Twenty years ago this month the 3000 community got its biggest assurance of a long future. The system's lab manager said that the IA-64 architecture would be fully supported in future 3000 models. New compilers would be built for a new MPE/iX. Channel partners and resellers got the news from the labs two months earlier at a swell retreat.
The man delivering the news for that July article is a familiar name with 3000 customers. He was Winston Prather, head of R&D at the time—and three years later, the person who decided the 3000 community didn't need the computer. A weak ecosystem was supposedly the reason Prather could be the Sandman and help put HP's MPE/iX business to sleep.
Three summers earlier, all the news we could report in the NewsWire was good.
"The 3000 customers who experienced the move from Classic to MPE/XL know exactly what they’ll be looking at as they move forward,” Prather said. “One thing that makes me feel good about it is that it’s something we’ve done before. I think we pulled it off pretty successfully, and we learned quite a bit. We’ll use some of the same learning and techniques as we move to the new architecture."
Prather said that by early in the 2000s, 3000 customers would be able to buy and use an operating system to run with both PA-RISC and IA-64 processors, "Customers who need the additional performance of IA-64 will then be able to buy IA-64 processor boards to plug into HP 3000 processor slots on the new systems." It was an audacious design. HP bragged of the bold move.
“Prior to the IA-64 boards or chips," Prather said, "there will be complete new boxes available at the high end and the midrange, and then potentially at the low end.” The new 3000s would use new IO systems, giving customers a way to step into new hardware technology incrementally. The IO arrived, late out of those labs, in the form of new PCI bus architecture. IA-64 on MPE was put to sleep.
Watching the whiplash as HP first promised a future, pre-Y2K, then took away the hope of 3000 site, in 2001, baffled a lot of us. The system deserved a big tech investment. Then it didn't.
Although CSY is working to delay it, a time will come in the next decade when there will be two versions of MPE/iX, one for the PA-RISC systems and one for the IA-64 boxes. Customers work in such an environment today if they support Classic (CISC) HP 3000s running MPE V alongside MPE/iX.
HP has already been working on bringing 64-bit features such as large memory space and large files to MPE/iX long before IA-64 is ready for the 3000. The technical discussions taking place over the last year inside CSY include methods to keep from dividing MPE/iX into two camps.
“We don’t want to fork the operating system,” Prather said. “We have had internal debates about how we could provide all of this new functionality, the 64-bitness, and in the future avoid forking the operating system. We have a strong desire to avoid that. I’m not sure we’ll be able to avoid it forever.”
The departure of HP from MPE/iX futures is a fact that changed thousands of careers. Now the future demanded costly migrations for many firms. Everyone could be forgiven for being flummoxed by the leadership flame out, considering how certain the 3000 lab leader sounded in 1998 about building for the future.