RE: [PATCH] Performance Improvement in CRC16 Calculations.

Joe, Doug, Nicolas,
The CONFIG patch change suggested by Joe and Doug makes sense to do.   I'll=
 do some additional testing to verify the performance on my systems.  =20


Jeff Lien

-----Original Message-----
From: Joe Perches [mailto:joe@perches.com]=20
Sent: Saturday, August 11, 2018 10:06 AM
To: dgilbert@interlog.com; Nicolas Pitre <redacted>
Cc: Jeffrey Lien <redacted>; linux-kernel@vger.kernel.org; linux-c=
rypto@vger.kernel.org; linux-block@vger.kernel.org; linux-scsi@vger.kernel.=
org; herbert@gondor.apana.org.au; tim.c.chen@linux.intel.com; martin.peters=
en@oracle.com; David Darrington [off-list ref]; Jeff Furlong <j=
eff.furlong@wdc.com>
Subject: Re: [PATCH] Performance Improvement in CRC16 Calculations.

On Sat, 2018-08-11 at 02:04 -0700, Joe Perches wrote:

On Fri, 2018-08-10 at 22:39 -0400, Douglas Gilbert wrote:

quoted

but below is a copy and paste of a table 27 from draft SBC-4=20
revision 15 in chapter 4.22.4.4 on page 87.

=20
The posted code returns the proper crc for each=20
CONFIG_CRYPTO_CRCT10DIF_TABLE_SIZE value from
1 to 5 for these arrays.

Jeff, could you please test the suggested patch with your comparison framew=
ork again with each CONFIG_CRYPTO_CRCT10DIF_TABLE_SIZE from 1 to 5?

I get on a very limited test framework here:
(runtime average of 10 runs)

1: 4.32
2: 1.86
3: 1.31
4: 1.05
5: 0.99