From: gg on
I am using pure RAID 5 on barracuda drives but using NVIDIA technology. for
up to 16MB the write is fast. read is fairly fast. copy on the same raid is
slower but across different volume be IDE and Sata are fast

when using matrix 5 with mixed Raid on the same set of spindles, I can see
performance degradation when accessed simultaneously.

"John E" <jce(a)myhouse.invalid> wrote in message
news:Z%8Hj.27221$%N1.4741(a)newsfe3-gui.ntli.net...
> "Mr. Bro" <bbro(a)en.com> wrote in message
> news:mdmpu39o9qji92f8qf3g6dporsfqqkubp6(a)4ax.com...
> > Have 2 SATA drives. NOT planning to set up in a RAID configuration so,
> > In the BIOS should they be set up as IDE or AHCI. The drives will be
> > partitioned. I'm new at this type of HD so a bit confused. According
> > to the manual to gain anything, drives should be set to AHCI. The
> > first partition of the first drive will be the OS boot drive. Any help
> > appreciated
> >
> > Mr. Bro
>
>
> I have set up my pc (using Vista x64 Sp 1) to run my SATA drives as either
> AHCI using Intel matrix manager or as 'Sata configured as IDE'. (It was
> necessary to install the Intel matrix manager drivers when installing the
> OS, but there is a work-around, if you have already set up the OS).
>
> I found that my SATA drives (Samsung Spinpoint and Seagate Barracuda) are
> not able to use the 'write caching' and 'enhanced performance' settings
> under windows (device manager, drive settings). Consequently disk reads
and
> writes are slower with AHCI than they are with IDE. (Significantly
slower -
> as in a 50% improvement with IDE rather than AHCI).
>
> There are advantages in using AHCI. The drive is then hot-swappable, and
you
> get NCQ which should speed up random access - but my tests showed that
even
> random access was faster with the drives configured as IDE.
>
> So if you are planning to use Windows as your OS, you may be better off
with
> Sata configured as IDE, rather than AHCI.
>
> John
>


From: RobV on
Mr. Bro wrote:
> Have 2 SATA drives. NOT planning to set up in a RAID configuration so,
> In the BIOS should they be set up as IDE or AHCI. The drives will be
> partitioned. I'm new at this type of HD so a bit confused. According
> to the manual to gain anything, drives should be set to AHCI. The
> first partition of the first drive will be the OS boot drive. Any help
> appreciated
>
> Mr. Bro

Interesting. I have had the opposite experience from John E, for the
most part.

I have 2 IDE drives and a SATA drive (OS is on SATA) and the SATA set as
IDE or AHCI is pretty much equal as well as being nearly twice as fast
as the IDE drives. There is a slight increase in overall read speed
(very slight) when set as IDE, but when set as AHCI, random reads is
somewhat faster, while average read speed is nearly identical, using HD
Tach.

However, in practical use, with the software I use, AHCI is much faster,
as the data is not laid down sequentially and the NCQ really speeds up
writing and reading of data (this is with music creation software
[Cakewalk Sonar 6]).

It really depends on what you use the system for.


From: DevilsPGD on
In message <Z%8Hj.27221$%N1.4741(a)newsfe3-gui.ntli.net> "John E"
<jce(a)myhouse.invalid> wrote:

>I found that my SATA drives (Samsung Spinpoint and Seagate Barracuda) are
>not able to use the 'write caching' and 'enhanced performance' settings
>under windows (device manager, drive settings). Consequently disk reads and
>writes are slower with AHCI than they are with IDE. (Significantly slower -
>as in a 50% improvement with IDE rather than AHCI).

Write caching is potentially still supported, but under the control of
the Intel Matrix Manager rather then Device Manager.

You can test this by performing a file copy using a technique which
permits caching (Explorer's "copy" feature in Vista does not, in SP1 it
should -- In either case, a command prompt copy should do the trick) and
watch when the application believes the copy finished vs when the drive
lights indicate activity has finished.

