[cctalk] OLD V NEW (Was: Old UNIX Systems

On 06/04/2013 09:13 PM, Liam Proven wrote: Well I suppose I should've qualified that. Yes, it has to be "better", but that's defined in different ways for different people. For most people today, I say through gnashed teeth, it means "cheaper". For people in upper management of most companies, it means "whatever will get the salesman to give me those tickets to the big game". To me, along with a disappointingly vanishingly small list of other people, it means "does the job better". That can be in terms of performance, reliability, manageability, or (more commonly) a combination of those things. (you know all this; I just wanted to spell out my thoughts) An *unambiguous* yes? Of course not. Neither would I. Yes, reference what I typed above. Yes. *grumble* This is certainly a valid approach in my view. I don't think it's the *best* approach, and it's not what *I* do, but I can't see myself faulting you for it, because it works for you. My approach is a bit different, and is summed up by the old line, "I'm not rich enough to use cheap tools." Yes, I'm an idealistic perfectionist, but I can function that way because I'm very lucky to have the knack of being able to find amazingly good deals on top-end hardware, be it HP/Tektronix/Fluke test equipment, Sun/HP/etc computers, Metcal soldering equipment, etc. It's not impractical to use top-notch stuff if you can get it without paying list price, and it pays in the long run. I get lucky, and I appreciate that because I know that not everyone else does. I understand. No worries. Yes. I acknowledge that this is a failing of mine. I am the most "trollable" person around. But here, there are good reasons for my reactions. As you well know, this forum is a common-interest one based on classic computers. It is not a list of technical professionals, industry-experienced engineers, or even people who have ever used what we consider to be a classic computer, or any other kind of computer. It's just people who are interested. And as it turns out, Very few of the people here have actually used these "classic" computers before this became a recognized hobby. Even fewer have used them when they were current technology. They weren't THERE, seeing and using them in "real life". Also, very few here have actually written software or designed hardware. Very few have designed a large-scale computer network, or even a small-scale one. I do not consider this to be a "fault", or anything even approaching a negative thing. It's just the way it is. One needn't have done any of the above to have an interest in, or a love for, classic computing, and thus be able to contribute meaningfully to this community. I *have* done these things. Now, before you take that as a blatantly egotistical assertion...I know I'm not the most knowledgeable person here, and I know I'm FAR from the most experienced. There are a few old-timers here whose experience and expertise I practically worship (they likely don't even know it), and I learn something from practically every one of their posts. (I wish my own S/N ratio were as high! ...something to aspire to) But a n00b I am not. I've done these things, I do them now, I'm a qualified, respected professional in this industry and I'm proud of my work. But this being a forum in which many (most?) people do NOT have that level of experience, I feel as if some people here (you in particular, when we get into these things) are talking to me as if I'm one of the ones who wasn't actually THERE, and who doesn't actually DO this stuff as anything more than a hobby. Lumping me in, as it were. That bothers me. It probably shouldn't but it does. So my "bludgeoning" reactions (thanks to Ken for that great image ;)) are somewhere between: "Yes, thank you, no need to explain further, I already know that." and "Don't try to feed me that line of bull, boy, because I've been there and done that." ...depending on the situation and, more specifically, the presentation. I particular, toward the second reaction, people have been trying to tell me that "serious computers" are "going away" for a very long time. We had a 386 box in the lab where I worked at DoD ~23 years ago, in a roomful of SGIs, and a couple of the guys just would not stop talking about the "SGI dinosaurs" that would be "replaced by PCs any day now". Those "dinosours" were brand-new, top-of-the-line visualization workstations with $50K price tags, and that's where all the work got done...because they were the best tools for the job, being used by scientists who understood why. Yes, PCs grew up (a bit) and replaced the SGI workstations...but fully TWENTY YEARS LATER. That's a long time. And SGI eventually became irrelevant not due to PCs being better in some way, but to their own management being absolute idiots. Huh?? Wow, I'm surprised that you see me that way. I don't think that way (at least not consciously) at all. I generally embrace new things, I love learning, and I have no issue with being proven wrong. But if someone tries to tell me something that I know damn well to be patently false, then yes, I react poorly. As above, no worries, I understand. Of course; we've all seen that trend. I think, though, to consider it to be 100% universal would be a mistake. Lots of people shop at Wal*Mart, but not everyone puts up with the garbage. Oh yes, you're a Brit...Wal*Mart is, by far, the largest retail chain here in the US. They have muscled themselves into every town and forced every other store out, and they carry the absolute worst, cheap Chinese plastic garbage you'll ever see. And they sell it the cheapest, so people buy it. And they don't seem to mind when they have to replace it when it breaks. They don't understand the false economy. And when Wal*Mart is the only store left, if they don't have what you need, you're screwed.

And most people just don't get it. High process and thread granularity server workload, mostly network I/O, effectively zero floating point. I ascribe it to a balanced architecture that was *designed* rather than slapped together. More specifically, I ascribe it to excellent memory bandwidth, multiple I/O paths, no significant bottlenecks, and a kernel that handles threading better than any other OS I've seen. Yes, that last item is an OS thing, not a hardware thing, but I treat this at the "system" level, including the OS. I'm right there with you on that. It's "better" than some machines based on metrics other than performance.

-Dave -- Dave McGuire, AK4HZ New Kensington, PA