Author: Vincent Diepeveen
Date: 08:26:36 08/04/03
Go up one level in this thread
On August 04, 2003 at 09:43:40, Bo Persson wrote: >On August 04, 2003 at 07:59:48, Vincent Diepeveen wrote: > >>On August 04, 2003 at 07:43:04, Bo Persson wrote: >> >>>On August 04, 2003 at 05:41:55, Vincent Diepeveen wrote: >>> >>>>On August 04, 2003 at 04:39:56, Bo Persson wrote: >>>> >>>>hi, >>>> >>>>it's not so difficult to get a look&feel for it. >>>> >>>>a 5 line tool of mine that just prints text out of binary files. so not a >>>>dissassembler at all. >>>> >>>>Kernel32.dll : nothing what i searched for >>>>winfax.dll : nothing what i searched for >>>>gpedit.dll : >>>> >>>>"normal program termination R6009 - not enough space for environment R6 >>>>008 - not enough space for arguments R6002 - floating point not loaded >>>> Microsoft Visual C++ Runtime Library Runtime Error! Program: .. >>>>. <program name unknown> IsProcessorFeaturePresent KERNEL32 " >>>> >>>>Do i need to spell it out? >>>> >>> >>>Yes, please. >>> >>>Do you mean that kernel32 does not link to the C library? Surprise. I wonder >>>what services the C library is linking to... >>> >>>On my NT4 machine I can also see than kernel32.dll is 0.3 MB out of the 358 MB >>>in the /winnt directory. That's about 0.1%, or so. >>> >>>I can also see that the hardware interface hal.dll is 51k. Might also contain >>>lots of assembler code. >>> >>>What about the other 300+ MB ? >> >>So in your eyes the kernel is 300MB????? >> >>are you out of your mind? > >:-) > >> >>important is functionality like a program that asks a system function like >>GetTickCount() >> >>Then you want that as quick as possible and C code won't do that for the kernel! >> >>Anyway this is all theoretic discussion. >> >>It's assembly. > >Vincent earlier wrote: > >>>I guess they are working hard and basic problem is not only that m$ is 32 bits >>>in some respects (file systems and such already long period ago 64 bits in 1995) >>>but especially that their kernel stuff is written in assembly. >> > >>> It is hard for me to understand why so much is in assembler and why there >>> seemingly is no x bits 'C' version which they can compile on any hardware >>> with little effort. > >Ok, we agree that there is "some assembly required". I can see that it might be >0.1%, or 0.5%. or even 1.0% of the code base. I wouldn't call that "so much", or >see that it would be a problem to move it to x86-64 when they have already tried >it out for Alpha, Itanium, and others. > >If you think otherwise, fine. you are just guessing a % here. the entire kernel is assembly though. let's be clear. i'm not saying that the entire kernel being assembly is a problem to move it. seemingly reading nalimov's words correctly they have a C version too, which for the itanium i can consider as being a good idea. writing assembly for it is a horror. For the x86-64 i bet they want an assembly version too, because the platform is going to be very important. You cannot waste registers with a kernel simply. Using regurarly 1 register just for a processortickcount is already wasting 1 register. So for CPU's with just 16 registers you definitely need assembly for a kernel like NT. just consider how much software most people have installed. it only adds up and they all regurarly query the kernel. linux does not have such problems. there is hardly software working for it! >Bo Persson >bop2@telia.com
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.