DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Wang, Yipeng1" <yipeng1.wang@intel.com>
To: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>,
	"Richardson, Bruce" <bruce.richardson@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
	"michel@digirati.com.br" <michel@digirati.com.br>,
	nd <nd@arm.com>
Subject: Re: [dpdk-dev] [PATCH v2 0/7] hash: add extendable bucket and partial key hashing
Date: Sat, 29 Sep 2018 00:46:15 +0000	[thread overview]
Message-ID: <D2C4A16CA39F7F4E8E384D204491D7A6614E9719@FMSMSX151.amr.corp.intel.com> (raw)
In-Reply-To: <AM6PR08MB3672ED8AC62B0C573756E85F98140@AM6PR08MB3672.eurprd08.prod.outlook.com>

>-----Original Message-----
>From: Honnappa Nagarahalli [mailto:Honnappa.Nagarahalli@arm.com]
>> Second, the patch set changes the current hashing algorithm to be "partial-
>> key hashing". Partial-key hashing is the concept from Bin Fan, et al.'s paper
>> "MemC3: Compact and Concurrent MemCache with Dumber Caching and
>> Smarter Hashing".
>I read this paper (but not the papers in references). My understanding is that the existing algorithm already uses 'partial-key hashing'.
>This patch set is not adding the 'partial-key hashing' feature. Instead it is reducing the size of the signature ('tag' as referred in the
>paper) from 32 to 16b.
>Please let me know if I have not understood this correct.
[Wang, Yipeng] Currently two signature values are stored in hash table: sig_current and sig_alt.
They are used to derived the index of two alternative buckets.
Partial key hashing avoids storing two values, but stores only one. The alternative bucket
Index is derived by XOR this single signature with the current bucket index.
So, this commit not only reduces the signature size from 32-bit to 16-bit, it also
reduces the number of signatures stored from two to one.
As a result, we can use one 64-byte cache line for the bucket instead of two.

>> Instead of storing both 32-bit signature and alternative
>> signature in the bucket, we only store a small 16-bit signature and calculate
>> the alternative bucket index by XORing the signature with the current bucket
>> index.
>According to the referenced paper, the signature ('tag') reduces the number of accesses to the keys, thus improving the performance.
>But, if we reduce the size of the signature from 32b to 16b, it will result in higher probability of false matches on the signature. This in
>turn will increase the number of accesses to keys. Have you run any performance benchmarks and compared the numbers with the
>existing code? Is it possible to share the numbers?
>
[Wang, Yipeng] 
>From our test it is very unlikely that two different keys would map to the same bucket and also have the same 16-bit signature.
Even we reduce the signature size from 32-bit to 16-bit, it should be very very rare assuming a good hash function.
For performance numbers, it could vary depending on the test case. Since the speedup comes from the 2X memory efficiency,
If your hash table is large (e.g. over last level cache), it will give much higher speedup. For the existing unit test, I generally seen 10% speedup. 

  reply	other threads:[~2018-09-29  0:50 UTC|newest]

