I was told (on Intel Developer Zone Forum) the culprit was my wrong usage of /Qxsse4.1 instead of /archSSE4.1 , the first checks whether the CPU is Intel whereas the second uses the extension without checking:
Hi to all,
please help me to understand what causes the denial of running my code.
I asked a very friendly guy owning a machine with AMD Ryzen 1800X to run my single-threaded textual benchmark and for my BIG NASTY SURPRISE I found that Ryzen denies running it.
On my laptop with i5-2430M it runs fine:
see the attachment below
For my regret I lost 2 hours not of my time, no problema, but Gabe's time as well, NOT GOOD! He tried to run the benchmark over and over again but Ryzen refuses to execute SSE4.1 code made by Intel C optimizer/compiler v.15 and GCC code. WHY SO?!
The actual dump from the command prompt:
C:\Benchmark_Bible_(Washigan+_vs_lzbench_vs_TurboBench)_2017-Mar-27>call "_BENCH_a_file.BAT" The_Project_Gutenberg_EBook_of_The_King_James_Bible_kjv10.txt
The current date is: Tue 03/28/2017
Enter the new date: (mm-dd-yy)
The current time is: 15:21:36.48
Enter the new time:
C:\Benchmark_Bible_(Washigan+_vs_lzbench_vs_TurboBench)_2017-Mar-27>"Nakamichi_Washigan+_(1xQWORD+1xXMM)_Intel_15.0_64bit_SSE41.exe" The_Project_Gutenberg_EBook_of_The_King_James_Bible_kjv10.txt.Nakamichi /bench
Please verify that both the operating system and the processor support Intel(R) X87, CMOV, MMX, FXSAVE, SSE, SSE2, SSE3, SSSE3 and SSE4_1 instructions.
His video, in there we went back and forth trying to localize the problem:
Benchmark_Bible_(Washigan+_vs_lzbench_vs_TurboBench)_2017-Mar-29.zip (10,441,150 bytes) downloadable at:
On my slow laptop (i5-2430M) the whole benchmark took 28 minutes:
The actual resultant dump on my laptop (i5-2430M) running Windows 7 64bit:
Message was edited by: Georgi Marinov
Results_i5-2430M.txt.zip 5.1 KB