MY guess is that Intel has better burst speeds and thats why it wins the short "veg test" - but AMD's are work horses and power thru anything you through at it, at a constant speed where as the intels start to bog down in long renders...

Just my theory
I'm running dual Athlon MP2000+

I have a computer at work with a P4 2.4Ghz and 1GB of Ram and at home I have a Athlon XP 2500+ and 1GB of RAM, they are basically identical systems except one is Pentium and one is AMD and I have always thought the AMD renders faster althought I have never done an actual test with identical project files.

Without going into lots of details AMD and Intel chips handle some tasks differently. The results you get with the "official" render test file on SPOT's site are for sure much longer than "real world" rendering. Because it is easy to get at and uses only what's build into Vegas, no external source materials, its a "fair" test to compare apples to apples, not to compare apples to oranges.

A more practical test would be to have a few seconds of DV source, probably AVI then give some specific insructions how to stretch that source to a longer span on the timeline and apply some specific filters, etc., then compare to that.

Speaking of render times, I currently have an AMD 1.2 gig machine and was thinking what would be the gain in render times if I were to upgrade to the latest AMD chip? Would the performance justify the upgrade? Thanks!

I upgraded from a Athlon XP 1400 to a Athlon XP 2500+ and my render times are a LOT faster, it was definately worth the upgrade, but I had to get a new MoBo too since the 2500 is a 333Mhz FSB I also added an extra 256MB of RAM as well.

As Spock would say: "Fascinating..."

Seriously, if you can document the fact that the machines are virtually identical with the exception of the processor chips, I think AMD would be very interested to hear from you. Right now they are getting hammered by Intel on all the video benchmark tests. Such tests typically measure the time it takes to encode an avi to MPEG2 or Divx. If it can be shown that "real world" video processing is an entirely different kettle of fish, it would be major news.

For that matter, I think Tom's Hardware Guide (www.tomshardware.com), PC World, and PC Mag would make this a feature story, since it goes contrary to the established Conventional Wisdom.

I have always been fond of AMD, primarily because they are the underdog and as such have been responsible for accelerating the pace of development as well as reducing the prices for CPUs. Can anyone seriously doubt that Intel would be charging $1k for a 1GHz processor if AMD hadn't been around to poke them in the eye with the Athlon a couple of years ago?

ReneH: I would guess that going to the faster chip, along with getting faster memory and a faster channel to that memory would reduce your render times by 25 to 33%. In other words, if something rendered in 10 minutes before, it would take about 7 minutes with the new hardware. This is just a guess based on what I've read here and elsewhere.

Would love to see a proper comparison by tomshardware or similar - maybe using various softwares, all with video editing in mind. Be nice to see how Prem does on the same hardware compared to vegas etc

I am in the market for a new PC so i will be watching this very closely - i wasnt even prepared to think about getting AMD, but this does put a new spin on it.

Reading spec sheets the Intel is more attractive, but not if there will be no real advantage.

Then again ANYTHING will be better than my current P866 for rendering 30 - 60min of video :(

Some one else tells me that with a good fan these 2500+ chips can be overclocked and HAPPILY run at 3200+ 24/7

I've had a stable overclocked system for a couple months or so now all with the regular fan and heatsink that came with the P4. Took a P4c 2.8 and I'm getting 3.360 Ghz out of it. As far as overheating, if its going to, it probably will quickly, like within ten minutes or so under load, surely within a half hour.

I take a more overall approach. It isn't just the CPU throwing off heat, though that's the main problem. Also a graphics card with lots of memory on it, the RAM itself, if you got a bunch of hard drives and probably the most overlooked, weak or too few case fans. It does nothing to to buy a good heatsink fan and run it at high RPM for the CPU if all you're doing is recircualating the hot air inside the case. You got to get rid of it and multiple case fans help. I have one far to suck cool air in, and three to exhaust out the hot air, two in back, one in the side panel. Also for a few dollars more I got my memory with a built-in copper heatsink.

I've always why noone every did the "real" real world test: Boss's and people aways ask me if I can build them a computer. i answer by asking them how much they want to spend. I ask them that because prices between parts are extreamily different. Example: I can get a 1ghz P3 chip of a pricewatch site for $100. i can get an AMD XP 1800 and motherboard for $100-110. the AMD is, of course, faster.

So... has anyone ever done a review of AMD vs Intel with a price limit? Say $2,000? so, you get all you can for your AMD for 2g's, and all you can for your Inter for 2g's. go for Rambus 800 or DDR 400? Big price difference. Duel or single? Big price difference. What good is a Duel P4 3.6ghz if it costs twice as much as the a duel AMD MP 3000? Is the extra grand worth the 3-4 seconds fasterrender time? Is a duel processor worth the extra $ if you can buy another AMD box for the same price and have 2 people working at once?

