IBM 610 workstation computer 3447
Who is the I in this last sentence? I am stating that the I is the OS. It cannot be hardware. It is only the OS who knows which data is the "correct" to use. In some cases, the "old" data is the correct one...they're bizarre cases and usually concern exceptions but this situation still can exist.
You have yet to achieve single photon wires. As soon as you do achieve that, people are going to build system that need 1-2 photon-sized wires. This is just typical human gear evolution. As soon as you make X easy, it will be used past its limits and a demand will ask for an X'.
You sure as hell do. There is one case where this is necessary. That is STOP! RIGHT NOW! DO NOT DO ANYTHING! broadcast.
ARe you kidding!!! Nobody know what the market is because nobody has had a change to use these things. Can you imagine what kinds of thing are going to be developed once any one who has a compute intensive calculation can just go typeity type on his laptop and not have to have been a grad student at a university who owns Cray for ten years?
IBM 610 workstation computer 3450
ref: the problem was that with one cpu ... the cache ran at full-speed. with two cpus, the cache ran at full-speed ... but...
IBM 610 workstation computer 3452
Using any approach requiring a global system lock seems to prevent any system from scaling to more than 4-5 CPUs...
Kick,kick,kick emoticon aiming at arse
Yes, I see that as problem, too. But I see it as the implementor's mindset as the cause.
I do. The whole mindset of the soft-hardware biz says so.
Yes, and you are speaking with somebody who has watched it all... most of it.
IBM 610 workstation computer 3448
Sigh! Who is the it in that sentence? This question is the most important section...
No, it doesn't. It works as badly as it used work.
Yea, well. If I could write the equation, we could then go to an interpreter who will change it to words.
HAH! I just a.f.c.'ed the thread.
And the art of a timesharing OS is to be able to do this synchronous stuff asynchronously.
EXACTLY!!! Now you're starting to think.
And the "person" to correct that will be the OS, not the hardware. Think hot swap. Think catastrophic hardware recovery. Think of any transaction system whose top priority is to continue working and its second priority is correct data.
But it's not. In our day, this problem was disk based. This thread is talking about the problem but is CPU cache based.
That's right. You sent a smiley but this is exactly what I've been talking about. It is you, the human, who knows what bits on your system, are important. There is no way the hardware can decide this. Yet you hardware types are saying that the gear can determine this. A gal with a crystal ball will have a better hit ratio than any hardware.
IBM 610 workstation computer 3453
this is somewhat dependent on the traditional spin-lock approach to global system lock. this was almost totally eliminated with the VAMPS bounce lock approach (reducing the cost of having lock...
Whose order? The CPU architect's? The OS developer? The app developer? The governement? The hog? (There's lots of computing in the farming biz.)
Yes. Now, is that device tied to one, and only one, CPU? Or is there a multiple pathway to that device such that any CPU can deal with it? Think about done interrupts, e.g. If the device is tied to one CPU, only that CPU can detect and deal with the done interrupt. If the device is tied to any CPU, then any CPU can respond to the done interrupt. This is what I call "increased thruput". Because the user who is waiting for the command prompt after type EXIT will get his prompt "faster" wallclock time and be able to issue the next command.
And I've got 2006-1968= goodgriefing emoticon looks for gray hare 38 years. And I was intimate with real timesharing.
You are pbutting the f***ing buck. I said you talk to them. Learn.
With a new hardware generation coming out every two years, no software guy has the time to deal with it. Part of the problems we see today in this biz is that OS has become production line work with all art form taking out of it. Nobody has time to play. By the time an OS guru has become intimate with a piece of gear, all that knowledge has to be poocanned because a new one has become top priority.
IT IS NOT ANOTHER TOPIC ENTIRELY. IT IS EXACTLY WHAT i'VE BEEN TALKING ABOUT ALL THIS TIME.
That depends on whether the human want's it to add correctly or add incorrectly.
emoticon watches analogy idea fizzle into the ground in a poof of purple smoke buttume that no other CPU will handle interrupts. It buttumes that all events on the system will funnel through itself. So quality control is ensured because each piece will be examined by the single CPU. When you add a second, now you have two things that can make a decision. And we all know what happens when there are too many chiefs and not enough Indians.