4 ms·

I don't know, but it's the obvious next step. There are certainly rumours that ARM is going to go that way. I'm not plugged in to the chip manufacturer gossip n

I don't know, but it's the obvious next step. There are certainly rumours that ARM is going to go that way. I'm not plugged in to the chip manufacturer gossip network but my guess is that ARM doesn't want to start an open fight with Intel before they have 1) Tested the server market. A 32-bit chip has its limitations, but for things like caches that are IO limited it is fine, and 2) Got their 64-bit architecture nailed down. I expect they don't want to give Intel a long lead time to respond.

mansr 16y ago
When power is a major concern, and that is the applications ARM is targeting, using a 64-bit system when 32-bit would suffice is counter-productive. A 64-bit CPU naturally requires more transistors and thus more power than a 32-bit one. There are other overheads (in OS etc) associated with a 64-bit system. The difference may not be great, but for many applications it is a cost with no reward. Comparing to MIPS, they have had a 64-bit ISA variant for a very long time, yet all the recent cores have been 32-bit as this is all their current target markets require. A full 64-bit system is really required only for HPC-type applications, a domain currently owned by Intel and IBM. If ARM are interested in eating into this market segment, it is most likely a very long-term goal, and obtaining a foothold in the 32-bit server space would serve as a first step in that direction.
- froydnj 16y ago
  Networking equipment is another place where 64-bit systems are useful. See chip offerings from places like NetLogic and Cavium, for instance: their multicore MIPS chips are 64-bit. Databases are another place where 64-bit systems shine. Maybe you already assumed that was included under "HPC-type applications", though. When I hear HPC, I think number-crunching, not databases, though.
Tamerlin 16y ago
40-bit addressing is probably enough for a lot of applications that Intel isn't a significant player in yet, also -- like console gaming, streaming video, and that sort of thing. That would also explain the target 2.5 GHz clock speed, 8 cores, and media and vector instructions. I'd also guess that limiting the addressing to 40 bits is partly for power consumption, as mentioned elsewhere. You're probably right about not wanting to start an open fight with Intel, but it's hard to not see this as a shot across the bow, especially compared to Atom. If nothing else, it should make the competition between Atom and Bobcat more interesting :)

mansr 16y ago

When power is a major concern, and that is the applications ARM is targeting, using a 64-bit system when 32-bit would suffice is counter-productive. A 64-bit CPU naturally requires more transistors and thus more power than a 32-bit one. There are other overheads (in OS etc) associated with a 64-bit system. The difference may not be great, but for many applications it is a cost with no reward. Comparing to MIPS, they have had a 64-bit ISA variant for a very long time, yet all the recent cores have been 32-bit as this is all their current target markets require. A full 64-bit system is really required only for HPC-type applications, a domain currently owned by Intel and IBM. If ARM are interested in eating into this market segment, it is most likely a very long-term goal, and obtaining a foothold in the 32-bit server space would serve as a first step in that direction.

froydnj 16y ago
Networking equipment is another place where 64-bit systems are useful. See chip offerings from places like NetLogic and Cavium, for instance: their multicore MIPS chips are 64-bit. Databases are another place where 64-bit systems shine. Maybe you already assumed that was included under "HPC-type applications", though. When I hear HPC, I think number-crunching, not databases, though.

Tamerlin 16y ago

40-bit addressing is probably enough for a lot of applications that Intel isn't a significant player in yet, also -- like console gaming, streaming video, and that sort of thing. That would also explain the target 2.5 GHz clock speed, 8 cores, and media and vector instructions. I'd also guess that limiting the addressing to 40 bits is partly for power consumption, as mentioned elsewhere. You're probably right about not wanting to start an open fight with Intel, but it's hard to not see this as a shot across the bow, especially compared to Atom. If nothing else, it should make the competition between Atom and Bobcat more interesting :)