Видео

Keep going 💪

The speed of a processor is how many clock cycles a processor runs per second, not how many instructions one can execute. The same goes for RAM and video cards. Back in the 1990s and before when we needed to use assembly/machine language for parts or all of our project's code, we had to know how many clock cycles each instruction took on the specific processors our projects were coded for, calculate how many clock cycles each algorithm in our code executed, and had to get creative to reduce the number of those clock cycles our algorithms took through best optimization practices and through any extra benefits the system provided us like when the math coprocessor was first introduced on the Intel processor. This was very important in areas like speeding up the graphic renderings in our projects, especially games. Thankfully today we can use the GPU on the video card to do most of the work for us instead of trying to do it ourselves on the CPU. If I remember correctly, a simple division instruction in machine language on the Intel 386 processor took 300-600 clock cycles, and multiplication took 60 or more clock cycles, which is why we used to have the saying, "Multiply if you must but never divide". So, we really had to get creative with trying to find crafty ways around having to use division in our algorithms if we could. Of course, even back then, like it still is today, the arrangement of our machine language instructions is critical to how fast our algorithms execute, since if the instructions are sent in the "wrong order", it would cause the processor to stall for a number of clock cycles before the next instruction is executed. On modern Intel processors, we have multiple instruction pipelines that allow us to execute multiple machine language instructions in one clock cycle. With the event of multi-core processors, more instructions can be executed concurrently. This is why the speed of the processor does not indicate how many instructions it executes per second. On a further note, modern Intel processors, with cache memory, read in chunks of data and code and stores them on the processor for extremely faster execution since it doesn't have to transfer the information on the data bus as often. This is why it is important in coding to make sure you place functions that work more together, closer to each other so the CPU does not have cache hits that cause the processor to fill the caches again with more data and code. It may be beneficial if you have a look at the 5,000+ page PDF of the "Combined Volume Set of Intel® 64 and IA-32 Architectures Software Developer’s Manuals" located at: www.intel.com/content/www/us/en/developer/articles/technical/intel-sdm.html The Intel Software Developer's Manual will tell you almost everything you need to know about most of the processors they have built. I hope this information helps. I would be happy to help any who has any questions on what I stated.

Good job

fantastic video!

Technically, that actually says there is only one kind of person in the world. 0 isn't really a number, it's more like saying, "not one".

Well explained 👏

Good one.. Keep sharing

keep going bro.. you are one of the best

I will say, it would have been more accurate to show a sine wave, rather than a triangular wave, in the initial explanation of a Hertz. Every fundamental frequency is a sine wave. Other wave forms are composites of sine waves. And, clock inputs are usually square waves, with the device being clocked triggering on either a leading or trailing edge, usually. The clock is a bit more than just synchronization, which is more a matter of using the same clock throughout, or clocks based on that clock. No data moves anywhere without a clock pulse. The data going to a register is presented at its inputs, but the state doesn't change until it is clocked. There may be different parts of a system that use different clocks, but they'll usually be divided down or multiplied up from the same base clock source, providing the synchronization for the whole system. So, yeah, a computer instruction is going to take more than a clock pulse to execute. Every move the data makes along the way, is a clock. But, some instructions will take more cycles than others. Your measurement of MIPS or whatever sort of depends on the mix of instructions measured with. The clock, however, is more or less the heartbeat of the system and has a direct relation to speed of execution. More efficient instruction sets would also increase speed.

Such a great explanation 👏 👌

Dude this was such a clear and succinct description! Great video.

👏👏

👍🏻

Excellent

Great efforts.