Thread overview: 107+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-09-06 17:09 [dpdk-dev] [PATCH v1 0/5] hash: add extendable bucket and partial-key hashing Yipeng Wang
2018-09-06 17:09 ` [dpdk-dev] [PATCH v1 1/5] test: fix bucket size in hash table perf test Yipeng Wang
2018-09-06 17:09 ` [dpdk-dev] [PATCH v1 2/5] test: more accurate hash table perf test output Yipeng Wang
2018-09-06 17:09 ` [dpdk-dev] [PATCH v1 3/5] hash: add extendable bucket feature Yipeng Wang
2018-09-06 17:09 ` [dpdk-dev] [PATCH v1 4/5] test: implement extendable bucket hash test Yipeng Wang
2018-09-06 17:09 ` [dpdk-dev] [PATCH v1 5/5] hash: use partial-key hashing Yipeng Wang
2018-09-21 17:17 ` [dpdk-dev] [PATCH v2 0/7] hash: add extendable bucket and partial key hashing Yipeng Wang
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 1/7] test/hash: fix bucket size in hash perf test Yipeng Wang
2018-09-26 10:04     ` Bruce Richardson
2018-09-27  3:39       ` Wang, Yipeng1
2018-09-27  4:23     ` Honnappa Nagarahalli
2018-09-29  0:31       ` Wang, Yipeng1
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 2/7] test/hash: more accurate hash perf test output Yipeng Wang
2018-09-26 10:07     ` Bruce Richardson
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 3/7] test/hash: fix rw test with non-consecutive cores Yipeng Wang
2018-09-26 11:02     ` Bruce Richardson
2018-09-27  3:40       ` Wang, Yipeng1
2018-09-26 11:13     ` Bruce Richardson
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 4/7] hash: fix unnecessary code Yipeng Wang
2018-09-26 12:55     ` Bruce Richardson
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 5/7] hash: add extendable bucket feature Yipeng Wang
2018-09-27  4:23     ` Honnappa Nagarahalli
2018-09-27 11:15       ` Bruce Richardson
2018-09-27 11:27         ` Ananyev, Konstantin
2018-09-27 12:27           ` Bruce Richardson
2018-09-27 12:33             ` Ananyev, Konstantin
2018-09-27 19:21         ` Honnappa Nagarahalli
2018-09-28 17:35           ` Wang, Yipeng1
2018-09-29 21:09             ` Honnappa Nagarahalli
2018-09-29  1:10       ` Wang, Yipeng1
2018-10-01 20:56         ` Honnappa Nagarahalli
2018-10-02  1:56           ` Wang, Yipeng1
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 6/7] test/hash: implement extendable bucket hash test Yipeng Wang
2018-09-27  4:24     ` Honnappa Nagarahalli
2018-09-29  0:50       ` Wang, Yipeng1
2018-09-21 17:17   ` [dpdk-dev] [PATCH v2 7/7] hash: use partial-key hashing Yipeng Wang
2018-09-27  4:24     ` Honnappa Nagarahalli
2018-09-29  0:55       ` Wang, Yipeng1
2018-09-26 12:57   ` [dpdk-dev] [PATCH v2 0/7] hash: add extendable bucket and partial key hashing Bruce Richardson
2018-09-27  3:41     ` Wang, Yipeng1
2018-09-27  4:23   ` Honnappa Nagarahalli
2018-09-29  0:46     ` Wang, Yipeng1 [this message]
2018-09-26 12:54 ` [dpdk-dev] [PATCH v3 0/5] hash: fix multiple issues Yipeng Wang
2018-09-26 12:54   ` [dpdk-dev] [PATCH v3 1/5] test/hash: fix bucket size in hash perf test Yipeng Wang
2018-09-27 11:17     ` Bruce Richardson
2018-09-26 12:54   ` [dpdk-dev] [PATCH v3 2/5] test/hash: more accurate hash perf test output Yipeng Wang
2018-09-26 12:54   ` [dpdk-dev] [PATCH v3 3/5] test/hash: fix rw test with non-consecutive cores Yipeng Wang
2018-09-27 11:18     ` Bruce Richardson
2018-09-26 12:54   ` [dpdk-dev] [PATCH v3 4/5] test/hash: fix missing file in meson build file Yipeng Wang
2018-09-27 11:22     ` Bruce Richardson
2018-09-26 12:54   ` [dpdk-dev] [PATCH v3 5/5] hash: fix unused define Yipeng Wang
2018-09-28 14:11   ` [dpdk-dev] [PATCH v4 0/5] hash: fix multiple issues Yipeng Wang
2018-09-28 14:11     ` [dpdk-dev] [PATCH v4 1/5] test/hash: fix bucket size in hash perf test Yipeng Wang
2018-10-01 20:28       ` Honnappa Nagarahalli
2018-09-28 14:11     ` [dpdk-dev] [PATCH v4 2/5] test/hash: more accurate hash perf test output Yipeng Wang
2018-09-28 14:11     ` [dpdk-dev] [PATCH v4 3/5] test/hash: fix rw test with non-consecutive cores Yipeng Wang
2018-09-28 14:11     ` [dpdk-dev] [PATCH v4 4/5] test/hash: fix missing file in meson build file Yipeng Wang
