- I need a CPU core exclusively for one thread - 18 Updates
| Bonita Montero <Bonita.Montero@gmail.com>: Jul 11 07:36AM +0200 Am 10.07.2023 um 22:16 schrieb Vir Campestris: > Even so I was writing code to run on server farms. The fastest computers > we could get. All linked together with high speed networking. And even > so the simulations took _days_ to run. Maybe, but not doing calculations and on data which are constantly fetched across the network. > What he has described is entirely consistent with my experience. ... He's lying. |
| "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com>: Jul 10 10:40PM -0700 On 7/10/2023 10:36 PM, Bonita Montero wrote: > fetched across the network. >> What he has described is entirely consistent with my experience. ... > He's lying. The Troll Bait 666 ? Or is the the 665 version? |
| Keith Thompson <Keith.S.Thompson+u@gmail.com>: Jul 10 10:56PM -0700 > On 7/10/2023 10:36 PM, Bonita Montero wrote: >> Am 10.07.2023 um 22:16 schrieb Vir Campestris: [...] >>> What he has described is entirely consistent with my experience. ... >> He's lying. > The Troll Bait 666 ? Or is the the 665 version? Have you considered *not* feeding the troll? -- Keith Thompson (The_Other_Keith) Keith.S.Thompson+u@gmail.com Will write code for food. void Void(void) { Void(); } /* The recursive call of the void */ |
| "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com>: Jul 10 10:56PM -0700 On 7/10/2023 10:56 PM, Keith Thompson wrote: >>> He's lying. >> The Troll Bait 666 ? Or is the the 665 version? > Have you considered *not* feeding the troll? This might be a special kind of troll? Sorry. |
| Bonita Montero <Bonita.Montero@gmail.com>: Jul 11 11:40AM +0200 Am 11.07.2023 um 07:40 schrieb Chris M. Thomasson: >>> What he has described is entirely consistent with my experience. ... >> He's lying. > The Troll Bait 666 ? Or is the the 665 version? The cache has access times from four to five clock cycles on L1. Memory has access times from 20-30ns (page hit) to 50-60ns (page miss). I guess that a block access to a block on a remote SSD is about 1ms. Do you really think you could do constant calculations on data fetched that way without waiting on I/O nearly all the time. Please find some project on the net that is "number-crunching" that way and which has this kind of access pattern on data fetched from I/O. I'm too stupid for that. |
| Muttley@dastardlyhq.com: Jul 11 03:43PM On Mon, 10 Jul 2023 17:52:09 +0200 >W2K was also good, but I did find NT4.0sp3 quite solid. Once the Win9x >line died out, Windows stability became a lot better - it all depends on >what you do with it. A server OS marketed to business shouldn't crash no matter what you do with it, end of. It should also cope with much more than 1 application at a time. Back in the 90s proper server hardware cost 5 or 6 figures so you wanted to get your moneys worth from it which means just running it as a file server is a non starter. |
| Muttley@dastardlyhq.com: Jul 11 03:44PM On Mon, 10 Jul 2023 18:23:46 +0200 >>> drivers for VMS. >> They couldn't all have been bad. >There is a reason why Microsoft started with Microsoft certified drivers. Yet linux has 3 party drivers too. |
| Muttley@dastardlyhq.com: Jul 11 03:50PM On Mon, 10 Jul 2023 18:52:39 +0200 >>> His assumptions are felt competence. >> And again in english? >You say things you don't know but that you feel. Why is it you think anyone whose experiences in IT don't match your own is either lying or an idiot? Back in the 90s I was doing contract work on Sun Sparc and IBM RS/6000 kit for major corporations including 2 banks. None of them even looked twice at NT4 for anything serious except for one small company who thought it would be a cheap alternative to AIX once they'd ported they DBs. It didn't end well. |
| Bonita Montero <Bonita.Montero@gmail.com>: Jul 11 06:28PM +0200 >> You say things you don't know but that you feel. > Why is it you think anyone whose experiences in IT don't match your own is > either lying or an idiot? I bet my right hand that you never had any experience with VMS. |
| David Brown <david.brown@hesbynett.no>: Jul 11 06:29PM +0200 > Back in the 90s proper server hardware cost 5 or 6 figures so you wanted to > get your moneys worth from it which means just running it as a file server > is a non starter. Alternatively, if all you needed was a file server, then NT 4 server was a reasonable choice, at a hundredth of the cost of what you term "proper server hardware". Our NT 4 file server did not crash. I don't actually care if it might have crashed had we tried to run a database on it at the same time - because we didn't run a database on it at the same time. I agree with you entirely that you would not use it for running a bank database, but most small companies do not run bank databases. Different use-cases have different requirements. (As that server got old and its disks got full, we moved to Linux for the servers, and have stuck with that ever since.) |
