Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262824AbUCOWOm (ORCPT ); Mon, 15 Mar 2004 17:14:42 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262786AbUCOWOm (ORCPT ); Mon, 15 Mar 2004 17:14:42 -0500 Received: from mtvcafw.sgi.com ([192.48.171.6]:21364 "EHLO omx2.sgi.com") by vger.kernel.org with ESMTP id S262824AbUCOWLy (ORCPT ); Mon, 15 Mar 2004 17:11:54 -0500 Date: Mon, 15 Mar 2004 14:11:27 -0800 To: Andrew Morton , axboe@suse.de Cc: linux-kernel@vger.kernel.org Subject: Re: 2.6.4-mm2 Message-ID: <20040315221127.GA23743@sgi.com> Mail-Followup-To: Andrew Morton , axboe@suse.de, linux-kernel@vger.kernel.org References: <20040314172809.31bd72f7.akpm@osdl.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20040314172809.31bd72f7.akpm@osdl.org> User-Agent: Mutt/1.5.5.1+cvs20040105i From: jbarnes@sgi.com (Jesse Barnes) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5490 Lines: 95 On Sun, Mar 14, 2004 at 05:28:09PM -0800, Andrew Morton wrote: > - Added the new per-address_space block unplugging code. This is designed > to reduce the locking contention against the global plug list's lock and it > also allows us to avoid unplugging all the queues in the machine when we > want just a single queue to kick off its I/O. Looks pretty good, same benchmark as last time (though hopefully the profiles look a little nicer--sorry David, I wasn't able to get qprof working today). 10 qla2200 fc controllers, 64 cpus, 112 disks ------------------------------------- 2.6.4 ~43252 I/Os per second [root@revenue sio]# readprofile -p /root/profile.out -m /root/System.map | sort -nr | head -20 926930 total 0.1601 508304 snidle 1323.7083 170017 cpu_idle 354.2021 148308 default_idle 4634.6250 21765 blk_run_queues 37.7865 16914 __make_request 4.6775 16897 scsi_request_fn 8.9497 6478 scsi_end_request 11.9081 6111 schedule 1.1233 4750 dio_bio_end_io 12.3698 3749 qla2x00_start_scsi 1.5215 3475 qla2x00_queuecommand 0.8484 3062 sn_dma_flush 4.5565 1313 scsi_dispatch_cmd 0.9118 997 scsi_device_unbusy 2.3966 917 dio_await_one 1.5920 910 get_request 0.5687 890 __end_that_request_first 0.8428 873 mempool_alloc 0.9094 834 __mod_timer 0.8145 vmstat: procs memory swap io system cpu r b swpd free buff cache si so bi bo in cs us sy id wa 4 113 0 252927360 12112 4288 0 0 14649 0 73289 56926 0 7 45 48 3 114 0 252927424 12112 4288 0 0 23187 1 75524 91055 0 13 1 85 5 111 0 252927424 12112 4288 0 0 23142 0 75440 90893 0 13 0 87 5 112 0 252927552 12112 4288 0 0 23581 0 75750 92907 0 12 0 88 9 108 0 252927552 12112 4288 0 0 23224 0 75698 91226 0 14 0 86 3 114 0 252927680 12112 4288 0 0 23474 0 75912 92509 0 12 1 87 2 114 0 252927680 12112 4288 0 0 22651 0 75092 88730 0 15 0 85 7 114 0 252927872 12112 4288 0 0 22915 0 75298 89525 0 15 0 84 4 112 0 252927872 12112 4288 0 0 23106 0 75348 90598 0 15 0 85 0 0 0 252951488 12112 20688 0 0 21193 0 89210 85060 0 17 5 79 ------------------------------------- 2.6.4-mm2 ~46655 I/Os per second [root@revenue sio]# readprofile -p /root/profile.out -m /root/System.map | sort -nr | head -20 935122 total 0.1656 543734 snidle 1415.9740 182199 cpu_idle 379.5813 158268 default_idle 4945.8750 6850 scsi_request_fn 3.6282 6067 scsi_end_request 11.1526 4695 __make_request 1.3338 4347 dio_bio_end_io 12.3494 3782 schedule 0.6677 3323 sn_dma_flush 4.9449 3314 pci_cacheline_size 4.5027 1753 sd_rw_intr 1.5652 1313 dio_await_one 1.7840 1259 scsi_device_unbusy 3.0264 1201 scsi_dispatch_cmd 0.8340 847 sd_open 1.2031 671 __mod_timer 0.5825 661 __end_that_request_first 0.6259 593 mempool_free 1.1582 568 scsi_io_completion 0.2863 vmstat: procs memory swap io system cpu r b swpd free buff cache si so bi bo in cs us sy id wa 2 112 0 253452832 12128 69872 0 0 7703 17 77455 25398 0 2 94 4 4 112 0 253452640 12128 69872 0 0 24694 0 76935 84163 0 8 79 13 2 114 0 253452512 12128 69872 0 0 24744 8 76523 86890 0 8 71 21 6 110 0 253452512 12128 69872 0 0 24614 0 76418 86889 0 8 70 22 3 113 0 253452448 12128 69872 0 0 24652 0 76427 87022 0 8 68 24 4 112 0 253452640 12128 69872 0 0 24504 0 76552 87179 0 8 67 25 2 114 0 253452640 12128 69872 0 0 24540 0 76559 86955 0 8 66 27 5 111 0 253452640 12128 69872 0 0 24525 17 76521 87415 0 8 66 26 7 109 0 253452640 12128 69872 0 0 24528 0 76526 87092 0 8 66 27 2 114 0 253452768 12128 69872 0 0 24488 0 76619 87209 0 8 66 27 0 0 0 253498848 12128 69872 0 0 5789 0 84614 21581 0 2 91 6 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/