SATA pci card performance increase over ata100?

[Folkert Rienstra]

Before you run to any conclusions do a good performance analysis to find
botllenecks in his system.
Check CPU, memory, paging and disks performance when he runs Nuendo.
Performance monitor in Win2000 is a good start. Based on the results you can
judge what to do.

If Nuendo writes separate tracks as separate files, you may have disk random
access write performance issue.

 

[Rod Speed]

Rod Speed wrote

to pick your brain a little deep (if you don't mind?)

Sure, that's what these technical groups are for.

as these things interest me once they are an issue for my needs, why is that?

The sectors per track and the RPM determine how fast the sectors
move under the heads, and thats what determines the speed of the
drive for reads and writes of decent chunks of data.

The 20G drive is very likely a single platter drive and its basic maths
to work out the GB/surface and thats roughly the sectors per track
after you allow for a higher track density with the 250G drive.

they are both ATA 133 at 7200 rpm's, although the 250GB HD
does have 16MB cache compared to 8MB cache for the 20GB,
but the 20 does have a slightly faster documented avg. seek
time of 8ms versus 9ms for the newer 250.

Sure, but that seek time isnt very relevant for your use
of the drive, for the ops that matter speed wise, anyway.

also, I have the spec sheets for the older drive and it also
documents 63 sectors/track, the same as the 250GB drive.

That isnt the physical sectors per track, thats the logical sectors
per track, what the drive pretends to have to keep primitive
systems like the bios happy. The physical sectors per track
varys in bands across the drive with modern drives and its
simpler to give logical data to primitive systems like the bios
that cant handle variable sectors per track across the platter.

Most ops dont use the cylinder heads sectors numbers
anymore, they use the logical block numbers instead.

The physical sectors per track, platters and tracks
per surface should be in the datasheet too.

the only difference is the amount of heads,

Those are fake numbers too. The 20G
drive likely only has 1 platter, so two heads.

but that's understandable, and the cache. even
though its a older, smaller drive, it still seems to have
current IDE drive specs, comparably speaking???

Bet it doesnt with the physical sectors per track.

You should be able to see a real difference with HDTach too.

------------------------
sorry, if it sounds like I'm answering my own questions now, I
apologize for you time as I'm not a Usenet newbie by any stretch
of the imagination so I guess I should have done some simple
homework first, then asked you to fill in some of the uncertainties.
your feedback and time is much appreciated! found this, which
turned out to be a good place to get a better feel for all this.

no, at least I don't think so? thus the questions. whatever
WinXp and HP has set up. I do notice the BIOS still uses
PIO mode references so that is what has me still wondering
in that direction. the BIOS setting list the HD as auto with
a setting of PIO mode4 and Ultra DMA-5.

OK, that's just saying what the best PIO mode the drive
can support if you choose to use it in PIO mode.

out of curiosity, is has a 32 bit transfer rate
disabled, would that be something to tweak.
Nope.

the optical drives are both ATAPI with PIO mode 2 and UDMA-2
and now digging a little deeper - from the article I reference above
- into DMA modes with ATAPI, probably a good reason to keep
them together on a separate IDE channel.
----------------------------
"Hard Disk and ATAPI Device Channel Sharing: There are several
reasons why optical drives (or other ATAPI devices) should not be
shared on the same channel as a fast hard disk. ATAPI allows the
use of the same physical channels as IDE/ATA, but it is not the
same protocol; ATAPI uses a much more complicated command
structure. Opticals are also generally much slower devices than hard
disks, so they can slow a hard disk down when sharing a channel.

That's just the hogging of the channel while the optical drive seeks that
I already mentioned. Not relevant if you arent doing the time critical
stuff on the hard drive simultaneously with USING the optical drive.

Finally, some ATAPI devices cannot deal with DMA bus mastering
drivers, and will cause a problem if you try to enable bus mastering
for a hard disk on a channel they are using."

In practice that isnt a problem in your situation either.

---------------------------------

btw, how do I know if the system is taking full advantage
of drives speed as I see really nothing in system settings to
confirm the BIOS settings as far as DMA options acknowledgment,
like in the device manager for any of the drives, HD or CD

The device manager does show if DMA is being used for each drive.

except for what is probably a separate issue,
'enable write caching' for the HD, which it is?

