Content

• How To
• Configuration
• Time Measurement
• Tasks
• Results

How To

1. Create github account (if not exists);
2. Make sure SSH clone & commit is working (Connecting to GitHub with SSH);
3. Fork this repo (just click Fork button on the top of the page, detailed instructions here)
4. Clone your forked repo into your local machine, use your user instead of username:

git clone git@github.com:username/oneapi-2025.git
cd oneapi-2025

5. Go to your group folder, e.g.:

cd 3821B1FI1

6. Go to needed task folder, e.g.:

cd 1_integral_oneapi

7. Create new folder with your surname and name (make sure it's the same for all tasks), e.g.:

mkdir petrov_ivan

8. Copy your task source/header files (including main program) into this folder (use copy instead of cp on Windows), e.g.:

cd petrov_ivan
cp /home/usr/lab/*.cpp .
cp /home/usr/lab/*.h .

8. Push your sources to github repo, e.g.:

cd ..
git add .
git commit -m "1_integral_oneapi task"
git push

9. Go to your repo in browser, click Contribute button on the top of page, then Open pull request. Provide meaningfull request title and description, then Create pull request (see details here).
10. Go to Pull Requests page in course repo, find your pull request and check if there are no any merge conflicts occur. If merge conflicts happen - resolve it following the instruction provided by github.

Time Measurement

The following scheme is used to measure task execution time:

int main() {
    // ...

    // Warming-up
    Task(input, size / 8);

    // Performance Measuring
    auto start = std::chrono::high_resolution_clock::now();
    auto c = Task(input, size);
    auto end = std::chrono::high_resolution_clock::now();

    // ...
}

Configuration

• CPU: Intel Core i5 12600K (4 cores, 4 threads)
• RAM: 16 GB
• GPU: Intel UHD Graphics 770 (8 GB)
• Host Compiler: GCC 11.4.0
• oneAPI: 2025.0

Tasks

Task #1: Permutations

To train modern C++11 skills, the following task is suggested.

There is a set of strings, each string is unique and contains small English letters only. The goal is to make a dictionary of permutations - for each string in a set one should find all other strings from the same set that are permutations of this string.

The following function should be implemented:

using dictionary_t = std::map<std::string, std::vector<std::string>>;
void Permutations(dictionary_t& dictionary);

Initially, dictionary will contain key strings only (all vectors will be empty). After function completion, the same dictionary should additionally keep the lists of permutations for each key string. Each list of permutations should be sorted in reverse alphabetical order.

The following example will illustrate the idea. Let's consider the following set of strings as an input:

aaa
acb
acd
ad
adc
bac
bc
bcc
bd
bda
bdc
caa
cad
cb
cc
ccb
cd
dac
db
dc
dca
dcb
dcc
dd

As a result, one should get the following:

aaa :
acb : bac
acd : dca dac cad adc
ad :
adc : dca dac cad acd
bac : acb
bc : cb
bcc : ccb
bd : db
bda :
bdc : dcb
caa :
cad : dca dac adc acd
cb : bc
cc :
ccb : bcc
cd : dc
dac : dca cad adc acd
db : bd
dc : cd
dca : dac cad adc acd
dcb : bdc
dcc :
dd :

Two files are expected to be uploaded:

• permutations_cxx.h

#ifndef __PERMUTATIONS_CXX_H
#define __PERMUTATIONS_CXX_H

#include <map>
#include <string>
#include <vector>

using dictionary_t = std::map<std::string, std::vector<std::string>>;

void Permutations(dictionary_t& dictionary);

#endif  // __PERMUTATIONS_CXX_H

• permutations_cxx.cpp

#include "permutations_cxx.h"

void Permutations(dictionary_t& dictionary) {
    // Place your implementation here
}

Task #2: Double Integral Computation

In many cases there is no analytical solution for the integral, but one may use approximation of any kind. One of such approximations could be retrieved with the help of Riemann Sum.

