OK.. let's do a benchmark on everyone's system... these are just my idea's that we can discuss then maybe have a new posting for the actual times/files/info? Now we have to keep it kinda reasonable for those who can't tie up thier system for hours on end. so I'm just trying for an "even ground" here between getting the most info and having it done in a reasonable time frame. 1. Size 1024 x 768 @ 72 dpi is good.. (1600 x 1200 is kinda big for some of the slower systems) 2. Render quality set to "Final" (good inbetween setting) 3. File, needs to be a standard that's on the install discs. (the Waiting Room on mine doesn't have the plants for some reason) but maybe it should have some veriety.... (thinking "tommies" the Vue title scene?) These are just my thoughts and totally open for discussion. what do you think?

I had already done some tests last night and this morning using foresters modified waiting room (which simply replaces the extra plants with some of the vue standard ones. Here are the results -Waiting Room (forester modified) -1600 x 1200 Final Render Settings (72 dpi) -Vue 4 Esprit on a Compaq Presario 2100 (1.8 ghz) AMD XP mobile Processor 512mb RAM, Windows XP Home -1 hour 44 minutes 37 seconds Waiting Room (forester modified) -1600 x 1200 Final Render Settings (72 dpi) -Vue 4 Professional on a HP workstation xw4100 (3.0 ghz) Pentium 4 Processor 512mb RAM, Hyperthreading Enabled, Windows XP Professional -1 hour 23 minutes 55 seconds I won't have time to run more tests until tommorrow at the earliest so hopefully this works.

can't use that file... don't have "kentia" file on my copy.. not on install disc or the extras disc

I didn't have it either, it didn't seem to affect the render one way or another (it's outside in the back, if it can't find it it will just render the leaves as a solid color so... won't crash anything or otherwise impact the render time).

well if the leaves are a solid color it's faster than rendering them as a texture? right? that's why I was thinkig the Tommies file cause EVERYONE has that, it's E-ons title scene... has a little of everything so that any variables will be there at least to some degree!

Message edited on: 08/18/2004 17:55

Mostly because the tommies scene renders out in 3 minutes 58 seconds on my athlon and 2 minutes 58 seconds on my p4 using the settings you suggested (1024 x 467 Final), which really isn't long enough and complex enough to gauge whether one system is better than another, which is kinduv the whole point, no offense, I just think we need a more complex scene here that really shows what the time differences are.

OK.. yup that's a little fast.. but an hour and a half on the faster machines is going to be REALLY long on the slower, making it hard foor those who have older machines to compare/ see what they may get from a new machine... maybe just kick it down a size??

Knocking it down to 800 x 600 should shave it down to about 35-40 minutes or so. That would probably be sufficent for a decent test.

well... I just did the Waiting room (forester version) 58 mins, 13 secs Vue 'd Esprit 4.22.002 (Mac OS) settings - final/1600x1200 Computer (basicly a stock unit) Mac G5 dual 1.8 gig ram.. blaa blaaaa blaaa post "system profile" for detailed specs on attached image if anyone really wants to know... Actually i'm kinda suprised it's that much faster. We did something like this in the Carrara forum and the P4's in the 2.5-3 range were right with mine if not faster. thinking Vue is set up better for Mac OS right now than it is for P4's (some of those chips/motherboards are very new on the wintel side so maybe software hasn't caught up?)

if you want to do a real test someone make a nice scene wich isn't large(in mb's) and upload it somewhere. evryone will have the same scene then

That is probably the best thing to do; and take a little time to get some things that will stress the system set up (funky lighting, volumetrics, etc).

Attached Link: http://www.geocities.co.jp/Playtown/1785/About-Vue.html

There are results of rendering "Cactus motel" in final mode, 800 x 450.

For the record folks, when I did the original Waiting Room scene (my thread is the one that started this whole shebang), I also was missing the plant textures, but I ignored that and rendered the scene without it. I simply wanted a relatively complex scene to test the rendering speed. So don't worry if you're missing the textures. If everyone wants to be onan even keel with the original reference, simply use Waiting Room without the textures at the resolution and render settings I used. If you go through the original thread, you'll see that the problems apparently lies somewhere in either the chipset and/or the Presscott based P4 and how Vue interacts with it. There maybe similar problems with the Northwood P4 core, but that's not clear at this time. nanotyrannus, what are the motherboard chipset and buss specs for those systems you worked on, do you know? You've gained roughly a 20% increase in rendering speed, which while better than the figures I've had, doesn't appear to be that great a difference considering the obvious hardware diffferences. thomllama, considering your hardware specs, I would imagine seeing something in the upper 40s at least, but you may be experiencing multiprocessor overhead. Either that, or Vue isn't all that well compiled for the new Mac architecture either. It'd be interesting to see what a dual Xeon sytem with similarly clocked CPUs would give.

Radalat, Unfortunately I'm not a hardware person so I couldn't tell you what the chipset and buss specs are (is there a properties dialog box that might tell me this?) You're right about the gain in performance being negligible as well, especially given that some of the time difference can be attributed to improved render times in Vue Pro (which is the one on the pentium 4). Unfortunately I can't load esprit on the p4 or pro on the athlon as one is my personal computer (the ahtlon) and the other is my work computer (the p4).

