Where is SATA 300

[YanquiDawg]

SATA 300 was supposed to come about a year from the original SATA 150. What
happened?

 

[~misfit~]

SATA 300 was supposed to come about a year from the original SATA
150. What happened?

Ever hear of the term "Vapourware"? It's used to refer to technology that is
announced with a big fanfare and then either takes five times as long as is
first quoted to appear or never appears at all. It's often used as an
incentive to get early adopters to jump on a slower version of the new tech
(*even though it might be no better than existing tech) to get a big
user-base and sell the first iteration of the tech.

(*I say this as I'm yet to see an ATA drive, P or S, that can sustain
anything like the 100MB/sec that P-ATA can handle easilly. In fact I was
looking at the specs for a Maxtor SATA drive the other day for a guy on a NG
(could have been here) and sustained throughput was only 50-something
MB/sec. ATA66 could handle this around six years ago or more.)

The only advantage SATA has over P-ATA is the little cables. The
disadvantage of only being able to have one device per controller and not
being able to run optical devices from the bus (therefore meaning you still
have to have P-ATA anyway) make it a bit of a dog. It may be of value when
we migrate to solid-state storage media but until then it's been very
successful at what it was designed to do. Part people from their money. It's
sooo geek-chic to have SATA drives.

You may see SATA 300 in another year but I wouldn't count on it. Also, why
the hell would you want it? The only advantage over ATA66 is when it's
reading from the drive cache. Mechanical drive technology is *never* going
to get close to 300MB/sec sustained throughput, there are unavoidable
physical limitations inherent to the design that won't allow it.

 

[Derek Baker]

Ever hear of the term "Vapourware"? It's used to refer to technology
that is announced with a big fanfare and then either takes five times
as long as is first quoted to appear or never appears at all. It's
often used as an incentive to get early adopters to jump on a slower
version of the new tech (*even though it might be no better than
existing tech) to get a big user-base and sell the first iteration of
the tech.

(*I say this as I'm yet to see an ATA drive, P or S, that can sustain
anything like the 100MB/sec that P-ATA can handle easilly. In fact I
was looking at the specs for a Maxtor SATA drive the other day for a
guy on a NG (could have been here) and sustained throughput was only
50-something MB/sec. ATA66 could handle this around six years ago or
more.)

The only advantage SATA has over P-ATA is the little cables. The
disadvantage of only being able to have one device per controller and
not being able to run optical devices from the bus (therefore meaning
you still have to have P-ATA anyway) make it a bit of a dog. It may
be of value when we migrate to solid-state storage media but until
then it's been very successful at what it was designed to do. Part
people from their money. It's sooo geek-chic to have SATA drives.

You may see SATA 300 in another year but I wouldn't count on it.
Also, why the hell would you want it? The only advantage over ATA66
is when it's reading from the drive cache.

http://faq.storagereview.com/tiki-index.php?page=BufferSizes

Yes, that's about size, but it shows how important cache is.

Mechanical drive

technology is *never* going to get close to 300MB/sec sustained
throughput, there are unavoidable physical limitations inherent to
the design that won't allow it.

What limitations? I don't now that there aren't some, but predictions
something can't be done with technology are always perilous.

 

[~misfit~]

http://faq.storagereview.com/tiki-index.php?page=BufferSizes

Yes, that's about size, but it shows how important cache is.

Yeah, the size of the drive cache *is* important in a lot of instances.
However, until the drive is *all* cache (ie. solid-state) then *sustained*
speeds of 300MB/s are way above and beyond what HDDs are capable of.

What limitations? I don't now that there aren't some, but predictions
something can't be done with technology are always perilous.

The limitations of accelerating and deccelerating the read/write heads, with
accuracy, once the correct track has been located. The only way that HDD
technology is going to increase sustained throughput is with higher aureal
density (more data per square mm) and faster rotation speeds. However the
positioning of the heads will always be the limiting factor.

