bugid reports are not a mail list

[Tom Pusateri]

> On Apr 28, 2019, at 5:45 PM, Dennis Clarke <dclarke at blastwave.org> wrote:
> 
> On 4/28/19 5:38 PM, Tom Pusateri wrote:
>>> On Apr 28, 2019, at 5:24 PM, Dennis Clarke <dclarke at blastwave.org> wrote:
>>> 
>>> 
>>> Merely wanted to suggest that bugid reports contain more substantive
>>> content than open hand waving discussion. I am guilty as charged in
>>> this respect also.
>>> 
>>> Also ... follow up to comment left by Tom Pusateri on 235060 : yes
>>> the powermacs are cheap and everywhere and not much exists elsewhere.
>>> Unless one wants to spend $10k as a minimum. The situation is worse
>>> for RISC-V wherein nearly nothing exists other than QEMU and Spike.
>>> So in that hardware architecture one will need to spend $55M or more
>>> to get your own 9nm fabrication done. Maybe less. Not much less.
>> My apologies. I guess I should take this to the ARM mailing list.
> 
> No this is a good place and input is appreciated. I have arm on hand
> here also and it is a bugger to deal with.
> 
> Dennis

Agreed. But at least it is improving and isn’t stagnating.

I powered up my Mac Pro PowerPC G5 recently because I submitted pull requests to a project on github and the continuous integration testing done by the project used QEMU to test the code on big endian linux and it failed under QEMU but I was sure the code was fine. Trying it on the same version of linux on real hardware worked fine and so there was QEMU bug that prevented my pull request from being merged.

In addition to linux, I tried my favorite operating system (12.0-RELEASE) on the G5 and after some trouble installing, my code worked great there too. Next I tried it on MIPS64 FreeBSD which was very slow (ERL). I looked around for other options and couldn’t find much except that both the ARM64 could be programmed in either Big Endian or Little Endian but there do not seem to be any operating systems using Big Endian mode.

I think it would be useful to have modern inexpensive hardware that can run in Big Endian mode and if there’s enough others that feel this way, maybe we could have an ARM64 Big Endian version in addition to the current Little Endian version.

Tom