Answer the question
In order to leave comments, you need to log in
Why are new processors better than old ones, although weaker?
Example:
FX 8350 and Ryzen 1200
8 cores - 4 GHz and 4 cores - 3.1 GHz
https://overclockers.ru/lab/show/88687/podvodim-it...
What is the secret?
Answer the question
In order to leave comments, you need to log in
More frequency and more cores - right. But this is how to compare modern cameras in terms of the number of megapixels. Modern processors are very complex and have not been "stupid" executors of the written program code for a long time. For starters, no assembler instruction (machine opcode) is executed "directly" on the processor. The firmware parses x86 opcodes and turns it into even lower-level code. The CPU registers you've probably heard of (eax, ebx...) are nothing more than a virtual layer. The microcode of a modern processor can be thought of as an emulator of the x86 architecture; real operations are performed on a completely different architecture. Herein lies the rub. For example, the integer division operation in x86 is indivisible. But it is converted into several lower-level instructions and, as a result, not one processor cycle is executed, but several. More modern processors may have faster versions of the same operations. As a result, less gigahertz, but in fact it works faster.
Next, a whole world of code execution optimizations. Branch prediction, speculative execution of operations (the command is executed even before execution reaches this point, if at all), rearrangement of unrelated commands in places and their parallel execution on several ALUs, pipelining ... In new CPUs, these algorithms are being improved and improved, and now the acceleration of the processor is mostly due to this.
More and more new SIMD "extensions" of the architecture are being added, new sets of commands, with which you can greatly speed up the execution of the same video encoding (SSE, AVX, etc.)
Finally, the new CPUs have faster hardware in terms of memory (cache) and connections (data bus), the exchange with RAM is much faster, memory read / write commands slow down much less. Communication between the cores in Ryzen is completely different (Infiniry Fabric).
If so, then the Pentium 4 was also released with a frequency of 4 GHz, the limit is somewhere in this area. A further increase in frequency is only possible by moving from silicon to graphene. So they suffer with the addition of cores and software optimizations.
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question