IRQ's are a very small part of how an OS works and interacts with the hardware. There's much much more involved than "when they get their turn to talk."
This is an overview of the first NT release of Windows and how the OS itself is constructed to create many layers between what you see and the computer itself. Benchmark software is designed to test hardware. Sending commands through a specific set of channels to see how well they perform with mathematical equations.
Real life use and diverse application testing, along with diverse code execution on differing subsystems and kernel segments is going to paint a different picture. Benchmarks are not the end all, be all of performance. You find them everywhere because people who don't understand the finer details like to look at pretty graphs and turn around and give their own anecdotal interpretation of two colored bars. They were good when you compared two different Video cards while every other variable remained the same. Putting a tractor on different ground may give you a good idea of how stable the soil is when you it move back, then forwards and dig a small hole, but once the real work begins, your driver may have different feedback.
That is completely false. Describing a microprocessor in business management terms while injecting a few jargon words is pretty risky too.
A task is never a task in computing these days. A task is a list of dependencies and their dependent tasks. You could have a few lines of code take longer to execute than a 3Mb batch file. Not because the complexity, but because of the exponentially larger number of other calls that need to be made.
Pipeline management is important, I agree; we're already there though. Unless you're running a lab like CERN, these little things are not going to matter much considering the bottlenecks in mainstream PC's exist in many other places. This makes the techniques or managing the flow of information and requests a moot point until such a time as everything else is upgraded.
Speed of the CPU is very similar to the above. Mainstream computing doesn't require it like everyone thinks it does. The bottlenecks exist elsewhere. You can't honestly think that being able to reach 4Ghz 14 years ago was an accident that can't be reproduced easily/cost effectively right? The speed is not as big of an issue as everyone thinks. People buy computers under misinformation... because processor A runs at 2.2Ghz, and processor B runs at 2.4Ghz, B is automatically better. Again, these little things have specialized purposes, and what you do every day on your computer is not one of them.
Last edited:
