Sometimes the on-board RAID on motherboards can be flaky and corrupt your data- this is not a good thing!! The Intel chipset RAID should be decent, although that isn't what you have.

2- PATA and SATA as interfaces are about the same speed *effectively*, since the interface is not where your bottleneck lies. SATA drives tend to be newer models and hence slightly faster.

You should be fine with either.

I thought there wasn't much of a difference, but felt I may have missed something.

Farss, I've been lucky so far, but it was cheaper at the time to just use a couple of 160GB drives I hade to build a RAID0 setup than to buy a single 300GB drive, although that has changed recently - I may still opt for a single larger drive, but for now, the Rosewill (SIIG) hardware based Raid card I have is quite nice.

Thanks again for the insights...

Cliff

I would think you can also better take advantage of speed (RAID vs. one single drive) if you have multiple streams of video, such as four tracks of HDV on your timeline. That said, it might be even faster for this scenario if instead of being in a RAID those two drives were separate with the four streams of HDV divided amongst the two drives.

Which brings me to another point.. If you have two standalone PATA drives on the same cable, you can't access them both simultaneously. With SATA, they have their own individual channel, thus you can hit them both at the same time.

I actually have each drive on it's own channel on the RAID card - I was seeing bottle necks when accessing the drives when they were on the same cable as master and slave. I put each drive on the card with a single connection cable on each port and all is well...

I am now almost wishing I HADN'T gotten AMD after reading the last couple of days reviews of Conroe... :-/

I have had more BSOD's in the last month than I have had in the last 3 years... Not sure what else to do at this point...

Try troubleshooting your computer. A new computer shouldn't BSOD at all... any hardware that causes BSODs should be sent back for warranty work.

Try:
http://sp2004.fre3.com/
This will stress the RAM and CPU. CPU errors will show up pretty quickly (<15min). RAM errors will show up somewhat quickly, although subtle problems might take a day to pop up.

Also run a motherboard monitoring utility. Hopefully your mobo manufacturer provides such a utility.. otherwise try motherboard monitor or speedfan.

Well, I did install all current updates as of late last night from M$ and this seems to have solved whatever was happening - I am running x64 XP Pro not sure what in the updates fixed my issue. I am currently 15 minutes into using the utility and is humming along just fine. Here are the results:

CPU1:
Type: Blend - stress CPU and RAM Min: 8 Max: 4096 InPlace: No Mem: 3839 Time: 15
CPU: 2010MHz FSB: 200MHz [201MHz x 10.0 est.]
CPU: 1005MHz FSB: 200MHz [201MHz x 5.0 est.]
9/29/2006 3:27 PM Beginning a continuous self-test to check your computer.
Press Stop to end this test.
Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
Test 2, 4000 Lucas-Lehmer iterations of M19922943 using 1024K FFT length.
Test 3, 4000 Lucas-Lehmer iterations of M19374367 using 1024K FFT length.
Test 4, 4000 Lucas-Lehmer iterations of M19174369 using 1024K FFT length.
Self-test 1024K passed!
Test 1, 800000 Lucas-Lehmer iterations of M172031 using 8K FFT length.
Torture Test ran 17 minutes 57 seconds - 0 errors, 0 warnings.
Execution halted.

CPU2:
Type: Blend - stress CPU and RAM Min: 8 Max: 4096 InPlace: No Mem: 3839 Time: 15
CPU: 1809MHz FSB: 200MHz [201MHz x 9.0 est.]
CPU: 1005MHz FSB: 200MHz [201MHz x 5.0 est.]
9/29/2006 3:27 PM Beginning a continuous self-test to check your computer.
Press Stop to end this test.
Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
Test 2, 4000 Lucas-Lehmer iterations of M19922943 using 1024K FFT length.
Test 3, 4000 Lucas-Lehmer iterations of M19374367 using 1024K FFT length.
Test 4, 4000 Lucas-Lehmer iterations of M19174369 using 1024K FFT length.
Self-test 1024K passed!
Test 1, 800000 Lucas-Lehmer iterations of M172031 using 8K FFT length.
Torture Test ran 18 minutes 5 seconds - 0 errors, 0 warnings.
Execution halted.

The only other thing I could think of is that my primary browser for the last year has been Opera on Win32 XP Pro, and am running it on x64 XP Pro. Seems like I got BSOD's regarding memory issues after running it. Firefox seems to run faster on x64 XP Pro so am forcing myself to use it (I love the hotkeys in Opera - they are different on Firefox).

Not sure if this is the culprit but I'll give it a day or so to see if I still have any issues...

Yeah software issues can definitely be it.

I tend to try to diagnose the hardware first since it's easier.

Just last month I upgraded my own system from an XP2100 to an AMD 3800x2. Now I'm obsolete again.

Seriously, though, this system is much faster than the one before, and although Conroe is the current king, it will no doubt be obsoleted by somthing else later on. Just yesterday I commented on a 4-core chip that blew the doors off Conroe, etc...

