Crappy SATA performance on asus p5q

[Dan Harris]

Hello Guys

Finally upgraded my p3 700 to a quad core 2.66 4gig ram 750G HDD with asus p5q MB Everything seems to work but the performance from the sata HDD is pretty crap. In fact I get the same performance from a ata 133 ide disk connected to a udma66 HDcontroler

dolly:/home/davmul# hdparm -tT /dev/sda

/dev/sda:
Timing cached reads: 4904 MB in 2.00 seconds = 2453.04 MB/sec
Timing buffered disk reads: 262 MB in 3.02 seconds = 86.67 MB/sec


dolly:/home/davmul# hdparm -tT /dev/hde

/dev/hde:
Timing cached reads: 4894 MB in 2.00 seconds = 2447.39 MB/sec
Timing buffered disk reads: 166 MB in 3.04 seconds = 54.63 MB/sec


Nobody tells you that when they are reviewing the board.

At the moment the sata 300 drive is set to udma6 but I cannot change that to udma7. Does anybody have any ideas how to improve performance.

PS debian lenny amd64-2.6.26
Anybody got a similar mb please post your hdparm tests

Thanks Dan

[ivanovic]

Uhm, you do know that "Timing cached reads" means that it is basically testing your ram, don't you? The only relevant part regarding your harddrive is "Timing buffered disk reads". And there you do get a difference between sata and ide. Beside this: those values look completely normal to me. Do *NOT* expect to really get 300MB/s just because marketing says that SATA2 is specified with a speed of up to 300MB/s.

With modern harddrives it is possible to get ~100MB/s read speed. But this always depends on the drive series/manufacturer. Current "mid range" is somewhere around 90MB/s, so your values for the SATA drive looks really fine and okay to me.

[ivanovic]

To provide you with some data from my system:

200GB Maxtor 6B200M0 (SATA1, "ancient")

~ # hdparm -t /dev/sda

/dev/sda:
 Timing buffered disk reads:  184 MB in  3.03 seconds =  60.79 MB/sec

1000GB SAMSUNG HD103UJ (SATA2, really new)

~ # hdparm -t /dev/sdb

/dev/sdb:
 Timing buffered disk reads:  322 MB in  3.01 seconds = 107.13 MB/sec

250GB SAMSUNG SP2514N (PATA, newer than Maxtor SATA1)

~ # hdparm -t /dev/sdc

/dev/sdc:
 Timing buffered disk reads:  196 MB in  3.02 seconds =  64.88 MB/sec

As you can see, just because a drive is SATA does not make it by far faster (cf the SATA1 drive and the PATA drive). And as you can see, too, the 1TB Samsung drive gets up to 107MB/s, basically the maximum what most current hdds are able to achieve. You should not be disappointed about the ~87MB/s you get.

[Dan Harris]

/dev/sdb:
Timing buffered disk reads: 322 MB in 3.01 seconds = 107.13 MB/sec[/CODE]

Yes but that is still 20 MB/Sec faster than my Samsung Spinpoint 750 Gig SATA2?

Dan

[ivanovic]

/dev/sdb:
Timing buffered disk reads: 322 MB in 3.01 seconds = 107.13 MB/sec[/CODE]

Yes but that is still 20 MB/Sec faster than my Samsung Spinpoint 750 Gig SATA2?

Yes, it is, *BUT* your 750GB version is built on a different basis. That is it does have three smaller pattern while mine has three high patter with a higher density. This makes a difference in this case. If you had a benchmark of the 640GB version, you would see it is as about as fast as my 1TB drive since it has 2 of the patter with higher density.

[deanjo]

Yup, when it comes to harddrives, the interface type is not presently a bottleneck for any conventional harddrive. As ivanovic pointed out, other items such as areal density, cache, and read write methods are more of a factor for disk performance then the interface it uses. Even the Velociraptor drives cannot saturate a Sata 1 interface with it's sustained throughput.

[Dan Harris]

Yes, it is, *BUT* your 750GB version is built on a different basis. That is it does have three smaller pattern while mine has three high patter with a higher density. This makes a difference in this case. If you had a benchmark of the 640GB version, you would see it is as about as fast as my 1TB drive since it has 2 of the patter with higher density.

Are you sure about this? This is my model SAMSUNG HD753LJ. If this is the case i am pretty pissed about this and will change it immediately.

Thanks Dan

[ivanovic]

Are you sure about this? This is my model SAMSUNG HD753LJ. If this is the case i am pretty pissed about this and will change it immediately.

Yes, 100% sure. The 750GB model is based on 3 250GB pattern where the 640GB is based on 2 320GB pattern. The size per patter might be a little higher and not deactivated to serve as "backup space" for those areas that tend to get damaged over time, so in fact the patterns might be 340GB or something like this.

What stays is that the 320, 640 and 1000GB drives are built upon the same patters. So they are more dense and the header has to move less while reading (and moving the read head takes a huge potion of time needed for reading/writing).

[Dan Harris]

Its going back then. Whats the best way to dalete all partitions/data on a drive?

[Dan Harris]

Yes, 100% sure. The 750GB model is based on 3 250GB pattern where the 640GB is based on 2 320GB pattern. The size per patter might be a little higher and not deactivated to serve as "backup space" for those areas that tend to get damaged over time, so in fact the patterns might be 340GB or something like this.

What stays is that the 320, 640 and 1000GB drives are built upon the same patters. So they are more dense and the header has to move less while reading (and moving the read head takes a huge potion of time needed for reading/writing).

Thanks for the info, much appreciated. It looks like I have made the wrong choice.

Dan