Choose processor generation
Once we have chosen the processor with the highest possible number of cores, we must make the next decision, which is to choose its generation and therefore its architecture.
- AMD processors have the name spelled as follows: AMD Ryzen x yz000.
- Instead, those from Intel are named Core ix-xy000
Where:
- Where “x” can be 3, 5, 7 or 9. Where this marks the range within a family of processors and therefore the highest, most powerful, but also the most expensive.
- Where “y” refers to the generation of the processor and therefore to which microarchitecture it belongs.
- Where “z” refers to the pattern of each price segment. This is because we can find the same processors, but they differ in things like clock speed.
What interests us is the second point and that is that from one generation to another changes are made to increase its performance, which translates into the average number of instructions released per unit of time and , therefore, the processes that are resolved in the processor. For what a new generation processor in the same price range is always better.
Is the number of cores important when buying a processor?
We have to assume that each core is technically a full processor. The chips today are so called because visually we have a single piece or, failing that, several on the same socket. In the past, when we talked about a multiprocessor system, each core had its own chip. We now have a multi-core chip with a structure that intercommunicates them and an interface with memory in common.
Each of the programs running on your PC are multiple processes running simultaneously, some coordinated with each other and some asynchronous, but all are directed by the operating system. Since it is the main program and the one that tells the processor which core each process goes to. So, if we have few cores available, then a conflict or a bottleneck is created, since there are not the necessary resources to perform said process with the necessary ease.
It is therefore important to have more hearts. However, it must be assumed that there are parts of the programs that cannot be executed in parallel and depend on a powerful processor. By simplifying a processor, we can put many more cores in an area, but the part of each program that runs serially will be negatively affected. The point is reached where having many cores no longer gives performance. That’s why Intel and AMD focused more on more and more powerful cores and not on their huge quantity.
Hyperthreading and SMT
Both are two trade names for the same thing, the idea that each core can act as if it were virtually two different cores, but not actually. The concept is to use what we call bubbles in execution. These are periods during which, due to the latency between an instruction and its data, the CPU can spend time doing nothing. The solution? Make it context switch and run another much simpler process during th is time. It is for example when one of us takes advantage of the time to do small tasks like going to throw the garbage.
For what HyperThreading or SMT will give you an improvement, but it will be very slight because it will depend on the resources available to the processor. Moreover, since it requires the duplication of certain parts of the processor, its activation increases the energy consumption and reduces the clock speed. However, all operating systems today know how to use this feature and it is included in all processor cores.
Small or efficient kernels
Since SMT or HyperThreading cannot be used on low power systems for obvious reasons, the solution was to allocate bubble time to smaller cores. At the beginning alternately and today with these working in parallel. In the PC world, it was the Intel Core 12 that popularized this trend. Although this time the problem is not how much they consume, but the area. Where we can see size differences between 4 and 5 times inside the chip. Although the question is whether or not this influences when buying a processor.
Small or efficient kernels are a better solution than HyperThreading and SMT when running simple processes, they therefore contribute to the overall performance of the system. However, and as we mentioned before, if a process needs a powerful kernel, it will not benefit from it. If the operating system is not optimized for these types of tasks, it may mistakenly assign processes to inappropriate cores. The goal in the future is for such mapping to be performed at the hardware level, but at the time of writing this is not yet the case.
On the other hand, to say that they don’t add up is a waste of time, any additional resource that adds to the total computing power is welcome. It is better for a process to run on a less capable core than to be starved of resources by competing for resources on a larger core.
How important is clock speed when buying a processor?
Clock speed is how often a processor performs each of its tasks, so the higher it is, the more powerful it is. So once you have chosen the number of cores and the generation of the processor, you will have to look at the clock speed to choose which processor to buy. Because the higher the clock speed, the harder it is to get them, you find processors that are identical in everything except how many GHz or MHz they run at. The difference between them? A considerable price difference.
Our choice when buying a processor? We think that paying more for extra speed is dumb if you leave the rest of the PC neglected, since you can use this money to improve the rest of the components. On the other hand, it is necessary to start from the fact that at certain speeds it is necessary to use more advanced cooling systems. It is therefore necessary to choose a box with a good ventilation system and even a good heatsink for the processor. Sometimes having the fastest CPU doesn’t necessarily mean having the best PC, so keep that in mind when shopping for a CPU.
With or without integrated graphics?
We finally have the right model for our PC, but we find that we have two different variants where the only difference is the integrated graphics card or iGPU. Our point of view ? Opt for the version without iGPU, given that you are going to install yes or yes a graphics card in the PC and iGPUs unless we have chosen an AMD APU in general tend to be quite poor in power. In other words, they serve to move the desktop of the operating system and nothing else. In addition, many applications today use the power of the graphics card to speed up certain processes, so it is recommended to have a dedicated graphics card. It is therefore advisable to allocate part of your PC budget to the purchase of one.
Table of Contents