What's nice about the 3800x2 chip is that it is so overclockable. Stock is 2GHz, and I'm running at 2.4 effortlessly (could probably go to 2.6-2.7), so I essentially have a 4600x2.

OC'ing is greek to me - I've tried to get a grasp of what to do, so far - it's too technical for me (This coming from someone who understands the zone system from my b/w 4x5 film days) - I would love to get some laymans term info on oc'ing my machine - nothing fancy,

Your motherboard has a clock generator that determines what speed your computer runs at. Many of your parts' speed is determined by taking that clock speed and *multiplying* it be particular numbers.

The system clock is sometimes synonymous with the FSB's (front side bus) clock speed.
For the CPU, the CPU's clock speed is FSB clock multiplied by the "multiplier". The multiplier is just some number, like 11.
So a 200mhz FSB and a multiplier of 11 = a 2200 mhz CPU (i.e. 2.2 Ghz)

The memory's speed is similar, and has its own multiplier.

The three key speeds are:
Front Side Bus
CPU
Memory

By manipulating the multipliers, you can get the CPU and memory running at particular speeds. FSB speed is not that important, but it usually determines one or two of the other speeds.

2- You want to run the CPU and Memory at particular speeds since one overclocks more than the other. Usually, you want to overclock your CPU as much as possible as it affects performance the most. Your RAM may not be optimally overclocked, and that's ok.

3- You can increase the memory timings on the RAM to allow it to overclock further.

Basically, that helps you overclock the CPU even more.

4- Clock speed is a tradeoff between performance and stability. The higher the clock speed, you edge towards poorer stability. It's like the speed limit... a higher speed limit increases the chances of collisions.
At a certain point, the system will get calculations right (practically) all the time. i.e. if you set the speed limit at 10 miles per hour, you won't get any auto collisions.
You could run the speed limit higher and still not get any auto collisions. Overclocking is kind of like this.

5- Increase the speed high enough, there becomes a point where there is an error once/day or thereabouts. This would be the grey area. And then even higher, your computer won't boot properly. You want to avoid both. The grey area is insidious and may not be noticeable. This is why you should stress test your computer. Figure out the point where your computer is barely unstable, and then back off.

6- As you get to the higher speeds, heat builds up more. If you cool your parts better (mainly the CPU), then it will have more overclocking headroom.
There are different levels of effort you can go to.
Use thermal paste instead of the waxy stuff that comes with the retail CPUs. i.e. Artic Silver 5 (read the instructions!!)
Use a better heatsink + fan; anything beyond this is probably wasting your time
Use water cooling instead (i.e. some of the G5s do this)
Use Peltier cooling (really exotic)
Use dry ice (impractical, but some people are crazy)

7- If using any fan-based system (this includes most water and peltier cooling systems), cooling is a tradeoff between cooling efficiency/power and noise generated. So the really fast-spinning fans will cool well, and make a racket. Some fans are more noise-efficient than others.

Ok I think that explains the basic concepts.

8- What to do with the various multipliers is dependent on what CPU you are using (and sometimes dependent on chipset). The information is specific to the CPU, so you have to find the relevant information.

9- Similar, motherboard setting information is specific. Some motherboards do not give you the full range of options available.

Resources to try would be:
Forums for particular motherboards
Enthusiast sites like hardocp.com (very relevant motherboard reviews)
Google; there may be guides for overclocking your particular CPU type

9- General strategy for overclocking (the way I'd do it):
Set the memory to be as loose as possible, so that your memory will not limit your overclocking. Adjust the relevant multiplier, and set the memory latencies to as loose as possible.

Raise the FSB speed incrementally, until your computer is obviously unstable (i.e. doesn't load Windows, doesn't boot). Clear the CMOS if necessary (via the CMOS jumper).

Back off on the FSB speed until the computer seems it may be stable.

Stress test the computer. Use something like Stress Prime 2004.
http://sp2004.fre3.com/

Let it run overnight. Also install (beforehand, with no overclocking) a motherboard monitoring utility. The one from your motherboard manufacturer will work. If they don't provide one, try your luck with Motherboard Monitor or Speedfan. You want to monitor your temperatures to see that they look ok (i.e. that it looks like your parts are being cooled properly). What you want to look for is parts that aren't being cooled well... i.e. there is a big difference between the idle temperature and under stress temperature. Sometimes, you may install the CPU incorrectly (and it won't be cooled very well in that case).

Figure out the point where your system is marginally stable/unstable.

Repeat the process for the memory. For video editing, optimize for memory bandwidth (i.e. use the loosest/slowest memory timings).

If you don't want to risk instability, just back off on the overclock. Leave some headroom.

10- For unknown reasons, your computer may still be unstable (and not generate errors in Prime95). I suspect that Prime95 may not find every kind of error.