Yeah, and I have older and slower Nvidia GTX960 2GB and it takes several hours (2 - 6)
Buy GTX1070 or GTX1080
Buy GTX1070 or GTX1080
- Apr 26, 2017
- 83
- 330
Just remember that Iray has to load everything to Vram to start rendering. If Vram get's full it will switch to CPU and Ram and slow down. The more Vram your GPU has the more you can load in a scene. Also if you plan on using 2 GPUs you won't have more Vram, Iray will "copy" all the data to each GPU. The rendering speed is always bound to the slowest GPU used. Finally, depending on the Operating System you are using a portion of Vram is bound to it for display.
Let me give you an example. I have a GTX 1080 with 8GB and a GTX 750 Ti with 2GB. If I use both for rendering I'll have 2GB of Vram available and the speed of the rendering will be as fast as the 750 Ti can render. However, I can use 750 Ti for diplay, unplug 1080 from any monitor and have all of the 8GBs of it available for Iray.
Let me give you an example. I have a GTX 1080 with 8GB and a GTX 750 Ti with 2GB. If I use both for rendering I'll have 2GB of Vram available and the speed of the rendering will be as fast as the 750 Ti can render. However, I can use 750 Ti for diplay, unplug 1080 from any monitor and have all of the 8GBs of it available for Iray.
- Jul 8, 2017
- 763
- 13,057
I'm getting a GTX 1060 with 6 GB VRAM. Also upgrading my RAM from 8 to 16 GB. Moving my old GTX 660 with 2 GB (which makes it useless for rendering anything more than maybe a Utah teapot in Iray) to the office PC, but to do that it will need a PSU upgrade, so another $60 for that too.
The most important thing, after VRAm, is how many CUDA core has card. GTX1060 has 1280 GTX1070 has 1920. Mine GTX960 has 1023 core and it's sloooow.
- Jul 8, 2017
- 763
- 13,057
When you say sloooow, how slow? I'm getting 12 hours for a 1080 HD render on a 2.3 GHz Core i7.
I looked at CUDA core count vs. prices, hoping I'd see one model that stands out as a good value in $$ per CUDA. But it surprised me to see graphics card prices scale almost perfectly linear when it comes to the CUDAs: expect to pay 20 cents per CUDA on most current models, with higher models having the added benefit of more RAM. All of these except the lower 1060 model have 4 GB or more.
- Jun 21, 2017
- 54
- 251
TL: DR - buy a gtx 1060 6 gig
one thing to remember is that between "generations" of graphics cards, cuda performance can not be directly compared. 1 cuda core on a 900 is not comparable to 1 on a 1000 series. I'd advise to look at direct benchmark comparisons between your current GPU and the potential purchases. Google is your friend, simply put in your current GPU + "cuda benchmark" and then look for the same benchmark software running on a potential purchase.
i had a 670 gtx mini that i loved, and i thought it was turning out "decent" render time for a bit, and when it died, i picked up a cheap, and yes for 230 bucks on amazon at the time was cheap for a good GPU, 1060 6 gig and i was shocked at the difference in render times and over all performance.
just remember that if you aren't running iray, and instead are running Reality, octane, or luxrender, you can probably run an AMD card on OpenCL, however if you are runnign IRAY, you need Nvidia.
personally, if you aren't running enterprise, missian critical stuff, and you aren't a "gear slut" i'd advise in the 1060 range, they are generally targeted at the "sweet spot" of price/performance ratio, and you'll get decent FPS in games @ 4K, nothing buttery smooth, and at the sub 250 dollar range, if you want to, in the future, it's easy to add a second one in for SLI.
one thing to remember is also "video card memory" can be a cap on your IRAY renders. if the scene you are rendering is larger than your graphics card memory, IRAY defaults to the CPU rendering. so the 6 gig 1060 again is a nice sweet spot, it's not gonna do heavy polygon, heavy texture 4K renders, but it has enough overhead ram wise to get what you need from a hobbyist perspective done.
i hope this helps
Last edited:
- Jul 8, 2017
- 763
- 13,057
- Jun 21, 2017
- 54
- 251
yeah, but i also wanted to try and give a technical perspective and rationale, lots of people get the idea that features are identical generation to generation, a core 2 quad is comparable to say an i5 running at the same frequency, a 780 and a 980 running at the same frequency is identical, 1 MHZ = 1 MHz, 1 cuda core = 1 cuda core, (and yes this conversation occurs in businesses to this day, and even in families.... facepalm) et all. and i can state from someone that's been running the non SSC evga 1060 6 gig that it's lovely bang for the buck.
- Jul 2, 2017
- 1,217
- 4,480
Now that we know most of the used assets. Which one could be used, to achieve good results on G8 models. And which G8 Models would you use, to replace the original G3 character models?
Lisa I would use Aurora:
Kira is a perfect candidate for Victoria8, the MakeUp styles are spot on:
Mom, Kate, Alicia and Olivia I have no clue right now.
Max and other male characters have to wait till november (It is assumed that the High Quality G8M characters will be released there).
Lisa I would use Aurora:
You must be registered to see the links
Kira is a perfect candidate for Victoria8, the MakeUp styles are spot on:
You must be registered to see the links
Mom, Kate, Alicia and Olivia I have no clue right now.
Max and other male characters have to wait till november (It is assumed that the High Quality G8M characters will be released there).
I never thought about this simply because user love already created characters
- Jul 2, 2017
- 1,217
- 4,480
I didn't say, that the original models aren't good. But I, for example, couldn't use them to their fullest, hence my whole collection of expressions, morphs, poses and so on, are bought with G8F and G8M based models in mind. XD
And the PowerPose tool too, plays way more better out with G8 than with G3. So G8 models are a little bit easier to work with, when you haven't a lot of experience with legacy or G3 models. That's were my thoughts are comming from.
And the PowerPose tool too, plays way more better out with G8 than with G3. So G8 models are a little bit easier to work with, when you haven't a lot of experience with legacy or G3 models. That's were my thoughts are comming from.
If I'm not wrong, you can always try to convert G3 models to G8, or if you don't have G3, ask someone to do this
Maybe result will be similar to original.
You must be registered to see the links
and other tutorials on his site
- Apr 22, 2017
- 497
- 7,656
- Apr 26, 2017
- 83
- 330
I'm uploading more BB related assets to:
If anyone wants to contribute PM me. Here's a quick list with the files I have collected so far. The files on the list should be synced with the folder within the hour, so be patient.
You must be registered to see the links
If anyone wants to contribute PM me. Here's a quick list with the files I have collected so far. The files on the list should be synced with the folder within the hour, so be patient.
- Jun 25, 2017
- 2,556
- 7,346
