On Fri, Dec 8, 2023 at 11:47 AM Waldek Hebisch <[email protected]> wrote:
> On Fri, Dec 08, 2023 at 10:50:13AM +1100, Hill Strong wrote: > > If the daase is so small that it only takes 23 Mbytes to load into > memory, > > it would be worth it if it took 100 Mbytes to load into memory. How many > of > > our machines would be constrained by this? My current machine is old and > it > > has 7.7 Gbytes usable. It is nothing special - a stock standard HP > laptop. > > > > The question to ask is: why has this not been done sooner? > > Well, memory is there but it does not mean that we should waste it. > I mean, if we gain something valuable, then sure use what is > If the memory is available for such a small quantity then by all means use it. Even an extra 100 Mbytes is relatively inconsequential compared to the usual amount of memory in these machines today. Having to do I/O is an extraordinarily slow process and any activity that requires this and can be mitigated by use of a small amount of memory achieves a significant speed up with little effort. It was true back in the '80's when main memory was measured in Mbytes and it is even more applicable today. I think one of the bigger speed ups and more effective directions for change is to look at alternatives to using any form of Common Lisp. That however is a personal opinion. needed. But using memory just becuse it is there is antisocial > it is only antisocial for multiuser machines or for machines that have a single user where large memory systems are concurrently in play. As far as FriCAS is concerned, how much memory is being used should be user configurable. and may lead to troubles. As an example look at GCL: it checks > how much memory machine has and then feels free to use most of > it. So on 64 Gb machine it feels that there is no need for garbage > collection before memory use hits something like 45Gb. One > If GCL is so bad that it cannot garbage collect under the constraints put on it by the user, then why use it in the first place? This is an implementation issue and shows a failure on the part of the developers of GCL to handle user defined constraints. That is something to take up with them. This is not an argument for not being able to use in memory databases for the FriCAS application. trouble is that in defualt settings GCL can not cope with that much > of memory, it gives errors because some parts can only handle 2Gb. > Once this is resolve there is another trouble: machine has 40 > logical cores (20 physical) and could usefully run 40 copies of > GCL (each on its own logical core). But at 20 copies machine > is thrashing as 20 copies try to use about 800 Gb and this is > much more than what the machine has. > This is a known problem that has a known fix. Of course, few systems today use this known fix. It is called separation of executable code from user data. If multiple instances are required for some piece of code, there's only a single copy of the executable code in memory which all processes use and each process has its own data store. > > To put is differently, programmers should exhibit common decency: > do not use megabytes when kilobytes would do, do not use gigabytes > when megabytes would do. > Which programmers? The programmers who develop the compilers, the programmers who develop the operating systems or the developers who are using these systems to solve the specific problems that they are writing code for? Or should the marketing and sales people for the companies producing the operating systems or compilers be the ones we get rid of first by putting them up against a wall and beating the living daylights out of with cricket bats? If you have had to deal with the last lot, you would understand just how much of a bane they are in this respect. > > -- > Waldek Hebisch > > -- > You received this message because you are subscribed to the Google Groups > "FriCAS - computer algebra system" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/fricas-devel/ZXJnixPwLn2Ql4jt%40fricas.org > . > -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/CAEnaMTG__N0JMuPgx6QWtvzeu5ZhUZfT30Q%3DEFhLrXB6Pr833A%40mail.gmail.com.
