View previous topic :: View next topic |
Author |
Message |
lintoo n00b
Joined: 04 Mar 2004 Posts: 12
|
Posted: Fri Mar 05, 2004 8:23 pm Post subject: |
|
|
I'm having troubles with my new Gentoo setup. I have a Samsung 5400rpm drive. I know for a fact it is ATA100 with DMA ability. In the BIOS I have it set for PIO mode 4 and DMA mode 5.
When I do something like:
Code: | hdparm -d1 /dev/hda |
It tells me the old "Not available". (And yes, of course I'm doing this from root )
My motherboard is an MSI KT3Ultra2. It has the VIA8235 chipset. I did a kernel recompile last night with the VIA8xxxx options, but I still cannot set DMA. I think this would be a major detriment to my system performance? It is not terribly slow at the moment, but speedups are always welcomed.
Any help would be appreciated. |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Fri Mar 05, 2004 9:30 pm Post subject: |
|
|
Run:
Code: |
hdparm -tT /dev/hda
|
If the Buffered Disk Reads are <20MB/sec, then you probably have room for improvement, otherwise, you might not.
What's the result from:
|
|
Back to top |
|
|
lintoo n00b
Joined: 04 Mar 2004 Posts: 12
|
Posted: Fri Mar 05, 2004 10:02 pm Post subject: |
|
|
Code: | hdparm -I /dev/hda
/dev/hda:
ATA device, with non-removable media
Model Number: SAMSUNG SV4002H
Serial Number: 0379J1FRC54530
Firmware Revision: QP100-07
Standards:
Used: ATA/ATAPI-6 T13 1410D revision 1
Supported: 6 5 4 3
Configuration:
Logical max current
cylinders 16383 4047
heads 16 16
sectors/track 63 255
--
CHS current addressable sectors: 16511760
LBA user addressable sectors: 78242976
device size with M = 1024*1024: 38204 MBytes
device size with M = 1000*1000: 40060 MBytes (40 GB)
Capabilities:
LBA, IORDY(cannot be disabled)
bytes avail on r/w long: 4 Queue depth: 1
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Recommended acoustic management value: 128, current value: 0
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
|
Code: | hdparm /dev/hda
/dev/hda:
multcount = 16 (on)
IO_support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 8 (on)
geometry = 4870/255/63, sectors = 78242976, start = 0
|
Code: | hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 856 MB in 2.01 seconds = 426.51 MB/sec
Timing buffered disk reads: 18 MB in 3.07 seconds = 5.87 MB/sec
|
This is extremely poor. |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Fri Mar 05, 2004 10:17 pm Post subject: |
|
|
After some poking around, I think I found some very interesting differences in HD performances:
Seagate ST94811A, laptop ATA100 40GB drive, 5400rpm, 8MB cache:
- Interface Transfer rate: 100MB/sec
- Internal transfer rate: 386Mb/sec
- Avg. sustained transfer rate: > 34.5MB/sec
- Theoretical Max Sustained transfer rate: 48MB/sec
Seagate ST340014A, desktop ATA100 40GB drive, 7200rpm, 2MB cache:
- Interface Transfer rate: 100MB/sec
- Internal transfer rate: 683Mb/sec
- Avg. sustained transfer rate: > 58MB/sec
- Theoretical Max Sustainted transfer rate: 85MB/sec
Seagate ST380013AS, desktop SATA150 80Gb drive, 7200rpm, 8MB cache
- Interface Transfer rate: 150MB/sec
- Internal trasfer rate: 683Mb/sec
- Avg. sustained transfer rate: > 58MB/sec
- Theoretical Max Sustained trasnfer rate: 85MB/sec
Seagate ST336753LW, desktop/server SCSI-320 80GB, 15Krpm, 8MB cache
- Interface Transfer rate: 320MB/sec
- Internal transfer rate: 609-891Mb/sec
- Sustained transfer rate: 57-86MB/sec
- Theoretical Sustained Transfer rate: 76-111MB/sec
Which, I think, basically proves my thoughts that the transfer rates we're seeing in hdparm are going to max-out at the avg sustained transfer rate of the disk, which is the theoretical transfer rate, minus some overhead. From my math, it seems there's about a 25-30% overhead, so the realistic max's are about 70-75% of the internal transfer rates for the drives.
