2.10 Optimization management

✅ Optimize the computer

• File Management: FAT32
• RAM management: Segment, Paging
• Process management: FIFO, SJF
• IRQ/DRG: DMA, Polling, PIO
• 👉🏻 All this make the computer overloaded
• 👉🏻 We need to optimize the computer

✅ Priority of computer components

1. CPU
2. Main Memory RAM
3. Secondary Memory (HDD, SSD)
4. Peripherals

• 👉🏻 We should optimize the computer components following this order

1️⃣ Lets help the CPU

🤒 Problem that CPU suffers

• 1️⃣ speed problem, being slow ➡️ bottleneck problem
• 2️⃣ temperature problem 💊 better refrigeration problem, better ventilation

💊 Solve the bottleneck problem

• Use CPU cache
• CPU cache is inside CPU
• The CPU cache is divided into 4 potions

💡 CPU cache is stable

• in general, CPU cache is not deleted
• CPU cache is stable(not deleted after session finishes)
• unless we manually clean them

4 types of CPU cache

• L1D: Level 1 Data cache
• For data that CPU uses very very often
  • 👀 power of 2, as in computer everything is in binary, it is stored in cache for frequent access
  • 👀 1s and 0s
• ☝🏻 one L1D per core
• it needs to be close to the CPU, as it is used very often to the nueman machine

• 🐁 size: comparatively small, we just want them for super frequent things

• L1I: Level 1 Instructions cache
• for instructions that we use very very often
  • 👀 standard mathametical instructions, like sum, minus, mul, div, move(move data to another space), sqrt(square root)
• ☝🏻 one L1I per core

• 🐁 size: comparatively small, we just want them for super frequent things

• L2: Level 2 Cache
• Both instructions and data mixed together that we use quite often
  • 👀 mathametical operations that are not so frequent: log, cos function...
• 🐇 size: bigger than level 1, need to save both instructions and data
• 🗺️ location: located further away from the neuman machine, compared to level 1

• ☝🏻 one L1I per core

• shared L3: Level 3 Cache
• Both instructions and data we use sometimes
  • 👀 loops, like for, while
• 🐳 size: bigger, very big
• 🗺️ location: far from neuman machine
• ✋🏻 Shared level 3: Not one per core, shared among all the cores
• Level 3 cache uses front side bus

❓ If I have 8 cores, how many CPU caches do I have?

- L1D: 8
- L1I: 8
- L2: 8
- L3 : one shared L3 per core

👉🏻 in total 25 CPU cache memory

💡 ISA

• Instructions Set Architecture
• set of instructions that the CPU needs to exectute
• the instructions that come by default when I buy my computer
• factory settings of my CPU

• the instructions that the neuman machines do by default

• CPU cache helps ISA, as some of ISA instructions are stored in the CPU cache

❓ If I reset my computer to factory settings, pure nueman machine, am I deleting the CPU cache memory?
👉🏻 yes, we are deleting

w❓ hich element of the CPU cache helps the following?
- IPC to the kernel
- ISA
- IRQ
- DRQ
👉🏻 ISA

2️⃣ Lets help the RAM

🤒 Problem that RAM suffers

• 1️⃣ problem of space: even smaller bc of process table, pages table, vectors table, service routines and above all, processes!

💊 Sharing

• there are some common portions that several proecesses share
• 👀 menu bar in word, excel, ppt are more or less the same
• If some processes have some elements in common,
• then put the same elements just one time on the RAM

• and share it among the processes

• 👎🏻 the shared elements would be less secure
• as it can be accessed by several processes
• sharing can be deactivated

💊 Virtual Memory

✔️ Some processes are ok to be slow and is not used so frequently

• can be slow
• is not used all the time
• we do not need the process to be there all the time in the RAM

✔️ Page File and SWAP area

• we can leave the process in a specific part of the Harddisk(secondary memory)
• called in Page File windows/mac/android and SWAP area in linux
• 👉🏻 in page file/SWAP area we store the pages of the process that
• we don’t use very often
• and we do not need so fast
• like having an extra storage
• 👍🏻 makes the RAM feel bigger than it seems