| "Fred. Zwarts" <F.Zwarts@HetNet.nl>: Jul 11 06:36PM +0200 >>> They couldn't all have been bad. >> There is a reason why Microsoft started with Microsoft certified drivers. > Yet linux has 3 party drivers too. I know. I used them. The kernel got stained and every now end then there was a kernel crash. |
| scott@slp53.sl.home (Scott Lurndal): Jul 11 05:04PM >> is a non starter. >Alternatively, if all you needed was a file server, then NT 4 server was >a reasonable choice, So was unix and later Linux with NFS. A far superior protocol compared with SMB. |
| scott@slp53.sl.home (Scott Lurndal): Jul 11 05:07PM >> Why is it you think anyone whose experiences in IT don't match your own is >> either lying or an idiot? >I bet my right hand that you never had any experience with VMS. Who will wield the scmitar? Can't speak for Muttley, but given my six years of VMS work (some of which was in the kernel itself), he has much more credibility than you ever will. |
| David Brown <david.brown@hesbynett.no>: Jul 11 08:21PM +0200 On 11/07/2023 19:04, Scott Lurndal wrote: >> a reasonable choice, > So was unix and later Linux with NFS. A far superior protocol > compared with SMB. Yes, unless the clients were Windows machines - which was the case for virtually all workplace computers at the time. Technical superiority is rarely the major deciding factor in such matters. |
| scott@slp53.sl.home (Scott Lurndal): Jul 11 06:49PM >>> a reasonable choice, >> So was unix and later Linux with NFS. A far superior protocol >> compared with SMB. There were third party NFS clients available for windows. Many were using netware until it became nonviable on windows. Services for Unix (SFU) also included an NFS client IIRC. https://en.wikipedia.org/wiki/Windows_Services_for_UNIX |
| David Brown <david.brown@hesbynett.no>: Jul 11 09:48PM +0200 On 11/07/2023 20:49, Scott Lurndal wrote: > were using netware until it became nonviable on windows. Services > for Unix (SFU) also included an NFS client IIRC. > https://en.wikipedia.org/wiki/Windows_Services_for_UNIX I know you are very experienced in some areas, but I suspect you are quite unfamiliar with how IT was (and is) handled in most companies where people don't need high-end workstations, top-notch security, compute farms, five nine's uptimes, etc. Good enough is good enough - anything beyond that is a waste of money. Why would a small company (we were perhaps a dozen people at that time) want to spend significantly more money than their entire collection of client PCs in order to buy a Unix system as a server, then buy third-party add-ons (which would mostly not have worked on the diverse range of PC's and systems we had), just because NFS is "far superior" ? The NT file server worked flawlessly, did the job without needing to change client PCs or software, and cost little. /Nothing/ is better than working flawlessly - the best a Unix monster could hope to achieve would have been "just as good". And that logic applied to almost all small and medium businesses. Once the IBM PC with MSDOS entered the scene, it was not until Linux became mainstream on commodity hardware that any kind of *nix made sense outside a small proportion of niche uses. Those use-cases have always been vital, of course, but are numerically insignificant. |
| "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com>: Jul 11 02:26PM -0700 On 7/11/2023 12:48 PM, David Brown wrote: > became mainstream on commodity hardware that any kind of *nix made sense > outside a small proportion of niche uses. Those use-cases have always > been vital, of course, but are numerically insignificant. Fwiw, Ken Williams had to blow some serious coin on some silicon graphics workstations to create the game Phantasmagoria. Iirc, it was tens of thousands per machine. |
| kalevi@kolttonen.fi (Kalevi Kolttonen): Jul 11 10:05PM > Fwiw, Ken Williams had to blow some serious coin on some > silicon graphics workstations to create the game > Phantasmagoria. Iirc, it was tens of thousands per machine. Commercial UNIX workstations were not cheap, that is quite true. A friend of mine, a real UNIX freak once told me that some company wanted to get rid of their big SGI machine. It had become obsolete, but was in full working order. They said he could have it for free, but it was his responsibility to move it away. I was asked to help with the moving operation. I am pretty sure that the computer was SGI Onyx2: https://en.wikipedia.org/wiki/SGI_Onyx2 This thing is absolutely huge and must have cost a fortune back in the day! The machine we saw was fully loaded with all the goods. We worked our asses off for many hours, but the machine was simply too big. We could not get it out, no matter how hard we tried. We got it moving all right, but it was too tall. I suppose he got it out later with more friends. SGI Onyx2 runs IRIX. br, KK |
| You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to comp.lang.c+++unsubscribe@googlegroups.com. |
No comments:
Post a Comment