Benchmarking - how 'bout a new round...

[paulison]

hey percy, the Dell I rendered on was still the dell format but I have disabled the extra stuff that may have been running in the taskbar and also the only application installed is viz2005 with vray.

i remember 1.46.14 was much slower than .15 so that probably doesnt help the time

I think the suggestion about the Processors not being seated properly could be a valid point. Those dells (You have the same model right?) have massive passive heatsinks that weigh quite a lot and im sure theres some potential for them to move slightly off the face of the CPU in transit

Also check if you have any network virus scanning running? I was a naughty boy and uninstalled the antivirus software from mine because im paranoid about virus scanners slowing computers down to an unusable speed. if all or more than 1 of your pc's have the same problem its most likely a software or configuration issue IMO

The only other thing I can see is that your CPU reports differently to mine. Even though they are both 2.8Ghz, possibly mine is a 533 FSB chip (northwood?) and yours a 800FSB chip (prescott?) or visa versa. Although I doubt there would be such a difference unless there is a known problem?

mine still seems unusually fast compared to the other dual 2.8 xeons but im putting this down to it being a netrender.

really really dissapointed that the AMD X2 4400+ isnt really any faster than a dual 2.8 Xeon. Can anyone else test one to see if its consistent?

Does anyone have a Pentium D 830 or 840 they can benchmark? Id be really interested to see if they are faster than the AMD X2 4400+ in vray....that would further illustrate a clear advantage to intel processors

PS: great thread, keep up the benchmarks!!!

[DaForce]

Thanks Chris.

Saves me from doing it... hhehehe
But thats quick 59.5 seconds dual xeon 3.0 was actualy a 3.6 might want to change that

paulison:
yeah im disappointed in the speed of the x2's as well.
That goes back to my previous posts about vray prefering Intels.
Fingers crossed for a AMD build
Cause the x2's beat everything but the dual core AMD's in scanline and mental ray. I would have thought it would ahve been similar in vray.

[RErender]

hehe, nice work Chris, One error though the 59.5 sec 3.0 should be a 3.6. Hate to see my errors compounding.

We should receive our 830's any day now but you'll have to wait till I get back from vacation in a week and a half.

[digitarch]

cornelltest 1h 49m 43s
dual athlon-xp 2,8 (2133 mhz)
1,5gb ram
max7
vray1.47.03

[priad]

An other test:

[DaForce]

man, talk about varying times

we have a dual amd 2800, which should be about the same as the dual pentium 2.8 give or take. but its no where near.

and we have a dual p4 3.2ghz which is far slower than it should be.

its crazy

[digitarch]

For those interested in more performance-tests here is a link to some more rendering benchmarks with scanliner:
http://www.anandtech.com/cpuchipsets...spx?i=2410&p=8

[trixian]

Tbh, this all comes down to the quirky nature of the windows installation. Most evident when booting. you can install windows (2k or xp) on the same machine 2 times in a row, and notice large discrepancies in boot time. I still haven't found out what causes this, and other friends of mine have noticed the same thing. Couple this with varying fragmentation of the discs, room temperature and such, these results confuse more than they clarify.....except in the extreme cases of the p4 2ghz and the dual opteron 275.
I guess this will be easier to compare once vray hits standalone....and if its ported to Linux. Then we can all boot on a very similar kernel, and should be much more consistent for the os part at least.

[Da_elf]

here is my try

[meanadam]

Dual 2.4 xeon = 1:28:35

What about the cache? I know they are different for every model of P4 and xeon... maybe some people have 512k vs 1mb vs 2mb? And L1 or L2?

Also - what is with the old school settings on this one? Can we do a new scene that takes about 20-30 mins?

[priad]

Also - what is with the old school settings on this one? Can we do a new scene that takes about 20-30 mins?

I'm agree with meanadam!

[percydaman]

yeah i agree its a little long. if someone wants to adjust the scene feel free at this point.

[RErender]

yeah i agree its a little long. if someone wants to adjust the scene feel free at this point.

Should probably change the wall colors or post the results in a new thread or something else it will probably get real confusing...

[meanadam]

I changed to QMC and Light Cache expecting great results..... same machine = 1:05:00. 23%.... could probably do better with more changes. Maybe that is what the next thread should be about?

[Gleb_ATL]

Intel SE7525GP2 Dual XEON 3.6 Irwindale 2GB PC2700