Arauna: Real-time Ray Tracing Benchmark - Post your Results

[Chickenfeed]

Score : 6412

[mAJORD]

wow, a train of new results hehe, I'll update shortly.

There should be absolutely no GPU usage, the bench is very sensitive to RAM Bandwidth, Cache (L3 speed), (so those uncore/CPU-NB clocks) so perhaps your other system isn't as fast there..?

[justin.kerr]

rig in sig, 24/7 settings
9056

ran it with HT off same settings otherwise, and got 8431

[bot@xs]

5349 points
aurena_01.JPG

[kl0012]

[gmat]

Whoa MajorD, you mixed me up with Pyr0:

----- #8 ---- 6460 - gmat -------------Intel Core i7 920--------- 3800Mhz -- 4/8-- DDR3-1810 8-8-8-24

should be:

----- #8 ---- 6460 - gmat -------------Intel Core i7 975EE------- 4142Mhz -- 4/8-- DDR3-2000 8-8-8-24

and i'll try to post a new bench next week after the WCG effort is over

[mAJORD]

sorry gmat, fixed up..

So many results to post this update, got a bit muddled.

If I can ask a small favor of you guys, can you please format as per OP but without all the spacer lines would be really handy if you have time..

[ Score-Nick-CPU-Mhz-Core/thread- Ram details if you want ]

much appreciated..

Couple of comments.

Rubycon.. your 2000 extra pts came from the 2 extra cores!

WOW @ Justin.Kerr . so close to 10K. congrats on a close 1st place there

and finally, thanks to Lightning man for the first Thuban results.. you can do 6k!

[Major_A]

The three computers I use the most:
Phenom X3

E5200

Core2Quad

[XII]

Nice tracer, runs smooth, load @80%.

Third run improved almost 1000 over second, that improved 500 over the first round.
Third posted.
(Improvement second on third seems to be a glitch of the memory - Got errors soon after, reseated the memory, and now 3-500 points gain first to second. More bring no real gain)

Run7 improved to 8951



Did several runs now, using same setup, different settings.
Mem-tuning seems to be as relevant in improving scores, as high clocks.

Works nice, and almost clock-equal to Linx for stability testing.
Prime95 needs 1Mhz FSB less, to run stable with the same settings.
Like it, this nifty little program

[XII]

Swapped mobo.
RexIII - sofar on par with Classy.

Again, great for quick stability testing. 1 click behind on LinX , 2 clicks behind on Prime95.

[crazydiamond]

heres an x6 result.. for lightman

click for 1900x1200

[purecain]

i must of missed something with this bench... if its ray tracing ...why is there so much aliasing... i dont get it...

doesnt ray tracing eradicate aliasing all together....???????????????

[Lightman]

i must of missed something with this bench... if its ray tracing ...why is there so much aliasing... i dont get it...

doesnt ray tracing eradicate aliasing all together....???????????????

No, it doesn't You need to supersample RT to get decent results.

[Chumbucket843]

i must of missed something with this bench... if its ray tracing ...why is there so much aliasing... i dont get it...

doesnt ray tracing eradicate aliasing all together....???????????????

no. ray tracing has the same problem because it determines visible surfaces in image precision (pixels) as opposed to object precision (primitives). from a mathematical view pixels are an x,y coordinate which are infinitesimally small but represent a finite area on a monitor. if you take one sample (a ray in this case) you are representing only one component of the information at that pixel.