Anyone think the same way as me?

Also, on the straight Intel vs AMD thing, the only thing I've seen Inter have the advantage at is that with simular chips (AMD 3000 vx intel 3.0ghz) the inter multitask's better (from a www.motherboards.org review of those 2 chips).

Check out this URL with a comparison of CPUs doing Video Rendering

http://www6.tomshardware.com/cpu/20030217/cpu_charts-26.html

When I upgraded from my Athlon XP 2000 with a GIG of ram to an Intel P4 3Ghz (actually I still have the whole athlon system, did not replace parts) I ran a benchmark between the two on a 5 minute project I was working on at the time, well the P4 won buy quite a margin of about 10 minutes faster the p4 took like 27 minutes and the athlon took around 36 minutes. What I did notice was that when the p4 was doing simple stuff during the render is where it really took off when it was just creating new video frames not having to render any FX it would really smoke, but when it came to heavy parts of the video chroma keying and real heavy compositing it would bog down to just a little faster than the speed my athlon was rendering the frames at in those sections. Then all of the sudden it would just start cranking them back out during the easy parts. The athlon pretty much kept chugging along at a pretty regular clip where the P4 would have dramatically different rendering speeds. My guess is that the P4 can really fly when it does some of the basic stuff (.ie copying video frames) and does slower when the heavy stuff comes along. Where as the athlon has a better IPC and a stronger FPU. The P4 has a longer pipeline which is great when you have semi predictable results but when it mis-predicts it stalls the longer pipeline which can slow it down. That's why there are tons of P4 optimizations that can help predict results to a large degree but never nearly 100% of the time. I think the P4 is great in video processing but if you start throwing lots of variables at it can cause it to bog down, whereas the Athlon with it's great FPU is a little more consistent in most areas and why allot of university's use them for serious number crunching with out waiting for or having to recode their algorithms. Just my 2 cents though.

Isn't the Athlon a RISC chip too? I know that the AMD 386 to K2 were RISC based with a converter.
I just know I can afford an Athlon and not an Intel. :)

No basically it is identical to an Intel in that it is a CISC based chip or X86 instruction set. I agree that the athlon is more cost effective but the Athlon 64 FX will change that price difference some initially.

I spent about a half hour on the phone yesterday morning with the venerable Vegas guru, Douglas Spotted Eagle. I guess he developed and wrote the render test. I asked him his opinion as to why I would be seeing the results I am, and he explained to me that unless you've got identical bios chips and identical motherboards except for the chip, any comparison between AMD and Intel CPUs is going to be inaccurate.

I asked him why my "real world" test would show such different results than the render test, and he said my definition of real world is going to be different from someone else's. I agree that perhaps not everyone who uses Vegas is going to be doing the exact same kind of work I am, but I at least hoped he would give me a clear answer either way.

He assured me that if I were to "optimize my Pentium 4 machine for video" that it would beat the AMD machine in every test. I don't really understand how this could be, since I haven't "optimized" the AMD machine at all. I did a base install of WinXP Pro on both machines, then installed Vegas 4.0 and updated to the most recent versions. If a P4 is going to not be optimized for Vegas out of the box, shouldn't we as end users have some kind of instruction from Sonicfoundry as to how to optimize it?

He also told me that since Vegas was developed on Intel, it's going to be more stable on Intel, and that AMD chips have "serious issues" when it comes to multimedia. Does anyone know what exactly he might be referring to? The argument sounded suspiciously similar to the old "Macs are better for video than PCs" argument, with which I take serious issue. Is there any truth to this claim that AMDs have problems with multimedia? I've been developing videos on my Athlon system for a few months and have had absoultely no problems. What is he referring to?

Obviously my test must not be testing the same things as his, which would explain the different results, but since the majority of our customers are going to be doing the kind of work I tested for, it looks like we'll save a few thou' and go with Athlons for the lab. Thanks for all your input and suggestions. I still haven't come to a firm conclusion as to why the results were so different, but if somebody as well-versed in Vegas as Douglas Spotted Eagle appears to be couldn't give me a clear explanation, I doubt anybody could.

I hope all my rambling helps other people looking for an answer to this age-old question. My suggestion: find an opportunity to test out both platforms with the kind of work you'll be doing and see how they perform for you. There must be vastly divergent tasks associated with Vegas, some of which the Athlon excels at, and some of which are dominated by Intel. Until you see for yourself, you may not know which is better for you. Good luck everyone!

DSE (Douglas Spotted Eagle) is a smart guy with a lot of experience but no one is correct 100% of the time, and since I know of lots of AMD machines that do very well running Vegas and other "multimedia" apps, I would suggest that perhaps he had a bad experience with AMD years ago, and that event continues to affect his opinions.

