Thread: sine curve representing Microsoft

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

I cannot wait for the next MS Windows release

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!)

Originally Posted by DavidP

(And it's meant more as a joke anyways, people!)

But that doesn't account for the missing OSes, does it?

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 . . . .

Originally Posted by DavidP

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!)

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.

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.

Originally Posted by Salem

> 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.

--
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.