E.g., for double integral the following formula could be used to get its approximate value:

$\int_a^b\int_c^df(x,y)dxdy=\sum_{j=0}^{n-1}\sum_{i=0}^{n-1}f(\frac{x_i+x_{i+1}}2, \frac{y_j+y_{j+1}}2)(x_{i+1}-x_i)(y_{j+1}-y_j)$

The task goal is to compute the following integral using Middle Riemann Sum:

$\int_{start}^{end}\int_{start}^{end}sin(x)cos(y)dxdy$

Hint: for $start=0$ and $end=1$ it should be equal to $0.3868223$.

Implement the function in SYCL with the following interface:

float IntegralONEAPI(float start, float end, int count, sycl::device device);

$Count$ means how many intervals one should use to split integration space (the same for $x$ and $y$). E.g. if $count=10$, one will have $10*10=100$ rectangles in total.

Two files are expected to be uploaded:

• integral_oneapi.h

#ifndef __INTEGRAL_ONEAPI_H
#define __INTEGRAL_ONEAPI_H

#include <sycl/sycl.hpp>

float IntegralONEAPI(float start, float end, int count, sycl::device device);

#endif  // __INTEGRAL_ONEAPI_H

• integral_oneapi.cpp

#include "integral_oneapi.h"

float IntegralONEAPI(float start, float end, int count, sycl::device device) {
    // Place your implementation here
}

Task #3: Jacobi Method (Accessors)

Systems of linear equations are basic apparatus (as part of mathematical model) for variety of problems in physics, chemistry, economics, etc.

There two methods of their solving - direct and iterative. Direct methods (like Gaussian Elimination) are able to get accurate solution, while iterative ones can only provide approximate results. At the same time, in practice iterative methods may be preferable, e.g. in case of huge matrices that have to stored in RAM while using direct approaches, or if one has close enough initial estimation for final result.

One of such iterative methods is Jacobi Method that allows to get accurate enough solution using the following formula:

$x_i^{(k+1)}=\frac{1}a_{ii}(b_i-\sum_{j \neq i}a_{ij}x_j^{(k)})$

Here $x^{(k+1)}$ is the next approximation of system solution, computed from the previous $x^{(k)}$. First approximation $x^{(0)}$ could be all zeros.

There are two ways to stop computations:

1. After $N$ iterations, where $N$ is some predefined constant;
2. If $|x^{(k+1)}-x^{(k)}|<Eps$, where $Eps$ is target accuracy.

Note: to ensure method convergence one should use it only for strictly diagonally dominant system, that means:

$|a_{ii}|>\sum_{j \neq i}|a_{ij}|$ for any $i$.

To complete this task, one should implement the function that computes the solution for the system of linear equations using Jacobi method:

std::vector<float> JacobiAccONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device);

One should implement both stop methods at the same time. Use $N=1024$ as maximum iterations count and $accuracy$ argument as $Eps$. computations have to be stopped when $|x^{(k+1)}-x^{(k)}|<accuracy$ first, and if it's not happening, when after 1024 iterations.

Matix $a$ is stored by rows. One should implement the algorithm using SYCL buffers & accessors approach.

Two files are expected to be uploaded:

• jacobi_acc_oneapi.h

#ifndef __JACOBI_ACC_ONEAPI_H
#define __JACOBI_ACC_ONEAPI_H

#include <vector>

#include <sycl/sycl.hpp>

#define ITERATIONS 1024

std::vector<float> JacobiAccONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device);

#endif  // __JACOBI_ACC_ONEAPI_H

• jacobi_acc_oneapi.cpp

#include "jacobi_acc_oneapi.h"

std::vector<float> JacobiAccONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device) {
    // Place your implementation here
}

Task #4: Jacobi Method (Device Memory)

This task assumes Jacobi method implementation using SYCL device memory approach (see all the details in Task #2).

