I, too, would like to see SF's answer to this, as it affects our buying plans for computers. What gives, SF, please?

I know this is a little off topic but do you guys who are running duel Xeons get smooth output to your NTSC monitors on real time previews---Just wondering if processer speed would make a difference---mine jitter until they are rendered.

Sorry, I can't help you, but would like to know more about it myself. I may get the job (blame) for the next round of purchases here at my departmental labs. Already I am pushing for VV, even though it is much more expensive (about double) the Pinnacle Studio software which some are suggesting. Educational pricing helps the ratio, but didn't solve the problem.

My knowing as much as possible about the good, better, best computer configurations would help my case a ton. It is not that I don't know computers, it is the interaction with the specific software, VV, that I have problems with. So little is explainined in any literature, website, help file, or other source, that I am really beginning to suspect that all this is supposed to be a secret.

Why doesn't SF toot their horn a bit more clearly? We can't here you, SF!

Hi All,

About the hard drives, 4 X 120gig ide (7200rpm) in raid 0+1 for video, 3 X 80 gig
for system drives. About Hyperthreading - it's 'supposed' to come in when a cpu is being underutilised, I believe it would be very OS dependant on how well this works in practice. VV would need to tell the Kernel processes in the OS what priority to give each thread depending on what you the user was doing at the time - some threads such as background processing are obvious, however others are not. This is relatively routine on a single cpu system, a little less so on a 2 cpu system - and downright difficult on a two cpu system masquarading as a 4 cpu system. Although XPPro does show 4 cpus when hyperthreading is on, I'm not sure if it is able to take full advantage of this psuedo hardware when push comes to shove. Nice in concept - perhaps not so easy in reality. I would, though, still be interested in VV's views on this as 2cpu systems with hyperthreading capability will only become more common as prices drop.

Best Regards

Robin.

PS I use PAL as I'm in Australia so I have no answer to the NTSC question - sorry.

Vegas uses multiple threads for audio FX processing, and a separate thread for DV rendering/writing. However, we're rarely CPU-bound, since video is largely an I/O bandwidth issue. Therefore, you won't see 100% utilization on both processors of a dual-proc machine. Probably more like 60% on each during rendering. UI will be more responsive since we can be drawing & interacting on one thread and rendering the video preview on another. Dual procs are also nice when you're running multiple instances of Vegas, or use other applications while Vegas is working. I can't make the call as to whehter dual procs are worth the price increase, though, since I don't have enough personal experience with them myself, and it depends on your own needs.

Hyperthreading is an Intel technology that simulates a dual proc machine using a single proc (read more here). I have no metrics on this, so I can't say at all how it affects Vegas.

///d@

To summarize HT: One Physical CPU acts as two logical CPUs by utilizing unused L2 Cache. Windows XP Pro is currently the only desktop level OS that supports more than 2 CPUs with purchasing additional licneses. Windows 200 Server products also support more than 2 CPUs.

In the testing I performed on a Supermicro P4DCE+ dual Xeon system board I found that two physical CPUs with HT disabled will outperform a single CPU with HT enabled, but not when HT is enabled with both CPUs installed. However, this was only using SANDRA 2002 benchmark utility, so the results are synthetic. I haven't tried timing renders in VV yet, but I will and post my results here.

Where do you control Hyperthreading from; the BIOS, OS or VV?

On my machine it is set in BIOS by pressing DEL when booting and making a sleection there.

Supermicro, dual Xeons, 1 gig

That's odd, I don't recall Intel making a 1gig HT Xeon! Do you mean 2gig?

Intel's current recommendation is to run the Xeon processor with Hyperthreading disabled. Many programs are not optimized properly to recognize the difference between a real second processor and a second thread, and you won't see any performance increases in some cases.

Having not benched Vegas directly with HT on and off, I would suggest you guys actually do test renders with both cases and decide how you prefer to run. HT enable/disable should be in your BIOS setup, and you must have support for the additional processors in your OS (at least on win2k, e.g. two physical Xeon processors with HT on will show up as 4 processors in windows, so you have to be running Windows2000 server), although I think the support is in XP Pro right now to support extra threads without extra processor licences, and I believe you can run 2 Xeon processors with HT enabled on XP pro (and it will show up as 4 processors in the task manager).

Briand,

Benchmarking with Madonion3Dmark on my system with HT on will slow down the system by about 4%, (GeForce4 128mb). This is obviously not 'real world' and it remains to be seen if VV is affected at all. I have a 20 min video I will render tonight using MCMpeg1 and 2 and time the difference with HT on or off. XP Pro does indeed display 4 cpu's if HT is on. When only one cpu was fitted, and HT was on , the second 'cpu' was, when under load, directly tracking the 'head room' which was not being used by the physical cpu. The technology works, however, as I said in an earlier post, it requires some clever work by the OS to fully utilise the 'bandwidth'.

Best regards

Robin.

Here's what I found. Since I only have a demo copy of VV3.0c on this system, I am rendering to Windows Media format.

Important System Specs
MoBo: SuperMicro P4DCE+
CPUs: (x2) Intel Xeon 2.0GHz w/512MB L2 Cache
RAM: 2GB RamBus
OS: Windows XP Pro

Project Information
Clips: 2
Transition: 1 (simple x-fade)
Video Effects: 0
Sound effects: 0 (removed all default effects)
Total clip length: 30sec
Render as: Best Quality, Windows Media Video V8 format, 256 Kbps Video template with Video quality changed to 100% and two pass encoding enabled.

Render Time Results
HT enabled: 2:34
HT disabled: 2:30

I know the sample file is pretty small, but this testing does show that it actually took longer with HT enabled. The difference can increase significantly with a full lenth movie.

It appears that, until VV is updated to utilize HT, it may be a good idea to disable HT.

Hee Hee,

Your results mirror mine other than the 2 pass encoding. I must be blind but where is this option, and is it available for the MC Mpeg 1/2 encoder as well?

Note: I gig of RDRAM - not 1 GHz CPUs (there are 2 ea. 2.2GHz installed at this time).