2018-09-28 14:11     ` [dpdk-dev] [PATCH v4 5/5] hash: fix unused define Yipeng Wang
2018-10-25 22:04     ` [dpdk-dev] [PATCH v4 0/5] hash: fix multiple issues Thomas Monjalon
2018-09-26 20:26 ` [dpdk-dev] [PATCH v3 0/3] hash: add extendable bucket and partial key hashing Yipeng Wang
2018-09-26 20:26   ` [dpdk-dev] [PATCH v3 1/3] hash: add extendable bucket feature Yipeng Wang
2018-09-26 20:26   ` [dpdk-dev] [PATCH v3 2/3] test/hash: implement extendable bucket hash test Yipeng Wang
2018-09-26 20:26   ` [dpdk-dev] [PATCH v3 3/3] hash: use partial-key hashing Yipeng Wang
2018-09-28 17:23   ` [dpdk-dev] [PATCH v4 0/4] hash: add extendable bucket and partial key hashing Yipeng Wang
2018-09-28 17:23     ` [dpdk-dev] [PATCH v4 1/4] hash: fix race condition in iterate Yipeng Wang
2018-10-01 20:23       ` Honnappa Nagarahalli
2018-10-02  0:17         ` Wang, Yipeng1
2018-10-02  4:26           ` Honnappa Nagarahalli
2018-10-02 23:53             ` Wang, Yipeng1
2018-09-28 17:23     ` [dpdk-dev] [PATCH v4 2/4] hash: add extendable bucket feature Yipeng Wang
2018-10-02  3:58       ` Honnappa Nagarahalli
2018-10-02 23:39         ` Wang, Yipeng1
2018-10-03  4:37           ` Honnappa Nagarahalli
2018-10-03 15:08           ` Stephen Hemminger
2018-10-03 15:08       ` Stephen Hemminger
2018-10-03 16:53         ` Wang, Yipeng1
2018-10-03 17:59           ` Honnappa Nagarahalli
2018-10-04  1:22             ` Wang, Yipeng1
2018-09-28 17:23     ` [dpdk-dev] [PATCH v4 3/4] test/hash: implement extendable bucket hash test Yipeng Wang
2018-10-01 19:53       ` Honnappa Nagarahalli
2018-09-28 17:23     ` [dpdk-dev] [PATCH v4 4/4] hash: use partial-key hashing Yipeng Wang
2018-10-01 20:09       ` Honnappa Nagarahalli
2018-10-03 19:05     ` [dpdk-dev] [PATCH v4 0/4] hash: add extendable bucket and partial key hashing Dharmik Thakkar
2018-10-01 18:34   ` [dpdk-dev] [PATCH v5 " Yipeng Wang
2018-10-01 18:34     ` [dpdk-dev] [PATCH v5 1/4] hash: fix race condition in iterate Yipeng Wang
2018-10-02 17:26       ` Honnappa Nagarahalli
2018-10-01 18:35     ` [dpdk-dev] [PATCH v5 2/4] hash: add extendable bucket feature Yipeng Wang
2018-10-01 18:35     ` [dpdk-dev] [PATCH v5 3/4] test/hash: implement extendable bucket hash test Yipeng Wang
2018-10-01 18:35     ` [dpdk-dev] [PATCH v5 4/4] hash: use partial-key hashing Yipeng Wang
2018-10-02 20:52       ` Dharmik Thakkar
2018-10-03  0:43         ` Wang, Yipeng1
2018-10-03 19:10     ` [dpdk-dev] [PATCH v5 0/4] hash: add extendable bucket and partial key hashing Dharmik Thakkar
2018-10-04  0:36       ` Wang, Yipeng1
2018-10-04 16:35   ` [dpdk-dev] [PATCH v6 " Yipeng Wang
2018-10-04 16:35     ` [dpdk-dev] [PATCH v6 1/4] hash: fix race condition in iterate Yipeng Wang
2018-10-04 16:35     ` [dpdk-dev] [PATCH v6 2/4] hash: add extendable bucket feature Yipeng Wang
2018-10-04 16:35     ` [dpdk-dev] [PATCH v6 3/4] test/hash: implement extendable bucket hash test Yipeng Wang
2018-10-04 16:35     ` [dpdk-dev] [PATCH v6 4/4] hash: use partial-key hashing Yipeng Wang
2018-10-10 21:27     ` [dpdk-dev] [PATCH v7 0/4] hash: add extendable bucket and partial key hashing Yipeng Wang
2018-10-10 21:27       ` [dpdk-dev] [PATCH v7 1/4] hash: fix race condition in iterate Yipeng Wang
2018-10-10 21:27       ` [dpdk-dev] [PATCH v7 2/4] hash: add extendable bucket feature Yipeng Wang
2018-10-10 21:27       ` [dpdk-dev] [PATCH v7 3/4] test/hash: implement extendable bucket hash test Yipeng Wang
2018-10-10 21:27       ` [dpdk-dev] [PATCH v7 4/4] hash: use partial-key hashing Yipeng Wang
2018-10-16 18:47       ` [dpdk-dev] [PATCH] doc: update release note for hash library Yipeng Wang
2018-10-17 20:09         ` Honnappa Nagarahalli
2018-10-25 18:45           ` Wang, Yipeng1
2018-10-25 23:07             ` Thomas Monjalon

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=D2C4A16CA39F7F4E8E384D204491D7A6614E9719@FMSMSX151.amr.corp.intel.com \
    --to=yipeng1.wang@intel.com \
    --cc=Honnappa.Nagarahalli@arm.com \
    --cc=bruce.richardson@intel.com \
    --cc=dev@dpdk.org \
    --cc=michel@digirati.com.br \
    --cc=nd@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 NNTP newsgroup(s).