wdmfiber :
hannibal :
rrbronstein :
PepitoTV :
Just curious on what will become of the almighty Titan, the gap isn't too big between the 780 and the Titan so how the 780 Ti will fit?
i think its obvious that the titan is irrelevant, it isnt even meant for gamers. The 780 Ti will be king and provide higher gaming performance than titan and 780, but will still lack compute.
Indeed! Titan is 10 times faster in Douple presisieon computing than 780 is... So if you want to have GPU for calculations, you take Titan or 290X. If you want to have gaming GPU get the best bang for the buck and it just became better with the release of AMD r series GPU and prize cuts from both companies!
The GTX Titan isn't even good at compute. I don't know why people think it is. Tom's tested it months ago against a Radeon 7970 and it lost. And by alot on some benchmarks.
http://www.tomshardware.com/reviews/geforce-gtx-titan-performance-review,3442-10.html
Let me know when toms runs a REAL benchmark with CUDA vs. OpenCL (or whatever for AMD). Their chosen benchmarks are stupid for CUDA using TITAN users. Take any app, 3dsmax, cinema4d, adobe ae/ps/premiere, blender, lightwave etc etc. Use a cuda plugin like Octane or furryball (or just directly supported by adobe for years) vs. any plugin or choice for AMD (luxrender). Then come back and claim that crap again. Only a fool runs opencl with NV cards. That is the whole point of CUDA. There is NO reason to run OpenCL on an Nvidia card. Everything you can do in opencl can be done with Cuda FASTER whenever there is a Cuda option available.
Until toms starts pitting Cuda vs. AMD, their benchmarks are invalid for PRO stuff. Anandtech pulls the same crap. Synthetic junk you don't make money from (like toms, bitmining, folding@home...LOL, Basemark and then tested in OpenCL for NV? ROFL...RUN ADOBE for vid/photo editing and test cuda vs. AMD). When they do test cuda, they don't include the AMD putz side to compare, just say stuff like "amd doesn't support cuda so isn't benched". BS, you can run luxrender vs. cuda in maya, lightwave etc.
Their double precision etc financial benchmarks in sisoft (again, not a money making app, synthetic crap), they run them in OpenCL for both sides...ROFL. Apparently they find it too difficult to turn on CUDA and run Sisoftware's benchmark in that for NV. YES Sisoftware has CUDA built in and it runs faster. But in an effort to even the playing field Tomshardware shuns CUDA vs. AMD...LOL
http://www.sisoftware.co.uk/?d=qa&f=gpu_finance_fp64
"Note: OpenCL was used as it is supported by all GPUs/APUs. The tests are also available through CUDA which provides better optimisations for nV hardware."
Gee tomshardware...Why not use CUDA, even Sisoft's own page does this crap...I don't care that something is supported by all gpus. I want the FASTEST THING I CAN GET MY GPU TO DO no matter which brand I own. Right? How about all of you? I WANT CUDA if I buy an NV card. Because I know they won't support OpenCL fully until they are forced and Cuda dies (never will - taught in 600+ universities - it will be with us for a long while and already in 200+ pro apps like named above).
But if toms actually tested cuda you'd get the REAL picture and it looks quite a bit different than the one they represent. I laugh every time I see NV tested with OpenCL. WHY? Run CUDA.