Calculating Latency

[RadBrad]

Well, how do I calculate latency (without the x( BS) formula, preferable) Isn,t latency actually a product of bad code? I,m sure that Intel and Amd didn,t design it into their chips, or maybe they did!
I am sick and tired of not being able to prove a point because "you gotta account for latency". There must be an end to where it begins and, as such, a mathmatical equasion to express the distance between the two.

I,ll check back

 

[Shalmanese]

Originally posted by: RadBrad
Well, how do I calculate latency (without the x( BS) formula, preferable) Isn,t latency actually a product of bad code? I,m sure that Intel and Amd didn,t design it into their chips, or maybe they did!
I am sick and tired of not being able to prove a point because "you gotta account for latency". There must be an end to where it begins and, as such, a mathmatical equasion to express the distance between the two.

I,ll check back

Uh bad code? I dont think so. It's a physical limitation related to a myriad of factors such as the speed of light, capacitance etc.

 

[Peter]

There's always a latency in getting stuff done. This is all natural and quite obvious that outside of sales and marketing, getting stuff done "in no time" doesn't ever happen.
What's your point, and what's your question again?

 

[Mday]

Originally posted by: RadBrad
Well, how do I calculate latency (without the x( BS) formula, preferable) Isn,t latency actually a product of bad code? I,m sure that Intel and Amd didn,t design it into their chips, or maybe they did!
I am sick and tired of not being able to prove a point because "you gotta account for latency". There must be an end to where it begins and, as such, a mathmatical equasion to express the distance between the two.

I,ll check back

first, there are two terms commonly used: latency and delay. latencies are typically a sum of delays.

there is no end. latency is a function of the speed of electrons and the delays in circuits. you can simulate all the delays you want, and you can get statistical data for latency. that is all.

latency is not nor was it ever a product of bad code. yes, they did design delays into their processors. why? because delays are required for things to be synchronized. otherwise you'd be putting socks over your sneakers. things that go fast often has to be delayed so that things that go slow dont cause problems.

second, you need to think more of what you want to ask, because you're either asking a lot of questions, or just ranting about on things you dont fully understand.

 

[RadBrad]

Originally posted by: Mday

you can get statistical data for latency.

Where or how? This is what I need. I don,t have self esteem issues, so what would be the "point" of asking a question if I had a full understanding of what I ask? Maybe this question is more than highly technical.

I will be looking forward to your direction

 

[Tuan]

1) Buy a stopwatch
2) Buy a laser pointer
3) Point the laser at a wall
4) Start your stopwatch and turn the laser pointer on at the same time
5) Stop the stopwatch as soon as the light hits the wall

Congratulations, you've just measured latency.

p.s. If you want a real answer from anyone, try asking a coherant question.

 

[ZeroNine8]

Testing actual processes would be one way to find statistical data about latency. For example you can find that the delay between X and Y, where X and Y are two steps in a process, is normally distributed with some mean value and standard deviation. Because there are so many significant factors that contribute to latency, some which are in your control and some which are not, it is impossible to say the latency is exactly any particular value. Even if you can rigidly model many steps of your process, there are still many which you will be unable to model except with statistical probability. This is true for most systems, actually, not just processors. With larger and slower systems, however, the random variance tends to be much much smaller with respect to the values you are dealing with, so it is usually neglected.

 

[Matthias99]

The problem is you're asking a very vague question. Several, in fact.

You get statistical data for latency by hooking up high-speed oscilloscopes, etc. to a production sample and analyzing how a circuit behaves. You can try to model it in SPICE or other circuit simulation programs, but they're never 100% accurate.

You ask "how do I calculate latency?", but you never specified what kind of latency you are, in fact, talking about. If you have a more specific situation in mind, asking about that might get better answers for you.

I am sick and tired of not being able to prove a point because "you gotta account for latency". There must be an end to where it begins and, as such, a mathmatical equasion to express the distance between the two.

What points are you trying to prove? What do you mean by "an end to where it begins" and "the distance between the two"? Ask clearer questions and ye may receive clearer answers.

 

[Witling]

"second, you need to think more of what you want to ask, because you're either asking a lot of questions, or just ranting about on things you dont fully understand."

An amateur sees many solutions. An expert sees few. Zen saying.

It's frustrating to both parties when one doesn't know enough to ask a question in terms that can be answered.