From mboxrd@z Thu Jan 1 00:00:00 1970 From: Loic Dachary Subject: Re: fecpp C++ forward error correction library Date: Tue, 30 Apr 2013 10:11:59 +0200 Message-ID: <517F7CCF.1020107@dachary.org> References: <517538D7.6050706@dachary.org> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig387928F38C1A858519CFEA84" Return-path: Received: from smtp.dmail.dachary.org ([86.65.39.20]:39846 "EHLO smtp.dmail.dachary.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757879Ab3D3IMD (ORCPT ); Tue, 30 Apr 2013 04:12:03 -0400 In-Reply-To: Sender: ceph-devel-owner@vger.kernel.org List-ID: To: Christopher LILJENSTOLPE Cc: Ceph Development , Jack Lloyd This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig387928F38C1A858519CFEA84 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable On 04/30/2013 02:08 AM, Christopher LILJENSTOLPE wrote: > Supposedly, on 2013-Apr-22, at 06.19 PDT(-0700), someone claiming to be= Loic Dachary scribed: >=20 >> Hi Christopher, >> >> Jack Lloyd is the author of fecpp ( http://www.randombit.net/code/fecp= p/ ) and he tells me someone sent him a new SIMD approach a few weeks ago= =2E I'm not sure what SIMD means yet, but I'll figure it out ;-). I tend = to favor fecpp because it is more self contained and may be easier to emb= ed than https://pypi.python.org/pypi/zfec >=20 > I'll defer, provided that we don't incur a performance penalty (which i= s one of the negatives in sharding). As far as SIMD: Single Instruction = Multiple Data. Basically, if you think of a tiled processor (like a GPU)= , then you set all the tiles to do the same instruction, and stream diffe= rent data to each tile. As opposed to MIMD, where each tile runs a diffe= rent instruction on different data. You don't need parallel processing = CPUs to make use of SIMD, but you can get them going screamingly fast if = you do=E2=80=A6 Hi Christopher, Thanks for explaining :-) Although it seems disk / network I/O are the bo= ttleneck, it's good to be able to reduce the processing footprint to a mi= nimum. Cheers >=20 > Christopher >=20 >> >> Cheers >> >> --=20 >> Lo=C3=AFc Dachary, Artisan Logiciel Libre >=20 >=20 > -- > =E6=9D=8E=E6=9F=AF=E7=9D=BF > Check my PGP key here: http://www.asgaard.org/cdl/cdl.asc > Current vCard here: http://www.asgaard.org/cdl/cdl.vcf --=20 Lo=C3=AFc Dachary, Artisan Logiciel Libre --------------enig387928F38C1A858519CFEA84 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAlF/fM8ACgkQ8dLMyEl6F22LPQCaAhno1eoRVOq+fBHRVn0nMuVz gnIAoMLxNSUJUwadhOrnZGdcCNlCvPFo =UiUx -----END PGP SIGNATURE----- --------------enig387928F38C1A858519CFEA84--