TC schrieb:
> Could you run the benchmark with 32-bit PoyRay binaries? It should not take 
> too much time to download - and less then two minutes for you to run. ;-)

Not really:

- The 32-bit beta won't install unless I uninstall the 64-bit version 
first (which, as you can guess, I'm rather reluctant to do ;-))

- Although I compile patched versions of beta.33 almost daily, and it 
wouldn't be a big deal to just run a "non-installed" 32-bit binary once 
I got it built, I never tried before on my 64-bit machine. So, for 
instance, I didn't have 32-bit boost libraries ready.

Well, you can see that I'm using past tense here, so obviously I went 
that extra mile ;-)

Note however that the results must be used with caution:

- Windows XP x64 is a native 64-bit operating system; I guess any 32-bit 
application will suffer some slowdown due to overhead for switching 
between 32-bit and 64-bit every time OS functions are called (such as 
mutexing and other multithreading related stuff, file access, etc). So 
part of the speed difference could indicate not an advantage of 64 bit, 
but a disadvantage of running a 32-bit app on a 64-bit OS.

- I found that the SSE version didn't compile in Visual Studio 2005 for 
some obscure optimization-related issue, so I had to turn off some 
optimization option. (The non-SSE version does compile "as is", but 
would of course distort the results because the 64-bit version always 
uses SSE.)

I also ran a "home-brewn" 64-bit build, to make sure I'm not comparing 
compilers (official releases are built with Visual Studio 2008) or the 
effect of some experimental patches presently in my working code.

The results I got are as follows:

32-bit:     150 s
32-bit SSE: 123 s
64-bit:      74 s

So at least it can be concluded that using the 64-bit binary /does/ give 
a significant advantage /when running a 64-bit OS anway/.

Given the impressive observed difference, I also /think/ that the 
overhead from running a 32-bit app on a 64-bit OS and/or lack of some 
optimization can't possibly account for all of it, and that a "64-64" 
system indeed gives an advantage over a "32-32" system. However, I can't 
guarantee.

Post a reply to this message

Thank you for taking the time and the trouble.



I find the results rather shocking - I would not have guessed that switching 
from 32 to 64 bit would make such a difference. I am thinking of installing 
a 64bit Linux and a 64bit PovRay and doing the benchmark again.



However, this explains the difference between this benchmark and the 
benchmarks I studied when deciding which processor to buy. On 32 bit OS the 
difference between Q9550 and i7 is significant but not too big. For daily 
work I do not need this much computing power and memory. But it is nice to 
have - experimenting with our favourite ray tracer is really fun now. Until 
last month I was still using a 32 bit Athlon 2.4 GHz processor, on which the 
benchmark took 989 seconds.



Since you seem to be intimately involved in Pov Ray beta testing, maybe you 
can answer the question if the benchmark makes extensive use of memory. 
Maybe the difference between 3 and 6 GB would account for the difference 
between 32/64 OS.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> TC schrieb:
> > Could you run the benchmark with 32-bit PoyRay binaries? It should not take
> > too much time to download - and less then two minutes for you to run. ;-)
>

I became curious about how fast my machines would run and the differences
between Windows and Linux.

1.6Ghz Celeron, 1GB, Vista Home basic(32-bit) - 12m 40s
2.2Ghz Amd x2 4200, 1GB, XP Media Edition(32-bit) - 6m 17s (12m 37s on 1 cpu).
2.33Ghz Intel Quad 8200, 6GB, Vista Home Premium (64-bit) - 1m 59s (7m 49s on 1
cpu)
2.4Ghz Intel Quad 6600, 5GB, Vista Home premium (64-bit) - 1m 58s (7m 46s on 1
cpu)

I also have Mandriva 2008(32-bit) installed on the Q6600 - 1m 55s (7M 40s on 1
cpu).


So on the same hardware 32-bit Linux is very slightly faster than 64-bit Vista.

As soon as I get another hard drive for the Q8200 I plan to install a 64-bit
Linux in a dual-boot configuration.  I doubt that the 64-bit OS would make very
much difference in performance. The additional memory would help if I had an
extremely large number of objects, other than that there wouldn' be much
difference.

Isaac

Post a reply to this message

Mike Raiford wrote:
> Impressive!
> 
> Just for fun I ran the benchmark, thinking the Phenom was one heck of a 
> processor, Rendered the benchmark in under 2 minutes?
> 
> My Core2 Quad system rendered it in 1m47 s :-D
> 
> But then, my processor is a 2.8 ghz.

Yes, I know.  Essentially, the Phenom 2 from AMD barely outpaces the 
Core 2 architecture in terms of performance, but IIRC it has better 
power usage and is generally more affordable than the Intel counterparts.

Still, it's my first quad core machine, and I'm excited about it :)

...Chambers

Post a reply to this message

Mike Raiford wrote:
> Bill Pragnell wrote:
> 
>> Has anyone ever run the benchmark on an abacus?
> 
> No, but now I have the urge to create an electromechanical computer and 
> run the benchmark on that.
> 
> 

Ah, the Babbage Computational Engine :)

...Chambers

Post a reply to this message

clipka wrote:
> So you think 10s less is a big win? Then dig this - Core i7 Quad 920, 
> 2.67 GHz (probably running at 2.8 GHz though, too):
> 
>     1m 15s !!

Yes... unfortunately, I'm on a budget, so I pretty much had to go with 
AMD :(

...Chambers

Post a reply to this message

clipka wrote:
> (I'll look up the speed next time I boot)

http://www.cpuid.com/download/cpuz/cpuz_152_setup.exe (with install, 
includes 32 & 64-bit binaries).

Click on Memory tab, for current memory performance status, and SPD tab 
for SPD memory config. variants. Your memory should be config. like one 
of the SPD variants if it isn't over-clocked.

Post a reply to this message

Are your memories in Dual Channel mode? and the speed of them? here, to 
know the memory speed: http://www.cpuid.com/download/cpuz/cpuz_152_setup.exe

Click on Memory tab, for current memory performance status, and SPD tab 
for SPD memory config. variants. Your memory should be config. like one 
of the SPD variants if it isn't over-clocked.

Post a reply to this message

Mike Raiford schrieb:
> Bill Pragnell wrote:
> 
>> Has anyone ever run the benchmark on an abacus?
> 
> No, but now I have the urge to create an electromechanical computer and 
> run the benchmark on that.

Are you aware how many relays you'll need to even store the rendered 
image? :P

Even a device able to compute a b/w image of a RSOCP would be enough of 
a challenge I guess...

Post a reply to this message

Chambers wrote:
> So I finally built a new machine, meaning I don't have to lease one any 
> more (long story).
> 
> Anyway, I'm happy with it... Phenom 2 X4 at 2.6gHz, 4gB of RAM, and a 
> 1TB hdd :)
> 
> Beta 33 renders the benchmark with 4 threads in 1m58s :) :) :)
> 
> ....Chambers
Congratulations, nice system man, I'm finally upgrading my gear so this 
makes a great reference as well as the others who posted their specs. If 
I were to freely choose my specs would be exactly what you have choosen, 
but since my budget is smaller I think I'll have to saddle for an Phenom 
X3 2.33 Ghz, 4GB DDR2-1066, SEAGATE/MAXTOR 500GB 7200 RPM SATA II HDD.

Post a reply to this message