Thats a separate issue and that should be set like that.

in summery, what you were suggesting is its not worth
the trouble of using the 20 GB drive and just use the
newer 250 GB, 16 MB cache btw, to do everything.

Yes, unless you dont have enough physical ram so
the swap file is being used during your recording ops.

I already have created a partition that I use
just for recording so that should help a little.

Not necessarily. It can make things worse because it can
require a larger head movement between the swap file
and where the data is being written when recording.

but given the extra data, would you tackle it differently now?

I wouldnt bother with a separate partition for the swap
file myself and I'd make sure I had enough physical
ram so the swap file isnt being used when recording.
You're already doing that.

but, upon further review and more data for you to
process, would it be worth the 20 bucks for an
adapter to take full advantage of this smaller -
and equally fast throughput, so it seems - drives

No it isnt.

and independence of a separate channel, away from the ATAPI devises
or slave master setup, as I have been reading and find that an IDE
channel can only access one drive at a time on a single channel?

You have enough physical ram so you wont be
accessing both drives simultaneously while recording.

------------------------------
Master/Slave Channel Sharing: By its very nature, each IDE/ATA channel
can only deal with one request, to one device, at a time. You cannot
even begin a second request, even to a different drive, until the
first request is completed. This means that if you put two devices on
the same channel, they must share it. In practical terms, this means
that any time one device is in use, the other must remain silent. In
contrast, two disks on two different IDE/ATA channels can process
requests simultaneously on most motherboards. The bottom line is that
the best way to configure multiple devices is to make each of them a
single drive on its own channel, if this is possible. (This
restriction is one major disadvantage of IDE compared to SCSI). An
add-in controller like the Promise "Ultra" series is a cheap way of
adding extra IDE/ATA channels to a modern PC.
-----------------------

In practice its better to have enough physical ram so that you arent
using both hard drives simultaneously and if you are too poor to buy
the physical ram to ensure that, just have the two drives on different
motherboard IDE channels because you wont normally be using the
optical drives while recording.

If you do need to use the optical drives while recording, I'd still
spend the money on physical ram rather than a separate controller.

I did just added a GIG of ram to the system
for 1.5, the overall capacity being 2 GB.

Then the swap file shouldnt be being used at all when recording
unless the software you are using for recording is very badly written.

as for the swap file, what would you suggest for a swap file size.

With that much physical ram, its best to leave it to
the OS. Its only going to be used at boot time and
when starting an app by modern OSs like XP.

 

[Odie Ferrous]

Hi, a friend of mine has an Intel D845wn motherboard which has
onboard ata100 drive interface and wants to install a SATA pci card
and drive. He is doing digital audio and hopes to obtain more
simultaneous tracks. Will there be a significant performance increase
over ata100 or will the pci interface be a bottleneck? Has anyone a
similar experience to share? I know he would be better off upgrading
his entire computer but he doesn't have the cash at this time and
wants the cheap fix for now.
TIA
Darrell

The interface is meaningless with todays disks. ATA is as fast as SATA
for the same disk model, unless there is some other bottleneck. I have
a pair of matched Samsung 160GB HDDs, one with ATA and one with SATA
and both are pretty much the same speed: 56MB/sec for ATA and 59MB/sec
for SATA ('hdparm -t'). However SATA goes over the SCSI layer in
Linux and I suspect that is a bit faster than the ATA layer, so both
disks may have exactly the same raw speed. Even if not, that is about
5% difference, i.e. meaningless.

PCI should not be a bottleneck with one (!) drive, unless a lot
of other data is pushed over it.

Arno


Interesting info Arno, a salesman at the local electronics store told
him that he would see vast speed improvements with the card and drive.

I figured he might see a little more speed but not major as stated,

A Seagate Barracuda8 can do 72MB/s, an ATA 100 is limited
to 45MB/s if it is from a mfgr that also has ATA133 drives.
That's a 50% improvement.

In real life, Folkert, I doubt many people with single drives get a
"true" [1]
throughput of much more than about 40MB per second on a new drive.
(Sustained.)

[1] I'm not talking about benchmarking software - I'm talking about
real-life
scenarios - actually, physically copying the data and measuring it.

Thanks for that completely superfluous statement, smelly.
The gain is still substantial.

Now, don't get personal, Folkert - only my wife is allowed to call me
smelly.


Odie