TheAnonymousPoster

I am writing a 3D application in webgl, and need to provide my own depth data that is contained within a texture, my current code does this: VS: varying vec2 vUv; void main() { vUv = uv; } FS: uniform sampler2D depthTex; varying vec2 vUv; void main() { gl_FragDepth = texture2D(depthTex, vUv).r; } however gl_FragDepth is disabled in opengl-es (and therefor webgl) is there anyway to somehow enable it, or any way to provide my own depth data that doesnt involve heavy manipulation of render targets?

THanks for the suggestions, i wasn't looking to spec out a personal machine at the moment, iv'e built many machines and i'm some what picky / I have contacts to get some lovely pricing. whilst the magnum titan is a HUGE chassis, i'm sticking to a single system just very high end, using a Quadro K6000 and a pair of 780ti's on a dual boot system (As far im aware you can't run two sets of drivers for two cards on the same system) on a dual socket board. I will most likely use a caselabs chassis or if i do go the two system route i'd be using a Red harbinger desk. CPU wise ill be using E2670V2's (10 cores, 20 threads). Most of my rendering at the moment is done on the CPU side of things and only really IPR / raytracing is handled by the gpu. I'm also a modder / getting much deeper into modding so i'll be using a custom water loop(s) rathe then all in ones, much quieter too. Your gaming system is similar to what i'm currently running bar the second 780ti though!

Can't quote on my phone that's annoying! I've looked into online cloud rendering and its about $0.30 per core per hour from what I've seen which isn't huge and does buy me a lot of render time, an offline render farm would be a preferred option however it's a hefty investment as I'd rather get an iray when I've got the money hence my original solution to buy a dual socket system and add a farm later on! Back on topic I've heard from Asus they're launching dual socket x99 boards! No more info though

Thanks for the advice, if i were to do that i'd most likely stick with the current system as a gaming rig (add a second 780ti in there perhaps) Or push comes to shove, buy a Red harbringer? Desk and fit two systems inside of that. I'll have a look into render farms, i do deal with dense meshes (Upwards of 27million polygons in some instances, large scenes not so much however iv'e done one or two) Perhaps spending as you said $5-8,000 on the workstation (Not including the already aquired quadro) and pairing it with a render farm may be a solution, i'll have a speak with the better half to see what space I can aquire for such a thing. Having a seperate gaming pc though isn't a huge deal for me most of my waiting times comes from rendering. I'm a huge user of NukeX too, i'll drop the foundry an email to see if they can enlighten me on a dual socket support for NukeX. Space is a premium which is annoying for the time being but its why i was aiming for a dual socket system to get the most power / performance from a single chassis. A render farm would also open up the chance for me to use my Ultrabook to make minor changes on small scale scenes whilst at the clients and have it being rnedered, time dependent there and then on the farm. It's certainly an appealing idea however its nothing that can't be done on a single system! I shall do some more research but your help has certain been insightful! I'm now thinking a single high powered Xeon workstation / gaming system and see how I get on over my current setup ($8,000-$9,000 mark) which I can mod and make spectacular! , if needs be purchase a render farm in the future as its something which can always be added too when I have more space.

I wasn't planning on using my I7 in the system, I'm moving over to Xeon E5 a pair of E5-2670V2's to be specific ^.^ I was only able to afford a K6000 due to PNY offering me one very cheap. I'll have about $25,000 to spend on the system overall (However it'll be fully water cooled, sleeved and the chassis modded extensively along with a second 780ti for gaming needs) and from what iv'e read and been speaking to Autodesk, maya does benefit largely from a second cpu, especially as both mental ray, vray and its ndynamics use CPU(s). Nvidias Irays start at around $50,000 from what iv'e read which is well out of my price range. A render farm was an option however i'd love to keep it all in one machine if at all possible, i move around a lot, although slogging it over to backburner using various other pcs i have lad laying around is something iv'e done in the past. I'm aware Intel have announced a 16 core 32 thread cpu which although isn't comparable to the 2670's it should be cheaper for not a huge loss in performance, the money saved could go towards a second K6000.

Don't do this to me ;-; Aye, very rarely (only one or two projects) and thankfully they were stills and on an older system! I still think to this say I set the point intop to 0.001 rather then 0.01 which would explain it!