So for my laptop, with it's 240-333Mb/s internal transfer rate, I shouldn't expect more than 22-29MB/sec of measure transfer rates. And it falls right into that at 22-26, depending on lots of stuff.
Of course, this only is for sustained reads. Any kind of burst traffic is likely to be much faster, but if the read-ahead buffer can't cope with it, it'll fall back to seeks and reads.
Which again seems to imply that you want rpms for low seek times, and high internal transfer rates to get it over the wire, and your interface to the system is much, much less important than your HD's internal capabilities are. |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Fri Mar 05, 2004 10:21 pm Post subject: |
|
|
lintoo wrote: | Code: | hdparm -I /dev/hda
Capabilities:
LBA, IORDY(cannot be disabled)
bytes avail on r/w long: 4 Queue depth: 1
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Recommended acoustic management value: 128, current value: 0
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
|
Code: | hdparm /dev/hda
/dev/hda:
multcount = 16 (on)
IO_support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 8 (on)
geometry = 4870/255/63, sectors = 78242976, start = 0
|
Code: | hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 856 MB in 2.01 seconds = 426.51 MB/sec
Timing buffered disk reads: 18 MB in 3.07 seconds = 5.87 MB/sec
|
This is extremely poor. |
yah, you betcha, it's poor.
Since DMA is disabled, and you can't seem to enable it, it's probably a chipset issue. Latest kernel sources and sure that you've included the right chipset support? Some of the chipsets were, IIRC, blacklisted for bad behavior, or perhaps it was drives... So that might be an issue
Next, make sure that you're actually using your newly built kernel (mounted boot before you copied it over?).
And also go through the output to dmesg. I'm sure there's something in there that's explaining why dma is being disabled. |
|
Back to top |
|
|
Strukt n00b
Joined: 11 Oct 2002 Posts: 56
|
Posted: Fri Mar 05, 2004 11:01 pm Post subject: |
|
|
Ok people, lookie here!
dmesg, looks ok huh?
Code: | VP_IDE: IDE controller at PCI slot 0000:00:07.1
VP_IDE: chipset revision 6
VP_IDE: not 100% native mode: will probe irqs later
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
VP_IDE: VIA vt82c686b (rev 40) IDE UDMA100 controller on pci0000:00:07.1
ide0: BM-DMA at 0xd000-0xd007, BIOS settings: hda:DMA, hdb:DMA
ide1: BM-DMA at 0xd008-0xd00f, BIOS settings: hdc:pio, hdd:pio
hda: WDC WD800BB-75CAA0, ATA DISK drive
hdb: WDC WD800BB-00CAA1, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hda: max request size: 128KiB
hda: Host Protected Area detected.
current capacity is 156250000 sectors (80000 MB)
native capacity is 156250080 sectors (80000 MB)
hda: 156250000 sectors (80000 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA(100)
/dev/ide/host0/bus0/target0/lun0: p1 p2 p3
hdb: max request size: 128KiB
hdb: 156301488 sectors (80026 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA(100)
/dev/ide/host0/bus0/target1/lun0: p1 |
Yep, two almost the same drives. One WD80GB as a master, and another WD80GB as slave.
Look beneath, drives look okay, right?
Code: | h0mer / # hdparm -i /dev/hda
/dev/hda:
Model=WDC WD800BB-75CAA0, FwRev=16.06V16, SerialNo=WD-WMA8E1317211
Config={ HardSect NotMFM HdSw>15uSec SpinMotCtl Fixed DTR>5Mbs FmtGapReq }
RawCHS=16383/16/63, TrkSize=57600, SectSize=600, ECCbytes=40
BuffType=DualPortCache, BuffSize=2048kB, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=156250000
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5
AdvancedPM=no WriteCache=enabled
Drive conforms to: device does not report version:
* signifies the current active mode |
And the other drive...
