Path: csiph.com!eternal-september.org!feeder.eternal-september.org!nntp.eternal-september.org!.POSTED!not-for-mail From: Lynn Wheeler Newsgroups: comp.os.linux.misc,alt.folklore.computers Subject: Re: what's a mainframe, NYT and Stratus (Re: [OT] Dead DEC Punch card times) Date: Tue, 10 Nov 2026 07:29:52 -1000 Organization: Wheeler&Wheeler Lines: 58 Message-ID: <87eccsppbj.fsf@localhost> References: <18dee4b6cab427fa$865$2574533$802601b3@news.usenetexpress.com> <11cr1ub$q796$1@dont-email.me> <11creup$trgi$1@dont-email.me> <11ctde2$2ibp$1@gal.iecc.com> <5O-dnQvUAMIEF2_3nZ2dnZfqn_SdnZ2d@giganews.com> MIME-Version: 1.0 Content-Type: text/plain Injection-Date: Tue, 10 Nov 2026 17:29:58 +0000 (UTC) Injection-Info: dont-email.me; logging-data="2454781"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX1/Dp1otccC8z4HNEKv3m4Rbtt/3/6n3m3U="; posting-host="79a5ac4974d049d0c74487c0d80bb5d5" User-Agent: Gnus/5.13 (Gnus v5.13) Cancel-Lock: sha1:dU8u4eC/OJQVNrjH+XcJ/RL1EJA= sha1:pwnLd/VpZnus38w5xoCOnXOtiFk= sha256:CSa627S7fR2FIjriB8Q85iLSsJvcMCbRSS91W6T3944= sha1:+DtlSek528AB8IvmCRkEFKSYjgw= sha256:uOYRxItFmk9S6eDWJyZ0RONBB63Vc45jqQh0NoRPh1I= Xref: csiph.com comp.os.linux.misc:95431 alt.folklore.computers:236729 Alexander Schreiber writes: > No. The operating system and application software ran on the mainframe > CPU. But - as described above - a key design aspect of mainframes back > then was that I/O was not done by the CPU (like it was done on home > computers were all I/O was entirely handled by the CPU), but fully off- > loaded to peripherials that were essentially limited computers of their > own: > - CPU tells I/O interface (channel): "Do this I/O" > - CPU goes back to computing, channel does some I/O heavy lifting > - channel has completed I/O and sends an "Hey, I'm done" interrupt > to the CPU Original 360, there was limited processor and storage, but a lof of excess channel capacity ... as a result things like finding data was offloaded to channels ... which would "search" for desired data ... and as a result, channels ran half-duplex ... with channel dedicated to single operation ... constantly pulling search argument from processor storage. Move into 80s, storage got lot more abundant ... and channel architecture increasingly channel I/O increasingly became system throughput bottleneck. Early 80s, I wrote tome that 60s era 360/67 could get 70-80 CP67/CMS users on system ... and early 80s, with IBM processors getting 40-50 times faster and storage increasing 40-50 larger, should be able be see 3000-4000 users, but was only seeing 300-400 users (because disks & channels throughput only increased 3-5 times (partly because half-duplex operation increasingly played a latency role in fully being able to see benefits of any speed increase). A IBM disk division executive took exception and assigned the division performance group to refute the claim. After a couple weeks, they came back and said I had slightly understated the issue. 1988, one of issues of being able to see significant increase in throughput with LLNL serial stuff ... becoming fibre-channel standard was running full-duplex being able to concurrently transmit and receive on 1gbit speed links (w/o having to wait for full round-trip hand-shaking). Note: 1980, IBM STL (since renamed SVL) was bursting at seams and 300 people/terminals from IMS group were being moved to offsite bldg, with dataprocessing service back to STL datacenter. They had tried "remote 3270" but found human factors totally unacceptable. I get con'ed into doing channel extender support, allowing placing 3270 controllers at offsite bldg with no perceived difference in human factors. Unanticipated side-effect was the associated systems' throughput increase 10-15 percent. STL had been speading 3270 controllers across all channels with 3330 drives. The channel-extenders (operating full-duplex) significantly reduced the channel busy interference with the 3330 drives (for same amount of 3270 traffic; there was some consideration moving all 3270 controllers to channel-extenders). -- virtualization experience starting Jan1968, online at home since Mar1970