Much like the Z9PE-D8 WS from asus, a dual socket system is my next end goal, I was wondering if there is any word on the street for some updated dual socket motherboards? I'd happily wait, iv'e seen some info over on OCN but nothing official, i'm pestering my contacts too! (Before the flame, I'm a CGI / previs artist, whilst my dream would to have a completely separate render farm the convenience of a single desktop solution has a much higher priority right now, render times currently vary from 40-50 minutes to upwards of 27 hours a frame. I do have a Quadro K6000 on the way to deal with this though however simulation is becoming an ever bigger part of what I do and i need the cpu horse power to deal with it!) Anyway, thoughts and any info!

Here is my code NSManagedObjectContext *context = [[NSManagedObjectContext alloc] init]; Memory *newMemory = (Memory*)[NSEntityDescription insertNewObjectForEntityForName:@"Memory" inManagedObjectContext:context]; And here is the error message "Terminating app due to uncaught exception 'NSInternalInconsistencyException', reason: '+entityForName: could not locate an NSManagedObjectModel for entity name 'Memory" From the sounds of it is unable to find the entity 'Memory'. However I have no idea why. Any help? Cheers for any help,. Tom

Hey, Currently looking for a router with Sim card support for GSM 3G. We get a meer 1mb internet here (92kb/s downloads) so we're buying a unlimited mobile internet plan, the mobile dongles are some what useless so we're looking for a router with sim support at a decent price. We've seen them at £300 ideally under £100 is best for us! Please no portable ones a proper router! Thankyou in advanced

This isn't a "AMD nvidia fanboy battle" this was purely for my choice of card what would be best. Maya also relies heavily on PhysX for its simulation and Ndynamics, mental ray is also made by nVidia along with their MIP shaders so for my purposes, and not using a workstation card, Nvidia was the best choice.

Nope, i'm a CGI artist, Maya, Mental and Vray at my tools. I went with a 780ti in the end and its phenominal. Vray RT is about 20% quicker then my 3960x now with the gpu. Rendering iv'e not had chance to test yet ( You have to config some dll's fo rmental ray gpu rendering) however there is a VAST improvement over the 670 as expcted. I'll post up some benchmarks soon, the extra cuda cores and higher clock definitely helps more then the double precision from the titan. EVGA are launching a 6gb 780ti soon too.. i'll be "stepping up" to that.

I'm going with the Refernece cooler but a superclocked edition, my chassis is a small form factor with some-what limited airflow at the moment. the blower with it pushing air out the back *Should* be the best choice..hopefully

Well ive been talking to Nvidia and Autodesk, they've both said the same thing, although the Titan has a higher frame buffer and vram maya can borrow from the system memory to compensate if need be, hitting the 3gb is extremely rare too unless you're pushing 4k renders with a crap-tonne of textures, even so it pushes it into system memory (which is slower and hence why the QUadro series gpus have 12gb+ of memory) Push comes to shove, the 780ti has the same amount of Cuda cores as a K6000, without the memory or insane precision / performance, so words from Nvidia Theoretically a 780ti should be quicker then a Titan as rendering engines at the moment dont take advantage of double precision cards, it has higher clock speeds and more cores. Soo... 780ti it is....EVGA... just...which...version!

Sums up what I found exactly, the black is the perfect solution but it's out my price range which is a bit of a bummer. The 780ti seems to be the better solution with the fully unlocked cores and although I do 4k encoding its dumped to ram with my hex core the 670 rarely struggles as is with view port only really heavy scenes. I think a nice evga acx 780ti sc will do!

Iv'e come to a semi-conclusion, there are currently no 3D rendering engines which use double precision for their calculations, and there probably never will be. Whilst the double memory on the titan is favorable the fully unlocked core and higher clock speeds of the 780ti would prove to be more of an advantage for the smaller scale renders. Benchmarks ive seen for rendering, not gaming, the 780ti out performs the titan, a 780 outperforms a K4000 so that's more then good enough for me. I'll run some standard vray rt tests too and compare it to the 670 just to see how much of an improvment there is, and i'll do some nuke 4k encoding too on a simple vfx shot! feel free to comment if you think its the wrong thing!