OT: Good place to ask about XP memory problems

[Man at B&Q]

[e-mail]...

No, that's not what I mean at all.

MBQ

[John Rumm]

Ow now let me think.... how about Kernighan and Ritchie's at Bell Labs. You may have heard of them - the clue was in the word "original".

C's reliance on a pre-processor enforces at least two passes even on many modern compilers. Some can precompile header files, but quite often that only works well when the same set of files are included in the same order in every module.

[Man at B&Q]

If the data needs to be portable then it is a requirement of the application to provide a method to export/import the data in a portable format. Whether the app is written in C, or..., is irrelevant.

MBQ

[dennis]

well anyone can make a mistake

[dennis]

messagenews: [e-mail]...

So you mean don't use any C data types then.

[John Rumm]

"everything" as long as you ignore the fact that the majority of CPUs on the planet are ARM derivative I presume you mean?

[dennis]

Now you have ruined it! ;-)

[John Rumm]

They also fixed the 8088/6 double prefix bug which broke some apps...

(don't think a slightly faster 8086 would have dug this particular project out of the mire anyway - it was still a tad sluggish in places on the order of magnitude faster '386)

[John Rumm]

Yup, done loads of commercial stuff in Turbo pascal and later in Delphi.

[John Rumm]

I always have to smile when you meet IBM types who think that "assembler" is an IBM specific language and can't get their heads round the concept that everyone else has a far more generic definition of the term.

8051 is kind of weird with its bit addressing. Might be quite handy for some apps if it were not for the fact that the bus cycle is so damn inefficient as to render any coding performance gain pointless.

[John Rumm]

16 bit external bus...

[John Rumm]

Mad even worse by Cyrix with things like the 486 SLC which were narrow bus (and being realistic more like a 386)

[Bob Eager]

Funnily enough, I mentioned to someone this week (while offering to teach a course on assembler for some geeky types) that one should call it 'assesmbler', not 'Assembler' - the latter being the IBM term. They used lots of generic terms for their specific products - JCL and VM come to mind.

[John Rumm]

The DX became the original 486 when the SX was introduce. The DX2 and DX3 were the multiplier versions.

[Bob Eager]

Yes, it was really good on benchmarks because of the improved multiply/ divide performance. I have a feeling the pipeline was different too, but the stuff is all on a shelf at work. Overall, not that different - I did fit one to my PC though. (in place of an 8086, I had one of the first Advance 86 machines)

[Bob Eager]

I went on a four day Intel course on the 80286, and they mentioned it there. But marketing may not have bothered in the end.

[John Rumm]

In the early days of 386 PCs there was not much use made of the MM capabilities. However that changed with utilities like QEMM and 386MAX plus windows 3.1, also the use of LIM EMS/XMS memory managers that could do away with the bank switch hardware initially required to page in EMS blocks in the upper memory area/EMS page frame.

Everything the 286 had was present in the 386... It added translated memory paging, virtual 8086 support, 32 bit ness, and hardware debug registers etc. Along with much greater addressing capability (both physical and logical)

Yes you can. The 386 architecture just extended them to allow for virtual 8086 segments and the like.

[Clive George]

DX2 and DX4, x2 and x3 multiplying. Can anybody remember Intel's justification for the DX4 name other than to try and pretend things were faster than they were?

[The Natural Philosopher]

Jesus H Christ.....

[The Natural Philosopher]

I think there was a crappy Turbo C thing that did maybe one pass..well it appeared to be one pass anyway..

Dennis sounds like a Borland twit.