Cubieboard is even faster, but a little bit more expensive. I'll have to find some time to tinker around.

Looking at this one.

No - everything I have is what's documented as officially supported, apart from the Chromebook which I took the soldering iron to and didn't yet manage to unlock.

I think someone else has with the ODROID though - I'm sure that I have seen it mentioned on this forum.

Have an Odroid XU-4.
I'm not fond of it, setting up a rootfs is a pain as you need to copy data on empty space at the start of the SD/MMC. The ventirad is also a joke as the CPU hangs your system when overheating. Needs to be changed if you want to use the really powerful processor.
That said, you can compile and use a mainstream kernel, which is a good thing.
That's why you should stay away from boards with allwinner chips.

The Orange Pi, Banana Pi both have Allwinner CPUs. The way that the system is configured is up to the manufacturer.

I thought I read somewhere that the company that develops the Allwinner CPUs is in violation of the GPL?

That rings a bell, but I don't see how it'd have anything to do with the hardware. It's probably some Kernel modules that were modified and released in binary form only, or something like that.

Regarding gcc -- I'm rebuilding gcc without any of the Debian patches: One of the issues with including so many patches is that you cannot submit bug reports to GCC upstream, but if it still breaks without the patches applied, you can submit a bug report safe in the knowledge that it's the original source (with just one patch that's included with x86 Slackware).

As it turns out, I didn't rebuild libtool when I changed the toolchain target name, which caused all gcc builds to fail. Took me a while to figure out that it was libtool!

I upgraded gcc to the latest version of -current. I tried building mame again, and I finally cleared the previous error. However, I was getting a different set of errors and the build would fail at different points. It was never consistent. This lead me to believe that I might have been running out of swap space. So, went ahead and added another Gigabyte of swap and that finally allowed me to build it. Thank you so much for your help drmozes!!

Good to hear! The "killed" part of the log normally means it was killed by the memory manager (I didn't notice that first time) -- checking dmesg would confirm that.

Hmmm...very possible and I should have thought of that earlier. Anyhow, removing your Debian patches did not break anything here at least. So, that's good.

If anyone is interested, I can post my mame package along with the source if anyone is interested.