Linked by Thom Holwerda on Wed 18th Mar 2009 11:48 UTC, submitted by PLan
In the News In a move that would certainly shake up the computer industry quite a bit, IBM is reportedly in talks with Sun Microsystems about the possibility of IBM acquiring Sun. Sun is going through hard times at the moment, and has been actively looking for someone to be acquired by.
Permalink for comment 353632
To read all comments associated with this story, please click here.
SPARC future
by John Bayko on Wed 18th Mar 2009 14:14 UTC
John Bayko
Member since:

Sun made a mistake in SPARC development with UltraSPARC III, when they decided to do a conservative, quick optimisation design without any new technologies like the competitors were doing. It turned out that developing a whole new design takes the same amount of time to debug whether you use new technologies or not, so it ended up being a generation behind in performance without the advantage of being earlier than the others (it was actually later).

In contrast, Fujitsu implemented it's next generation SPARC64 using the same performance techniques as the competitors, and ended up with a competitively performing CPU - which is why Sun is now using Fujitsu's version.

Sun is now going in the opposite direction and has plans to incorporate very ambitious new technologies in the "Rock" version of their SPARC CPU. In contrast, all the competing CPUs, such as Intel and AMD, and POWER, have few new performance features, and are now mostly just adding cores or multithreading (to catch up with a different, highly parallel version of SPARC which Sun bought from another company and made its own).

The result is that, of the next generation of CPUs, Sun is the only one that is likely to have any sort of genuine technology breakthrough. Among the potential technologies is the idea of a "scout thread" which can predict and pre-fetch instructions and data into caches, and transactional memory, which could make multithreading vastly faster and more reliable (a database taking advantage of it, like MySQL, could really fly). Sun has also done a lot of work on asynchronous logic, which lets faster operations complete without always waiting for a clock signal to proceed - I don't know if there will be much of that in the "Rock", but it has a lot of potential to vastly reduce power consumption while increasing speed.

I'd like to see this introduced, if only to get some innovative experimental technologies into the real world where they can do much good. Genuine processor and architecture development seems to have come to a standstill over the past few years, even though there are still potential new developments. The problem is that many are not incremental, and need a "leap of faith" which, right now, only Sun seems to be up for - possibly out of desperation (it still has Fujitsu to fall back on if "Rock" doesn't work), but it's doing it nonetheless.

Reply Score: 7