Hi all, I am aware That sometimes you receive a question like that, but at this time, it is worth spending money on a wx5100 or is better to p2000? Nvidia on Linux has a good shaped driver and AMD historically has a lot of issues, reading the wiki (and in my personal experience) Cuda is ready for production, but I would like to use OpenCL.
Any experience or suggestions are appreciated.Thanks, regards.
Unless you have a specific reason, I wouldn’t suggest Quadro. The only reason I personally use an M4000 is because I use a wide range of applications all of which all are Quadro certified (which Blender is not, specifically, AFAIK). I also want to keep my Dell workstation within Dell specifications. Quadro grants you a degree of stability and reliability, as well as some OpenCL-specific improvident on Linux that is shared with some higher-end GTX and Titan cards.
But you’re paying a pretty big premium for a bit of stability assurance, and plenty of professionals don’t even bother with it. I don’t know about firePro/Radeon Pro, i’d venture to guess though that their Workstation drivers are a bit more robust on Linux than their consumer drivers are.
What exactly do you find appealing about OpenCL? Are you a developer? Do you use Houdini or CAD/CAM? There are advantages to using OpenCL with Quadro, but it’s very unlikely you’ll see that performace improvement with a 2000-series. I just BARELY see an improvement on my M4000 over CPU (2xE5-2530v2, 12c/24th @ 2.6ghz) with OpenCL in Houdini, and even then only on very, very large simulations. If you’re planning on using Blender, at this time stick with Nvidia.
If you really do have an OpenCL need, the GTX 1070 is probably the best option from a price/performance standpoint. While it has been true that Nvidia lacked in OpenCL performance in the past, that isn’t really true anymore.
I agree. Unless you need the “Workstation” certification for some application or you are commonly creating scenes that use over 8 gigs of ram (1070-1080) or 11 gigs (1080ti), I wouldn’t bother with them. 1070 or 1080ti would be my preference (I have two 1070’s). Actually, if you’re looking at workstation class cards, that means you have the money to buy one. With that kind of money you could get two or three 1080ti’s! Ha! Now that would be crazy fast.
I’m looking at quadro/firepro because blender is not my main application. DaVinci (now on linux) and AfterEffects (my last win app) and other application/plugin rely on OpenGL/OpenCL, so a quadro (or firepro) like p2000 with good amount of vram (usefull in 4k project) it seems to me a good and versatile choise. Least but not last, is a single slot expansion card and (my fault) I do not trust at 100% in gaming hardware (overclocked, overpumped, overover), to avoid any trouble. I do not need so much power to use a sli conf, but perphas a 1050ti 4gb is a better choise?
Ok. Well, we sound like we’re in a similar situation then, multiple applications that are Quadro-certified and needing a single-slot GPU in a production environment. I also understand what you’re getting at with the consumer hardware - it’s hard to really know what’s hype and what could potentially be problems, though if you get the so-called “reference” cards, then that isn’t so much of a problem.
If you make your living from a GPU, and your software is Quadro certified, I think it might be beneficial to use Quadro.
FYI, SLI is a gaming technology and generally isn’t used outside of gaming and game development. Professional applications address each card separately, less Resolve uses it to synchronize displays.
I’m really thinking back to AMD/ATI for linux. I read that Steam hired a couple of AMD developers to improve performance on Linux. Also that AMD is seriously supporting Blender.
I am angry with nVidia, they do not give importance to Linux. Yes, the driver works well in many cases, but if you have a problem, you better start praying. I really do not know how nvidia has done to create a functional driver for Linux, sometimes I have the feeling that nvidia linux developers are not even Linux users. Once communicating with the nvidia linux support I had to explain them basic things related to the usage of Linux.