Skip to contentSkip to conversation

ide/disk perf?

DISCUSSION Mohammad A. Haque 3 participants 6 messages Oct 29, 2000 – Oct 30, 2000

lore
Share & tools
lore

Mohammad A. Haque

author
#0 lore raw

Could someone who knows ide and drive inside and out (Andre?) please take a look at these figures? Am I forgetting to do something (or doing something I'm not suposed to) to get the best numbers? I thought I'd be able to get more than ~4MB/sec off the HPT366 and a UDMA66 drive.

Kernel: 2.4.0-test10-pre6

Onboard PIIX4 (Abit BE6-II):

[mhaque@viper mhaque]$ sudo hdparm -X34 -d1 -u1 -c3 -m16 /dev/hda

/dev/hda:
setting 32-bit I/O support flag to 3 setting multcount to 16
setting unmaskirq to 1 (on)
setting using_dma to 1 (on)
setting xfermode to 34 (multiword DMA mode2)
multcount = 16 (on)
I/O support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 1 (on)

[mhaque@viper mhaque]$ dmesg|grep '^hda:'; sudo hdparm -Tt /dev/hda
hda: IBM-DJNA-371350, ATA DISK drive
hda: 26520480 sectors (13578 MB) w/1966KiB Cache, CHS=1650/255/63,
UDMA(33)

	/dev/hda:
	 Timing buffer-cache reads:   128 MB in  1.12 seconds =114.29 MB/sec
	 Timing buffered disk reads:  64 MB in  4.94 seconds = 12.96 MB/sec

	[mhaque@viper mhaque]$ sudo hdparm -i /dev/hda

	/dev/hda:

	 Model=IBM-DJNA-371350, FwRev=J76OA30K, SerialNo=GM0GMFE4929
	 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs }
	 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=34
	 BuffType=3(DualPortCache), BuffSize=1966kB, MaxMultSect=16,
MultSect=16
	 DblWordIO=no, OldPIO=2, DMA=yes, OldDMA=2
	 CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=26520480
	 tDMA={min:120,rec:120}, DMA modes: mword0 mword1 *mword2 
	 IORDY=on/off, tPIO={min:240,w/IORDY:120}, PIO modes: mode3 mode4 
	 UDMA modes: mode0 mode1 mode2 mode3 mode4

Onboard HPT366 (Abit BE6-II):

[mhaque@viper mhaque]$ sudo hdparm -X66 -d1 -u1 -c3 -m16 /dev/hde

/dev/hde:
setting 32-bit I/O support flag to 3
setting multcount to 16
setting unmaskirq to 1 (on)
setting using_dma to 1 (on)
setting xfermode to 66 (UltraDMA mode2)
multcount = 16 (on)
I/O support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 1 (on)

[mhaque@viper mhaque]$ dmesg|grep '^hde:'; sudo hdparm -Tt /dev/hde
hde: Maxtor 53073U6, ATA DISK drive
hde: 60030432 sectors (30736 MB) w/2048KiB Cache, CHS=59554/16/63,
UDMA(66)

	/dev/hde:
	 Timing buffer-cache reads:   128 MB in  1.24 seconds =103.23 MB/sec
	 Timing buffered disk reads:  64 MB in 16.18 seconds =  3.96 MB/sec

	[mhaque@viper mhaque]$ sudo hdparm -i /dev/hde

	/dev/hde:

	 Model=Maxtor 53073U6, FwRev=DA6207V0, SerialNo=K608NRYC
	 Config={ Fixed }
	 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=57
	 BuffType=3(DualPortCache), BuffSize=2048kB, MaxMultSect=16,
MultSect=16
	 DblWordIO=no, OldPIO=2, DMA=yes, OldDMA=0
	 CurCHS=65535/1/63, CurSects=-4128706, LBA=yes, LBAsects=60030432
	 tDMA={min:120,rec:120}, DMA modes: mword0 mword1 mword2 
	 IORDY=on/off, tPIO={min:120,w/IORDY:120}, PIO modes: mode3 mode4 
	 UDMA modes: mode0 mode1 *mode2 mode3 mode4
Signature · 12 lines
=====================================================================
Mohammad A. Haque                              http://www.haque.net/ 
                                               [email protected]

  "Alcohol and calculus don't mix.             Project Lead
   Don't drink and derive." --Unknown          http://wm.themes.org/
                                               [email protected]
=====================================================================
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
#1 lore raw

Le 30-Oct-2000, Mohammad A. Haque écrivait :

Quoted 4 lines

Could someone who knows ide and drive inside and out (Andre?) please take a look at these figures? Am I forgetting to do something (or doing something I'm not suposed to) to get the best numbers? I thought I'd be able to get more than ~4MB/sec off the HPT366 and a UDMA66 drive.

It could be unrelated, but I had problems several times with Maxtor drives recently; Their performances are usually high (some models give >20 Mb/s both reads and writes), but under some conditions that I couldn't narrow down yet, the read throughput is stuck to the floor (a few megabytes/sec) until next reboot. The write performance is always ok.