All that being said, at least with the Seagate (And possibly the
Samsung), did you remove the jumper? If not, the drive is significantly
hobbled and you won't come near it's true performance.
From: John E on
"DevilsPGD" <spam_narf_spam(a)crazyhat.net> wrote in message
news:6n3uu39opqtveojtideofh6i449bu27d70(a)4ax.com...

> Write caching is potentially still supported, but under the control of
> the Intel Matrix Manager rather then Device Manager.
>
> You can test this by performing a file copy using a technique which
> permits caching (Explorer's "copy" feature in Vista does not, in SP1 it
> should -- In either case, a command prompt copy should do the trick) and
> watch when the application believes the copy finished vs when the drive
> lights indicate activity has finished.
>
> All that being said, at least with the Seagate (And possibly the
> Samsung), did you remove the jumper? If not, the drive is significantly
> hobbled and you won't come near it's true performance.


Thanks for the comments. I have no jumpers on the drives (they were bought
OEM and came without jumpers). I tried your suggestion of copying a big file
(about 4GB) from one drive to another using the command prompt and the
transfer speed worked out at
57MB/s - using AHCI. With IDE the same transfer was almost identical:
58.86MB/s The drive light stopped flashing about 20 seconds or so after the
transfers - similar in each. I compared the transfer speed with that from
the benchmark program I'm using (Passmark's performance test 64bit
version - HD Tach doesn't run) and
it gave the following results (best results shown):

Seq. read: IDE: 76.8 AHCI: 66.1
Seq. write: IDE: 79.0 AHCI: 58.1
Random R/W: IDE: 3.04 AHCI: 3.37

(All in Megabytes per second)

The AHCI values were quite a bit better this time than when I previously ran
the tests,
especially the Random R/W which was slower than for IDE last time.

Intel matrix manager reports that the drives are using NCQ, but I did read
on the net that the performance benefits from that are associated with long
command queues, which are unlikely to be achieved on a desktop machine.

So, overall, I have to say that I am still not convinced that AHCI offers
any significant improvement - although that small random access improvement
should give a small benefit in day to day use. My major disk activity is
associated with video editing, which involves moving and copying large
files - and it would appear that I'm a bit better off (at present) using
IDE.

John

From: SumGuy on

"John E" <jce(a)myhouse.invalid> wrote in message
news:Z%8Hj.27221$%N1.4741(a)newsfe3-gui.ntli.net...
> "Mr. Bro" <bbro(a)en.com> wrote in message
> news:mdmpu39o9qji92f8qf3g6dporsfqqkubp6(a)4ax.com...
>> Have 2 SATA drives. NOT planning to set up in a RAID configuration so,
>> In the BIOS should they be set up as IDE or AHCI. The drives will be
>> partitioned. I'm new at this type of HD so a bit confused. According
>> to the manual to gain anything, drives should be set to AHCI. The
>> first partition of the first drive will be the OS boot drive. Any help
>> appreciated
>>
>> Mr. Bro
>
>
> I have set up my pc (using Vista x64 Sp 1) to run my SATA drives as either
> AHCI using Intel matrix manager or as 'Sata configured as IDE'. (It was
> necessary to install the Intel matrix manager drivers when installing the
> OS, but there is a work-around, if you have already set up the OS).
>
> I found that my SATA drives (Samsung Spinpoint and Seagate Barracuda) are
> not able to use the 'write caching' and 'enhanced performance' settings
> under windows (device manager, drive settings). Consequently disk reads
> and writes are slower with AHCI than they are with IDE. (Significantly
> slower - as in a 50% improvement with IDE rather than AHCI).
>
> There are advantages in using AHCI. The drive is then hot-swappable, and
> you get NCQ which should speed up random access - but my tests showed that
> even random access was faster with the drives configured as IDE.
>
> So if you are planning to use Windows as your OS, you may be better off
> with Sata configured as IDE, rather than AHCI.
>
> John

NCQ is only a benefit if you are running very intensive disk operations. It
is of no use to your average home user.