UPDATED BENCHMARK (ITS FASTER I PROMISE)

[Da_elf]

wait a second. an x2 is a specific model of amd? i thought x2 was just a nice short way of saying dual core

[DaForce]

hehehe
Athlon64 X2
They come in 4 flavours
4200
4400
4600
4800

I think x2 is meant to represent the times 2 cores, but thats its actual name Athlon64 X2

http://www.amd.com/us-en/Processors/..._13041,00.html

[percydaman]

you CAN get dual dual core processors if you go with the opteron flavor. Im sure they're pretty spendy though....

[neoVisual]

to Paul Oblomov

Hello Paul

Did you make another rendertest with A64 3200+ on Windows 2003 server 64 bit with correct results ? Your image from first try is much darker than reference image.

Time around 9 minutes with single 2000MHz AMD64 seems very unbelievable. It is error in your MAX / Vray settings or it is due to 64 bit server operating system ?

Could someone else try to render this scene on any AMD64 (or X2) on Windows 2003 server environment ?

thanks Deck

[Paul Oblomov]

Hiya
Well hard to explain, but don't know I can make a video or rerender to prove. And my system is really hardcore - I mean it's tweaked a lot.

Now I'm downloading again this test file...

...

So rerendered it (with a lil more higer clocks - dunno where it came from must be someone tried to OC my machine) and now it as bright as others and 11minutes long
There must be smth I can't understand...

[neoVisual]

To Paul Oblomov

Thanks Paul for a second try, it seems more clear now. Time around 11 minutes is what it has to be with 2200MHz AMD64. I don't know what you think with "hardcore" system. You have some specific memory settings or what ? Speed around 2200MHz on AMD64 is just an ordinary overclock value, nothing to be considered as "hardcore".

That problem with changing MHz settings can be due some contemporary "auto overclocking engine" - it is integrated in new nForce4 motherboards. Maybe you check your system configuration first

That darker image from your first post seem to be somewhat "uncompleted" or caused by error in rendering. Didn't you merged whole benchmark scene into some other scene with adjusted QMC settings in Vray ? That can explain visual and time difference in your benchmark.

Deck

BTW: Intel P4 Northwood 3.0@3.4GHz with HT and 1GB RAM is giving me time 9m 5s on this benchmark.

[Da_elf]

well. you only have a single processor and 1gb ram correct? thats probably why its as slow as 11 minutes

[meanadam]

Ok.... here is the thing:

The first test might have been a better indicator (sorry I whined about the time). It seems that these things are on an exponential curve. That is, the time difference actually gets smaller as the render times increase. I am sure at some point again, the difference will start to increase. So getting to the bottom of the $ / render time becomes more challenging.

Likewise all of the published benchmarks are under 5 mins.

I am currently running an animation here, this is an example:
Dual 2.4 xeon - 2gb vs single p4 3.2 w/ HT – 1 gb.
@ 20 mins a frame (dual xeon) – the p4 3.2 is 40% slower.
@ 40 mins a frame, the p4 is only 20% slower.

I suppose if we look at an average of frame times based on the median range of times we might get closer to the answer? But then also more of the other issues would come into account: bus speed, ram, etc.

I don’t know, I am just an architect…never had more than caluc 1 (believe or not) – are there some math whiz’s out there that can better figure out how to do this?

[Paul Oblomov]

Nope... nothing like "NOS" "A.I." in my ASUS A8N-E turned on.
Well on Winchester 2.2 from 2.0 is indeed ordinar, but for Venice - not.
Hardcore - I told you that it's just registry tweaks, memory timings (2-3-3-10@1T), voltmod on video card, kinda "smart" and efficient cooler solution... well nothing outstanding, but it's not that ordinary user would do

Scene was downloaded and then opened and then - rendered. Nothin' more. When you open file in max (file->open) does it merge into your previous scene or etc. ?

[RErender]

Finaly got a chance to bench one of our new dual core 830's, pretty good bang for the buck. Not as good as the X2's though.

Pentium D830
Asus P5LD2-VM i945
2 GB DDR2 pc4200
WIN XP64

[CCS]

So who's in charge of charting the results?

[n6]

1) The beast :















2) The specs :

The beast is a dual Xeon workstation originaly 2x2.4Ghz overclocked 100% stable for more than a year to 3.4 Ghz with also a FSB higher than 200Mhz x4=more than 800Mhz FSB....

1Gb of PC4000 CORSAIR RAM


4 SATA HD mounted in a RAID 5 array.















3) The results :








4) The price :

Contact me in PM if you need such a monster at half the price of a Compaq or Dell toy
(kidding)

[DaForce]

The main problem with the dual core pentiums is that they have basically are just 2 cpus stuck together, where as the x2's actually have 2 cores built into them.
So the only way the 2 cores in the pentium can talk together is via the FSB. Which is why they dont quite perform on par with the x2's.

[RErender]

It's interesting that most of our renders render a bit faster on the 830's than our older dual 2.8's but this scene really slows the 830's down, maybe it is all the reflections?

n6's rig looks like the way to go, that things fast

[davexl23]

athlon x2 4200

Hey Serge,

how are you getting a 2.2 GHz to run at 2.6???!!



Sure that is a 4200?