I don't think it's a chipset problem, with a box filled with several Maxtor drives and I observe the problem with some disk and not others at the same time.

I don't think it's a Linux problem either, since it can be observed with Windows benchmarking tools too.

As a side note, I usually get better results tweaking disks using kernel compiling options ("default to DMA", "autotune chipset", ...) than using hdparm -d1 and friends, give it a try.

Signature · 7 lines
Linux blade.workgroup 2.4.0-test10 #1 Sat Oct 28 18:00:09 CEST 2000 i686 unknown
  1:09pm  up 1 day, 19:01,  2 users,  load average: 1.04, 1.08, 1.09

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
#4 lore raw
In reply to Mohammad A. Haque
Quoted 7 lines · Pierre Etchemaite

Le 30-Oct-2000, Mohammad A. Haque écrivait :

Quoted 4 lines

Could someone who knows ide and drive inside and out (Andre?) please take a look at these figures? Am I forgetting to do something (or doing something I'm not suposed to) to get the best numbers? I thought I'd be able to get more than ~4MB/sec off the HPT366 and a UDMA66 drive.

Quoted 5 lines

It could be unrelated, but I had problems several times with Maxtor drives recently; Their performances are usually high (some models give >20 Mb/s both reads and writes), but under some conditions that I couldn't narrow down yet, the read throughput is stuck to the floor (a few megabytes/sec) until next reboot. The write performance is always ok.

I had the same problem with Seagate ST313021A (13 gig) drives on BP6/HPT366/UDMA66. Initial throughput reported by `hdparm -t' was 22 megs/sec which would slowly wilt to 5 megs/sec.

I discovered that sending _any_ reconfiguration command to the drive - even one which was not supported by that particular drive - would bring the performance back.

So when it goes slow, try running, say, `hdparm -A1' and see what happens.

Andre and I scratched each others heads for a while, suspected a firmware bug. He sent an email to a contact at Seagate. This was in April, so I guess that person is a very slow typist.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

author
#5 lore raw
In reply to Mohammad A. Haque

Interesting theory, but no go here.

I just remembered one other thing that may have an impact. I have a IBM-DTLA-307045 (~45 GB i think) hanging off the same channel as slave. The thing with that one is that if I try to do a lot of i/o on the disk..my machine locks up. Hard. alt-sysReq doesn't even work.

I'll try disconnecting that drive when I get a chance but I welcome any feedback/theories before I get home to do this.

Quoted 15 lines · Andrew Morton

I had the same problem with Seagate ST313021A (13 gig) drives on BP6/HPT366/UDMA66. Initial throughput reported by `hdparm -t' was 22 megs/sec which would slowly wilt to 5 megs/sec.

I discovered that sending _any_ reconfiguration command to the drive - even one which was not supported by that particular drive - would bring the performance back.

So when it goes slow, try running, say, `hdparm -A1' and see what happens.

Andre and I scratched each others heads for a while, suspected a firmware bug. He sent an email to a contact at Seagate. This was in April, so I guess that person is a very slow typist.

Signature · 12 lines
=====================================================================
Mohammad A. Haque                              http://www.haque.net/ 
                                               [email protected]

  "Alcohol and calculus don't mix.             Project Lead
   Don't drink and derive." --Unknown          http://wm.themes.org/
                                               [email protected]
=====================================================================
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
author
#2 lore raw
In reply to Pierre Etchemaite

At the bottom is the IDE/ATA part of my .config. let me knwo if I am missing something. Should I worry about the Multi_Mode configuration?

Quoted 8 lines · Pierre Etchemaite

As a side note, I usually get better results tweaking disks using kernel compiling options ("default to DMA", "autotune chipset", ...) than using hdparm -d1 and friends, give it a try.

--
Linux blade.workgroup 2.4.0-test10 #1 Sat Oct 28 18:00:09 CEST 2000 i686 unknown
1:09pm up 1 day, 19:01, 2 users, load average: 1.04, 1.08, 1.09

