Intel发布PPT能效超越M1Max

有个进步，这次没用降压Cinebench，用了SPEC2017 Rate-N，图上35w时多核领先12.7%。

编译器ICC对比Clang。

Based on superior performance of 12th Gen Intel Core i9 12900HK against Intel Core i9 11980HK, AMD Ryzen 9 5900HX, and Apple M1 Max. Intel processor performance is estimated based on measurements with Intel Reference Validation Platforms. AMD processor performance is estimated based on measurements on a Lenovo Legion R9000K with RTX 3080. Apple M1 Max performance is estimated based on public statement made by Apple on 10/18/2021 and measurements on Apple M1 Max 16" 64GB RAM Model A2485. Best available compilers selected for all processors. Binaries compiled with ICC for Intel/AMD, binaries compiled with Xcode 13.1 for Apple. The metric used is the geometric mean of an n-copy SPECrate run of the C/C++ integer benchmarks in SPEC CPU 2017.

Benchmark:

SPEC* CPU2017 is a benchmark from the SPEC consortium (www.spec.org) that measures computer performance and throughput using compute intensive application subtests. SPECint*_base2017 provides a comparison point as to how fast a device completes a series of single integer compute tasks. SPECint*_rate_base2017 provides a comparison point for throughput, or how many integer computes tasks a device can accomplish in a given amount of time. The SPECrate® 2017 Integer and SPECrate® 2017 Floating Point suites measure the throughput or work per unit of time.

x86最大的问题在于待机，不知道这代有没有解决。

14核打10核比多核能效。。。编译器还不一样。。。

来说说槽点。第一个是英特尔称各自平台都用了最佳编译器
而对于M1 Max实际上Clang还不如GCC。根据我们之前对M1的测试，跑SPEC2017 Int GCC11对比A站的Clang就有13%的提升。

真tm 绝了，ICC比Clang

第二个是功耗。12代实际的功耗不清楚，暂时认为Intel公布的没问题。而对比M1 Max呢，跑SPEC2017 Int RateN对内存带宽需求不大，用同样10核的M1 Pro跑出来的结果不会有区别。这是我测试M1Pro的功耗。显然平均Package大概只有30w。

至于能发挥M1Max内存带宽优势的fp嘛，自然是不能对比的

经典icc

amd也用的是icc？

另外我们可以对比ppt上的65w的11980hk。icc下领先clang的m1 max 8.4%。而同编译器m1 max领先53.38/38.82=37.5%

intel的制程还是劣势，这样看苹果的CPU设计水平也没想像中那么强。

这是intel吹自家编译器的牛，这时候对比就有大问题了

英特尔的外围和制成都是劣势

笔记本晶体管还是用桌面版那一套，英特尔的ppt高密度产品为什么不上？笔记本又用不上高频，ppt高密度低功耗，多合适啊，恐怕是没有这东西。