I agree with that statement. I made many rendering tests lastly and , basically, CPU speed rules entirely the rendering time.

XP 1800 render 2x faster than Duron 900, 2,8 Gh render 1,6 x faster than 1,8, etc...

This is something I'm really interested in, too.

When you say "CPU speed rules entirely," do you mean Quantispeed (or whatever fancy term AMD has, i.e., the 1800, 2700 designation, etc.) or the clock speed? For example, my Athlon is designated a "2700," but the clock speed is 2.167 Ghz. In contrast, some of the "XP 3000"s have a clock speed of 2.0 Ghz.

Also, does your analysis take into account hyperthreading on P4 chips? How much does this help?

Brian, I was under the impression that if the app is not written for hyperthreading, it won't make any difference.

Anyone else know for certain... one way or the other?

J--

http://www.tomshardware.com/cpu/20040322/prescott-13.html

Toms Hardware has a Main Concept MPEG rendering benchmark. Pretty much clock speed.

Gordy, get the AMD XP. For anything else you''ll need a new MB, which means you might need to upgrade somrthing else (ie video card, or loose the ability to use something built into your current MB). You will need faster DDR RAM to (333) to use to use the faster CPU though (that's if you have something lower then 333 right now).

Plus, AMD is actuatly the bus speed x 2 (133mhz on each clock cycle, or something, so it's 266, etc). and the P4's are 4x (800 is 200x4, 533 is 133x 4 etc, same thing as AMD). I'd say that any bus speed above 166 isn't true now (but still, the multipliing DOES help!) :)

I thought Vegas was one of those programs that CAN take advantage of hyperthreading.

AMD are right in their designations.

Last time i compared, an XP1770+ was about exactly on par with a PIV 1,7, and have a proportionnal rendering time to my PIV 2,8.

Keep in mind that we talk exclusively about rendering porformance here.

CPU speed is important. So is the buss speed. So too memory timing. Of the three raw CPU speed is probably the most important. However if the FSB (front side BUSS) isn't high then you'll never get the horsepower your CPU could otherwise deliver. The higher the FSB setting the faster your CPU can access memory. The fast that happens the less time it takes to do any task, including rendering videos.

The Intel/AMD debate has been going on for years. Both make good chips. My advice spend a few hours reviewing the various chips/boards you're considering to see reviews on OVERALL perforamance that really determines how "fast" your computer is.

The issue of faster memory is also imporant. Everything else being equal faster memory will squeeze still more out of your system then slower memory will.

Example: I have a Intel 4 HT on a ASUS P4P 800 board. I used rather ordinary 3200 memory. If I could have found 3700 memory when I bult the system, then I could have overclocked another 10-15% easily.

I further suggest you pick up a free copy of CPU-Z. It will give you a snapshot of how well things are running under the hood, including raw CPU speed and the multipler value. So if you have a multipler of 14 and a FSB set to 220 that yields a CPU clock frequency of 3.080 Ghz. By further cranking up the FSB and/or using faster memory a FSB of 240 would bump up CPU speed to 3.36 and so on.

Buying the fastest clock speed chip you can find is not the smartest approach. It is for sure the most expensive. Buying a somewhat slower rated chip and overclocking can yield a faster chip at far less cost. Its a dirty little secret that the chip makers don't tell you. There is no such thing as a 3.0 or a 2.4 CPU. There are only chip famlies. Any chip within a family will run from X to Y speed within design specs. When they are tested, its much like grading eggs, like the biggest eggs command a higher price, they're still just eggs. Ditto for CPU chips. The ones that test the fastest without failing get labeled at a higher frequency and command the highest price.

You with a little knowledge can take a slower chip and make it perform like a faster one and save a hundred bucks or more. That's what overclocking is all about.

The hype with Intel chips is the so-caled hyperthreading. It just means it can hand off some tasks to a sub part of the chip while the main core does something else. The rub is this may or may not increase performance dpending on what tast you're asking of the chip. To further muddy the waters a lot of other numbers are out there. In additon to BUSS speed there is bandwidth. In the Intel 4 cases that's a factor of 4. So a FSB of 220 really means the potential bandwidth is a factor of 4 times that or 880. But again that's misleading and should not be considered a 4 to 1 advantage of chips that use somewhat different means.

The issue here is some chips and motherboards/BIOS combinations are more forgving then others. With some you can really crank thinks up with overclocking. Others you can't easily or at all.

That's why if speed is important, do your homework first.

Thanks for the replies. One more question. I have 512MB of Kingston Hyper-X. Will another 512MB do me any good or have any performance gains (most specifically with rendering)?

To BillyBoy. I've had my system overclocked in the past and had many problems with the machine shutting down for no reason. I've since put it back to normal specs. I bought the AMD 2100+ Thoroughbred "Revision B" for the specific intent of overclocking it because it was supposed to overclock well. Didn't work how I planned. Maybe because I'm using onboard graphics which is the which is the Nvidia Geforce 4 MX440.

I OC's my CPU a couple months ago (November i think). Ran great! I only beafed it up about 800mhz, but some things ran faster. Then I get a new 8x AGP card. Well, my OCing didn't work anymore. Needed to mess with the AGP voltage.

So many things can go wrong with OCing... I don't have any spare cash to risk it right now.

Oh, i'll buy your AMD XP 2100 for $18 if ya want someone to take it off your hands. I have an 1800. :)

As far as overclocking how successful you are depends on the chip, motherboard and BIOS. Some take to it like a duck to water, others seem unstable. If that happens its usually the memory timing/voltage settings. A minor tweak there can make things stable. Why you need to do research on the particular chip/board/memory you want. In other words let some other guy do the grunt work and sweat the details. There are hundreds of such web sites. That's their hobby and they go to great pains to test various setups.

How exactly am I supposed to read and interpret the MainConcept 1.4.1 chart on Tom's Hardware page? I see that's it's a 1.2GB file they are rendering from DV to MPEG2. But how do I do the math for say a half hour or hour long program?