> 84,589 |        6.62        |
> +---------------------+---------+---------+---------+---------+---------+=
---------+-------
> -------------+
> | Numbers 64 - 255    |  16,865 |  15,909 |   9,209 |   7,363 |  12,496 |
> 10,046 |        0.86        |
> +---------------------+---------+---------+---------+---------+---------+=
---------+-------
> -------------+
>                                                     |  Weighted Multiplie=
r [*]    |        1.44        |
>                                                     +--------------------=
---------+--------------------+
> There was no evidence of AVX-512 frequency throttling from perf data, whi=
ch
> stayed steady during the test.
>=20
> Feedback on this proposed improvement is appreciated. Some questions:
> 1) This AVX-512 ISA-L derived code uses BSD-3 license [2]. Is this compat=
ible
> with the PostgreSQL License [3]? They both appear to be very permissive
> licenses, but I am not an expert on licenses.
> 2) Is there a preferred benchmark I should run to test this change?
>=20
> If licensing is a non-issue, I can post the initial patch along with my P=
ostgres
> benchmark function patch for further review.
>=20
> Thanks,
> Paul
>=20
> [0]
> https://www.researchgate.net/publication/263424619_Fast_CRC_computati
> on#full-text
> [1] https://github.com/intel/isa-l
> [2] https://opensource.org/license/bsd-3-clause
> [3] https://opensource.org/license/postgresql
>=20
> [*] Weights used were 90% of requests less than 256 bytes, 10% greater th=
an
> or equal to 256 bytes.