Two files are expected to be uploaded:

• jacobi_dev_oneapi.h

#ifndef __JACOBI_DEV_ONEAPI_H
#define __JACOBI_DEV_ONEAPI_H

#include <vector>

#include <sycl/sycl.hpp>

#define ITERATIONS 1024

std::vector<float> JacobiDevONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device);

#endif  // __JACOBI_DEV_ONEAPI_H

• jacobi_dev_oneapi.cpp

#include "jacobi_dev_oneapi.h"

std::vector<float> JacobiDevONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device) {
    // Place your implementation here
}

Task #5: Jacobi Method (Shared Memory)

This task assumes Jacobi method implementation using SYCL shared memory approach (see all the details in Task #2).

Two files are expected to be uploaded:

• jacobi_shared_oneapi.h

#ifndef __JACOBI_SHARED_ONEAPI_H
#define __JACOBI_SHARED_ONEAPI_H

#include <vector>

#include <sycl/sycl.hpp>

#define ITERATIONS 1024

std::vector<float> JacobiSharedONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device);

#endif  // __JACOBI_SHARED_ONEAPI_H

• jacobi_shared_oneapi.cpp

#include "jacobi_shared_oneapi.h"

std::vector<float> JacobiSharedONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        float accuracy, sycl::device device) {
    // Place your implementation here
}

Task #6: Block Matrix Multiplication

General matrix multiplication (GEMM) is a very basic and broadly used linear algebra operation applied in high performance computing (HPC), statistics, deep learning and other domains. There are a lot of GEMM algorithms with different mathematical complexity form $O(n^3)$ for naive and block approaches to $O(n^{2.371552})$ for the method descibed by Williams et al. in 2024 [1]. But despite a variety of algorithms with low complexity, block matrix multiplication remains the most used implementation in practice since it fits to modern HW better.

In real applications block-based approach for matrix multiplication can get multiple times faster execution comparing with naive version due to cache friendly approach.

In block version, algorithm could be divided into three stages:

1. Split matricies into blocks (block size normally affects performance significantly so choose it consciously);
2. Multiply two blocks to get partial result;
3. Replay step 2 for all row/column blocks accumulating values into a single result block.

From math perspective, block matrix multiplication could be described by the following formula, where $C_{IJ}$, $A_{IK}$ and $B_{KJ}$ are sub-matricies with the size $block_size*block_size$:

$C_{IJ}=\sum_{k=1}^{block_count}A_{IK}B_{KJ}$

Each matrix must be stored in a linear array by rows, so that a.size()==size*size. Function takes two matricies and their size as inputs, and returns result matrix also stored by rows.

For simplicity, let's consider matrix size is always power of 2 and all matricies are square.

Two files are expected to be uploaded:

• gemm_block_oneapi.h:

#ifndef __GEMM_BLOCK_ONEAPI_H
#define __GEMM_BLOCK_ONEAPI_H

#include <vector>

#include <sycl/sycl.hpp>

std::vector<float> GemmBlockONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        size_t size, sycl::device device);

#endif  // __GEMM_BLOCK_ONEAPI_H

• gemm_block_oneapi.cpp:

#include "gemm_block_oneapi.h"

std::vector<float> GemmBlockONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        size_t size, sycl::device device) {
    // Place your implementation here
}

Task #7: Matrix Multiplication Using oneMKL

The most performant way to multiply two matrices on particular hardware is to use vendor-provided library for this purpose. In SYCL it's oneMKL. Try to use oneMKL BLAS API to implement general matrix multiplication in most performant way.

Each matrix must be stored in a linear array by rows, so that a.size()==size*suze. Function takes two matricies and their size as inputs, and returns result matrix also stored by rows.

For simplicity, let's consider matrix size is always power of 2 and all matricies are square.

Note, that in oneMKL BLAS API matrix is expected to be stored by columns, so additional transpose (or slightly different API) may be required.