Code: | h0mer / # hdparm -i /dev/hdb
/dev/hdb:
Model=WDC WD800BB-00CAA1, FwRev=17.07W17, SerialNo=WD-WMA8E2822537
Config={ HardSect NotMFM HdSw>15uSec SpinMotCtl Fixed DTR>5Mbs FmtGapReq }
RawCHS=16383/16/63, TrkSize=57600, SectSize=600, ECCbytes=40
BuffType=DualPortCache, BuffSize=2048kB, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=156301488
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5
AdvancedPM=no WriteCache=enabled
Drive conforms to: device does not report version:
* signifies the current active mode |
Then we run some tests on /dev/hda
Code: | h0mer / # hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 380 MB in 2.01 seconds = 189.08 MB/sec
Timing buffered disk reads: 6 MB in 3.33 seconds = 1.80 MB/sec |
And the same test on /dev/hdb
Code: | h0mer / # hdparm -tT /dev/hdb
/dev/hdb:
Timing buffer-cache reads: 484 MB in 2.01 seconds = 240.24 MB/sec
Timing buffered disk reads: 94 MB in 3.06 seconds = 30.74 MB/sec |
The performance differs dramaticly, but i do not understand why because the drives are almost identical, they are setup the same in bios. HDA is primary master, and HDB is primary slave, on the same cable.
I do not know how to solve this problem, ofcourse i've tried disabling dma, enabling dma and so forth. No good there. I also have the correct chipset compiled directly into the kernel. (2.6.3). I recently upgraded to kernel 2.3.6 from 2.6.0 in an attempt to get rid of this problem but it didnt help.
Help me |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Sat Mar 06, 2004 12:50 am Post subject: |
|
|
Try making the two HDs both masters (hda and hdc, and the other two drives as slaves (assuming these are opticals?)
Does that get you anything? Seems like it's not a cabling issue, unless the master drive doesn't like the cabling or the master-slave-cs option.
Another thing to try is to swap the drives, and see how that changes things.
And yes, you'll be mucking with Grub and fstab to do all the drive swapping.
Normally it's considered good practice to put to the two fast drives on separate channels so that they don't compete with each other for bandwidth on it.
If you have a liveCD, can you boot from that? And if so, different or the same results? It also might help to do the swapping around when using the livecd, so you don't need to much with grub and fstab to keep everything sane. |
|
Back to top |
|
|
Strukt n00b
Joined: 11 Oct 2002 Posts: 56
|
Posted: Sun Mar 07, 2004 7:44 pm Post subject: |
|
|
hi woody, or anyone else for that matter .
Ok, first thing first. My drive i had problems with works fine now. What i did was to take that particular drive, put it in another gentoo based computer, i ran the hdparm benchmark again, and i got the same lousy results. At that part i started to believe that the drives controllerboard was fried or something. Anyway.
I took the drive back into the original computer again, blowed away all partitions, made new ones. Made an ext3 filesystem, and after that ran some benchmarks again. And lookie... the speed is up to around 30MB / sec again.
I dunno what the problem was....but apparently a reformatted harddrive solved my particular problem... i guess the partition table was farked up somehow, we have had some powercuts here....could be related to that i presume.
Anyway, it works fine now! |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Mon Mar 08, 2004 6:01 am Post subject: |
|
|
While I have no idea how hdparm reads the data (which access mechanism), I'm surprised that it was changed by the reformat. Not saying it isn't true, just surprised.
Glad you've got it working. |
|
Back to top |
|
|
Stieltje Apprentice
Joined: 03 Apr 2003 Posts: 158 Location: The land where the sky is black
|
Posted: Mon Mar 08, 2004 8:18 am Post subject: |
|
|
Hi all.
I am building myself a new computer, and of course I want a nice fast disc to go with it..
