Thread: Amiga X5000?
View Single Post
Old 02 October 2015, 23:17   #31
Registered User
Join Date: Jan 2010
Location: Kansas
Posts: 876
Originally Posted by eXeler0 View Post
How future proof is each choice - might also be a factor to consider.
IMO, every door except the PPC looks future proof. PPC lost the desktop market to x86/x86_64 years ago, lost most of the console market to x86_64 recently and has gradually been losing market share in telecommunication and automotive embedded markets to ARM. There are unlikely to be any new PPC CPU designs but there may be modifications or enhancements of existing CPU designs, probably for embedded customers.

ARM is actually Thumb 2 CPUs (most existing ARM CPUs) and ARMv8 CPUs (newer more powerful 64 bit CPUs which are more like the PPC). Thumb 2 will continue to be used for low end CPUs and ARMv8 CPUs are likely to take market share from PPC.

The FPGA market is healthy and growing even in this down weak global economy. Intel recently payed a premium price for Altera because of its growth and future outlook. FPGAs should be faster, bigger, cheaper and have more options in 5 years. The embedded market is driving the demand where FPGAs give a cost savings so even if we enter another global great depression the technology is not going to go away.

Originally Posted by eXeler0 View Post
The 68k path is compelling but lets not forget that x86 and most other architectures have a 20 year head start and a couple of billion $$ worth of R&D between them.
This is true but x86 started with more baggage and added more ISA mistakes early on. The PPC ISA is too expansive with parts unsupported in hardware which varies from one CPU design to the next. ARM has so many different optional hardware options that it becomes difficult to support. The 68k ISA starts out cleaner anyway, although mistakes could be made. The 68k CPU designs have not been developed but we do have the 68060 to look at which does a lot of things right and mostly avoids what people would expect to be bottlenecks in the 68k (decoding and complex addressing modes). The 68k can beat x86/x86_64, ARM (including Thumb 2) and PPC in average instruction length (smaller is better for superscalar designs) and code density (compactness of code) which are great innate traits. It can be very powerful for its clock speed while probably not clocking as high as some other designs. I think it could be made competitive with other processors but no one wants it, not even A-Eon which maintains course with PPC despite the head winds and icebergs.
matthey is offline  
Page generated in 0.05181 seconds with 9 queries