3000 experts examine emulation plan

Project to mimic 3000 hardware gets nod from PA-RISC engineers

Engineering talent with more than 14 years experience developing parts of the RISC-based HP 3000 operating system, utilities and applications is taking a closer look at emulation for the system this month — and waiting for HP to make a statement on supporting such a project.

While OpenMPE Inc. organizes and presents a business plan to HP for its coming customer-owned organization, Allegro Consultants engineers feel confident they can create an emulator which would run MPE/iX applications on Intel-based hardware.

In such a project, HP officials would have to make the HP 3000’s operating system software available in a license to run on PC-based systems. The engineering would make HP’s software, as well as existing MPE/iX applications, run on Intel x86 systems.

Discussions being presented to HP include the prospect of creating a 3000 emulator that would be restricted to running on HP’s PC hardware until the end of HP’s MPE support in 2006. Emulation advocates want to be sure HP doesn’t feel threatened by the existence of a 3000 emulator.

Gavin Scott, vice president at Allegro, said his company has been talking with both HP and OpenMPE about the project, and he feels the software to mimic the 3000 hardware has a good chance of becoming a product available from Allegro.

“I think it’s most likely that it would be an actual Allegro product,” Scott said. “We’re talking about just doing the development ourselves and selling it as a commercial product. It’s in HP’s hands to say about the [MPE] licensing.”

An HP 3000 hardware emulator is much more easily created than emulating the operating system, Scott explained. Debugging an emulator for an MPE/iX which might run 3000 applications on PA-RISC Unix hardware presents major hurdles.

“You write a replacement for VPlus, IMAGE, all the intrinsics, the Command Interpreter language,” Scott said. “Essentially you end up having to rewrite MPE, or the top half of it, to provide every service an application might want. If you write this stuff and then the 3000 program doesn’t run quite right, the debugging effort and making it bug-compatible with MPE is an enormous effort. I don’t think it’s very viable.”

While re-creating MPE is still out there for someone else to do, Allegro isn’t interested in the work. “We don’t think it’s got a big enough benefit for the amount of effort you’d have to put into it,” Scott said.

HP’s November set of choices for its customers pointed the entire community at leaving the platform in five years or less. Major customers as well as smaller ones have replied they don’t want to be pushed off a computer that’s still working well for them.

“At this point I think HP’s been getting some feedback that everybody’s not happy with that idea,” Scott said. “I don’t really think that’s going to happen.”

OpenMPE Inc. is forming with the mission of developing another alternative to leaving the platform. Scott said his company has been examining how they might partner with the new company to engineer a solution.

“We’re talking with them because they’re thinking an emulator is a good idea,” Scott said. “There are possibilities for funding of the emulator that might involve OpenMPE Inc. providing funding to Allegro to develop it. It would be owned by the community, by OpenMPE, or something like that.

“We’d develop an emulator to run today’s MPE on other platforms. Until the 2027 problem with dates, there’s no reason why MPE/iX 6.0 or 7.0 won’t be able to continue to do what it’s doing today. Some of their key applications customers may decide to keep running on MPE forever, making sure they have enough hardware support they trust, or buy a couple of spare machines. Those are the customers an emulator will be a potential solution for.”

Customers would take an MPE/iX SLT tape and load it into the emulator, boot up, install and run the real MPE/iX from HP. The operating system will have no idea it’s not running on a real HP 3000 machine. Scott said the concept is equivalent to Virtual PC on the Macintosh, that runs PC applications on a Mac, or VMWare on Windows.

Scott added that performance of an emulator, even on current-day Intel hardware, would likely be enough to satisfy the needs of companies using small and midsize HP 3000s — anyone except the companies which need the performance from a multiprocessor N-Class system.

“You can get a 2.2-GHz multiprocessor PC for a small fraction of the price of an entry-level 3000, which has a reduced speed clock to begin with,” he said. “HP has kept MPE running on relatively low-performing chips anyway, so we’ve got a lot of room to do emulation on an x86 PC running Windows or Linux. All of the very small and medium-size [3000] customers could run their applications under an emulation environment.”

HP’s willingness to understand how such a solution could help its 3000 customers is essential to making an emulator viable.

“There’s still a chance they will support the emulation idea, which would be ideal for people who don’t want to or can’t get off MPE in the timeframe HP has provided,” Scott said. “That includes people who have regulatory requirements for keeping data around a certain number of years, like medical systems.”

Even a company migrating its applications might have a use for an emulator. Customers could find there’s 10 percent of their application that’s more cost-effective to run on a PC running an emulator, than replace it or write it from scratch.

HP’s technical support for the emulation project isn’t nearly as important as its licensing and legal concessions to the 3000 community.

“You have to get the MPE operating system from somewhere,” Scott said. “The emulator requires you to have the 3000 operating system and the rights to run it on your emulator. Today you can’t buy that from HP. None of this is going anywhere until HP says it’s a viable solution for their customers who are stuck on MPE, or choose to stay there.”