andrewt0301/hse-acos-course

Materials for the "Computer Architecture and Operating Systems" course taught at Faculty of Computer Science of Higher School of Economics

hse-acos-course's Issues

Timetable Draft

https://docs.google.com/spreadsheets/d/1-DzUiDC11i98lK6i1v870pKLCRdCYodcHtF_f0h1FQY/edit#gid=32028350

Code

F

OS Test Note

Powering a negative number into a non-integer power can result in NaN:

andrewt@LAPTOP-5GPL0IVV:~/src$ cat test.c
#include <math.h>
#include <stdio.h>

int main() {
   printf("x = %f\n", pow(-7, 1.23));
   return 0;
}
andrewt@LAPTOP-5GPL0IVV:~/src$ gcc test.c -o test -lm
andrewt@LAPTOP-5GPL0IVV:~/src$ ./test
x = -nan

We consider it is OK.

Quiz

https://forms.gle/g2brkRvwnqDwtbgZ6

Quizzes 2023

Quiz authors:

• Oleg Malchenko
• Daria Lapko
• Vladislav Kirichok
• Andy Xu
• Artem Borisov
• Nikolay Chechulin

CA quizzes:

1. Vladislav Kirichok
2. Oleg Malchenko
3. Andy Xu
4. Daria Lapko
5. Nikolay Chechulin
6. Artem Borisov

OS quizzes:

Timetable

https://docs.google.com/spreadsheets/d/1Pr5pdLjdnTu2h7BF2DzN3u9RpScxv78sWQCh8j0v6Qg/edit#gid=746855010

meeting

Andrei Tatarnikov is inviting you to a scheduled Zoom meeting.

Topic: Andrei Tatarnikov's Personal Meeting Room

Join Zoom Meeting
https://us05web.zoom.us/j/5131565311?pwd=U3hUS1NaRVRFNGVPVXAwSUQvQ3A5UT09

Meeting ID: 513 156 5311
Passcode: N3N4mH

Final Test Issue

Notes on Final Test Programming Task

Task

The programming task as specified here:

Write a RISC-V assembly program that inputs integer value x (assumed non-negative) and calculates the value of the f(x) mathematical function according to the specified equations. f(x) must be implemented as a function and must comply with RISC-V calling conventions.

The implemented function must use callee-saved registers (s0, s1, etc.) to store intermediate results of calculations. These registers must be saved to the stack and restored when the function returns.

Variant

Group 202, variant 7. Taken from here:

f(x) = 6 * x**5 + 4 * x**3 - 5 * x**2 + 7 * x if x < 5
f(x) = 4 if x >= 5
f(x) = 7**x + -2 if x == 6
f(x) = -1 * x + 10 if x > 9

This task means to create a RISC-V assembly program that does the same as this Python program:

def f(x):
    if x < 5:
        return 6 * x ** 5 + 4 * x ** 3 - 5 * x ** 2 + 7 * x
    if x > 9:
        return -1 * x + 10
    if x == 6:
        return 7 ** x + -2
    return 4  # if x >= 5


x = int(input())
print(f(x))

Expected results:

x = 1
f(x) = 12

x = 5
f(x) = 4

x = 6
f(x) = 117647

x = 7
f(x) = 4

x = 11
f(x) = -1

Solution

#попытка  f(x) = 6 * x**5 + 4 * x**3 - 5 * x**2 + 7 * x if x < 5
.text
li a7, 5 
ecall
mv t0, a0  

addi t1, zero, 6 
addi t2, zero, 4
addi t3, zero, 5
addi t4, zero, zero

	j compare
loop:
	mul t0, t0, t0
	addi t4, zero, 1
compare:
	blt t4, t2 loop

bge t0, t3, endif
mv a0, t4
li a7, 1
ecall 
endif:


# f(x) = 4 if x >= 5
.text
li a7, 5 
ecall
mv t0, a0  

addi t1, zero, 5
addi t2, zero, 4

blt t0, t1, endif
mv a0, t2
li a7, 1
ecall 
endif:

# f(x) = 7**x + -2 if x == 6
.text
li a7, 5 
ecall
mv t0, a0  

addi t1, zero, 7 
addi t2, zero, 2
addi t3, zero, 6

mul t4, t0, t0
mul t4, t4, t1
sub t4, t4, t2

bne t0, t3, endif
mv a0, t4
li a7, 1
ecall 
endif:

#  f(x) = -1 * x + 10 if x > 9
.text
li a7, 5 
ecall
mv t0, a0  

addi t1, zero, 9 
addi t2, zero, -1
mul t3, t0, t2
addi t3, t3, 10

bge t1, t0, endif
mv a0, t3
li a7, 1
ecall 
endif:

