Hmm, forgot about that. I have an Octane2 sitting here. Nowadays they can't claim squat, they sit on the backs of IBM and Novell's grunts. Ever since NVIDIA and cheap x86 processors became economical SGI has not even been viable. This was around 2001, last great machine with actual innovation, not just superficial speed upgrades, was the Octane2, which I have.
Anyways, I guess I'll agree there, even now I have less issues with transitioning apps between 32-bit and 64-bit on the SGI. The MIPSpro compilers only could compare to Sun Studio, GCC is just a slow horse on MIPS. It came at a price, this is the other reason they went down, they were competing with Sun, another HPC Server vendor. The Octane2 stock as $57,600 in 2001. I have upgrades worth an additional 20-25K. Heck, their RAM and optical drives cost an upwards of a grand alone. Hardly worth the price, the software is what kept them going for all the years they had their head up their butts trying to switch to the illfated Itanium. The Itanium actually borrowed ideas from SGI's most successful line of MIPS CPU's. James On Oct 25, 2007, at 8:26 AM, Casper.Dik at Sun.COM wrote: > > >> Now you're talking about the SunOS days. I'm referring to in >> comparison >> to the high level of innovation now to the lack of innovation abroad >> between Solaris 2.6-8. 9 is when they started doing new and >> important >> changes. I do respect the innovation done with the original SVR >> reimplementation, and I do enjoy the fruits of a mature ABI. It is >> second to none, most backwards and mostly forwards OS I've ever had >> experience with. One good innovation during the dark ages was being >> essentially the second vendor to have full 64-bit capability, this >> was >> back in 1997. The only other vendor to have mostly stable kernel >> API, >> and stable ABI was DEC's Alpha, actually I believe they were the >> first >> to engineer and deliver 64-bit microprocessors. > > > SGI/MIPS may have stronger claims to that crown, I'm not sure anymore. > > Digital made at leasst one mistake in their 64 bit ABI: 32 bit time_t > > Casper >