But reading this thread, it does not seem worth the extra $ to get SATA?
with my 80 GB Seagate 7.200 I get:
Code: |
PanzerFaust root # hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 1352 MB in 2.00 seconds = 674.56 MB/sec
Timing buffered disk reads: 164 MB in 3.04 seconds = 54.02 MB/sec
|
how much faster would a SATA seagate 80 GB be?
or any other SATA disk for that matter.
Thanks. |
|
Back to top |
|
|
der_maddin Apprentice
Joined: 22 Feb 2004 Posts: 209 Location: somewhere between my systems ..
|
Posted: Mon Mar 08, 2004 12:35 pm Post subject: |
|
|
don't waste your $$ for the SATA - drive, it'll give
you nothing more. Your ATA drives performance is
just great and you would not get any better values
using a SATA drive.
The most important feature of having a SATA - bus
system would be that you could have a bigger number
of disks in your system than with using simple ATA.
And the performance gain is simpy too small to pay extra money for. _________________ neoCortex maddin # uname -a
Linux neoCortex 2.6.14-gentoo-r5 #2 SMP Fri Jan 13 04:40:02 CET 2006 x86_64 AMD Athlon(tm) 64 Processor 3400+ AuthenticAMD GNU/Linux |
|
Back to top |
|
|
ktech Guru
Joined: 12 Apr 2003 Posts: 340
|
Posted: Mon Mar 08, 2004 5:10 pm Post subject: |
|
|
der_maddin, I have a nforce-2 motherboard from abit (nfs-7 v-2.0) and I have a P-ATA hard disc (ST380023A) Barracuda-V with 80 Gb.
I'm running 2.6 kernel and I only have:
Timing buffered disk reads: 98 MB in 3.00 seconds = 32.62 MB/sec
This is very poor, not to? I was thinking in going to SATA, because a person I know is getting 66 mb/sec out of a sata drive. It's double speed!!
What tip can you give me about this? |
|
Back to top |
|
|
der_maddin Apprentice
Joined: 22 Feb 2004 Posts: 209 Location: somewhere between my systems ..
|
Posted: Mon Mar 08, 2004 7:12 pm Post subject: |
|
|
i think you should be able to get at least
about 40 ~ 50 Mb/s out of your current P-ATA disk,
depending a bit on how big the buffer of your device
is and how fast it spins. ( and depending on your
m/b's chipset, if you've select kernel support for your one etc .. )
EDIT : from what i've looked up
http://www.seagate.com/support/disc/specs/ata/st380023a.html
it seems your disk has a speed of 7200rpm ( a fast one )
and 2M cache.
try this out ( being root )
Code: |
hdparm -X69 -u1 -c3 -d1 -A1 -m16 -a256 -k1
|
look at the output of this command and post any errors here.
after entering the command do this :
and report the values shown back here
( using the settings above, i get about 55 MB/sec for my samsung
sp1614n (160gb, 7200rpm, 8MB) thats not far away from
those values you know from this guy with the sata drive .. )
EDIT : but btw the sp1614n is a great deal, it is available around here
for about 100 euros ... that's not too much for a drive that big =] _________________ neoCortex maddin # uname -a
Linux neoCortex 2.6.14-gentoo-r5 #2 SMP Fri Jan 13 04:40:02 CET 2006 x86_64 AMD Athlon(tm) 64 Processor 3400+ AuthenticAMD GNU/Linux |
|
Back to top |
|
|
ktech Guru
Joined: 12 Apr 2003 Posts: 340
|
Posted: Mon Mar 08, 2004 7:29 pm Post subject: |
|
|
No improvements. It seems that this controller n-force IDE has a very bad driver because another user as commented me that the same PATA hd connected to S-ATA throught a pata-2-sata converter runs faster than in PATA.