Two files are expected to be uploaded:

• gemm_mkl_oneapi.h:

#ifndef __GEMM_MKL_ONEAPI_H
#define __GEMM_MKL_ONEAPI_H

#include <vector>

#include <sycl/sycl.hpp>

std::vector<float> GemmMklONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        size_t size, sycl::device device);

#endif  // __GEMM_MKL_ONEAPI_H

• gemm_mkl_oneapi.cpp:

#include "gemm_mkl_oneapi.h"

std::vector<float> GemmMklONEAPI(
        const std::vector<float> a, const std::vector<float> b,
        size_t size, sycl::device device) {
    // Place your implementation here
}

Results

1_permutations_cxx (10240 elements)

Group	Name	Result
3821B1FI2	isaev_dmitrii	0.0033
3821B1FI1	mirzakhmedov_alexander	0.0038
3821B1FI2	petrov_maksim	0.0039
3821B1FI3	durandin_vladimir	0.0039
3821B1FI1	borovkov_sergey	0.0049
3821B1FI1	shipitsin_alex	0.0053
3821B1FI1	bodrov_daniil	0.0054
3821B1FI3	pivovarov_alexey	0.0058
3821B1FI3	prokofev_kirill	0.0059
3821B1FI2	nogin_denis	0.0060
3821B1FI2	kostanyan_arsen	0.0060
3821B1FI1	kashirin_alexander	0.0060
3821B1FI3	kulagin_aleksandr	0.0060
3821B1FI3	savotina_valeria	0.0063
3821B1FI2	zakharov_artem	0.0067
3821B1FI1	balyasov_ivan	0.0067
3821B1FI3	simonyan_suren	0.0067
3821B1FI3	kuznetsov_artyom	0.0067
3821B1FI1	lysanova_julia	0.0069
3821B1FI2	soloninko_andrey	0.0070
3821B1FI3	sharapov_georgiy	0.0070
3821B1FI3	korablev_nikita	0.0073
3821B1FI3	kulaev_zhenya	0.0073
3821B1FI1	bonyuk_peter	0.0074
3821B1FI3	safarov_nurlan	0.0075
3821B1FI3	ryabkov_vladislav	0.0076
3821B1FI2	zawadowski_jan	0.0078
3821B1FI2	loginov_maxim	0.0078
3821B1FI2	kostin_artem	0.0079
3821B1FI3	ivanov_nikita	0.0079
3821B1FI3	vasilev_ivan	0.0079
3821B1FI2	kazantsev_evgeny	0.0080
3821B1FI3	kulikov_artem	0.0080
3821B1FI3	safronov_mikhail	0.0081
3821B1FI3	volodin_evgeniy	0.0091
3821B1FI3	sadikov_damir	0.0093
3821B1FI3	benduyzhko_tatiana	0.0093
3821B1FI3	tyulkina_olga	0.0095
3821B1FI1	shmelev_ivan	0.0099
3821B1FI2	travin_maksim	0.0104
3821B1FI3	polozov_vladislav	0.0107
REF	REF	0.8951
3821B1FI1	pushkarev_ilya	TEST FAILED

2_integral_oneapi (65536 elements)

