Cray1 Anniversary Event September 21st 4550

Someone paid for and published in part of the Datamation article after Seymour's rest an analysis that Seymour had the potential for 7 machines in his remaining life and that the men in his family averaged 90 years of age. I have no idea how they got that.

Parts of the Cray 4 CPU are at the CHM given by his widow and a colleague local to me who should be in the audience tomorrow. Recalling this from my memory (w-o opening up my laptop). The 4 was to be a 1 ns clock cycle minus local memory back to B- and T- registers to fit in the size of a NeXT Cube with spray cooling. So Jobs continued to have influence after various Macs. I forgot the memory size, and I didn't ask GAM: some number of GBs, no idea other than Toshiba memory. Everything was to be factors of 4 improvement: CPU and memory if possible (roughly). It was GaAS. Oh, and it was IEEE-754 (Kahan made his point).

So the 5 was supposed to be 0.5 ns and the 6 to be 0.1 ns. I don't believe the SRC 6 is 0.1 ns. And the interweave on the 6 was supposed to be 512-way. The 6 I thought was Pentium technology. Cray was very confident that all of this was achieveable. Many open issues exist like fiber or free space optical interconnect.

Let me boot up my PB.

Let's see what else can I recall: oh yeah, that to expect gains in performance we really had to go to things like bacteria laying copper. That's documented in Sterling's Petaflops book.

I don't know if the Cray 4 boards are being brought out tomorrow night. I have a close time conflict with another work meeting. I don't know if any of Cray's children or widows will be there. I don't know if Chen will be there (officemate regarded the X Y and C-90 and T-90 lines are merely cleaned up 1s and not architecturally significant excepting economics). And I am not clear of any of Cray's close Lts. will be there. The Museum likely didn't take the time to contact CUG and I I don't know what the attendance count will be (overflow happened on such talks as the Alto talk).

--