If you don't know any other tip or trick, this night I'll be testing if that SATA driver is better and report.
bash-2.05b# hdparm -X69 -u1 -c3 -d1 -A1 -m16 -a256 -k1 /dev/hda
/dev/hda:
setting fs readahead to 256
setting 32-bit IO_support flag to 3
setting multcount to 16
setting unmaskirq to 1 (on)
setting using_dma to 1 (on)
setting keep_settings to 1 (on)
setting xfermode to 69 (UltraDMA mode5)
setting drive read-lookahead to 1 (on)
multcount = 16 (on)
IO_support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 1 (on)
keepsettings = 1 (on)
readahead = 256 (on)
bash-2.05b# hdparm -t /dev/hda
/dev/hda:
Timing buffered disk reads: 102 MB in 3.00 seconds = 33.96 MB/sec |
|
Back to top |
|
|
der_maddin Apprentice
Joined: 22 Feb 2004 Posts: 209 Location: somewhere between my systems ..
|
Posted: Mon Mar 08, 2004 7:35 pm Post subject: |
|
|
sorry, i don't know anything else to improve your performance
but i keep wondering why you can't hit at least the 40Mb/s
barrier with your drive and controller ... _________________ neoCortex maddin # uname -a
Linux neoCortex 2.6.14-gentoo-r5 #2 SMP Fri Jan 13 04:40:02 CET 2006 x86_64 AMD Athlon(tm) 64 Processor 3400+ AuthenticAMD GNU/Linux |
|
Back to top |
|
|
ktech Guru
Joined: 12 Apr 2003 Posts: 340
|
Posted: Mon Mar 08, 2004 7:46 pm Post subject: |
|
|
yes, the same here.
I think it's a very poor performance even when I heard that some seagate drives have some issues that must be corrected at driver level in order for the drive to handle the data in a correct way.
If any of you have some ideas, I'm going to go for SATA in 2 - 3 hours.
Thanks to all of you |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Tue Mar 09, 2004 12:24 am Post subject: |
|
|
seagate seems rather vague on the performance specs for this drive...
INTERNAL TRANSFER RATE (Mbits/sec) _______up to 570
SUSTAINED TRANSFER RATE (MB/sec)__________27 to 44
EXTERNAL TRANSFER RATE (Mbytes/sec) ______up to 100
So perhaps you're getting the best you're going to get from it. Do an hdparm -I on each drive (yours and your friends that's faster) and see if the firmware is different, or anything else. Could be something like that. |
|
Back to top |
|
|
ktech Guru
Joined: 12 Apr 2003 Posts: 340
|
Posted: Tue Mar 09, 2004 8:44 am Post subject: |
|
|
I cannot understand nothing.
I'm actually on SATA and the same happens. 33 mb/sec.
I'm starting to think that this drive is a little piece of shit. |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Tue Mar 09, 2004 8:25 pm Post subject: |
|
|
That's probably 33MB/sec, not 33mb/sec. And it's probably as fast as the drive itself can read off the platters. And there's no way to make that faster. It's within spec per seagate (see above comments). |
|
Back to top |
|
|
Stieltje Apprentice
Joined: 03 Apr 2003 Posts: 158 Location: The land where the sky is black
|
Posted: Wed Mar 10, 2004 7:47 am Post subject: |
|
|
Maybe I am just lucky with my Seagate then ?
anyway, I got a ST380011A drive attached to a epox8rda+ mobo (nforce2), 2.4.23-ck1 -sources
http://www.seagate.com/support/disc/specs/ata/st380011a.html
PanzerFaust root # hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 1324 MB in 2.00 seconds = 661.53 MB/sec
Timing buffered disk reads: 154 MB in 3.02 seconds = 51.04 MB/sec
/etc/conf.d/hdparm:
