Member Avatar for baldwindc

ARM's ASM is a breed in it's own. AMD and Intel, although different, have more closely related assembler than most. Someone mentioned the other day that AMD is just an Intel repeat, basically copying engineering designs and repeating them with reduced cost (as they don't pay to have it designed, just built).

Do you believe that there is any truth in that? Is Intel consistently ahead of the curve, or is it a more even balance?

On a personal note, I've always used Intel.

  • 3 Contributors
  • 2 Replies
  • 243 Views
  • 8 Hours Discussion Span
  • Latest Post Latest Post by caperjack

Dani AI

Generated

— short answer: no, AMD is not “just an Intel repeat,” but at the assembly level they look very similar because both follow the same x86 / x86‑64 instruction set. That shared ISA makes portable assembly possible, but the microarchitecture underneath (pipelines, caches, branch predictors, SMT/HT behavior, etc.) differs and is what drives real performance differences. x86-64 history and compatibility.

was right to flag the memory-controller story: AMD put an on‑die memory controller and x86‑64 (AMD64) into Opteron/Athlon64 in the early 2000s, which changed platform design and performance assumptions; Intel adopted a similar 64‑bit model later and moved its own memory controller on‑die with Nehalem (2008). Those moves illustrate that each vendor both borrows and innovates. (See AMD Opteron / AMD filings and the x86‑64 timeline.)

The “Intel pulled ahead with Core 2” memory is also accurate in context: Intel’s Core microarchitecture (Conroe/Core 2 Duo) changed the performance landscape around 2006 and restored Intel’s single‑thread efficiency advantage for several years. For a retrospective on that era see AnandTech’s Core 2 coverage. https://www.anandtech.com/show/10525/ten-year-anniversary-of-core-2-duo-and-conroe-moores-law-is-dead-long-live-moores-law

Fast-forward: both firms continue to diverge. AMD’s Zen family (Ryzen/EPYC) rebuilt competitiveness and introduced large‑scale chiplet packaging; Intel moved to hybrid P‑core/E‑core designs with Alder Lake and beyond. These are architectural choices, not ISA changes.

Practical takeaway for assembly or low‑level tuning: detect features with CPUID, avoid hard‑coding latencies, and measure on target silicon. Use Agner Fog’s instruction tables and uops.info / microbenchmarks to get per‑CPU latency/throughput numbers before hand‑tuning.

(Also: ’s memory is right in spirit — there were eras of pin/slot compatibility and clone CPUs; modern socket and memory designs changed that.)

Recommended Answers

All 2 Replies

Member Avatar for crunchie

Intel have probably copied AMD, particularly with the onboard memory controller.
Since C2DUO Intel have been out in front............for a price.

Member Avatar for caperjack

in the early days the the amd cpu chip fit into the same cpu socket as the intel

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.