Group	Name	Result
3821B1FI1	shipitsin_alex	0.0334
3821B1FI1	bonyuk_peter	0.0356
3821B1FI1	borovkov_sergey	0.0356
3821B1FI1	bodrov_daniil	0.0366
3821B1FI3	pivovarov_alexey	0.0367
3821B1FI1	kashirin_alexander	0.0370
3821B1FI2	petrov_maksim	0.4503
3821B1FI3	volodin_evgeniy	0.4515
REF	REF	0.4682
3821B1FI2	nogin_denis	0.9784
3821B1FI3	durandin_vladimir	0.9917
3821B1FI3	kulikov_artem	0.9977
3821B1FI3	safronov_mikhail	0.9996
3821B1FI2	travin_maksim	1.0000
3821B1FI3	ivanov_nikita	1.0015
3821B1FI2	zawadowski_jan	1.0018
3821B1FI3	tyulkina_olga	1.0048
3821B1FI3	korablev_nikita	1.0086
3821B1FI3	kulagin_aleksandr	1.2853
3821B1FI3	simonyan_suren	1.2950
3821B1FI2	kostin_artem	1.3396
3821B1FI2	kazantsev_evgeny	1.3527
3821B1FI3	savotina_valeria	1.3929
3821B1FI3	ryabkov_vladislav	1.3933
3821B1FI1	lysanova_julia	1.3934
3821B1FI3	prokofev_kirill	1.4003
3821B1FI2	isaev_dmitrii	1.4004
3821B1FI3	vasilev_ivan	1.4037
3821B1FI2	loginov_maxim	1.4044
3821B1FI1	mirzakhmedov_alexander	1.4051
3821B1FI2	zakharov_artem	1.4051
3821B1FI1	balyasov_ivan	1.4059
3821B1FI2	soloninko_andrey	1.4059
3821B1FI3	safarov_nurlan	1.4079
3821B1FI3	kuznetsov_artyom	1.4179
3821B1FI2	kostanyan_arsen	1.4367
3821B1FI3	sharapov_georgiy	1.4472
3821B1FI3	polozov_vladislav	1.4476
3821B1FI3	sadikov_damir	1.4479
3821B1FI3	kulaev_zhenya	1.4486
3821B1FI3	benduyzhko_tatiana	1.6310
3821B1FI1	shmelev_ivan	TOO SLOW
3821B1FI1	pushkarev_ilya	BUILD FAILED

3_jacobi_acc_oneapi (4096 elements)

Group	Name	Result
3821B1FI1	mirzakhmedov_alexander	0.2204
3821B1FI2	petrov_maksim	0.2334
3821B1FI2	kostin_artem	0.2355
3821B1FI1	borovkov_sergey	0.2373
3821B1FI3	safronov_mikhail	0.2539
3821B1FI1	shmelev_ivan	0.2551
3821B1FI1	shipitsin_alex	0.2593
3821B1FI3	safarov_nurlan	0.2594
3821B1FI2	soloninko_andrey	0.2597
3821B1FI3	prokofev_kirill	0.2649
3821B1FI1	balyasov_ivan	0.2651
3821B1FI1	bonyuk_peter	0.2655
3821B1FI2	nogin_denis	0.2655
3821B1FI1	lysanova_julia	0.2671
3821B1FI3	kulagin_aleksandr	0.2671
3821B1FI3	savotina_valeria	0.2673
3821B1FI1	pushkarev_ilya	0.2676
3821B1FI2	loginov_maxim	0.2677
3821B1FI3	pivovarov_alexey	0.2686
3821B1FI2	kostanyan_arsen	0.2692
3821B1FI3	sharapov_georgiy	0.2747
3821B1FI2	isaev_dmitrii	0.2791
3821B1FI3	kulaev_zhenya	0.2825
3821B1FI3	durandin_vladimir	0.2841
3821B1FI2	zakharov_artem	0.2855
3821B1FI3	kuznetsov_artyom	0.2886
3821B1FI3	ryabkov_vladislav	0.2902
3821B1FI3	ivanov_nikita	0.2926
3821B1FI3	vasilev_ivan	0.2932
3821B1FI1	bodrov_daniil	0.3032
3821B1FI3	korablev_nikita	0.3170
3821B1FI2	kazantsev_evgeny	0.3232
3821B1FI3	volodin_evgeniy	0.3332
3821B1FI2	travin_maksim	0.3474
3821B1FI2	zawadowski_jan	0.3533
3821B1FI1	kashirin_alexander	0.3647
3821B1FI3	kulikov_artem	0.3709
3821B1FI3	benduyzhko_tatiana	0.4267
3821B1FI3	tyulkina_olga	0.5634
3821B1FI3	polozov_vladislav	0.5856
3821B1FI3	sadikov_damir	0.6114
REF	REF	0.6595
3821B1FI3	simonyan_suren	0.9227

4_jacobi_dev_oneapi (4096 elements)

