Printable View
So I was originally going to post this on the "Thinking about upgrading to Linux" thread, but I figured that it would be too much of a tangent, and would probably take the thread way off topic, which I didn't want to do, so instead I decided to start a new thread about this.
I decided that Microsoft can be represented using a sine curve.
Here is sin(Microsoft) plotted on a graph:
Attachment 7723
Meh. Interesting pattern. I'm really not that petty btw.
I cannot wait for the next MS Windows release :p
It's wrong.
You missed windows 2000 which proceeded XP. That would put XP on the lower end and Vista riding high - not quite reality
I don't know if Windows 95 is supposed to be on the low curve - maybe more like 98?
Windows NT4.0 was also missed out.
This graph doesn't include anything from the NT line (except for XP...which can be considered NT). (And it's meant more as a joke anyways, people!)
*points to his indifferent expression*
But that doesn't account for the missing OSes, does it? :DQuote:
Originally Posted by DavidP
The sine curve works better with Intel CPUs. Starting with the 486 on the high position, then the Pentium on low. Pentium Pro had a considerably better design than the Pentium - high. Pentium II was meh. Pentium 3 brought new improvements - most importantly SSE. The Pentium 4 was here for a long time and reached high frequencies - but was usually totally smashed in overall performance by AMD's chips. Then, finally, the Core architecture, abandoning the high-Hertz philosophy and bringing out the uncontested best chip on the market. Only with the Core2 did Intel manage to break out of the curve.
No, that's all wrong, you have to put the Y axis just after Windows Vista, to represent time. :)
And you could add a third axis to represent number of users or something . . . .
Actually XP 'IS' NT 5.0, they just changed the designation. 2000 was part of the 95/98/ME line, the final one I thought until Vista came out, seems they never throw away code, no matter how buggy. I think they have to pay IBM royalties for parts of NT, since its based off the joint venture of OS/2. Maybe this is why they keep trying to get their fully proprietary IP to work.Quote:
Originally Posted by DavidP
And actually, 95 OSR2.1b wasnt that bad, I still keep a copy around for test purposes, along with NT 4.0 Workstation.
> 2000 was part of the 95/98/ME line,
Are you sure?
My 2K machine clearly states "based on NT" technology.
ME was the last of the 16-bit DOS wrappers.
Yes, having written drivers for 2K/XP, I know that this is a quite different driver architecture than that of 95/98/ME line of OS's. 2K is NT 5.0, XP is 5.1 - I guess vista is really NT6.Quote:
Originally Posted by Salem
--
Mats
Im positive. 'Based on' could mean anything from completely dirivative work to 'we copy and pasted the idle thread to make the marketting people happy'. I suspect its more of the latter than the former. I agree with the original poster though, MS does seem to go through cycles of crap/gold/crap. Its not just their OS's its their compilers as well, 6.0 gold, 7.0 (2003) crap, 8.0 (2005) gold. The only problem I have with their newer compilers is they have their head up their ass and refused to impliment inline assembly for 64 bit applications. This is THE reason it is taking hardware companies so long to develop new drivers for 64 bit, since drivers are heavily coded in assembly. Linksys has flat out refused to support 64 bit until there is 64 bit inline assembly. MS doesnt want to spend the time adding that feature because its not low hanging fruit, and they think it will only serve a niche market, but it really effects a lot more people than just driver writers. The game industry has traditionally used assembly to speed up critical sections of code and engineers very often do the same. Intrinsics just dont cut the mustard for HPC optimizations. MS would rather you just use CopyMemory(), even though it is about 35% SLOWER than an assembly implimentation.
Overall I think the problem is a matter of mangement being excellent at business but worthless at R&D. I just dont think they can get out of the comodity mindset.
But still wrong. Win2k is NT 5, XP is 5.1. There isn't really anything from 9x in 2k.Quote:
Originally Posted by abachler
I agree, there should be inline assembler in the Visual studio compilers - unfortunately, the men in control of that don't agree for some reason.Quote:
Originally Posted by abachler
When I say "based on", I don't mena "a small part of the kernel", but rather that the majority of the kernel is identical or slightly "improved" version of. The kernel architecture of Win2K is the same architecture as NT4, which is dramatically different from the Win9x kernels.
For the most part, drivers are NOT written in Assembler. The reason there is fewer 64-bit drivers is simply that most machines don't use 64-bit OS, and thus the demand for 64-bit drivers is small.
If you do memcopy to a large extent in your driver, you probably do something a bit wrong - but if you can't avoid it, then you can still write a memcopy function that is faster, just use MASM - which IS supported for 64-bit too.
As to 2K being "more similar to 9X" is probably based on the fact that the 9X look'n'feel was (sort of) moved to 2K, along with Direct3D, which wasn't supported by NT4.
--
Mats
I'm not certain about the time base... any chance of getting the service packs in there too?
You'll also see that the product name for 2000 is NT5 and XP is 5.1.