Hi. I'm making a C++ game engine and interested in supporting the A20 board. I wonder what features are available there. What's the gcc/g++ version? Is there a std::uint64_t type defined? Can it do 64 bit floating point?
On Sunday 2. October 2016 16.37.00 FaTony wrote:
Hi. I'm making a C++ game engine and interested in supporting the A20 board. I wonder what features are available there. What's the gcc/g++ version? Is there a std::uint64_t type defined? Can it do 64 bit floating point?
I imagine that the supported gcc version will be that provided by the distributions being offered. For Debian, you can find out versions by searching on the packages site:
https://www.debian.org/distrib/packages
Searching for "gcc" should provide results that employ descriptive package versions corresponding to the actual gcc version involved in each case (although you may need to strip off prefixes like "4:" and suffixes like "-1" as in "4:6.1.1-1" to get "6.1.1").
I see gcc 6.2.0 on Debian unstable, myself, but that's on i386. Nevertheless, the supported architectures tend to offer the same versions of things if at all possible. My cross-compilers seem to be 5.4.0, but an update is apparently pending with those.
I can't answer the specific questions about the A20, but maybe the gcc ARM- related notes can provide some answers. For example:
https://gcc.gnu.org/onlinedocs/gcc-6.2.0/gcc/ARM-Options.html
Paul
I imagine that the supported gcc version will be that provided by the distributions being offered. For Debian, you can find out versions by searching on the packages site:
Hmm, what version of Debian is preinstalled in the Debian card? I've read that it uses custom linux-sunxi kernel so thought it's some kind of a custom distro.
On Sunday 2. October 2016 18.42.00 FaTony wrote:
I imagine that the supported gcc version will be that provided by the distributions being offered. For Debian, you can find out versions by
searching on the packages site:
Hmm, what version of Debian is preinstalled in the Debian card? I've read that it uses custom linux-sunxi kernel so thought it's some kind of a custom distro.
I don't know the official situation, but the kernel need not have any impact on the distribution. For example, I compile my own kernel for a MIPS-based device I occasionally use, but it still runs the standard distribution of Debian.
People have "spun" their own custom distributions for various devices, but that usually happens out of necessity (using an unsupported architecture variant, for example) and/or under some impression that things can be "optimised" for a particular piece of hardware. But it should be the last resort if only because distributing the software will generally bring with it a significant support obligation. (The distributor will need to offer the software semi-permanently with corresponding sources and ongoing updates.)
I imagine that Luke will try and stay as close to the upstream distributions as possible. That's what I would do, anyway.
Paul
P.S. The effort of "spinning up" a distribution is arguably a lot greater than just getting a bootloader and kernel working for a device, which is why I was surprised that he dropped the jz4775 card, given that it only needed confirmation that u-boot and the Linux kernel were compatible. But then again, he also wanted FSF "Respects Your Freedom" certification for the distribution, and none of the certified distributions support mipsel: Debian does (and is not certified), but gNewSense (being Debian filtered for RYF certification purposes) does not presently (and without suitable hardware probably won't in future, either).
On 02/10/16 15:37, FaTony wrote:
Hi. I'm making a C++ game engine and interested in supporting the A20 board. I wonder what features are available there. What's the gcc/g++ version?
Depends on what release of what distro, generally these things are more of a distro thing than a board. Thing.
Is there a std::uint64_t type defined?
I would be very surprised if there wasn't, uint64_t has been arround for years.
Can it do 64 bit floating point.
Assuming you mean double precision floating point, of course it can. We aren't talking microcontroller cores here.
arm-netbook@lists.phcomp.co.uk
-
FaTony -
Paul Boddie -
peter green