I've been running a BMD card on dual Xeons for 2 years now, no problems.
No interest in anything AMD, too much grief, it's nice that AMD are there to keep Intel on their toes I'd certainly admit.

I have one Core Duo system, very low power, fast, quite etc.

Yes the Core 2 Duos smoke anything that's come before and they're only the beginning but they're a way off yet however I can run P4 or existing Core Duo on the same mobo and then fit Core 2 Duo later

This is to upgrade a system I use mainly for audio editing and plain DV editing, the high end stuff I do on my dual Xeon system.

The question really relates to the benfit, if any, of HT with Vegas. V6 seems to work very well with HT however I tended to do all my audio work in V5 as V6 just eats up more resources for no gain on the audio side. When I upgrade I'm hoping to retire V5 for good but how will V6 fly on a system without HT?

Bob.

Yes, dual Xeons on certain mobos were BMD-approved long before AMD.

I have been an Intel bigot since the 4004 (their first CPU), except when Zilog kicked their butt for a few years, and recently when AMD got good support chips from nVidia.

All problems with past AMD CPUs were caused by bad support chips, especially those from VIA.

The current generation of AMD CPU support chips from nVidia are regarded as being quite a bit more advanced than anything from Intel, and they have also been 100% reliable.

The current Core Duo is well priced (to try to recover some of the market share lost to AMD), and the socket compatibility is obviously cool, except Core 2 Duo will probably need new chips around it to get max. performance.

AMD also has some news later this year, there will be more butt kicking coming...

Still nothing wrong with getting a Core Duo system today for what you suggest.

I'm actually writing this on a Core Duo... :O)

Remember HT is just a fake multicore, I would expect the real thing to work better (and not much gained in Vegas from having both, but perhaps somebody else has actually tried it).

I'm also on V5, got scared off after seeing the bazillion V6 serious bug reports for a long time.

As near as I can tell, HT is just faux dual core. However, I just inherited an HT system at work and I have to say that video captures are much less error prone than they were on the non HT system before it. I never drop frames and can go about my other work.

In fact, I'm pretty happy with the general snappiness of this HT system. I'm sure real dual cores would be even better.

Rob Mack

Whatever you do, wait until the end of July....

http://forum.cakewalk.com/tm.asp?m=791430

http://www.dailytech.com/article.aspx?newsid=2800

I just saw some bench's on P4's, Core Duo, Core 2, Xeon, Opteron, AMD 64 X2, AM2 & FX's. The Core 2's are the fastest & even Core 2's that are inthe same mhz range as the X2/AM2/FX/Opteron's are faster, BUT... they are also more expensive them the X2's/AM2's in that speed range & not THAT much faster.

So, once again, is it worth an extra $100 when you're talking a few extra seconds off a render?

I personally used to be an Intel guy until I got an AMD XP 1800 chip. That blew my friends P4 out of the water. Never any problems with it eigther (infact, it's now my DVR machine!). My 486, Pentium MMX & P3 were awsome but a P4 was slower & more expensive then an XP chip. Not 1 problem with it ever (and my current 64 3000 has had 0 problems too).

People just don't realize that it's no longer Intel compatible & IBM compatible. :D

$403 (retail) for an AMD X2 5000+ sounds good...

And also the new Opterons...

Interesting that they're saying July 23 release to wholesale and Aug. 1 for retail.

Down here:
X2 5000 2.6Ghz $977.00
Pentium D 3.6GHz $694.55

However what I'm really interested in is the value of HT (fake dual core, I guess) against having two real cores with no HT running Vegas 6. Take into account that Vegas 6 seems to be optimised for HT.

Bob.

Bob:
I read somewhere an explanation of what HT is (and posted it in another thread). Basically, Intel found a bottleneck in the pipeline that was causing the CPU to waste bus cycles while waiting for something. So they found a way to utilize those wasted wait cycles by executing another thread.

I saw that fldave did some testing on a P4 w/ HT and posted results. Buried in his data you can see that he found that doing a render with threads = 1 and threads = 2 got the same results. This means that HT didn't benefit him at all during the render of that veg file.