Unfortunately I'm not a hardware person so I couldn't tell you what the chipset and buss specs are...<< You can download the freebie app AIDA32 from the site below. Aida32 is, as the blurb says: "AIDA32 is a professional system information, diagnostics, and benchmarking program. It extracts details of all components of the PC and can display, print, or save it in various formats like HTML, CSV, or XML. For corporate users, it offers command-line switches, network audit and audit statistics, remote system information, and network management." Aida32 can create, among other formats, an HTML report of you system and what makes it tick. It will tell you more about your machine(s) than you ever knew, and of course it will give you the lowdown on what kind of CPU, CPU core, ram, ram type, chipset, graphics card...well, you get the idea. While Aida32's creator now has "a real job" and there is no more software development on it, this version is still recent enough to give you the info that you need to know about your machine. You can download a copy here: http://fileforum.betanews.com/detail/1033800563 Here's some mirror sites if the above link craps out on you (the two top links at the mirrors are dead, but others underneath are active): http://www.filemirrors.com/search.src?type=contains&file=aida32ee_393 Another great little freebie app is CPU-Z. If you ever read hardware reviews at places like Tom's Hardware, motherboards.org, etc., you've seen a screen grab of CPU-Z. While not as extensive as Aida32, CPU-Z will tell you what kind of CPU, ram, and motherboard chipset, you're running, among other things. You can download CPU-Z here: http://www.cpuid.com/cpuz.php I recommend using Aida32 so you can see what you buss speeds are as well. So download these and let us know what you're running. Enjoy.

Radalat, here are my motherboard and bus specs

Also, here is the motherboard chipset and cpu specs (this screenshot and the previous one are for the P4, it may be tommorrow before I get to posting the specs for the athlon) Thanks for the links too, very easy to use program.

OK, so it appears you may have a Northwood-based CPU. I think you can get that info nn Aida32, but if you downloaded CPU-Z it definitely tells you. After reviewing my own times, I realized that overall you're getting the same results I did. Your Athlon is a slower chip than mine, so the time difference I initially saw aren't really any different from what I got. So the song remains the same. Vue just seems to have a real problem with P4s for some reason. If not the P4 itself, it may be an issue with the 875 chipset. By the way, if you want to generate a full report on your system, click on the little fouth icon on the upper right of the Aids32 screen and follow the steps. I typically save 'em as HTML docs.

It isn't just the 875 chipset. I reported that my P4 3.0 prescott had an 875 chipset but actually it has the 865G chipset. So I think it is just the P4 chip.

I reported that my P4 3.0 prescott had an 875 chipset but actually it has the 865G chipset<< It's thin line between those two, so I wouldn't rule out the chipset yet.

Results for the AMD 64bit FX53, with 1 GB of Muskin Black Label RAM (the high-end stuff) and the ASUS A8V Deluxe Motherboard. (And the Waiting Room test I posted, in 1600 x 1200 at "Fine" grade resolution - and with the missing plants) With hyperthreading turned on (interesting that this exists for the AMD 64bit CPU isn't it?)- 1 hour, 13 minutes, 40 seconds With hyperthreading turned off - 53 minutes, 13 seconds I have not optimized this machine yet. Still building it out, and have not instituted any overclocking. Rushed to put enough parts of it together to make this benchmark. (But my Lord, this thing screams through Truespace renders, even with hyperthreading turned on! Whoa! I'm glad I kept saving my pennies and waited the 9 months to get this. A great investment!)

forester; Actually, what is happening there is that the HT goody is there for the upcoming 'Toledo' chip...the one with the dual core. The Hammer cores were designed from the get-go for dualmultiple core capability; there are Hypertransport links and at least part of the a crossbar sitting in each Hammer chip (design wise it literally is a matter of dropping another core on the pin pad, and making the needed interconnects between them). Probably what's happening is the crossbar is being invoked, finding nothing, and booting things back to the single core. But AMD definitely has a winner on its hands, doesn't it? As for the P4.... Off the top of my head, I would say one issue is the 31 stage processing pipeline. That is a lot of potential latency issues if the code isn't tweaked to perform just right on that one kind of chip. A general app like Vue that is cross platform adhere's more to standard coding methods, not the custom ones that Intel suggests.

Just call me a booby! Both the wireless and the "Asus Probe" constantly poll the system on the AMD 64 FX53 box. By turning these and Netscape off as well, you can subtract about 18-20% to the render speed for this particular machine. (Was in too big a hurry to set this up, I think - or I didn't think.)

Of course, rendering complex scenes that take over an hour give more precise measurements. But I dno't have the patience for that. Tommies seems to be OK for a decent indication. My results on Tommies (Vue4Pro 4.53-02):

P4 2.8 HTT enabled, FSB800, 1.5 Gb DDR400 Dual Channel, ATI Radeon 9600 Pro 128 Mb, Intel 865G chipset: 3'22"

Athlon XP 2700+, 1 Gb DDR333 (Kingston, fastest timings), Geforce Ti4200 128 Mb, Via KT400: 3'24"

Both systems run WinXP Pro SP2.

I think render speeds depend both on processor type (Athlon being faster than P4, due to that pipelining issue) and memory speed. My systems appear to be roughly equal in render speed (and the Athlon system is a full year older than the P4 system. Go figure!)

Message edited on: 09/11/2004 18:26