Group	Name	Result
3821B1FI3	safronov_mikhail	0.1772
3821B1FI1	bonyuk_peter	0.1829
3821B1FI1	borovkov_sergey	0.1830
3821B1FI3	volodin_evgeniy	0.1848
3821B1FI1	shipitsin_alex	0.1855
3821B1FI1	shmelev_ivan	0.1874
3821B1FI2	nogin_denis	0.1921
3821B1FI3	simonyan_suren	0.1952
3821B1FI2	soloninko_andrey	0.1982
3821B1FI3	savotina_valeria	0.1986
3821B1FI3	kuznetsov_artyom	0.1987
3821B1FI1	balyasov_ivan	0.1988
3821B1FI2	zakharov_artem	0.1993
3821B1FI3	durandin_vladimir	0.1998
3821B1FI2	loginov_maxim	0.1999
3821B1FI2	petrov_maksim	0.2010
3821B1FI2	kostin_artem	0.2025
3821B1FI3	pivovarov_alexey	0.2033
3821B1FI2	kostanyan_arsen	0.2042
3821B1FI1	lysanova_julia	0.2052
3821B1FI3	sharapov_georgiy	0.2053
3821B1FI3	vasilev_ivan	0.2054
3821B1FI3	kulaev_zhenya	0.2069
3821B1FI3	safarov_nurlan	0.2079
3821B1FI1	mirzakhmedov_alexander	0.2098
3821B1FI3	prokofev_kirill	0.2100
3821B1FI2	isaev_dmitrii	0.2148
3821B1FI3	ryabkov_vladislav	0.2151
3821B1FI3	kulagin_aleksandr	0.2270
3821B1FI3	kulikov_artem	0.2559
3821B1FI2	kazantsev_evgeny	0.2707
3821B1FI3	sadikov_damir	0.2827
3821B1FI3	polozov_vladislav	0.2890
3821B1FI3	korablev_nikita	0.3057
3821B1FI2	travin_maksim	0.3224
3821B1FI3	tyulkina_olga	0.3359
3821B1FI3	benduyzhko_tatiana	0.3370
3821B1FI1	bodrov_daniil	0.3388
3821B1FI2	zawadowski_jan	0.3598
3821B1FI3	ivanov_nikita	0.3635
3821B1FI1	kashirin_alexander	0.3994
REF	REF	0.6662
3821B1FI1	pushkarev_ilya	0.6729

5_jacobi_shared_oneapi (4096 elements)

Group	Name	Result
3821B1FI3	sharapov_georgiy	0.1545
3821B1FI2	zakharov_artem	0.1564
3821B1FI3	durandin_vladimir	0.1606
3821B1FI2	soloninko_andrey	0.1675
3821B1FI1	borovkov_sergey	0.1680
3821B1FI2	kostin_artem	0.1704
3821B1FI1	pushkarev_ilya	0.1750
3821B1FI1	shmelev_ivan	0.1770
3821B1FI3	kuznetsov_artyom	0.1792
3821B1FI3	tyulkina_olga	0.1798
3821B1FI3	safarov_nurlan	0.1873
3821B1FI3	safronov_mikhail	0.1876
3821B1FI3	benduyzhko_tatiana	0.1879
3821B1FI2	loginov_maxim	0.1883
3821B1FI3	savotina_valeria	0.1897
3821B1FI1	lysanova_julia	0.1906
3821B1FI1	balyasov_ivan	0.1912
3821B1FI2	nogin_denis	0.1925
3821B1FI2	petrov_maksim	0.1926
3821B1FI3	pivovarov_alexey	0.1932
3821B1FI1	bonyuk_peter	0.1937
3821B1FI2	isaev_dmitrii	0.1963
3821B1FI3	prokofev_kirill	0.1971
3821B1FI3	vasilev_ivan	0.2001
3821B1FI3	ryabkov_vladislav	0.2002
3821B1FI1	shipitsin_alex	0.2007
3821B1FI3	kulaev_zhenya	0.2021
3821B1FI3	simonyan_suren	0.2047
3821B1FI3	kulagin_aleksandr	0.2055
3821B1FI1	mirzakhmedov_alexander	0.2076
3821B1FI3	korablev_nikita	0.2633
3821B1FI3	kulikov_artem	0.2676
3821B1FI3	ivanov_nikita	0.2710
3821B1FI3	polozov_vladislav	0.2711
3821B1FI2	kazantsev_evgeny	0.2721
3821B1FI3	sadikov_damir	0.2724
3821B1FI1	bodrov_daniil	0.3076
3821B1FI3	volodin_evgeniy	0.3227
3821B1FI2	travin_maksim	0.3601
3821B1FI2	zawadowski_jan	0.3608
3821B1FI1	kashirin_alexander	0.3695
REF	REF	0.6341
3821B1FI2	kostanyan_arsen	0.8133

