Path: csiph.com!eternal-september.org!feeder.eternal-september.org!mx02.eternal-september.org!.POSTED!not-for-mail From: Anne & Lynn Wheeler Newsgroups: alt.folklore.computers Subject: Re: You count as an old-timer if (was Re: Origin of the phrase "XYZZY") Date: Mon, 14 Mar 2016 17:17:40 -0700 Organization: Wheeler&Wheeler Lines: 55 Message-ID: <87k2l44msb.fsf@garlic.com> References: <79s9m8$lqh$1@nnrp1.dejanews.com> <1999Feb10.185011.19925@lorelei.approve.se> <79v589$6vf$1@nnrp1.dejanews.com> <36C2D90A.2CC6D420@trailing-edge.com> <7a1n71$kpf@nnrp2.farm.idt.net> <7a1v4p$mdi$1@nnrp1.dejanews.com> <36C55BB5.1F16DCBD@tnglwood.demon.co.uk> <8a7a1cfd-a80e-4723-ad46-a33da2e5f56e@googlegroups.com> <16m4ebpgco28bt6d5t8bj2qt0ah7urpo9h@4ax.com> <1f5de20f-5ed5-4ebc-ba13-bf034ccd8abb@googlegroups.com> <2b76adee-9e0a-457c-9762-950d2b2e226a@googlegroups.com> Mime-Version: 1.0 Content-Type: text/plain Injection-Info: mx02.eternal-september.org; posting-host="9a6c8ce24fcd2caeb84dd64e0fa6007c"; logging-data="18709"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX1/8ypeX9J67/IjydBZlLgBGZkoHYkihFyE=" User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.5 (gnu/linux) Cancel-Lock: sha1:tjcnJbjhffEMnrJ2/XfoWvYRYFI= sha1:9IPefcC/fi7416GWEmJv8EbwAYY= Xref: csiph.com alt.folklore.computers:160981 Jon Elson writes: > Higher level machines had wider data paths and somewhat faster memories. > We had a 360/50 with LCS (Large Core Storage). It gave you a LOT of memory > (for the time) but it was DOG slow, 8 us cycle time. We later got a 360/65 > and knew not to make that mistake again, it had internal memory, and they > could be interleaved, giving a substantial speed advantage. some of the LCS installations instead of treating it like extension to memory, treated it as a kind of (3090) expanded store ... or simulated electronic disk. They would simulate read/write transfer operations moving data back&forth between LCS and "fast memory". trivia ... 360/67 single processor was basically 360/65 with hardware address relocation (virtual memory). 360/65 multiprocessor was basically two single processor 360/65 with memory laid out as single entity ... but both processors had their own dedicated channels. all channels and both processors would contend for memory access. SMP I/O would be simulated by having two channel controllers ... that would be configured at same address on channels going to the two different processors. 360/67 multiprocessor had multi-ported memory original 360/60 & 360/70 were going to have 1mic, interleaved memory. before they ship, memory is improved to 750ns (8byte) interleaved memory and the models are changed to 360/65, 360/67, and 360/75. GA27-2719-2, http://bitsavers.trailing-edge.com/pdf/ibm/360/funcChar/GA27-2719-2_360-67_funcChar.pdf pg29: Four independent storage buses can be attached to the 2365-12: one from each CPU and Channel Controller in a full duplex system. pg36, figure8, model 67-1 has single bus to storage units pg37, System Data Flow In a Model 67-2 system, each CPU and each Channel Controller is connected, with separate bus, to each storage unit in the system. For example, a system with two CPUs and two Channel Controllers has four buses. Each bus can connect to as many as eight storage units. Conflicts that occur among the several buses connected to each storage unit are resolved at the storage unit. pg38, figure10, model 67-2 two processors, four buses to storage units pg47, Instruction Times goes into detail about additional latency added when storage isn't in the first 512kbytes (1st two storage units). -- virtualization experience starting Jan1968, online at home since Mar1970