|
Akasa
Round ATA100 IDE Cables
|
|
Author : Wayne
Date : 24th September 2001
|
3DVelocity would like to thank
Lapicon
Electronics Ltd and especially Adrian Young for their help
and courtesy in providing these cables for review.
Introduction :
The thirst for extreme
cooling and general case enhancements has taken many
forms over the past year. Not only are PC owners prepared to
go to any lengths necessary to improve airflow within their
cases, the fad for adding case windows means that not only must
these solutions work well, but they need to look good too.
The humble IDE ribbon cable
has been with us for many years, and has only really seen one
substantial design change, that of switching from the traditional
40 wire format to the newer 80 wire ATA66/100 design which we'll
look at in a moment. Next year should see the appearance of
the new serial ATA interface which promises considerably better
performance and infinitely more manageable cables, but for now
we must contend with ribbon cables that are bulky, inflexible
and unsightly slabs of grey plastic granted to disrupt efficient
airflow within your case.
Rounding of IDE cables had
been carried out by some enthusiasts at home long before machine
rounded cables appeared on the scene, but making a home rounded
cable look good was always a challenge, and the actual process
of separating the individual wires, particularly in an 80 conductor
cable often ended in mistakes and cables relegated to the trash
can. Machine rounded cables on the other hand not only work
well, they look good too, and despite the reservations some
people have with regards the crosstalk problems that may arise
from separating the data signal from its neighbouring ground
cable (another subject we'll address in a moment), rounded cables
are certainly something to consider if cooling and looks are
paramount.
The particular cables I
want to look at today are made by Akasa, yes, the same Akasa
that produce the Silver Mountain cooler. They are available
in yellow, green, black or white, and in 45 and 90cm lengths.
For the test I'll be using the 45cm yellow cable, and if you
expect me to benchmark the rounded floppy cable that was also
supplied you're gonna be disappointed, you'll have to settle
for me telling you that it worked.
ATA100-45 on the right,
floppy cable on the left
The cable :
From the standpoint of looks,
the Akasa cables offer nothing of note over the competition.
The outer shroud is extremely flexible and feels like a silicone
rubber rather than a conventional rubber or plastic. The flared
end caps are not fixed to the main shroud which also allows
greater flexibility when you need to bend the cable sharply
as it allows the caps to slide on the shroud itself. If like
me you have a nasty habit of removing your cables by tugging
on them, I'd strongly suggest you get into the habit of removing
these by pulling on the connector block instead. Pulling on
these cables in the same way many of us do with ribbon cables
is likely to put strain on individual wires, perhaps to the
point that they fail.
Potential problems :
As I mentioned earlier,
the rounding of ATA-100 cables may seem simple and even sensible,
but from a technological standpoint it is not considered the
wise thing to do. The first transition from ATA-2/3 to UltraATA
involved allowing data to be transferred on both the rising
and falling edges of the clock, much in the same way DDR operates.
This in itself was enough to prompt concerns about data integrity.
The switch to UltraATA-66/100 however and its increased frequency
pushed beyond the theoretical limit of the standard 40 conductor
(40 wire) IDE cable, and in a attempt to ensure reliable operation
and safe transfer of data, a new cable was designed that added
an additional 40 ground wires to the existing 40 data.
These 40 ground wires were
sandwiched between the data wires to eliminate ringing and crosstalk
between them, a phenomenon that occurs when data from one data
wire bleeds over to the adjacent wire. In theory, this crosstalk
isn't likely to introduce huge levels of data error, but what
it can do is increase setup times. Setup time is the time taken
for an individual wire to settle back to a normal state after
having carried a pulse of data, so if a wire is receiving external
interference, it may take longer to settle again and thus it
delays the sending of the next data pulse. This is a very simplistic
way of describing what happens, but it does offer an insight.
From the viewpoint of crosstalk
then it should be obvious that the separating of the individual
wires reduces the effectiveness of the ground wires as you may
now have data wires running right alongside each other. Well,
this is true, but as with most things in life there's a flipside
to the argument. Another major source of interference within
the IDE cable comes from external EMI (Electro Magnetic Interference)
created by things like your floppy drive, power supply and so
on. I've heard claims that as much as 75% of the errors on the
IDE line are as a direct result of external EMI, so although
separating the wires on your IDE cable may increase the risk
of crosstalk, the additional shielding provided by the shroud
can also reduce EMI within it. It's a contentious area that
could probably be argued about all day, but one fact that's
not in question is that the best way to ensure data integrity
is to use the shortest cable possible. If you can make do with
a 45cm rather than a 90cm cable, then do it.
Testing :
To begin with, I'm not going
to paste up a dozen graphs that show my case temperatures with
ribbon and then with rounded cables. I will tell you that my
case temperature actually dropped by 2 degrees Celsius, but
that means nothing if your cooling arrangement differs to mine.
I had two IDE ribbon cables that ran across the face of one
of my front intake fans, and eliminating these made for a worthwhile
improvement. Depending on your cooling layout, you may see different
results. What I was more concerned about was performance, and
I embarked on this test assuming I'd see identical or very slightly
lower scores, particularly as the ribbon cable I was replacing
was what I considered to be a good quality VSO Electric co.
cable.
First up I ran Sisoft Sandra's
drive benchmark using the standard ribbon cable
With the standard ATA-100
ribbon cable, Sandra returned a Drive index of 19309. This was
the average of three runs, none of which varied by very much.
Next I fitted the Akasa
round cable and ran the test a further three times. Here again
is the average run.
To say I was surprised is
an understatement. The score of 19854 may be just less than
a 3% increase, but as I was anticipating a reduction in performance
it was a real surprise. Wondering if this was just a fluke I
fired up HDTach to see what it had to report.
Again, the first run was
using the standard ribbon IDE cable. The maximum read speed
of 31047, minimum of 9753 and average of 28031 was the level
of performance I'd pretty much gotten used to.
Certain that this time I'd
see a different story, I slapped in the round cable and ran
it again.
Again there was an improvement.
The maximum had increased from 31047 to 31108. The minimum had
dropped very slightly from 9753 to 9750, while the average was
up from 28030 to 28110. Nothing spectacular granted, but an
increase none the less. Considering my ribbon cable was also
45cm, the same as the Akasa round cable this was all the more
impressive.
Concerned about possible
data corruption I ran a host of tests to try to pick up on examples,
but I found none. Ahead's DAE audio decoding tests showed zero
errors over about a dozen runs, large file transfers and program
copies from my primary hard disk to my secondary hard disk appeared
perfectly byte accurate, and the rounded cables even offered
slightly quicker transfers with a saving of eight seconds on
one particular 800MB file. Now whether the 90cm rounded cables
are able to offer the same performance and data integrity results
as the 45cm cables is something I can only speculate on as I
don't have one, but as I said earlier, if you can make do with
a 45cm cable, then that's precisely what you should do.
Conclusion :
PROs