6_gemm_block_oneapi (3072 elements)

Group	Name	Result
3821B1FI1	borovkov_sergey	0.8200
3821B1FI2	kostanyan_arsen	0.8220
3821B1FI3	tyulkina_olga	0.8278
3821B1FI3	simonyan_suren	0.8326
3821B1FI2	travin_maksim	0.8332
3821B1FI2	soloninko_andrey	0.8369
3821B1FI2	kazantsev_evgeny	0.8484
3821B1FI1	shipitsin_alex	0.8530
3821B1FI3	kulagin_aleksandr	0.8541
3821B1FI1	pushkarev_ilya	0.8616
REF	REF	0.8759
3821B1FI2	nogin_denis	0.8866
3821B1FI3	safronov_mikhail	0.8912
3821B1FI1	lysanova_julia	0.8928
3821B1FI1	shmelev_ivan	0.8931
3821B1FI2	loginov_maxim	0.8935
3821B1FI2	zakharov_artem	0.8939
3821B1FI3	ryabkov_vladislav	0.8944
3821B1FI3	benduyzhko_tatiana	0.8946
3821B1FI3	kulikov_artem	0.8975
3821B1FI3	ivanov_nikita	0.8976
3821B1FI3	safarov_nurlan	0.8977
3821B1FI1	mirzakhmedov_alexander	0.8982
3821B1FI3	sharapov_georgiy	0.8987
3821B1FI3	durandin_vladimir	0.8999
3821B1FI3	kulaev_zhenya	0.9010
3821B1FI3	pivovarov_alexey	0.9012
3821B1FI1	balyasov_ivan	0.9016
3821B1FI2	isaev_dmitrii	0.9018
3821B1FI3	vasilev_ivan	0.9022
3821B1FI2	zawadowski_jan	0.9025
3821B1FI3	korablev_nikita	0.9030
3821B1FI3	savotina_valeria	0.9033
3821B1FI3	kuznetsov_artyom	0.9044
3821B1FI2	petrov_maksim	0.9142
3821B1FI3	polozov_vladislav	0.9244
3821B1FI2	kostin_artem	0.9246
3821B1FI3	volodin_evgeniy	0.9248
3821B1FI3	sadikov_damir	0.9324
3821B1FI3	prokofev_kirill	0.9383
3821B1FI1	kashirin_alexander	0.9719
3821B1FI1	bonyuk_peter	0.9923
3821B1FI1	bodrov_daniil	1.0162

7_gemm_mkl_oneapi (3072 elements)

