Forum Coordinators: RedPhantom
Poser - OFFICIAL F.A.Q (Last Updated: 2024 Nov 09 8:30 pm)
I would agree that RDRAM isn't that great (at least for the extra cost). I got it because I bought my Pentium 4 machine around a month after P4's were launched and there was no DDR RAM option at the time and only two motherboards to choose from. If I were buying my computer today I certainly wouldn't get RDRAM.
P4 1.9Ghz, 2GB RDRAM, XP-Pro, P5SR3
time: 303 sec
One observation: If I render a second time in the same Poser session, render time will increase with more than 30 secs. Extra time seems to be spent between rendering the shadows and the actual rendering (new render window remains white). Is this specific to my system or do other people have the same experience?
Dual P4 1.7 Xeons 1 gig ram 100 meg HD, 50 free, last defragged 6 days ago (oops) Anti-Virus, Firewall, and three other apps running. 345 seconds. Just for kicks rendered again with the P4 renderer, all options on: 47 seconds. (of course the hair & reflection didn't come out). Reloaded the scene with the original render settings and replaced the dyanmic hair with Ziggy's Rockstar hair (closest thing I have to a mohawk). Render time was 225 seconds.
You can get to within a second using a watch by timing correctly. I hit the Render button at the "0" second on my watch, counting down the final ten seconds. I just counted minutes until the render neared completion. Then I counted seconds after that. So, I'd consider my time to be +/- 0.5 seconds. That's accurate enough for these magnitudes. :)
C makes it easy to shoot yourself in the
foot. C++ makes it harder, but when you do, you blow your whole leg
off.
-- Bjarne
Stroustrup
Contact Me | Kuroyume's DevelopmentZone
Well, just built a new system (the older's last choke was this test...), now I have a AMD 2600+ (real clock 2.083 Mhz), FSB333, memory bus 333 Mhz, and the old disks and peripherals and OS (with SP4).
Rendertime now is...210 secs!!!
I outperformed all the 2.8 CPUs minus one, and many P4 3.06!!!!!!!
Ok, the OS is clean now, but it isn't optimized yet; I have many tasks running, my systray is ten centimeters long :-DDD
I'm scratching my head trying to understand the means of that, I can only think about FSB and memory bus speed...
Time to try a clean render...stay tuned :)
OK, realized I could test my own theories on memory speed. I have a 100MHz AMD T-Bird, which I run at 133MHz. So I can adjust the memory bus speed from the BIOS - but only to be equal or faster than CPU speed, so I had to slow the chip back to 100MHz first. So I re-ran the tests at 3 combinations of CPU and memory bus speeds. Each was from a BIOS-level reboot, nothing else running except of course the necessary safe computing prophylactics! (The slight difference in CPU speed is closest I can get with chip multiplication factor with this BIOS.) All timing by stopwatch. 1.066GHz @ 133MHz, memory @ 133MHz, 1GB Ram, Win2K = 370 1.000GHz @ 100MHz, memory @ 133MHz, 1GB Ram, Win2K = 406 1.000GHz @ 100MHz, memory @ 100MHz, 1GB Ram, Win2K = 423 So with CPU at 100, and varying memory from 100 to 133 shows ~ 4% improvement. Whereas leaving the memory at 133 and adjusting CPU up from 100 to 133 shows a 10% improvement. So this would seem to NOT support the memory bus speed theory. Raw CPU - or FSB - speed has more than double the effect! Yeah, I know these are all slow speeds, but this should show up differences more clearly, as the increase is 33%.
My system is similar to Lucstef's XP2800+ Vs XP2600+, 2.13 GHz Vs 2.08 Ghz clock speed My system posted an average of 192 Sec vs 210 bye Luc's We both have the same FSB, assuming we both use the same amount of memory. The difference in time seems to support layingback's theory. JV
Software: Daz Studio 4.15, Photoshop CC, Zbrush 2022, Blender 3.3, Silo 2.3, Filter Forge 4. Marvelous Designer 7
Hardware: self built Intel Core i7 8086K, 64GB RAM, RTX 3090 .
"If you spend too much time arguing about software, you're spending too little time creating art!" ~ SomeSmartAss
"A critic is a legless man who teaches running." ~ Channing Pollock
I fooled with some settings on the file: Changed light properties from "depth shadow map" to "ray tracing" Changed bucket size from 32 to 256, changed texture filtering to 1.000, and enabled "remove backfacing polygons" and changed the texture size to 2048. p4 2.4 gh 1.5 gb ddr ram (266 mhz) win xp home render time went from 258 to 218 seconds
Yes, but bear in mind once you guys start changing things in Poser you aren't testing your computers any more, you are testing changes in Poser. wimvdb- I noticed that too, unlike Poser 4 which tends to get slightly faster on the second render. Does P5 access the drive at all during the render? (I can't tell as my HD light isn't connected yet.) I also noticed with the big bucket sizes P5 sometimes just sits there after the render is done, with the "rendering.." box still open, sometimes for 30 seconds or so. Sometimes it seems if you wiggle the mouse it will pop it out of this state. Weird.
