Kony, what about turning all this the other way around and asking you
if I would be able to run my PATA drives satisfactorily on a future
mobo by using a PCI adaptor on that new mobo to take PATA drives?
Yes, that is the preferred solution.
That "next" system will not have the past-generation Via
southbridge with the PCI problem, and it will have fewer
devices on 32bit/33MHz PCI too in almost every possible
case. Result- the PCI card and drive combo on a new board
will be faster than on your present board, and that PCI card
will allow reusing as many of your PATA drives (besides this
new one you consider) as possible.
PROBLEMS: (i) You were uneasy recently that my Via SV266A chipset
would provide decent throughput for a SATA drive attached by a PCI
card.
It is a fact that it will be slower than it otherwise would
be. I can't determine for you how much of a sacrifice to
make in this regard, but to pay MORE for LESS seems
outrageous to me, especially when that PCI card may effect
OTHER devices already in your system that aren't
malfunctioning (yet).
(ii) You say above that a USB connection may also be slow.
Yes. USB2 will be slower than the onboard PATA controller
and slower than the SATA card. It "might" not interfere
with other PCI devices though, "IF" the USB controller is
southbridge integral instead of a discrete chip.
However, if the USB is 1.1 not 2.0, it'll not just be slower
but dreadfully slow, unrealistic for most uses.
(iii) I would add that the cpu power & SD-RAM speed of my current
machine is not great.
True, only you can decide when that's a problem.
But a future mobo would have much better bandwidth, faster memory
than my SD-RAM and a faster processor. Would all these mean that I
would get reasonable throughput and response time if I put my current
PATAs on to a PCI adapter card on the new future mobo?
The limitation on your present system from a throughput and
disk response is almost exclusively due to the PCI bus and
further, that it's from that Via southbridge with a unique,
particular problem with high bandwidth devices and
especially multiples of PCI devices. Putting the exact same
PCI card in a newer system will be on a more effective PCI
bus to improve this even if you had an underclocked CPU and
memory to the effect of their being the same performance
level as your current CPU and memory.
To put it another way, you will get lower performance from
an SATA drive on a PCI card in your system than if the same
card and SATA drive were put into a Pentium 2, 450MHz system
with Intel 440BX chipset, because it didn't have the PCI bus
problem your particular board does.
I guess that PATA drives (I have several 160 GBs ones) attached like
this are not going to be as good as SATA drives attached to SATA
connectors on a future new mobo.
There is every reason to believe they will have nearly
identical peformance if roughly the same generation (platter
density, RPM, etc) of drive. SATA or PATA does not make
much difference at all when on a PCI card and with nearly
identical drives. The drive itself, not the interface to
the drive controller, is by far the larger bottleneck
if/when the PCI bus is working well.
But would this PATA performance be
hopelessly second class compared to the SATA drives?
No, it could as easily be that the PATA is still slightly
faster than SATA but both would be very close. In random
benchmarks, SATA sometimes beats PATA but in optimal
situations, not typically on PCI cards and not by much...
and, these are high end drives not the budget 160GB class.
My advice is to get the performance you can now... with a
PATA drive. When you do end up buying a new system, the
drives available _then_ will be higher performing than those
you have and the one 160GB you contemplate now, so at that
point in the future if utmost performance matters, THEN
would be the time to buy a new drive to address that
concern.