It is certainly true that a programmer can modify an application to speed it up on a particular platform. The Pentium 4 is an unusual chip in certain respects, and I have heard of instances where a modified program runs much faster on it than before.

I haven't downloaded the DSE render test to analyze what it does, but it sounds like possibly you are making more use of memory bandwidth in the DSE test and more FPU performance in the home made test. Does that sound possible to you?

According to your specs the Intel memory was running at 400MHz dual channel and the AMD was running at 333MHz.

Also I'm not sure what Douglas Spotted Eagle would be talking about as far as the AMD being bad at multimedia. From what I've always understood the AMD processors are the best game machines period. At least that was until the 800MHz FSB came out on the P4s. I don't know if anything changed because of that.

But, the reason AMD was considered a better gaming CPU was the it could handle more floating point operation than the Intel CPUs. So while Intel CPUs are the best for high data transfer, the AMDs are best with intensive calculations.

JBJones

That's partly the rub, there's so many variables at play it's extremely hard to compare and a comparison that holds true today can get blown away tomorrow.

Some time ago I was reading through what the author of VitualDub had to say about hyperthreading. He'd been using the uStuff compiler and didn't think much of Ht until Intel gave him a copy of their compiler. Things suddenly ran so much faster he produced two versions of the code, something that no progammer would do lightly.

So today the AMD chips could run VV renders faster, SoFo recompiles their code with a different compiler and the AMDs run at half the speed of the Intel chips.

I'd suggest seeing whose logo better matches the color scheme of the PC's case that your going to use :)

But joking aside how you do things in VV will have far more impact than anything else. Yesterday I did some work on video off 8mm film, I just had the raw tape from the telecine and took out all the crap and put black between each roll to backup to tape. Not really thinking and not having that much anyway I put multiple instances of the same generated media (the black) on the timeline. VV renders each one separately. If I'd rendered out the black to an AVI and put multiple instances of that on the timeline render time would have dramatically reduced.

Well, we're right back to apples to apples and apples to oranges.

SPOT's render test uses nothing but Vegas internally generated media. That makes it fair in the sense you can't fudge the results because of differences in your video source making it easier or harder to render depending on what you use. Of course the problem in that kind of testing is it isn't something you run across that often. Its like rendering a TV test pattern. Its useful to set up your TV, but not something you want to stare at for hours on end and therfore the results you get with it are only relative to what someone else gets rendering it not as a benchmark for other "real world" projects will render, which will be MUCH faster generally.

A more realastic test would be to render something truly real world. What someone needs to do (if they want) is volunteer a SHORT video, maybe 10 seconds worth or so that's neither static nor has too much motion. More truly real word. Then everyone can download that and see how it compares.

The issue is going to be for it to be a good test it should be digital quality and even a 10 second clip is going to be fairly big and whoever hosts the file is going to take a hit in useage. So before anyone steps up to the plate keep in mind we're talking maybe a file that's in the range of 500K at least and maybe several hundred people or more will download it. That may result in a surprise billing from your ISP for excess bandwidth to the tune of $$$$. So... seriously think about it.

As far as AMD to Intel its the never ending leapfrog thing. Right now Intel is a faster. For awhile AMD was. They will no doubt be faster again, then Intel will jump over them again.

There's more involved than just the CPU speed.

FSB (front side bus) or how fast the CPU is capable of grabbing data out of memory is a issue too. I seen that with my latest P4c that's over clocked so the FSB frequency is about 1100. That helps a good deal and helps more the bigger the project is.

Also the size of the Paging File and how often Windows swaps data in and out of it. Something pretty much out of your control beyond simply making it bigger.

What version of Windows you're using. Older versions of Windows especially can have their system resources depleted much faster because the heap size in two critical areas is so small. Much improved in XP.

The file system and if or not how badly your drive is fragemented, if you're using FAT 32 or NTFS all impact some, when you're rendering a longer project that's going to take hours.

What else is going on while you're rendering. If you haven't already, kill those screen savers. Especially the animated variety. The problem there isn't a problem when you're working and actively banging away on the keyboard. But if you start a render, go to bed and one of those fancy screen savers kicks in, it will be competing with Vegas all night long for CPU cycles maybe doubling the time it takes to render!

A friend of mine interned at Intel a couple years ago. After working there for the summer his pro-intel attitude fades somewhat. He said that the chips were almost identical in power, and the intel's high price is all advertising $$$, not much else. I've been using an AMD for 2 months with no problems with Vegas. Also, if the AMD XP is streamlines for Windows XP, woldn't it be better then Intel for XP anyway?

BillBoy,

I'd agree about screensavers but one thing that no one has asked SoFo for and I'm surprised is a 'Shutdown when render complete' button.

TMPGEnc has it and its mighty handy.