But for all I know that could be unique to his particular veg file. Don't know. And Cineform says they get great benefit from HT during capture and convert. So I can't really say anything conclusive about HT.

I can say from my own experience that my AMD X2 4400+ dual core is worlds faster at rendering than my 3.2 GHz P4 w/ HT ever was. And I found that as long as you have a good power supply, the system is extremely reliable. I am OC'ing it to 2.42 GHz, just a tad faster than the more-expensive 4800+ model. I haven't had any crashes.

The Intel processor (and most other modern CPUs) have multiple execution units which do different things. i.e. you have 1 unit just for floating-point math calculations, 1 or 2 for integers, etc. At a given time, not all of the execution units are being used.

So by using the hyperthreading trick, the CPU can masquerade as having two cores. The CPU can run another thread, taking advantage of the unused execution units.

Hyperthreading doesn't work well when both threads are trying to use the FPU unit... which is the case with most DAW/audio programs. Practically all audio programs do calculations with 32-bit floating point numbers, and only the FPU unit will handle this operation / these calculations. So hyperthreading won't work at all, and will slightly hurt performance (because there's overhead involved to get hyperthreading to work).

Hyperthreading doesn't work well when the program isn't good at splitting up the work equally into two processes. This way the case with Vegas 5, and has changed with Vegas 6. Actually I haven't tried lately with Vegas 6.

Anyways that's my Coles notes version. Better information can be found at
http://arstechnica.com/articles/paedia/cpu/hyperthreading.ars

Good point about the floating point use.

Intel right now is back to doing what they have been doing as long as they have had competition: they compare their integer arithmetic performance with that of other CPUs, and because that was always their priority, they did well.

AMD blows away Intel on floating point performance, mostly because Intel didn't think it was that important, but AMD did.

Bob,

Personally, I'm waiting for the Core 2 (conroe) for a few reasons... 1) it will kick butt in speed and performance per watt (like I really care about that). 2) This will be the beginning of the dev cycle for the new architecture so the systems based on them will be easy to upgrade as opposed to the current AMD's which are at the end of their archtiecture cycle. 3) I have had too many probs with AMD to even consider them.

Some independent benchmarks for the new intel chips are projecting 20% faster than anything else - including the newer AMD chips with DDR2. IMO, 20% will be a little more than a few seconds off render times. However, some people think AMD currently being a few milliseconds quicker spells the end of intel...

For my purposes I would only buy a new system now if it were to replace a failed one or for a scratch machine which would just be used to surf or email etc...

I know some people have had problems with intel but for me it is better the devil I know. When I have had to work on AMD's and they hang, there is always a suspicion in the back of my mind that it could be the chip. With an intel I don't have that suspicion and that is worth the few extra dollars.

DrLumen:
Well, there is one reason for getting a machine now: if you can get it really cheaply and it isn't your primary machine. While my AMD is my primary machine, I decided to upgrade my media server / file server PC.

I spent $123 on a dual-core Pentium D 805 processor and $200 on a mobo. This particular Pentium D has a low clock speed and a FS multiplier of 20. Prime for overclocking. For the additional expense of a $20 Arctic Freezer CPU cooler, I can overclock this thing to 3.6 GHz without a hitch and it runs extremely stable at standard voltages. Got the idea from a review on Tom's Hardware. So for a whopping $340 I got a dual core Intel that screams. (I already had all the other hardware for the machine).

I use it for a file server, media server (to an Avel LinkPlayer 2 that plays HD vids on my TV), print server, scanner server, and will use it as a network renderer too.

So, depending on what your use / need is, it's not necessarily better to wait on purchasing a new PC. There are bargains to be had...

And I'm guessing your skepticism about AMDs is based on bad experiences from some years ago. I had heard that they had heat problems, etc and also steered clear of them. But I bought my first AMD last March because of all the benchmarks and reviews I had read. This one ended my skepticism. And code is not only written for Intel as the reference. The 64 bit code was written for AMD first, and Intel "borrowed" the API for it's EM64T (at least that's what I've read here and there).