• 👍🏻 Feels like virtually, I have more RAM

• its called swap, bc when you need it,
• it seems the process is inside the RAM

• But actually, the process is inside the hardddisk

• So in harddisk, it is generally for closed program
• but in SWAP part of harddisk, there are opened process

✔️ Format for Page File and SWAP area

• This area is in the secondary memory
• ⚠️ the format is different from starndard( NOT FAT32, exFAT, NTFS, ext 2, 3, 4 ❌)
• the format of Page File and SWAP area is called swap
• as Page File and SWAP area is for helping the RAM, swap format is similar to the RAM

✔️ Virtual Memory

• VIrtual Memory = my RAM ➕ SWAP in Harddisk
• VM 🟰 MM ➕ SWAP area(SM)

✔️ Calculate the VM

• If RAM is small ⬇️ SWAP must be big ⬆️
• If RAM < 4GB, SWAP must be two times the RAM SWAP = RAM * 2
• If RAM = 4GB, SWAP must be same as the RAM SWAP = RAM
• If RAM > 4GB, SWAP must be half of the RAM SWAP = RAM / 2

❓ 3GB of RAM, my SWAP should be?
👉🏻 6GB

❓ Then how much is my VM?
👉🏻 9GB
------
❓ 4GB of RAM, my SWAP should be?
👉🏻 4GB

❓ Then how much is my VM?
👉🏻 8GB
------
❓ 16GB of RAM, my SWAP should be?
👉🏻 8GB

❓ Then how much is my VM?
👉🏻 24GB

⚠️ Error we can get if we do not follow the VM caculation rule

• If we follow these rules, we will never have a problem with the virtual memory
• However, if we do not follow these rules, we might get a problem

• cannot read instruction on 0x000000

• if you see this error, you have a virtual memory problem

• this error occurs bc many OS take SWAP smaller than needed
• many OS prefer more pure HD than more SWAP
• 💊 deactivate the automatic VM supply(administración automatica memoria virtual)
• 💊 and supply VM manually

✔️ Perfect limits for VM

• SWAP should always be between half and double the size of the RAM
• half of RAM < size of swap < double of RAM
• if my RAM is big, if my SWAP small, SWAP is half of RAM my VM would be RAM + RAM/2 = 1.5RAM
• if my RAM is small, if my SWAP big, SWAP is twice the RAM my VM would be RAM + RAM*2 = 3 RAM
• Thus, the VM size should always be

• RAM * 1.5 < size of VM < RAM * 3

• 👉🏻 This size of VM would avoid the problem of VM 0x error

❓ I have 8GB of RAM. Study the minimum and maximum limits of VM so I do not get the VM error.
👉🏻 8 * 1.5 < VM size < 8 * 3
👉🏻 thus, 12GB < VM size < 24GB
👉🏻 as long as I guarantee that my computer makes VM from 12 to 24GB, I will never get the 0x error.

3️⃣ Lets help the HardDisk

💊 Harddisk Buffer

• buffer: temporal storage for making profit of a data transaction

💡 proximity principle

• when you open a program, most probably you will need more than that
• 👉🏻 open more, also the ones next to the pointed program
• 👀 if you opened ejercicio 1, you will also open 2, 3,4…
• the extra can be stored in the harddisk buffer to have it closer to you.

💊 Web cache, Broswer cache

• when we download elements from the web and we visit web pages
• most probably, we will use the element/web page later
• store the elements in the broswer cache
• web cache is a folder in the harddisk
• when we empty the cache, we are emptying the web cache

4️⃣ Lets help the

✅

✅

✅

✅

1️⃣

2️⃣

3️⃣

4️⃣

• 1️⃣
• 2️⃣
• 3️⃣
• 4️⃣ 👍🏻 👎🏻

⭐️⭐️⭐️ EXAM ⭐️⭐️⭐️
❓
👉🏻

------
❓
👉🏻
❓
👉🏻
❓
👉🏻