Hypothetical question

[Dogbert]

Let's assume that someone comes with a new revolutionary free OS or Kernel patch of some kind that'll allow people to maximize their hardware potential. Let's assume this will mean 25%-60% better realworld performance.

Would it be good or bad for the industry ?

In the short term, such a thing would mean that nobody will need to upgrade which in turn will allow people to save money which they'd propably spend on other things.

Would the owner of such an invention survive long enough to finish it if it gets publishment before it's finished ?

[Jammrock]

In theory that would be good for the hardcore gamer. People who have no ph34r of formatting and setting up a dual boot, tweaking drivers, etc.

The idea of a 'GameOS' has been tossed around a lot, but the chances of a true GameOS succeeding is slim, so no one has really tried.

Jammrock

[Dogbert]

When I said "assume this will mean 25%-60% better realworld performance" I didn't mean only 3D, I'm talking about anything from pong to server farms and huge DBs.

I'm talking about a piece of code that'll revolutionise all the industry.
Think about a multimillion $$$ corpotation with a server farm that starts to get too slow. They're thinking about an upgrade when this kernel patch comes out and gives them 30% more performance from their current setup.
They've just saved huge amounts of money.

People will have no reason to buy the latest and greatest CPU which comes at a premium cause they can get now 50% more from any CPU they buy and no game is ready to fully utilize such a performance boost.
This could almost mean that the hardware industry will come into a halt till the softwre industry catches up with this boost of performance.

[Gurm]

Yeah, everyone claims to have written it, too.

The latest culprit:

http://www.dachsundsoftware.com

Their produce, "Hare", supposedly boosts your machine speed by gajillions of microticks, and one of their OTHER products will give you +10fps in every game regardless of whose video card you use. Their final product claims to halt all crashes in windows. Last I heard, Symantec couldn't pull this off with Win98, so they gave up. Nevermind hard crashes in Win2k/WinXP.

- Gurm

[Pace]

50% ain't that much. Even if I got a 50% boost it wouldn't be enough. To shake the industry, you'd probably have to come up with an alternative to the transistor/microprocessor/processor materials which will be ready to market within months, and offer 10 - 100x processing power for comparable prices.

As soon as anyone heard that, people would only buy PCs if absolutely necessary, or if they were sceptical of the new PaceyTech™ Processing Advances. Dell, Compaq/HP, Intel, AMD, NVIDIA, VIA would be left with masses of PCs and processors which will be obsolete within months.

Sound unbelievable? I'll be publishing my white paper on PaceTech™ Xaja technology soon, and products utilising it within the next few months.

P.

[thop]

unless you program a new OS in x86 assembler this is not going to happen IMO. and when you're ready with that OS everyone will have a 20Ghz CPU. thats probably why most people don't invest time to optimize their code but instead just throw cpu cycles on it as the computer speed increases so fast.
another thing is probably that marketing wont let them

[Pace]

He was speaking hypothetically thop - "if" someone made this revolutionary advance, what would happen - not "will it happen"

However, you mention it wouldn't matter because everyone will have a 20GHz CPU soon. Let's look at WindowsXP and OfficeXP....if Microsoft keep adding bloat at this rate then 20GHz CPUs will be mandatory pretty soon - and a 50% faster OS would *always* offer 50% more, doesn't matter if it's a MHz CPU, GHz CPU, or a THz one!

Software programmers are generally pretty lazy - I'm sure we all know that almost every large application we use isn't optimised too well. If Microsoft coded Win2000 from scratch (like they supposedly did the first time), making sure every section of code was extremely optimised (whether that be with algorithms or low-level programming), including DirectX, and if John Carmack coded more optimally as well (rather than adding features) - I'd guess we'd have a 50% boost in performance.

NVIDIA have shown they can do it with driver updates - are we to assume that Microsoft and ID Software get their first versions perfect?

P.

[Ribbit]

I think it would make no difference whatsoever.

Look at the abilities of modern PCs being sold today, and look at what the vast majority of them end up doing. Sure, a few get pushed near their limits running games and heavy server/workstation stuff, but most of them end up just running web browsers, email clients, and word processors. And yet people will still buy 2+GHz processors with 512MB of memory, and waste it all on the above. Dogbert's hypothetical invention would just mean there's more going to waste.