That's kind of interesting.. What was the CPU usage like between the two
drivers? Also, the NVidia drivers have a speed test button inside the
entry for the controller in Device Manager, might be interesting to see
what that shows.
NCQ mainly provides gains when there are multiple disk transactions
occurring at the same time - something like HD Tach likely doesn't
stress that very much. Then again, many common desktop applications
don't really do that either..
I bit the bullet and reinstalled all of the nForce driver package
6.53.
What a difference there is from the first time!
1 -- Previously, with nForce drivers installed, in DM I had no less
than NINE entries under "IDE ATA/ATAPI Controllers. With the MS
drivers, I had SEVEN. On reinstall with nForce the second time, I
have JUST THREE:
-- nVidia nForce4 ADMA Controller
-- nVidia nForce4 ADMA Controller
-- nVidia nForce4 Parallel ATA Controller.
Gee, that looks more elegant than that gemish that was there before.
2 -- Benchmarks are way better now than the first time nForce was
installed:
---Previously, with Microsoft-XP IDE controllers, the Raptor gave a
sustained read of about 64.5 MB/s and burst speed of 122 MB/s with CPU
usage of 4%, and that was about 20% and 5% better, respectively, than
the speeds previously measured with HD Tach on the nForce drivers the
first time they were installed.
---NOW, the 74GB Raptor gives the SAME sustained read but about 4%
slower burst speed, with 5% CPU usage. That's fine; there's no
significant difference but NOW I have NCQ enabled.
---The Seagate 7200.8 300GB drive now gives the same sustained-read
rate as the MS drivers (55.4 MB/s) but with a significantly greater
burst rate -- up from 116 MB/s to 137 MB/s (!!) which is VERY close to
the theoretical limit for SATA 1. Its CPU usage doubled from 2% to
4%, possibly within the reproducibility range of HD Tach and not,
therefore, necessarily statistically significant.
3 -- The Device-Manager "Speed Test" for the two drives is as follows:
--- Raptor: 71.5 MB/s sustained; 112 MB/s burst
--- Seagate: 71.5 MB/s sustained; 112 MB/s burst
It's interesting that in this nVidia test, the 7,200rpm drive gives
sustained-read rates the same as the 10,000rpm drive. I think the
results cast some doubt on the validity of the test.
In my case, the FOURTH IDE driver installation is the one that appears
to have finished correctly. I conclude that if everything is set up
optimally with nVidia drivers, and assuming that the SI3114 is not
enabled, there will be just three devices listed under "IDE ATA/ATAPI
Controllers." I think that any number greater than three indicates
the need to run the nForce setup program again.
Ron