# ATA/IDE/MFM/RLL support
#
CONFIG_IDE=y

#
# IDE, ATA and ATAPI Block devices
#
CONFIG_BLK_DEV_IDE=y
# CONFIG_BLK_DEV_HD_IDE is not set
# CONFIG_BLK_DEV_HD is not set
CONFIG_BLK_DEV_IDEDISK=y
# CONFIG_IDEDISK_MULTI_MODE is not set
# CONFIG_BLK_DEV_IDEDISK_VENDOR is not set
# CONFIG_BLK_DEV_IDEDISK_FUJITSU is not set
# CONFIG_BLK_DEV_IDEDISK_IBM is not set
# CONFIG_BLK_DEV_IDEDISK_MAXTOR is not set
# CONFIG_BLK_DEV_IDEDISK_QUANTUM is not set
# CONFIG_BLK_DEV_IDEDISK_SEAGATE is not set
# CONFIG_BLK_DEV_IDEDISK_WD is not set
# CONFIG_BLK_DEV_COMMERIAL is not set
# CONFIG_BLK_DEV_TIVO is not set
# CONFIG_BLK_DEV_IDECS is not set
CONFIG_BLK_DEV_IDECD=y
# CONFIG_BLK_DEV_IDETAPE is not set
# CONFIG_BLK_DEV_IDEFLOPPY is not set
CONFIG_BLK_DEV_IDESCSI=m
# CONFIG_BLK_DEV_CMD640 is not set
# CONFIG_BLK_DEV_CMD640_ENHANCED is not set
# CONFIG_BLK_DEV_ISAPNP is not set
# CONFIG_BLK_DEV_RZ1000 is not set
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_BLK_DEV_OFFBOARD=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_IDEDMA=y
# CONFIG_IDEDMA_PCI_WIP is not set
# CONFIG_IDEDMA_NEW_DRIVE_LISTINGS is not set
# CONFIG_BLK_DEV_AEC62XX is not set
# CONFIG_AEC62XX_TUNING is not set
# CONFIG_BLK_DEV_ALI15X3 is not set
# CONFIG_WDC_ALI15X3 is not set
# CONFIG_BLK_DEV_AMD7409 is not set
# CONFIG_AMD7409_OVERRIDE is not set
# CONFIG_BLK_DEV_CMD64X is not set
# CONFIG_BLK_DEV_CY82C693 is not set
# CONFIG_BLK_DEV_CS5530 is not set
# CONFIG_BLK_DEV_HPT34X is not set
# CONFIG_HPT34X_AUTODMA is not set
CONFIG_BLK_DEV_HPT366=y
CONFIG_BLK_DEV_PIIX=y
CONFIG_PIIX_TUNING=y
# CONFIG_BLK_DEV_NS87415 is not set
# CONFIG_BLK_DEV_OPTI621 is not set
# CONFIG_BLK_DEV_PDC202XX is not set
# CONFIG_PDC202XX_BURST is not set
# CONFIG_BLK_DEV_OSB4 is not set
# CONFIG_BLK_DEV_SIS5513 is not set
# CONFIG_BLK_DEV_SLC90E66 is not set
# CONFIG_BLK_DEV_TRM290 is not set
# CONFIG_BLK_DEV_VIA82CXXX is not set
# CONFIG_IDE_CHIPSETS is not set
CONFIG_IDEDMA_AUTO=y
# CONFIG_IDEDMA_IVB is not set
# CONFIG_DMA_NONPCI is not set
CONFIG_BLK_DEV_IDE_MODES=y

Signature · 12 lines
=====================================================================
Mohammad A. Haque                              http://www.haque.net/ 
                                               [email protected]

  "Alcohol and calculus don't mix.             Project Lead
   Don't drink and derive." --Unknown          http://wm.themes.org/
                                               [email protected]
=====================================================================
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
#3 lore raw

Le 30-Oct-2000, Mohammad Haque écrivait :

Quoted 2 lines

At the bottom is the IDE/ATA part of my .config. let me knwo if I am missing something. Should I worry about the Multi_Mode configuration?

I have it enabled, however I never used any HPT366 controller, so YMMV. The BX chipset is what I tested the most.

Quoted 11 lines

# ATA/IDE/MFM/RLL support
#
CONFIG_IDE=y
CONFIG_BLK_DEV_IDE=y
CONFIG_BLK_DEV_IDEDISK=y
# CONFIG_IDEDISK_MULTI_MODE is not set
CONFIG_BLK_DEV_IDECD=y
CONFIG_BLK_DEV_IDESCSI=m
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_IDEDMA_PCI=y

Quoted 8 lines

CONFIG_BLK_DEV_OFFBOARD=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_BLK_DEV_HPT366=y
CONFIG_BLK_DEV_PIIX=y
CONFIG_PIIX_TUNING=y
CONFIG_IDEDMA_AUTO=y
CONFIG_BLK_DEV_IDE_MODES=y

Looks very similar to my own settings

CONFIG_IDE=y
CONFIG_BLK_DEV_IDE=y
CONFIG_BLK_DEV_IDEDISK=y
CONFIG_IDEDISK_MULTI_MODE=y
CONFIG_BLK_DEV_IDECD=y
CONFIG_BLK_DEV_IDEPCI=y
CONFIG_IDEPCI_SHARE_IRQ=y
CONFIG_BLK_DEV_IDEDMA_PCI=y
CONFIG_IDEDMA_PCI_AUTO=y
CONFIG_BLK_DEV_IDEDMA=y
CONFIG_BLK_DEV_PIIX=y
CONFIG_PIIX_TUNING=y
CONFIG_IDEDMA_AUTO=y
CONFIG_BLK_DEV_IDE_MODES=y

But I only do

hdparm -qu1c1k1 idedevice

at boot time. Also, I've found

append="ide0=autotune ide1=autotune"

in my lilo.conf, left from earlier testing session. Remarks in ide.c say it only has influence on PIO modes, so unless the comment is obsolete I guess I could as well remove that line.

Best regards,
Pierre.

Signature · 7 lines
Linux blade.workgroup 2.4.0-test10 #1 Mon Oct 30 15:30:15 CET 2000 i686 unknown
  7:55pm  up  4:12,  5 users,  load average: 1.11, 1.20, 1.38

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/