For desktop use I don't really think rotation speeds are going to go up much
more. Have you heard a 15,000rpm drive in operation? Have you felt the heat
output? As for aureal density it will definitely increase, which will
increase throughput for the track currently being read, (as well as drive
capacity) however once the heads have to find a different track you are back
to the limitations of acceleration-decceleration. Also increased aureal
density will require that the heads be positioned far more accurately as the
data 'tracks' will be finer and closer together. This will make it more
difficult to accurately locate data tracks. Increasing aureal density is a
two-edged sword with HDD tech, you increase capacity of the platter but you
end up with a lot more tracks per mm, measured radially, meaning increased
precision is required to locate the correct track

So, while I'm sure sustained throughput of current-technology HDDs will
increase to a certain extent, I'm fairly confident that it won't increase
five-fold to the stage where we need 300MB/s-capable controllers. That will
require a whole new technology. Do you think Maxtor, Seagate, Western
Digital et al are relying on producing mechanical HDDs as their main income
for the year 2014? I seriously doubt it. One day (in the not too distant
future) we are going to remember the days of mechanical storage media in
much the same way we remember candles and lanterns. Or audio casettes. It is
a technology that has nearly reached it's limits. We have the technology now
for solid-state (SS) (non-volitile RAM basically) storage, the only reason
we're still using 'disks' is because they are cost-effective. Solid-state
storage is still very expensive. The future of digital storage could could
well be optical, as in lasers and crystals. <shrug> One thing I *can* tell
you is that mechanical HDDs aren't the future of digital storage, will never
reach sustained throughput of 300MB/s, and won't be around for much longer
(relatively speaking).

The only time I'll buy SATA is if it becomes the only (or cheapest) option.
Other than the small cables there is no reason to have it, and probably
won't ever be. (Unless it is kept as a transitionary interface when SS
storage comes in to make it backwards-compatible).

 

[DevilsPGD]

In message <[email protected]> "~misfit~"

(*I say this as I'm yet to see an ATA drive, P or S, that can sustain
anything like the 100MB/sec that P-ATA can handle easilly. In fact I was
looking at the specs for a Maxtor SATA drive the other day for a guy on a NG
(could have been here) and sustained throughput was only 50-something
MB/sec. ATA66 could handle this around six years ago or more.)

Have you looked at the WD Raptor drives? 10,000rpm, 5.5ms seek times. I
suspect (although I haven't tested) that they could exceed 100MB/s.

The other issue is the PCI bus, which would negate the need to go any
faster then SATA-150.

 

[Derek Baker]

In message <[email protected]> "~misfit~"


Have you looked at the WD Raptor drives? 10,000rpm, 5.5ms seek times.
I suspect (although I haven't tested) that they could exceed 100MB/s.

71.8MB/s sustained here:
http://www.storagereview.com/articles/200401/20040126WD740GD_2.html

The other issue is the PCI bus, which would negate the need to go any
faster then SATA-150.

If the controllers are in the chipset the limit will be the connection to
the memory controller.

 

[~misfit~]

In message <[email protected]> "~misfit~"


Have you looked at the WD Raptor drives? 10,000rpm, 5.5ms seek times.
I suspect (although I haven't tested) that they could exceed 100MB/s.

Yeah, I'm aware of them. I'd like to see some accurate benchmarks done with
one (or more). I've downloaded the WD data sheets on them but they don't
give any figures except data transfer rate; 150 MB/sec max.

I just did a quick look-see on the web. Storagereview.com (A very reliable
site for all things related to HDDs) give them a maximum sustained data
throughput of 72.2 MB/sec (marginally faster than an ATA66 controller could
handle) on the outer tracks, dropping to 54.1 MB/sec for the inner tracks.
They claim these figures are only exceeded by 15,000rpm SCSI drives in their
experience/benchmarks.

The other issue is the PCI bus, which would negate the need to go any
faster then SATA-150.

True, however I believe that some modern mobo's are available that don't run
the SATA interface off the PCI bus.

 

[kony]

SATA 300 was supposed to come about a year from the original SATA 150. What
happened?

If manufacturers make investments in SATA 150, why not enjoy a
price drop in that tech instead of paying more for SATA 300 which
has a theoretical benefit we won't realize for a long time?
Switching standards every year isn't good for anyone.

 

[Boba & Ilinka]

What about NCQ technology?

Boba Vancouver BC

If manufacturers make investments in SATA 150, why not enjoy a
price drop in that tech instead of paying more for SATA 300 which
has a theoretical benefit we won't realize for a long time?
Switching standards every year isn't good for anyone.