@Pascal4QM This bug occurs when the file title containing Chinese Characters.

@tomleslie when running the code in the first floor of this post, the acceleration ratio from multithreading and hyperthreading is 2，which is just the same as your case at the third floor of this post.

There's some difference in the sub version of OS and Maple.
My OS version is Win 10 @1909 and Maple version is 2019.2.1.

@Carl Love when running the code in the first floor of this post, the CPU usage can actually reach 100%, though the acceleration ratio from multithreading and hyperthreading is just 2.

@Carl Love any of the FPU benchmark in Aida64 can impose a 100% load to CPU 

@tomleslie The settings of the BIOS and OS are ok, which can be confirmed by CPU benchmark software, such as Aida 64. Actually, the double precision floating-point benchmark result of my i7-6700 is even higher than i7-3770k without overclocking.

@Carl Love 

Here is the introduction of the benchmark FPU Mandel in Aida64

This benchmark measures the double precision (also known as 64-bit) floating-point performance through the computation of several frames of the popular "Mandelbrot" fractal.  The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD, Intel and VIA processor core variants by utilizing the appropriate x87, SSE2, AVX, AVX2, AVX-512, FMA, and FMA4 instruction set extension.

FPU Mandel test consumes 4 MB system memory per calculation thread, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.

and Here is the result of the benchmark of my i7-6700 after lowering the voltage as much as 0.2V,

You can see it’s even higher than i7-3770k.

Aida64 can be obtained here, https://www.aida64.com/downloads

@Carl Love 

I’ve restored the voltage of my i7-6700HQ to its default and the result keeps the same.

Actually, it’s the operating frequency that determines the performance of a CPU with a given microarchitecture. For my i7-6700HQ, it can boost to the maximum operating frequency as designed even lowering the voltage as much as 0.2V. The score from the CPU benchmark software, such as Aida 64, won’t be affected by this action.

The consuming power of my i7-6700HQ is only around 20W when executing the code in this post and the noise from the fan is very gentle. My laptop has a very strong heat-removal system, since it’s actually a mobile workstation. Therefore, this is not a coincidence.

Here is a snap from intel XTU，

It’s really a little mysterious.

@tomleslie 

I just use the full license to reactivate my Maple2019 and find the benefits from hyperthreading appears.

But the acceleration radio is only close to 5. The CPU usage is only around 70% when executing the code.

@Carl Love 

I’ve purchased Maple2019 recently and been using the trial license since then.

I just use the full license to reactivate my Maple2019 and find the benefits from hyperthreading appears.

But the acceleration radio is only close to 5. The CPU usage is only around 70%.

@Carl Love 

Your i7-7700HQ is essentially the same as my i7-6700HQ except for the higher operating frequency.

This problem may be caused by the OS.

What's your OS, please?

Mine is Win 10 version 1909.

By the way, you can use Intel® Extreme Tuning Utility to lower the temperature of your CPU significantly via lowering its voltage.

Intel® Extreme Tuning Utility can be obtained from here:

https://downloadcenter.intel.com/download/24075

This operation is totally safe and I’ve used it to lower the temperature of my laptop ever since the summer of 2016.

Happy new year!

@acer 

Your technique based on shake( ) and evalr( ) really sounds wonderful!

Do you mind sharing this technique with others?

@Carl Love In Maple, an integer will be automatically converted to a float number when adding it to a float number, so it’s really of no necessity to write an integer in Maple with a float.

@Carl Love When executing the previously code, the CPU usage of my laptop is only a little higher than 50%. In another word, the hyperthreading doesn’t take part in. Have you tried the code on your computer and what's the result?

@vv Your example here seems to be the problem of large number calculation, which is one of major challenges in the modern numerical calculation.

@Carl Love This double nested version of evalf( ) is really efficient for calculation without large numbers involved.

By the way, it’s real glad it can be your inspiration.