Code: |
# You can either set hdparm arguments for each drive using disc*_args and cdrom$
# eg.
# disc0_args="-d1 -X66"
# disc1_args"-d1"
# cdrom0_args="-d1"
# Or, you can set hdparm options for ALL drives using all_args..
# eg.
# this mimics the behavior of the current script
all_args="-d1"
|
hdparm /dev/hda :
Code: |
/dev/hda:
multcount = 16 (on)
IO_support = 1 (32-bit)
unmaskirq = 1 (on)
using_dma = 1 (on)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 8 (on)
geometry = 9729/255/63, sectors = 156301488, start = 0
|
hdparm -i /dev/hda
Code: |
/dev/hda:
Model=ST380011A, FwRev=3.04, SerialNo=3JV0NWSH
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=unknown, BuffSize=2048kB, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=156301488
IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5
AdvancedPM=no WriteCache=enabled
Drive conforms to: ATA/ATAPI-6 T13 1410D revision 2:
* signifies the current active mode
|
ok, maybe this is a lengthy and unecessary post, but just some figures and such to compare to your own (seagate) disc.
cheers |
|
Back to top |
|
|
woody77 n00b
Joined: 06 Feb 2004 Posts: 69 Location: SC Mtns, CA, USA
|
Posted: Thu Mar 11, 2004 12:48 am Post subject: |
|
|
And in comparing specs on the two drives, the 7200.7 is faster than the ATA V is.
ATA V:
Code: |
INTERNAL TRANSFER RATE (Mbits/sec) _______up to 570
SUSTAINED TRANSFER RATE (MB/sec)__________27 to 44
EXTERNAL TRANSFER RATE (Mbytes/sec) ______up to 100
|
7200.7
Code: |
INTERNAL TRANSFER RATE (Mbytes/sec) ______up to 85.4
SUSTAINED TRANSFER RATE (MB/sec)__________up to 58
EXTERNAL TRANSFER RATE (Mbytes/sec) ______up to 100
|
Looks like it's faster to the tune of 32%. And this seems to be held up by the numbers we're seeing here. 30% faster than 40MB/sec is 52MB/sec. So it seems like you should be able to get up to at least 40 or so... But probably not beyond that. I wonder what the bottle-neck on your system is. |
|
Back to top |
|
|
dmouritsendk Tux's lil' helper
Joined: 22 Jun 2002 Posts: 138 Location: Denmark
|
Posted: Mon May 31, 2004 5:24 pm Post subject: |
|
|
Stieltje wrote: | Hi all.
I am building myself a new computer, and of course I want a nice fast disc to go with it..
But reading this thread, it does not seem worth the extra $ to get SATA?
with my 80 GB Seagate 7.200 I get:
Code: |
PanzerFaust root # hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 1352 MB in 2.00 seconds = 674.56 MB/sec
Timing buffered disk reads: 164 MB in 3.04 seconds = 54.02 MB/sec
|
how much faster would a SATA seagate 80 GB be?
or any other SATA disk for that matter.
Thanks. |
SATA have more bandwidth available per channel(and only one device per channel), allowing for faster disks. But ATM it seem that most SATA disks are exact replicas of IDE disk just with a evolved interface. The faster disk will come though, WD have already introduced a 10K-RPM RAPTOR line that reads about 70MB/s(which is faster than any ATA100/133 disk available, that I know of anyways ).
Also, one of the "big deals" about SATA is that it makes better use of the PCI bus with its serial nature. Allowing for 150Mb/sec transfers per channel/port, which makes it a good choice for RAID solutions with a lot of fast (+60MB/sec) disk is connected.
But is SATA "worth it"? Dunno really, depends on what you'll use it for. If you just need a new disk with good performance, there isn't much difference between the average SATA disk and the average ATA disk. Though, if you want to stripe more than two disks delivering ~60MB/Sec the SATA solution would probably be the best(non-scsi) way to go.
Me personally, went with a Maxtor SATA drive when i made my current system, mostly because the motherboard i had bought for it had these 4 SATA port with nothing connected to them. Since there was already a burner attached to one of the ATA133 channels, I felt that the SATA ports needed a device the most(also, SATA cables are cool lookin')
Since I only got one hard-drive in the box, I most likely wouldn't be able to notice any differences in speed as opposed to a ATA133 model of the drive. Which this head-2-head between two "identical" Maxtor drives with different interfaces clearly shows, the ATA133 is even a bit faster is some of the benchmarks. But the difference is tiny. If you try and throw a WD Raptor in to the Head-2-head, you'll get a hint of what the "future" is like for SATA. Then throw in a couple of SCSI320 models, and you got a pretty good overview what the different interfaces offer/are-good-at ATM. Its also clear from this that transfer speeds isn't everything, the 10K scsi drive transfers less data than the 10K SATA, but is a faster in 9/10 of the server benchmarks.
