[David Satz]

This is a follow-up to some messages I posted here several months ago.
I was trying to get an Alesis IO|2 to work reliably with a two-year-old
Toshiba laptop, but it was skipping samples even with simple 44.1 kHz
two-channel live recording. This was easy enough to see if I fed in a
300 Hz sine wave and recorded it--the resulting waveform had jagged
glitches in it every 10 seconds or so.

Recently I upgraded the RAM on the laptop from 256 MB to 512 MB and at
the same time, upgraded its hard drive from a 4500 rpm model to a 5400
rpm model. The average access time of the new drive is only around 20%
lower than the old one, but it has a much larger RAM buffer (8 MB).

Somewhere in all of this, the computer's performance apparently reached
the point at which there are no more glitches in the recorded waveforms
at 44.1 kHz--at least I haven't seen any so far in some brief tests,
which used to be enough to reveal the problem.

(Next I intend to record a long stretch of tone and write a program to
scan the resulting WAV file for glitches, since examining all the
waveforms visually is tiresome.)

What this tells me is that manufacturers' statements about the "minimum
hardware requirements" may need to be taken with a considerable helping
of salt. This laptop was already well within the stated requirements
before the upgrades, and frankly the upgrades weren't all that huge. I
seem to have started just on one side of the real line, and landed just
on the other side of that line with the upgrades.

There's _got_ to be a better way to find out what's wrong and what
needs doing in these situations ...

--best regards

[Mike Rivers]

David Satz wrote:
This is a follow-up to some messages I posted here several months ago.
I was trying to get an Alesis IO|2 to work reliably with a two-year-old
Toshiba laptop, but it was skipping samples even with simple 44.1 kHz
two-channel live recording.

Recently I upgraded the RAM on the laptop from 256 MB to 512 MB and at
the same time, upgraded its hard drive from a 4500 rpm model to a 5400
rpm model. The average access time of the new drive is only around 20%
lower than the old one, but it has a much larger RAM buffer (8 MB).

Somewhere in all of this, the computer's performance apparently reached
the point at which there are no more glitches in the recorded waveforms
at 44.1 kHz--at least I haven't seen any so far in some brief tests,

This crap is so unpredictable. I never had any trouble recording on my
four year old Dell laptop with 256 MB RAM and a 4200 RPM disk drive
until I got up to 8 tracks with the Mackie Firewire mixer. Two tracks
with a TASCAM US-122 worked just fine. I was getting regularly spaced
glitches with the Mackie until I unplugged the network cable. A
semi-permanent solution was a newer driver for the network card, more
reliable solution was to disable the networking hardware in a hardware
profile.

(Next I intend to record a long stretch of tone and write a program to
scan the resulting WAV file for glitches, since examining all the
waveforms visually is tiresome.)

That's the way I was testing. One of the guys here turned me on to a
freeware program called Wave Repair which does a good job of detecting
glitches, but it only works on a 16-bit 44.1 kHz stereo file. Good
enough for test purposes. Sound Forge has a "Find Glitch" tool, but it
has a couple of sliders to adjust it and I can't understand what they
really do and how they work, so I tried all combinations of sliders up
and down and didn't find any glitches until I purposely introduced one
to test the tester.

What this tells me is that manufacturers' statements about the "minimum
hardware requirements" may need to be taken with a considerable helping
of salt. This laptop was already well within the stated requirements

They don't have a clue. When I starrted shopping for a new laptop
computer, I saw several user reports of glitching on Toshiba computers
so I crossed that off my list. Manufacturers try their stuff out on
whatever computers they and their tech support folks have and don't
make any attempt to try to find "recommended" hardware. That's the nice
thing about a dedicated hardware workstation - the manufacturer knows
what the software has to deal with and makes adjustments accordingly.

There's _got_ to be a better way to find out what's wrong and what
needs doing in these situations ...

Yeah. Quadruple the cost of the inexpensive hardware so that they can
afford more extensive testing and sustaining engineering support, and
not be afraid to publish useful information like:

"If you have a Brand X model A320, Brand Y model B29, or Brand Z model
AK47 notebook computer, don't buy this product. We've found that it
doesn't work on those computers and we can't figure out how to make it
work. "

Like that's gonna happen.