Group	Name	Result
REF	REF	0.2893
3821B1FI3	sadikov_damir	0.4335
3821B1FI2	isaev_dmitrii	0.4872
3821B1FI2	kazantsev_evgeny	0.6009
3821B1FI2	nogin_denis	0.6340
3821B1FI2	loginov_maxim	0.6507
3821B1FI3	safronov_mikhail	0.6646
3821B1FI2	zawadowski_jan	0.6787
3821B1FI3	kuznetsov_artyom	0.6912
3821B1FI3	pivovarov_alexey	0.7002
3821B1FI3	tyulkina_olga	0.7044
3821B1FI2	soloninko_andrey	0.7344
3821B1FI3	simonyan_suren	0.7454
3821B1FI1	shipitsin_alex	0.7713
3821B1FI1	borovkov_sergey	0.7744
3821B1FI3	benduyzhko_tatiana	0.7849
3821B1FI1	bonyuk_peter	0.7913
3821B1FI3	kulaev_zhenya	0.7945
3821B1FI3	kulikov_artem	0.7959
3821B1FI2	travin_maksim	0.7961
3821B1FI1	pushkarev_ilya	0.7966
3821B1FI3	prokofev_kirill	0.8055
3821B1FI3	savotina_valeria	0.8079
3821B1FI1	kashirin_alexander	0.8137
3821B1FI3	sharapov_georgiy	0.8169
3821B1FI1	mirzakhmedov_alexander	0.8302
3821B1FI3	ivanov_nikita	0.8374
3821B1FI2	kostanyan_arsen	0.8474
3821B1FI3	volodin_evgeniy	0.8689
3821B1FI2	kostin_artem	0.9371
3821B1FI1	shmelev_ivan	0.9394
3821B1FI3	ryabkov_vladislav	0.9400
3821B1FI3	durandin_vladimir	0.9522
3821B1FI2	zakharov_artem	0.9576
3821B1FI2	petrov_maksim	0.9633
3821B1FI3	kulagin_aleksandr	0.9653
3821B1FI3	vasilev_ivan	0.9656
3821B1FI1	lysanova_julia	0.9668
3821B1FI3	safarov_nurlan	0.9733
3821B1FI1	balyasov_ivan	0.9790
3821B1FI3	korablev_nikita	0.9860
3821B1FI1	bodrov_daniil	0.9925
3821B1FI3	polozov_vladislav	1.0043

Tasks Done

3821B1FI1

Group	Name	Passed
3821B1FI1	balyasov_ivan	7/7
3821B1FI1	bodrov_daniil	7/7
3821B1FI1	bonyuk_peter	7/7
3821B1FI1	borovkov_sergey	7/7
3821B1FI1	kashirin_alexander	7/7
3821B1FI1	lysanova_julia	7/7
3821B1FI1	mirzakhmedov_alexander	7/7
3821B1FI1	pushkarev_ilya	5/7
3821B1FI1	shipitsin_alex	7/7
3821B1FI1	shmelev_ivan	6/7

Passed: 8

3821B1FI2

Group	Name	Passed
3821B1FI2	isaev_dmitrii	7/7
3821B1FI2	kazantsev_evgeny	7/7
3821B1FI2	kostanyan_arsen	7/7
3821B1FI2	kostin_artem	7/7
3821B1FI2	loginov_maxim	7/7
3821B1FI2	nogin_denis	7/7
3821B1FI2	petrov_maksim	7/7
3821B1FI2	soloninko_andrey	7/7
3821B1FI2	travin_maksim	7/7
3821B1FI2	zakharov_artem	7/7
3821B1FI2	zawadowski_jan	7/7

Passed: 11

3821B1FI3

Group	Name	Passed
3821B1FI3	benduyzhko_tatiana	7/7
3821B1FI3	durandin_vladimir	7/7
3821B1FI3	ivanov_nikita	7/7
3821B1FI3	korablev_nikita	7/7
3821B1FI3	kulaev_zhenya	7/7
3821B1FI3	kulagin_aleksandr	7/7
3821B1FI3	kulikov_artem	7/7
3821B1FI3	kuznetsov_artyom	7/7
3821B1FI3	pivovarov_alexey	7/7
3821B1FI3	polozov_vladislav	7/7
3821B1FI3	prokofev_kirill	7/7
3821B1FI3	ryabkov_vladislav	7/7
3821B1FI3	sadikov_damir	7/7
3821B1FI3	safarov_nurlan	7/7
3821B1FI3	safronov_mikhail	7/7
3821B1FI3	savotina_valeria	7/7
3821B1FI3	sharapov_georgiy	7/7
3821B1FI3	simonyan_suren	7/7
3821B1FI3	tyulkina_olga	7/7
3821B1FI3	vasilev_ivan	7/7
3821B1FI3	volodin_evgeniy	7/7

Passed: 21

Total Passed: 40