A7N8X Deluxe...Asus says it DOES support 150Mhz with it's onboard serial ATA..are they lying?

P

Paul

John Q. Public said:
A7N8X Deluxe...Asus says it DOES support 150Mhz with it's onboard
serial ATA. I've heard that the serial ATA on the A7N8X Deluxe runs
thru the PCI, and is therefore limited to 133....but Asus says no to
that...?

http://www.asus.com/support/faq/qanda.aspx?KB_ID=82959

Sometimes I don't know why Asus bothers with the FAQ page...

SATA performance has the following limits:

1) Media head rate - Raptor is 70MB/sec and is the fastest I know of.
2) Drive can have a RAM cache. Depending on where this sits (before
or after bridge), it can absorb the full data rate for short
intervals.
3) Drive uses IDE/SATA bridge chip - this is limited to 100MB/sec
or 133MB/sec. I don't know if there are any native hard drives yet
or not.
4) Cable rate is 150MB/sec and this is fixed. The cable actually
runs at 1500Mbits/sec, and is 8B10B coded. The coding reduces the
usable data rate to 1500*(8/10) = 1200Mbits/sec and this is the
stated 150MB/sec. I haven't read the whole spec, but it is
possible fewer symbols per second could be sent if flag
symbols (like JK or K28.5 are sent) are used to take the place
of data.
5) The SIL3112 has FIFO buffers inside the chip, which makes it
possible for the chip to absorb data at full speed (150MB/sec)
for very short intervals, from the SATA cable. I don't know how
flow control is achieved or whether flow control results from
limiting the size of data chunks used for transfers.
6) The PCI bus has a theoretical burst rate of 133MB/sec for the
32bit 33MHz version of PCI. When using reasonably short data
bursts on the bus, about 100MB/sec will be achieved. But the
SIL3112 datasheet says the chip can be clocked at 33MHz or
66MHz. It is unlikely a desktop motherboard would have something
faster than 32bit 33MHz for the PCI bus, so the 66MHz option is
extremely unlikely. (I would need Nvidia documentation to be
sure and they don't give it out.)

After all that mumbo-jumbo, most people will find bursting of
less than 100MB/sec for short intervals, is their limit. This
limit can be in several places, as stated above.

Higher than 100MB/sec has been achieved. Someone benched the
Intel SATA RAID on the ICH5R, and because the bus connection
inside the ICH5R is 266MB/sec, faster than 100MB/sec is
possible (with two Raptors in RAID mode).

I've also read comments recently from someone who has compared
the same IDE drive with and without a SATA adapter connected to
it. He finds the SATA way is slower, and this could be due to
the protocol overhead for sending commands to the drive. But until
you have seen several people try this experiment with different
brands of drives and PATA/SATA adapters, don't take this
suggestion as gospel.

HTH,
Paul
 
A

Andrew

SATA has a maximum speed of 150MB/s
PCI has a maximum spped of 133MB/s

SATA has a MHz value of 1500
PCI has a Mhz value of 33

it is all to do with the number of bits moved in each cycle.

for SATA, this is 1
for PCI, this is 32.

I currently ignore PCI-X or similar as they are not common. For
reference, the better PCI found on server boards is 66Mhz, 64bit per
cycle for a total of 533MB/s

and if you would like, think back to the old ISA format, 8Mhz and
16bits (if I recall correctly).

These speed limits are are expected to be around for along time, at
least until the newer PCI standard comes more common

A
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top