> Apparently, from the article referenced, it seems the basis of > the IBM lawsuit is that it's one thing to licence IBM's software > to run on a plug-compatible machine, but if you try to do so > for an Itanium-based machine using just-in-time recompilation > technology, then you're making an unauthorized translation > of the software to another processor's instruction set - hence > violating copyright law.
No logic. a) Amdahl machines used to do just that - infrequently used and new instructions were implemented in Macrocode that used essentially the same technique. As previous posts in this thread have also indicated, Itel's EXTEND and other packages did the same thing. b) Fundamental Software has sold over 1,000 machines using the JIT technique with IBM's full permission and sometimes assistance. IBM demonstrators are still using Flex-ES machines. > I had mentioned the company myself in a posting to alt.folklore.computers and comp.arch, having run across information on them in an Intel advertisement. I won't read either. I get upset easily. > Apparently, the versatility of the Itanium is such that it > makes all "proprietary architectures" obsolete, and this > was one example. Giggle. 21 November 2006 - http://www.theregister.co.uk/2006/11/21/itanium_q3_hp/ "Sun and Dell - the oddest couple - led the revenue charge. Sun's sales increased a stunning 25 per cent, while Dell enjoyed a nice turnaround on the back of an 11 per cent sales hike. IBM boosted sales as well by 7 per cent. HP joined Fujitsu as the biggest losers in the quarter. HP's sales slipped 6 per cent, and Fujitsu's fell 9 per cent. We'll let you guess which two of the five vendors mentioned sell Itanium-based servers." Hint - never believe what you read on a manufacturer's web site. _ESPECIALLY_ not PSI's - they're just as bad as UMX was. They've been a "leading supplier" for over two years despite never suppling anything? > Since this case raises the issue of the legitimacy of JIT emulation, it > seems that it will have wide-reaching implications. False premise - false consequence. > Of course, we could be really lucky, and some good could come out of > this. In order to do right by one of its showcase customers, Intel > might begin making microprocessors that handle the System/360 > instruction set directly. It would be nice to have some architectural > choice. Intel has considered it - Amdahl looked at outsourcing to them. But neither Hitachi nor Fujitsu believed they could make a profit given the incredibly high cost of developing zArchitecture CMOS processors - and they had a head start with years of knowledge of the architecture, their own validation engines, etc. It's even beyond IBM, and the next chip generation will have some POWER commonality. To imagine that Intel could fund such a massive operation for a couple of hundred chips a year is pure, unalloyed fantasy. This is one of the whacky bits of PSI's blurb - they're "using Amdahl technology" to implement zArchitecture on a chip that was available in testing quantities two years before Amdahl butted out of the business. So don't you think Amdahl would have rescued its business that way if it were possible? In fact PSI is a continuation of Stingray, which Amdahl vetoed as not cost effective. $60 million for the rights to CFCC code? How many systems are you going to sell? Even if you beat FSI's 1,000, it's still $60,000 per system. FSI benefited from IBM not having a product in the low end space - go head to head up at the top end against IBM's full basket of tactics and how many will you sell? Amdahl sold just one in its first year, Itel about four. Any business plan built on more than ten is bound to fail. You'll notice even PSI left out Parallel Sysplex claims. It's fun to look at what else they left out as well. Not viable in a modern data centre, even if IBM hadn't fired a shot across their bows. I cannot understand PSI's business model. Not only can I not make it work - I can't make it work by several orders of magnitude. I just don't know why they started. I'd love to know what they told their VC. We may not have seen the last lawsuit. -- Phil Payne http://www.isham-research.co.uk +44 7833 654 800 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
