From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hans Reiser Subject: Re: State of the Reiser4 FS Date: Wed, 15 Mar 2006 10:29:00 -0800 Message-ID: <44185CEC.3010103@namesys.com> References: <3aa654a40603140241s5d8a7430j6bf29a5ef7dd1bf5@mail.gmail.com> <1142336718.7415.55.camel@tribesman.namesys.com> <194f62550603140732j6ef10757j@mail.gmail.com> <4417BEC4.6060506@namesys.com> <20060315085745.GA21609@wintermute> Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Return-path: list-help: list-unsubscribe: list-post: Errors-To: flx@namesys.com In-Reply-To: <20060315085745.GA21609@wintermute> List-Id: Content-Type: text/plain; charset="iso-8859-1" To: =?ISO-8859-1?Q?Andreas_Sch=E4fer?= Cc: reiserfs-list@namesys.com Andreas Sch=E4fer wrote: >On 23:14 Tue 14 Mar , Hans Reiser wrote: > =20 > >>I generally believe that the per 4k >>approach used throughout the linux kernel is not as CPU efficient as >>sending larger groups of pages through the layers all at once. In other >>words, there is a reason we have bios, and we need to learn the lesson >>from them that they teach us, and abstract it into a general design >>approach. >> =20 >> > >This is a bit OT, but still: I think you're absolutely right there. This 4= k=20 >approach causes funny thing even in completely different areas.=20 > >I've recently been playing around with openMosix (transparent process=20 >migration between Linux hosts over the network) and there the write=20 >preformance is totally spoiled (>50% performance loss) just because=20 >copy_from_user is only called for 4k block when writing (and these=20 >requests then suffer from the high network latency).=20 > =20 > Tell the mosix guys we would be willing to cooperate with them regarding their problem. >If larger blocks would be used, this performance flaw wouldn't exist at=20 >all... Just my 2 cents > >-Andreas > > > =20 >