Slashdot Mirror
Sun & Fujitsu Team On SPARC Chips & System
An anonymous reader writes "Sun and Fujitsu just announced a 20-year partnership to jointly develop SPARC based technology and systems. It looks like the long-predicted partnership that was hinted at earlier has finally come to pass in a much more comprehensive manner than I've heard anyone predict, i.e. not just chips, but a unified range of systems. My guess: Sun drops Ultrasparc III to provide the Throughput computing chips for the low end / web / network stuff, and takes up the Fujitsu provided SPARC64 chips for the high end and workstation market. Will this spark a new RISC renaissance for Sun and Fujitsu? Or is it a last gasp before Opteron / PowerPC / Itanium crush them? I for one will be interested to see what systems and processors come out of this. This could really revitalize the SPARC system market, especially if Sun's work on Throughput computing proves out."
You were misled by the OP, but RTFA please. The press release said they're expanding their relationship that's already existed for 20 years. Not that they're announcing a 20 year partnership.
NEW YORK, NY -- July 8, 1987 -- Sun Microsystems, Inc., introduced today the Sun-4 family of 10-MIPS supercomputing workstations and servers that give users the performance of a VAX 8800 system at one-tenth the price.
...
...
Sun also announced that it will license the new SPARC architecture... SPARC licensees announced today are Fujitsu Microelectronics, Cypress Semiconductor, and Bipolar Integrated Technology.
You run Solaris on SPARC processors. Solaris is highly multi-threaded from the ground up. It's extremely fine-grained. It also has some sophisticated algorithms for migrating threads to the most appropriate processor based on things like memory locality and load. (Forgive me if my terminology isn't terribly accurate, I'm not an OS kernel expert). The Niagra and ROCK processors are designed to execute highly multithreaded loads. Fujitsu SPARC64 is more traditional, in that it is designed for loads with fewer concurrent threads. By adopting Fujitsu's high-end gear, Sun gets performance on less thread-intesive loads too. Now Sun and Fujitsu have a horse for every course, so to speak. If I were HP trying to sell itanic boxes, and cranky old (soon to be exterminated) PA-RISC kit, I'd be very worried.
Stick Men
Sigh.
Please try to remember that entry to the 'Top 500' list is as much about your interconnect topology and technology as the capabilities of the processors used.
It is a measure of one, and exactly one benchmark, LINPACK
Machines which are not well suited to this benchmark, or do not have network technologies/topologies well matching linpacks requirements will perform poorly at it, but possibly very well for their chosen purpose.
Good examples of this are the WETA digital clusters used in parts of the LOTR films, which are great for rendering, but hampered seriously in their linpack result by their 100MBit standard ethernet connections.
Another good example of this is the Virginia Tech G5 cluster, which gets a LARGE boost from it's infiniband interconnects (well, it will when Apple finish giving them the new machines... eventually..).
Not that I am defending SPARC's rather lackluster performance these days, just making a rather important point.
Those SPARC boxes better get a LOT cheaper VERY fast if they intend to find any real home in HPC.
Baloney. I've had power drop out and kill Sun machines and they just come right back. Are you talking about file system check? Just turn on logging in /etc/vfstab and even that goes away, just like the same reason you use ext3 instead of ext2 on Linux.