I don't really feel strongly about AMD vs. Intel. I've bought one each in the last 3 months and am happy with both.

[Conroe] will kick butt in speed and performance per watt (like I really care about that). 2) This will be the beginning of the dev cycle for the new architecture so the systems based on them will be easy to upgrade as opposed to the current AMD's which are at the end of their archtiecture cycle.

Intel measures the power consumption of their CPU against AMD's CPU.

Good comparison? Not even close. AMD CPUs contain the Northbridge inside, while Intel needs a major power guzzling separate chip for this.

AMDs X2 CPUs have barely started yet, and the Opteron architecture is good for another 3-5 years, perhaps longer (it's actually very old technology from DEC Alpha, but it's far more advanced than anything from Intel this year or next at least).

Do you think Lucasfilm and the other Hollywood studios are suffering a lot with their 5,000 AMD computers?

Do you think Google will collapse after switching from Intel to AMD?

Intel looks to be on track to recovery, but they're not quite there yet,

Well the existing Core Duos must not be too power hungry as ther're easily overclocked to 4GHz and beyond with water cooling (not for me though).

All that aside though the Core 2 Duo looks pretty promising, the site that Glenn provided a link to has lots of good oil on CPU design although they are very in depth reads. The one consistent theme that I see throughout all the articles is it all depends on the code and how it's written. I've seen this firsthand myself, recoding a few lines of Pascal into assembler that was CPU aware took execution time from hours to seconds for a simple diagnostic I wrote decades ago.

Now with Intel changing tack between designs and throw AMD into the equation and I'd feel pretty sorry for todays codesmiths trying to optimise code. Admittedly most of it's compiled however how do the compiler designers optimise and for what.

That's why I try to avoid the AMD Vs Intel debate, it's ultimately less fruitful than a FCP Vs Avid debate. On top of that for a lot of the work that I do I find that the CPUs aren't the bottleneck, the disk I/O seems unable to keep up and that's with SATA RAID 0.

Now I'm tryin to remember where I started from, oh yes, HT or Core Duo. Well if I can believe all that I've read HT wasn't a fudge entirely and Core 2 Duo will have HT as well as 2 cores just to squeeze the last ounce out of the CPU and it adds quite a bit more in the ALU dept as well.

So from what I've read the Core Duo without HT will run a bit better than the old P4 with HT (depending on the software) so that's the way I'll go. Yes, this is to revitalise an ailing system, the System HD died (finally) and the mobo only has USB 1 which is a real PIA. The only regret I have is the newer mobos at best seem to only have 2x1394 ports, I know, a cheapo card will fix that but still.

And of course this means a new video card but at least I'll get dual Dual DVI and a WMV / H.264 accelerator, sounds like something to brag about even if it'll be quite useless editing audio :)

Oh and from now on I'm only buying Samsung drives.

Bob.

Jayster,

RE: HT and Render Threads
If you reference the Vegas help on render threads, it defines it as "making part of your multiple cpus available for other tasks". Since I did nothing else during the test runs, it used both of my "virtual HT" cores for the render. So 1 vs. 2 changed nothing.

What I did find was that:
"This shows consistent treatment for Hyperthreaded as well as true Dual CPU systems. That is actually great news. " Great news because I hoped my HT functioned like a true dual core, and in those tests it did.

My HT showed the same CPU utilization as my true dual CPU machine, while monitoring the cpu using graphs. I'm sure a true dual core or dual cpu has advantages over the same x Ghz dual core, but HT mostly worked fine as a dual proc.

Dave

As you indicated, the existing Core Duo aren't too power hungry. Except when overclocking an 805 to 4 GHz+.

THG found that you needed a massive power supply (500W+) just for a barebones system with this, so it's good value for money, but you need to have good A/C unless you're at Ice Station Zebra.

(although perhaps you could hang the radiator out a window?)