Notes

1. There is no f(x) function: -2
2. Callee-saved registers are not saved/loaded to stack: -1
3. There are 4 programs:
   • f(x) = 6 * x**5 + 4 * x**3 - 5 * x**2 + 7 * x if x < 5 - does not work, -2
   • f(x) = 4 if x >= 5 - OK
   • f(x) = 7**x + -2 if x == 6 - wrong, 250 vs. 117647, -2
   • f(x) = -1 * x + 10 if x > 9 - OK

Result

10 - 2 - 1 - 2 - 2 = 3

ttttttt

Ejudge

Hi all!

This is Ejudge. Our contest system.

211: http://84.201.145.249/cgi-bin/new-client?contest_id=211

212: http://84.201.145.249/cgi-bin/new-client?contest_id=212

213: http://84.201.145.249/cgi-bin/new-client?contest_id=213

214: http://84.201.145.249/cgi-bin/new-client?contest_id=214

Links

https://docs.google.com/spreadsheets/d/1kFGcUtLRKk0gWtb5wSuvZNMLMVCo-orDJdcM54Vl2Jw/edit?usp=sharing

Key

a

Link

https://us04web.zoom.us/j/74252280655?pwd=wKv6dWnfhS6pi-yIGhJs-i31KlUIBy.1

Deadline Ejudge HW7

set deadline for compdc_1 HW 7 to 8.03.2024

OS Test

DSBA:

https://andrewt0301.github.io/hse-acos-course/part2os/os_final_test_dsba_2024
https://forms.gle/iwk4QTs9Zb9GT3sg8

COMPDS:

https://andrewt0301.github.io/hse-acos-course/part2os/os_final_test_compds_2024
https://forms.gle/vLXue9rES2T9gNxP9

Question

я продублирую что я хотел вам предложить насчет штрафов за просрочку: допустим за каждую неделю вы убираете на данный момент по 25%, мое же предложение состоит в том чтобы эти 25% поделить на 7 и считать проценты штрафа за каждый день, просто кто то может просрочить контест на пару часов и потерять существенное количество баллов, а данная система смягчит мелкие опоздания

Exam

DSBA:
https://docs.google.com/spreadsheets/d/1kFGcUtLRKk0gWtb5wSuvZNMLMVCo-orDJdcM54Vl2Jw/edit#gid=1775997570
COMPDS:
https://docs.google.com/spreadsheets/d/1nj-iDom2phr24SIESSsPhYvFh9nPEcd_dx1wgLftckg/edit#gid=1525021748

FLOPS

https://www.intel.com/content/www/us/en/support/articles/000057415/processors.html

Where Can I Find Information about FLOPS Per Cycle for Intel® Processors?

Summary
FLOPS per Cycle no longer available; instead there is GFLOPS (Giga-FLOPS).

Description
How many FLOPS per cycle does my Intel® Processor get?

Resolution
Please be aware that Intel no longer makes FLOPS (Floating Point Operations) per cycle information available for Intel® processors. Instead, Intel publishes GFLOPS (Giga-FLOPS) and APP (Adjusted Peak Performance) information. For details, see the Export Compliance Metrics for Intel® Microprocessors web page.

[Enhancement]: Rewrite benchmarks for Java and Python

While it's not a case for natively compiled languages like C, JITted languages (like Java) require much more careful benchmarking, taking into account JITting time and GC pauses (as well as possible standard deviation (due to many factors), tiered compilation, etc).

Broken link in HW1

In homework-1 we have a bunch of links to implementations of matrix multiplication. The last link is incorrect: the title says C incorrectly optimized (jki) while the link actually leads to ikj

Test

https://docs.google.com/document/d/1_N7JuKaqbtAqbrpKRRkad4Nkx3mIcRRD3LOZTxbXtXU/edit?usp=sharing

[Bug]: C seems to be compiled in default mode

C takes 2.7 sec at best according to the presentation (on my machine it takes 3.4 without flags).

Instead, if you supply optimization flags, it'd take 0.7 sec.

To benchmark fairly, one would need to compile it like this:

gcc matrix_jki.c -O3

Table with keys

https://docs.google.com/spreadsheets/d/1KVsUGHUl74Pfrx0GYiF2V-Cw0mFbVQ3428475FQKsDE/edit?usp=sharing

e

Hi all! Here is the link to OS Homrwork #1:
https://andrewt0301.github.io/hse-acos-course/part2os/02_C/lecture.html#homework
The deadline is 16.04 23:59.

test

test