Of-course more disks like the Raptor will come be introduced, which utilizes more of the SATA bandwidth. Making SATA show its advantages over ATA133 more clearly in disk-2-disk comparisons.
Try looking around storagereview.com(its a good resource), compare some drives and see the results and figure out what drive(SATA, ATA or SCSI) fits ur need the best. |
|
Back to top |
|
|
grzewho l33t
Joined: 31 Dec 2002 Posts: 626 Location: /home/g
|
Posted: Mon May 31, 2004 10:20 pm Post subject: |
|
|
der_maddin wrote: |
Code: |
hdparm -X69 -u1 -c3 -d1 -A1 -m16 -a256 -k1
|
|
what`s the purpose of the -a256 and -A1 switches ? _________________
Code: | USE="freedom -software_patents" emerge --deep --update world |
|
|
Back to top |
|
|
error` n00b
Joined: 20 Jun 2003 Posts: 2
|
Posted: Tue Jun 01, 2004 12:09 am Post subject: |
|
|
well, first my hdparm results:
Code: |
hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads: 916 MB in 2.00 seconds = 457.38 MB/sec
Timing buffered disk reads: 112 MB in 3.01 seconds = 37.15 MB/sec
hdparm -tT /dev/hdb
/dev/hdb:
Timing buffer-cache reads: 976 MB in 2.00 seconds = 487.10 MB/sec
Timing buffered disk reads: 166 MB in 3.03 seconds = 54.85 MB/sec
hdparm -tT /dev/hdc
/dev/hdc:
Timing buffer-cache reads: 936 MB in 2.00 seconds = 466.90 MB/sec
Timing buffered disk reads: 84 MB in 3.04 seconds = 27.62 MB/sec
hdparm -tT /dev/hdd
/dev/hdd:
Timing buffer-cache reads: 884 MB in 2.00 seconds = 441.18 MB/sec
Timing buffered disk reads: 158 MB in 3.02 seconds = 52.24 MB/sec |
but i got a strange problem, when im moving a file from hdb to any other drive i only get around 2-4mB/s. when im moving from hdd/hda to hdb i get around 30-45mB/s and when i moving a file from hdd to hda or otherway around i also get the full speed. so any ideas to fix the speed problem on hdb?
btw.: hdb is a samsung spinpoint 1614n (8mb cache), anyother hdd has 2mb cache and 7200rpm
best regards,
error` |
|
Back to top |
|
|
ikaro Advocate
Joined: 14 Jul 2003 Posts: 2527 Location: Denmark
|
Posted: Tue Jun 01, 2004 12:13 am Post subject: |
|
|
grzewho wrote: | der_maddin wrote: |
Code: |
hdparm -X69 -u1 -c3 -d1 -A1 -m16 -a256 -k1
|
|
what`s the purpose of the -a256 and -A1 switches ? |
-a get/set fs readahead
-A set drive read-lookahead flag (0/1)
so with -A1 you enable the feature and then use -a256 to set it.
you can use other values as well, depending of how big you want it ( i mean the -aXXXX)
my results now:
Code: |
/dev/hda:
Timing buffer-cache reads: 1100 MB in 2.00 seconds = 549.26 MB/sec
Timing buffered disk reads: 140 MB in 3.04 seconds = 46.07 MB/sec
|
i think i posted on this thread before with some results.
I use this in /etc/conf.d/hdparm
hda_args="-d1 -u1 -m16 -c1 -W1 -A1 -k1 -X70 -a64" _________________ linux: #232767 |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|