Jim: The slowing down of the second render does not seem to have anything to do with drive access. It looks like it is doing some cleanup of the previous render before it actually opens the new render window. This is directly after loading textures, object and calculating shadows: it just sits there for 20-30 seconds and then opens the new render window. This might account for some of the differences in the timing for equivalent machines.
It is STILL a good idea, igohigh, expecially when we can compare scores from very different specs (and if I recall correctly - just awoke... - there's no one spec similar to another).
Look at the last info from mountainmaster: he has the best specs stated until now, and has even one of the best score, so NOW I can say "CPU power AND memory bus speed does matter, and REAL Mhz among processors does matter".
Look, a 3.06 P4 can easily overrule all the AMDs, but AMDs are running at MUCH LESSER Mhz counts, and don't get so low scores.
Example: a 2.2 real Mhz AMD takes 192 secs to render this scene, when a P4 with 30% more Mhz takes 176 secs.
Is it worth the way more money spent on the P4, only to have 16 seconds less in this render???
Yes if you don't want only render power, no if you're a simple hobbyist who fire up Poser and play with DAZ's or Renderosity's toys, dreaming to be a photographer :-)
Well, this is only one of the miriads of conclusions that can be got.
But as we are talking of pure render speed, there's one other thing to keep in mind: the time for the final render is a small fraction of the time spent overall on the picture, unless you are an animator and you regularly make one hour lenght films (but I think at this time you aren't using a P4 and Poser, aren't you? ;-) ).
So, if you want render speed AND workflow, the 16 seconds for the first could be HOURS for the second...now it's up to you to decide how many budget to be allocated for your needs.
And this thread had convinced me that my last buy was a REAL bargain :-)))
CPU Type & speed (Overclock speed) , RAM amount, Operating system, RAM speed, AGP speed P4 2.4GHz (2.4GHz), 1.5GB, Windows 2000 Pro (patched with all latest patches except SR4), DDR (266), 4X (card maximum) 226 seconds to completion (rendering box vanished and render window has flashed white then back to render). Also running... Foreground apps: Outlook Express, Internet Exlorer (2 Renderosity pages), Windows explorer, Calculator (need math help, brain hurts) and CPU test readme. Background apps (that show up in ashtray): Sound Effects (default sound card utility) Atmoic Timesync MaxMem EXPERTool (default video card utility) CleanSweep Smart Sweep Cleansweep Internet Sweep Norton Internet Security (Enabled, natch) Norton Antivirus Local Area Connction status Clock (naturally!) A great deal of time was spent creating shadow maps - I can't help but think that faster RAM would have an impact.
igohigh- Gee, I was thinkinging the exact same thing! I'd like to complile all the results, only now I'm afraid now that some of them are tweeked, and not marked as such in the thread. I think, perhaps the results sould be averaged for each CPU anyway, maybe CPU and memory size (maybe "low" - less then 512, medium = 512Mb, high 1GB or more). Something like: P4 2.8Ghz High memory, average of 4 - 215 seconds P4 2.8GHz Med Memory, average of 2 - 221 seconds Any volunteers? ;-) I going to assume that stuff like 400 Hhz dual ported, DDR, Raid 0 and all that stuff doesn't really matter as a listed item, as "new" CPUs have it, old ones don't, nobody is building new P4 1Ghz machines, for example. So what I'm saying is while this stuff effects times, it is going to be reflected in the CPU it goes with, anyway.
Jim - actualyl some CPU's will work with a variety of RAM speeds... it still might be worth noting them, perhaps. BTW obviously I didn't tweak MY setting for rendering - tho first 2 renders were interrupted by telephone calls so I only timed 3rd one correctly (and that's the timing I gave - which I've been reasonably proud of given all the otehr stuff I had running).
Jim, for your info, the 176 seconds from my previous post were straight out of the box. With the bucketsize tweaked to 128 I get: Pentium 4 3Ghz, FSB800, 1GB dual channel DDR400 SDRAM, WinXP Home: 147 seconds Wow! I love my new computer! :) Sorry, I know this is not a contest but just could not resist ;)
Attached Link: http://www.tomshardware.com/motherboard/20030812/index.html
Jim, I think it's starting to look like OS, CPU (speed/make) and then FSB (front side bus) speed, in that order - with 2 caveats: The overriding caveat is that any memory capacity of less than 512MB is going to put you in a hole no matter what. And the second caveat is that you have to have the memory speed/type to make use of that FSB speed. FSB speed might almost give you more gains than CPU speed, but of course any system with really fast FSB is going to be a new system with a very new and thus relatively fast CPU. Anyone got a P4 2.4 or 2.6 overclocked to 3/3.25 on a 1GHz FSB yet? Seems to be a sweet-spot configuration (see link above), but hard to do with a P4 3.0, as it'd be pushing 4GHz!FSB must have much to do with it. Mine is only 533MHz FSB (with 2.66GHz cpu), giving 209 secs. mountainmaster, on the other hand, has 800MHz FSB (with 3GHz cpu) which cuts it down to 176 secs. His cpu/FSB combo gives about 5% (84% cpu ratio as compared 89% render time ratio) increase over cpu only. Hmmmm... Makes me wish I could have afforded a dual Xeon mobo with 800MHz FSB. :) Finally, my numbers are from the time I clicked Render until the actual image was displayed - no changing of Poser settings. I did try increasing Poser's priority, killing background apps, and so on, but there was no increase in performance.