I like the Samsung drives too, because they are so quiet.

I'm testing the new Seagate perpendicular drives for larger capacity though (up to 750GB currently). Any reason to think those are not good?

The only thing that's really starting to worry me about drives is the damn reliability of the things.
I'm typing this on an ancient PIII machine with a 40GB HD that takes a real pounding as I've got Indexing on and then there's the daily email archiving etc, etc. Never a beep out of this ancient 40GB drive. Back decades ago I've run old Maxtor 51/4" drives way past boiling point on a Netware server when the fan died and the only way I found out the fan had died was when someone complained about the smell. It was so HOT, even after it had cooled for some time I still got burnt on the thing and yet it ran until we replaced the whole system and those disks also took a pounding with about 50 users hung off them.

So far in the last few months I've had the HDD in my DVD duplicator go down, one of the drives in a 400GB RAID 0 system I use for 4:2:2 and now the boot disk in my Vegas audio system. In that same system some of the other drives are getting a bit flaky as well.

So how do I feel about a 750 GB drive, well that's a LOT of data to loose it one hit. Five of them in RAID 5 might be OK, just buy a few spare to have on hand.

I'm almost tempted to go down the SCSI path but at over $1,000 for around 100GB that's very expensive. It's not the cost of the drives that bugs me, it's the time to re-install Windoz, find drivers, install them, reboot a million times, find the drivers are the old version and have to update. And then install all the applications, find a zillion plug ins and script etc, etc.

One day I'll make the effort to find a 'Ghost' program that works and just keep spare backup drives. The ones I use for media are usually backed up to a second drive and that's a piece of cake compared to having to rebuild a system drive.

I'll stop ranting now, reboot #12669649 has just completed :)

Bob.

Drive robustness was easier when each bit was the size of a fist....

Cooling has become increasingly important for drives. It is no exaggeration to say that this is the #1 factor for data integrity.

Can't believe you're using Windows indexing. High load, low performance.

Google Desktop is unobtrusive and finds stuff indexing won't. Couldn't live without it.

I bought Acronis TrueImage 9 after seeing it recommended here, love it!

Saw it for $25 last week at Newegg too.

RE: HT and Render Threads

fldave:
On my dual core machine if I change threads from 2 to 1, then one of my two cores goes idle and render time goes up tremendously. If you are setting threads = 1 and seeing both of your CPUs (real and HT) pegged, then this surely meaans it works differently between HT and dual-core (and I assumed wrong about your results).

Jayster
"one of my two cores goes idle and render time goes up tremendously"

Not on both of my machines. Tried it on both machines (3.2 Ghz HT, and dual cpu 1Ghz PIII). Both 1 thread and 2 threads. With 1 thread, cpu 1 went from about 40% to up to 80%, while cpu 2 was around 80%-90%, while with 2 threads, they were about even between 80% and 100%.

Render time was about the same, very little difference. Tried it with MPG2 and NTSC DV avi. This was on a 20 second timeline of generated Gradient, key framed to shift to eliptical at the end of the clip.

I'm sure it depends on what effects are used and the output format.

So many variables to deal with!

Thanks for the tip on Trueimage 9, will get a copy asap, that could really save my bacon.

Bob.

you're referincing my post but you missed my point: The new Intels will be faster then intels, but at the lower end of the conroe 2 line they're not worth it. THOSE ones aren't any better then cheaper AMD chips. The fastest ones are, but those aren't even available in mass yet and by that time AMD may have a OC'ed Opteron/FX/64 to compete.

want to know the reasons the chip's hang? because of the software. I STILL find it amazing that people blame all their problems on AMD's when I've never personally met anyone who blamed a problem on them. It's always been a bug in the software or faulty mb/memory/HD/video card, etc. Never the AMD CPU.

I used AMD's & Cyrix's back in the 486 days (to replace my intel) and they ran faster & just as stable as the intel's. Only reason I went with an intel MMX chip was because it was faster then the AMD/Cyrix equivilent & the same price.