Linux Core Scheduling Nears The Finish Line To Avoid Flipping Off HT

[polarathene]

Going from previously 100% (150ms) to 33% (50ms) is 67% latency improvement, not 300%...

from 50 to 150 is 300%; from 150 to 50 is 66.666%. Both of those were 150 and are now 50

So the difference between 30 FPS (~32ms) and 60 FPS (~16ms) is 50%? Latency was halved, frames per second doubled.

If you go from 1 second to 1ms as an improvement is it 0.001% or 1000x better? (You can do it in 1000th of the time)

Slower units of time below one second are larger, the scale is flipped. 1 minute is faster than 60 minutes, 1 microsecond is faster than 1000 microseconds, but the difference is the unit boundary, 1 minute is 60 seconds, but 1,000 microseconds is 1ms.

In that sense 150 to 50 is equivalent to 50 to 150. What matters is the context:

- Latency reduced by ~66.7%, from 150ms to 50ms.
- Latency reduced down to just 1/3rd (33%) from 150ms to 50ms.
- Performance per ms improved by 3x (from 6.7 to 20 FPS).
- Performance per ms improved by 200% (implies a 100% baseline as 2x, so 200% == 3x).
- Latency is down to 50ms, a 300% improvement over the previous 150ms! (inaccurate % improvement, but reads better and understood just the same as a multiplier given the context)

[F.Ultra]

Going from previously 100% (150ms) to 33% (50ms) is 67% latency improvement, not 300%...

The latency is decreased by 67% which is a 200% improvement in performance.

[zboszor]

The latency is decreased by 67% which is a 200% improvement in performance.

Even that is not always true. The article said the software response latency to a keypress was reduced to 50ms from 150. But this won't make you suddenly type faster or decrease your perception latency (the delay between when you act on something you just heard or saw), which is a known human limitation and is about 200ms on average.

[Weasel]

What does a 300% improvement in keypress latency mean?

Let's say that previously keypress latency was 100ms (just making up a number). A 50% improvement would make keypress latency 50ms. A 75% improvement would make keypress latency 25ms. A 99% improvement would make keypress latency 1ms. 100% improvement is impossible. that's a 0ms latency.

So what, exactly, does 300% improvement mean in this context?

150ms / 300% = 50ms

I suggest you learn how to divide instead of just multiply. That applies to things where "less is better" such as latency. You divide.

[Weasel]

The latency is decreased by 67% which is a 200% improvement in performance.

It's a 200% improvement in latency as well.

When improvement means "less is better", you divide instead of multiplying.

You guys are the reason so many games have such shit math formulas for calculating damage and defense and why defense is always more overpowered to stack, and the idiot developers code workarounds for this instead of just fucking changing multiplication to division.

Division must be too hard for some people.

[F.Ultra]

It's a 200% improvement in latency as well.

When improvement means "less is better", you divide instead of multiplying.

You guys are the reason so many games have such shit math formulas for calculating damage and defense and why defense is always more overpowered to stack, and the idiot developers code workarounds for this instead of just fucking changing multiplication to division.

Division must be too hard for some people.

Not really sure what you mean by "instead of multiplying", both numbers are calculated by division ( (new-old)/old for the one and (old-new)/old for the other ), the only difference is that some people have a hard time understanding the difference between "decreased by" and "improved by".

No idea whatsoever how games calculate damage and defence, I always thought that they performed simple addition like D&D did (at least the rpg:s that I have seen seams to use the D&D logic).

[Etherman]

Guys, "Linux Core Scheduling Nears The Finish Line"!!!
Not "Learn Math with Photonix"

[angrypie]

I'd be wary of learning anything from Phoronix.

[Jaxad0127]

150ms / 300% = 50ms

I suggest you learn how to divide instead of just multiply. That applies to things where "less is better" such as latency. You divide.

150ms / 100% = 150ms. Is that a 100% IMPROVEMENT?

When talking about improving or 'more than', you need to take into account the base value, since you're adding to it (or subtracting from it).

10 + 100% = 20 -> 20 is 100% MORE THAN 10

10 * 200% = 20 -> 20 is 200% OF 10

[Weasel]

150ms / 100% = 150ms. Is that a 100% IMPROVEMENT?

When talking about improving or 'more than', you need to take into account the base value, since you're adding to it (or subtracting from it).

10 + 100% = 20 -> 20 is 100% MORE THAN 10

10 * 200% = 20 -> 20 is 200% OF 10

And? You just forgot to add the base 100%. That's your problem.

e.g. when something is 5% faster, it's the speed 100% + 5% of the original, aka 105%.

That has nothing to do with what I said, which is that for things where "less is better" (such as latency) you divide instead of multiply when it comes to improvement.