AMD performs well across many instruction sets.
AMD performs well across many instruction sets.
They mentioned AMD once promoted their 8-core chips as superior, but in reality they didn't outperform because they altered certain instruction sets used for CPU speed. These are Intel-specific instructions, and AMD adopted them mainly for comparison purposes. It was an edge from Intel's side. The issue is whether this situation still exists and if someone can run benchmarks comparing Intel and AMD with these sets applied. Also, whether these sets remain useful today.
Are you looking for a task that requires particular instruction sets such as SSE or AVX?
Intel leads with AVX 512, while AMD has matched or surpassed them in all other areas.
The instruction sets perform similarly in modern Intel chips like Ryzen. The only differences appear in AVX2 and AVX 512 scenarios, where Intel processors tend to be slightly quicker but use more power. You can adjust BIOS settings to limit core speeds when running programs that heavily use AVX512, helping manage power usage. This is due to design decisions—Intel offers a dedicated module for AVX512, while AMD provides two separate AVX 256 modules that can work together as an AVX512 unit. This flexibility allows AMD chips to support multiple cores using AVX256 modules, whereas Intel’s approach limits each module to a single core. This explanation is simplified and may not be entirely accurate.
AMD competes with them on the server side using straightforward power. (same or less capacity). The situation here is different, no Zen 2 yet and Zen+ is clearly slower. Why are we focusing on heavy-duty versions when it's about Zen? I don’t see any tasks where AMD instruction handling would be much worse than AVX.