OTOY Forums

hello.
i'm Impatiently waiting for open-CL version of Octane render.
My Wait is 2 years old!
when we can use Octane render in our system's?
Cpu : 3770K
Ram : 16GB
VGA : 2 * AMD 7970

open-CL is now a Huge and powerfull program language and it's not immature or in the start of way.
you can see the www.fireuser.com for this Claim.
this corporations presents the open CL Version of their Softwares :
Vray 3.0
Luxmark 3.0
SOLIDWORKS 2015
PTC Creo 3.0
All Adobe Software (Premiere Pro CC 2014,PhotoShop CC 2014,...)
OpenCL-based Bullet physics engine For Maya 2015
Maya 2015
3dMax 2015
ESI Visual-Environment 10 CAE / CFD
Simulia Abaqus and Siemens NX Nastran 9.1
CATIA V6
Ventuz
Siemens PLM
DaVinci Resolve 11
SCRATCH 8
NUKE 9
Modo
Mari
RealVNC
ESRI CityEngine
Eyeon Fusion 7
Lumiscaphe P3D
NVU Systems
and so on...
we will happy if you tenderness us and make an Open-CL vesion of your Product.
we hope to you.
at the end i thenk you are workink secure with amd???!!! Because :
http://fireuser.com/blog/the_future_of_ ... d_via_the_
We LOVE Octane Render.

Thank's a lot
Ya Heidar(A)

1) Octane 3 will support OpenCL.
2) A number of those programs you listed don't actually render using OpenCL.
3) This thread is about OctaneBench.

Hi there,

now my Rig is complete again and I did a new benchmark.
I read something about there is a 8 GPU Limit in Windows? Maybe 8 physical GPU Cards?
I am running 6 cards with 9 GPUs at the moment and it works fine.

Pretty decent results in terms of costs / performance and a almost 3 year old Rig (besides the GTX780ti), especially if 1.5 MB GPU Ram is enough for your work and you can get your hands on used GTX 590s now for around 200€. You can still build a high end Rig for a small budget with these cards today.
I can`t believe my Rig is almost as fast as 6xGTX Titans + 1xGTX Titan Black...

Octane version:
2.17
Operating system:
Windows 7 64-bit
Devices:
GeForce GTX 780 Ti
GeForce GTX 580
GeForce GTX 580
GeForce GTX 590
GeForce GTX 590
GeForce GTX 590
GeForce GTX 590
GeForce GTX 590
GeForce GTX 590


Scene, Kernel, Ms/s, GTX980 Ms/s, Ratio, Weight, Score
Idea (by Julio Cayetaño), info channels, 458.700628, 81.25, 5.646, 10, 18.818487
Idea (by Julio Cayetaño), direct lighting, 105.905251, 18.64, 5.682, 35, 66.285476
Idea (by Julio Cayetaño), path tracing, 93.376583, 15.99, 5.840, 35, 68.129673
Idea (by Julio Cayetaño), PMC, 14.603754, 2.88, 5.071, 20, 33.804986
ATV (by Jürgen Aleksejev), info channels, 179.767826, 32.48, 5.535, 10, 18.449079
ATV (by Jürgen Aleksejev), direct lighting, 69.073768, 12.40, 5.570, 35, 64.988761
ATV (by Jürgen Aleksejev), path tracing, 56.334364, 10.02, 5.622, 35, 65.592241
ATV (by Jürgen Aleksejev), PMC, 20.755434, 3.77, 5.505, 20, 36.702800
Box (by Enrico Cerica), info channels, 311.139482, 59.18, 5.258, 10, 17.525036
Box (by Enrico Cerica), direct lighting, 64.113927, 11.76, 5.452, 35, 63.605086
Box (by Enrico Cerica), path tracing, 65.154982, 11.74, 5.550, 35, 64.747995
Box (by Enrico Cerica), PMC, 18.150535, 3.83, 4.739, 20, 31.593620


Total score
550.243238

I hope we will see a really speed boost with the new Pascal Generation NVIDIA just revealed. But maybe we will have to wait 2 more years for that?

Hello guys.
Just made a new run.
My 4 gtx590(8gpus)made 414...nice result.

But as the default bench page shows the average tests from all users the other 2 results are not mine (lower), so the average score is 390.

I think the default page should be the maximum score, and then if we want we would change page to see average and minimum. My cards are running on air, and in default settings, no OC...so its strange to see lower scores then mine in the same cards.

4 gtx590 (8gpus) is 15th on the list

My Gpu´s total cost +/- 700€ .... the same as 1 GTX780ti....raw power/price/performance...gtx590 is still a beast.

cheers
resmas

resmas wrote:My Gpu´s total cost +/- 700€ .... the same as 1 GTX780ti....raw power/price/performance...gtx590 is still a beast.

cheers
resmas

Hi, friend.

GTX 780 Ti "fixes" optimized as Fermi but not total repair.

I have new EVGA GTX 780 Ti Classified but I have paid only 400 euros despite new card - very luck!

Can't wait for OpenCL and build a mini-cluster of Radeons! I started losing faith in Nvidia when the Kepler based 680 turned out to be slower in CUDA apps than a Fermi based 580, which was a compute beast. AMD seems to offer a lot more raw compute performance per dollar than Nvidia. Once Octane goes OpenCL, GPU rendering will really be available to the masses and go mainstream .

Is there a way to use network rendering? That would be interesting too.

PolderAnimation wrote:Is there a way to use network rendering? That would be interesting too.

No, we explicitly disabled network rendering, because it's too hard to measure accurately and we don't send all required information from the slaves to the master.

resmas wrote:I think the default page should be the maximum score, and then if we want we would change page to see average and minimum. My cards are running on air, and in default settings, no OC...so its strange to see lower scores then mine in the same cards.

Hi,
Yes there is a little confusion about it. But "max" will cause more confusions i guess since there is no detection if the cards are OCed or stock. Would be great if the OctaneBench could detect OC (at least testers can input notes) so we could compare stock cards and see OC capables.
cheers
Hüseyin

sadece wrote:

resmas wrote:I think the default page should be the maximum score, and then if we want we would change page to see average and minimum. My cards are running on air, and in default settings, no OC...so its strange to see lower scores then mine in the same cards.

Hi,
Yes there is a little confusion about it. But "max" will cause more confusions i guess since there is no detection if the cards are OCed or stock. Would be great if the OctaneBench could detect OC (at least testers can input notes) so we could compare stock cards and see OC capables.
cheers
Hüseyin

once You open results, let's say from TitanX - You can see how much cards been tested & what the result have been.. with time You should see more & more cards running stock & less of those that hit higher results =) in a way it's good, especially for those who think 'bout building some thing custom - highly OCed system or water cooled rig =) & You know exact range of how card is going to perform from best to worst conditions possible =) I believe all the other characteristics like speeds & such would only mess the page - it will be way too much info for general user, who don't know all these parameters..now everything is kind of clear =) for most..



for instance here higher performing TitanX were probably from someOne who watercooled them - heard Maingear put some test here..