OK how about these, which are asked from a Linux/open source user perspective:
1. When are you going to start treating Catalyst on Linux as a first-class citizen in the same manner that Nvidia does with its own drivers. By first-class, I mean as close to feature and performance parity with Windows as possible and having support for new versions of Xorg* BEFORE they are released instead of 6 months later. The reason I'm asking is that I'm basically stuck with Nvidia on Linux if I want a performance GPU (I have no overt love of Nvidia, but I also like things that work).
Pages like
this one from my distro (Arch Linux) don't inspire confidence. I do give AMD props for open-source support, but frankly it's only good for maintaining basal compatibility with obsolete cards. There's nothing wrong with that, but AMD shouldn't ignore performance Linux, especially when a large portion of the HPC world uses Linux heavily and Nvidia has made some real money in that realm. So: What is AMD's plan for the future of Catalyst on non-Windows platforms?
(* Oh, EGL support and Wayland support in Catalyst would be a huge win too).
2. FX is dead and we all know that AMD has been subtly telegraphing that message since at least late 2012. Now that AMD has gone APU-only, what kinds of improvements are we going to see with HSA that will translate into real-world performance and power/performance benefits? I'm asking because I've seen plenty of fragile, windows-only demos of OpenCL, but precious few real-world applications and day to day apps that use OpenCL. I'm not singling AMD out here, BTW, I defintely *don't* see CUDA applications on a day to day basis outside of esoteric compute workloads either (e.g. BOINC or things like that). What I am saying is that there is a big difference between a few benchmarks and real-world performance across a wide-range of applications.
This is much more critical for AMD because your own marketing department has pretty much written off high-end x86 performance (including vector-extensions to x86) in favor of OpenCL. Well, in the open source world there are even fewer applications that can benefit from OpenCL than on Windows. So, what is AMD's plan to push OpenCL software to a wider audience including cross-platform development?
3. Related in part to the above issues with Catalyst, what is AMD's policy going forward on devices with Android & ChromeOS? Chips like Temash have potential for high-envelope tablet and chromebook products that could have very good performance and graphics if there is driver support. It is encouraging to see that Temash is
no longer Windows exclusive but more support of Android and other alternatives is always welcome.
Here's a related question to tablets: Will AMD be willing to be better than the ARM licensees and Intel by working with partners to sell fully unlocked x86 tablets where the consumer (not some OEM or even worse a phone company) has control over the software that is installed? Is AMD willing to sell us unlocked tablets where we can choose a Linux distro that will actually run? Or windows 8? Or a non-OEM doctored version of Android? I really want to know because I'm more than willing to take back tablets from being toys into being real
personal computing devices if companies like AMD will step up and sell unlocked devices.