C makes it easy to shoot yourself in the
foot. C++ makes it harder, but when you do, you blow your whole leg
off.
-- Bjarne
Stroustrup
Contact Me | Kuroyume's DevelopmentZone
layingback- Yeah, I read that too, after I ordered the memory for my new computer (naturally!) I should have got faster memory, I've had a couple of lockups so I'm going to have to try standard speed again. Corsair (and maybe others) makes memory suitable for a 1000 Mhz FSB, my 2.8 would be running 3.5 at 25% overclocking, which might just fly.
Wish I could optimize the memory in my head... Anyway, my initial results were: P4 2.8(3.09)Ghz 1GB Ram Win2K 209 seconds I got a couple of lookups in Max, though, so I ended the 10% overclocking. However, I did play around with my DRAM settings, it turned out they were (in "auto") 4,4,8 CL 2.5, I changed that to 3,3,6 same CL, same burst length (8). New time is: P4 2.8Ghz 1GB Ram Win 2K 194 seconds, an 8% improvement that didn't cost anything. Hopefully, this will be stable (2,2,6 CL2 would be "radical", maybe I'll try that later if this seems O.K. after a week or so. Now to make some kind of a list from all these results!
Apple G4 (Silver Bullet for those who care) 867 MHz PPC, 1.5 G RAM, OS X 10.2.6 Poser 5: 980 seconds. this sucked. of course, my machine doesn't break the Gig barrier, BUT... when i increased the block size from whatever you had it at (i think it was 30) to 500, my render time improved to 654 seconds. not GREAT, but closer to reality.
Hey all :)
I thought I would put these times here, just as an FYI. I am running the CPU test on the new box we picked up to evaluate from some guys we know @ Maingear. I mention this because system speed has a lot of factors other than pure CPU.
Oh, and if this is one PC result to many, chalk it up to pride. Not only is this thing fast but it glows blue :)
Time:
Default Render time: 3:19 - that translates into 199 seconds - (avg of multiple runs)
Bucket size 128 Rendertime: 2:59 - that translates to 179 seconds (avg of multiple runs)
System:
CPU: P4 2.80G with HT
Memory: 1.0 gig "corsair" memory
OS: Windows XP, SP1, all updates applied. NO SPECIAL settings used. I run my system with all the UI "eye candy" turned on.
Resolution: 1600x1200x32 bits
Notes:
The system was not particularly clean, having been up for more than a week so there is some stuff running in the background.
Before hitting the render button, the system was using 388 meg of ram, during render it never paged to disk.
Poser only used 50% of my CPU during this render, as it is not HT aware. In effect I have a dual CPU system here and Poser couldn't use one of them. This may explain why background tasks and UI had no impact on speed.
Attached Link: http://www.keindesign.de/stefan
For that and just about any other question on P5 Renderer see www.keindesign.de/stefan.Attached Link: http://www.renderosity.com/messages.ez?ForumID=10139&Form.ShowMessage=2077843
update....i built an 'optimal poser rig, Jan 05, on advice of excellent Renderosity thread linked above.
Benchmark 159 seconds.
::::: Opera :::::
Almost the same rig as Opera: - Athlon64 3500+ - 4 GB Transcend dual channel @166 Mhz (stability issues...) - MSI K8N Neo2 (nForce3 Platinum chipset) - 2x 73 GB WD Raptor RAID 0, 2x 160 GB Hitachi ATA133 RAID0 - GeForce6800LE 157 secs (+/- 2) Steven.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
Tweaked my memory settings, now running at 200 Mhz: 148 secs (+/-2) Steven.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
This site uses cookies to deliver the best experience. Our own cookies make user accounts and other features possible. Third-party cookies are used to display relevant ads and to analyze how Renderosity is used. By using our site, you acknowledge that you have read and understood our Terms of Service, including our Cookie Policy and our Privacy Policy.
p4 2.4 w/1ghz pc333 2700ddr (2 512's) win2k pro sr4 p5sr3 with a whole bunch of stuff in the bg... kerio, aim, icq, printer,mouse,norton,soundmanager,freeRAM. time: 227seconds