Template for your README. Remove all unused parts and instructions

Authors (team):

• Mykhailo Bondarenko
• Bohdan Ruban
• Ostap Trush

Variant: No Variant for this lab

• gcc
• CMake
• boost
• Preferably a PC with multiple logical cores :D

Run ./compile.sh with the needed flags to compile the project(the binary will be in ./bin/)

1. Install all Prerequisites
2. Libs for python script:

pip install -r requirements.txt

The binary takes the config path as its argument.

./word_indexer <path_to_config_file>

The config file has path to the directory to be indexed, thread count and output paths.

indir="data/to_index"     # Директорія для індексації
out_by_a="data/res_a.txt"  # Результати, посортовані за алфавітом
out_by_n="data/res_n.txt"  # Результати, посортовані за кількістю
indexing_threads=3

None for now...

We have a word indexer program that can index words in parallel.

We tested it on multiple computers to see how it would differ from machine to machine.

The first one is: Intel Core i7-10510U(4-core 8-thread) @1.8Ghz | 16 GB DDR4 | 512 GB NVMe:

• We can already see that after 4 threads there is almost no improvement due to reading being a bottleneck in this lab.
• The threads index the files quicker than they can compute them.

The next machine perfectly shows us the bottleneck that is reading and, suprisingly, writing. Intel Core i5-4460(4-core 4-thread) @3.4 GHz | 8 GB DDR3 | SSD / partition(read) HDD /home partition(write)

• We can see that multithreaded indexing barely gives any improvement, because reading and writing is very long with writing longet than reading.
• This is because the /mnt folder, where the file is read from is a / partition which is on the SSD, but the results are written into /home, which is a partition on the hard drive.
• Hence, the reading and writing makes multiple threads almost useless.
• Also we see a spike in reading at 2 threads, that's because up to that point reading is not a bottleneck.

Додаткове завдання - 1 (2 бали): дослідити залежність часу виконання від кількості потоків на різних комп’ютерах і з різними наборами параметрів.