On Fri, May 2, 2008 at 12:30 PM, <ewood at izoom.net> wrote: > Would it not be a whole lot better to simply start making it work on G4 > and G5 Macs right now? Then there'd at least be a working OS to port to > something like POWER6 later on.
The *whole* idea from the very beginning has been to get some sort of port up and running to a point where we have a serial console and a shell and some tools. That would be enough to allow community people to keep on working. This process has been in a terrible state for years. Years my friend. We started with a community project with little or resources at Blastwave.org That then drew the attention of some people who brought in some heavy hitting talent and $money$ into the project all working on a platform that no one would ever be able to get again. The Open Desktop Workstation from Genesi based on the PegasosPPC motherboard http://www.pegasosppc.com/pegasos.php I have one of those little PPC processor daughterboards here now. The motherboard stopped working a while ago and there will never be a replacement because you can not get one unless you make it yourself. So we have had the big time blow out of money and with the help of a rock star consultant named Guy Shaw we have a pile of code that will boot up to a point and then .. well there we are. Stuck with code that is extremely locked to a platform which had great firmware and no one can ever get. The EFIKA is not the same thing but it has nice firmware also. Just FYI. To go forwards, again, we would need hardware that can be found anywhere. Better yet, we would need hardware with a future. I am thinking POWER6 gear from IBM of course. We do not want to go there because we would need IBM engineers and millions of dollars in R&D money to do the job. The rumour is that it took rocket science or something more tricky, actual computer science, to get Linux working halfway decent on the POWER6 gear because of serious time synchronization issues in the kernel for multiple threads of execution all running after the same blocks of memory. Please go look into the TSO ( total store order memory consistency issues ) with references like "Memory Consistency and Process Coordination for SPARC Multiprocessors" : Memory Consistency and Process Coordination for SPARC Multiprocessors Book Series Lecture Notes in Computer Science Publisher Springer Berlin / Heidelberg ISSN 0302-9743 (Print) 1611-3349 (Online) Also go looking for a paper by Arvind ( a one name guy? ) MIT CSAIL and Jan-Willem Maessen ( Sun Labs ) about Memory Model and Instruction Reordering + Store Atomicity and then more recent stuff at OpenSPARC such as : TSOtool: A Program for Verifying Memory Systems Using the Memory Consistency Model Written by Sudheendra Hangal, Durgam Vahia, Chaiyasit Manovit, Juin-Yeu Joseph Lu and Sridhar Narayanan IEEE Int. Symp. on Computer Architecture (ISCA04), 2004. Really, a machine with a very well understood cache process and detailed hardware docs is needed. Good luck with that from the Apple stuff. So the IBM stuff looks better but you need IBM to play along. Or we get some money together and build some ODW units ourselves. Either way, my friend, can you spare a million dollars ? Dennis
