No speed penalty, company claims. http://www.bit-tech.net/news/hardware/2016/03/11/otoy-cuda-compiler/1
Are they translating the code to OpenCL and then compiling or doing something else? When you say company x hasn't commented on something do you contact them and ask for comment or check things like twitter, press pages on websites and stuff like that. (I'm not lining up to criticize how you do things, I'm just curious)
If they are truly serious about their claims, they should open source it, because otherwise they are just extending the problem of CUDA to other GPUs.
Depends on the topic at hand. I've worked closely with these companies for years, so I have a pretty good handle on the questions they will or won't answer. For instance, there's no point asking any of them about unreleased product leaks 'cos you'll get the exact same canned statement every time: "Company X does not comment on rumours or speculation." General rule of thumb: if I say "Company X has not commented on the matter," I mean that I haven't asked it directly but I've checked the press releases and any public information outlets such as company blogs or social networking channels as well as any other coverage of the topic to see if a statement has been made; "Company X did not respond to a request for comment" or "Company X has declined to comment" means I've asked the question personally to no avail.
When Octane v3 was announced pretty much exactly a year ago, statements were made that Octane would work on OpenCL as well as on CUDA. Now, nine loooong weeks later, this is it? Is Octane using AMD's HCC and HIP C++ announced in Nov '15 for availability in Q1 '16, the "Boltzmann Initiative"? Probably not, as that seems limited to FirePro cards.
The only time I miss CUDA is in adobe products (PS and such), and they obviously could have coded AMD support but purposefully did not. So they won't add support for a thing like this either.