Windows History, Windows vs Linux Comparisons, Arduino Project Tutorials, Shop Projects, ESP32 Information and more in Dave's Garage. With DIY info and practical step by step tutorials. Uses visuals such as the ChipCam to make complex topics easy to understand. From hobbyist to engineer, everyone is welcome!
One thing that would be extremely cool would be if you could bring up two computers on the screen at the same time. One to the right and one to the left. And then, whichever side you drag the mouse to, that side becomes active and so the keyboard would be active on that side only. The monitor is wide enough. It probably could be done in software. I know there are third party products doing this but I would like it in the monitor itself. Also, how about a little device with buttons connected to a cable for quick switching if you are in full screen mode, connected to one of the USB ports.
For those with less hardware, the Llama 3.2 3B Instruct model is good for chat and requires a lot lower specs to run. I am able to run it on GPU using an Nvidia GTX1070Ti with only 8GiB of VRAM. So far it has been on par with Llama 3.0 & 3.1 for my use while being a lot faster. To get it, run the following: ollama pull llama3.2:3b-instruct-q4_K_M
innacuracy in description: you should also forward port 11434 when starting web-ui docker to be able to manage models from the web ui, so parameter `-p 11434:11434` also should be added to the last command
What I'd like to understand is why on most modern x86 operating systems, only two of the 4 rings are used. It seems like the other two rings would allow device drivers to run with more privileges without giving it full ring 0 power.
Just informative and entertaining? Man, I have learned so much from your channel. This is the best part of YT for many of us. Computing 101 explained to lightened and unlightened people. Keep your excellent work and I am very happy to see you a little bit more slim than in previous videos.
I worked on VAX/VMS for a while, there were 4 (!) access modes: user, super, exec and kernel. A problem in user mode crashed the program, in supervisor mode ended the process (sudden unexpected logout!) and both exec and kernel modes crashed the system. I never saw much point in having the exec/kernel distinction. The 'problem' was usually accessing memory address zero which was never a valid address on VMS. In Unix they said null pointers are so common so let's make zero is a valid address so programs with null pointers will keep running, while in VMS they said null pointers are so common let's be sure to hit the programmer with a big hammer every time. (Right approach IMHO.) VMS did have some 'free' memory; well-known memory addresses which any program could read and write. You could home brew some inter-process communication using that memory. Neat idea but I never heard of anyone doing that though.
In this example, the problem boils down to how math is taught and how we collectively fail to make the intent obvious. In many people's minds, when we see 2(2+1) we are grouping them together. Everyone would get the right answer if the problem were written this way: 6 ÷ 2 x (2+1). Done this way, no one can mistake the 2 outside the parentheses as being part of the parenthetical grouping.
Thank you Dave for your good content, in the same area, I recently started a project to organize my photos, using techniques similar to Google and Microsoft to auto-classify images with tags and descriptions. To achieve this, I combined Ollama with ExifTool and custom scripts to automate the tagging, titling, and description process. Processing was done on my 2011 machine, which took 15-20 minutes per image-a clear sign that it’s time to consider an upgrade! Nonetheless, I’m pleased with the results and look forward to optimizing performance further.Why not give it a shot? Here's a simplified script for setting up keywords (tags). (ollama and exiftool are required.) @echo off :: Set variables set "IMAGE_FILE=%1" echo. echo %time% Tags generation ollama run llava "create a list of 5 keywords separated by spaces for this image: %IMAGE_FILE%" | findstr /R /C:"^" | clip if errorlevel 1 ( echo Error: Failed to generate tags with LLaVA. exit /b 1 ) echo %time% Tags completed :: Get the clipboard content (once) for /f "delims=" %%i in ('powershell -Command "Add-Type -AssemblyName System.Windows.Forms; [System.Windows.Forms.Clipboard]::GetText()"') do set "TAGS=%%i" echo The generated tags are: %TAGS% echo %time% Write tags with exiftool exiftool -overwrite_original "-keywords+=%TAGS%" "%IMAGE_FILE%" if errorlevel 1 ( echo Error: Failed to write tags with ExifTool. exit /b 1 ) echo %time% Processing tags for file %IMAGE_FILE% completed
Why only two modes? Who not a system with 3 modes: "system modem" (god mode) where errors necessitate a BSOD/kernel panic, "driver mode" (direct UI mode), and then user mode. Perhaps then, a fault in driver mode could be recoverable by the systemmode code that could restart only the drivers and keep the user processes intact. This would seem to imply use of the MMU for drivers, to avoid that a buggy driver would trash the memory of another process.
Very interesting. Thank you. I would like to add that running a dedicated mini pc solely for added network security can add a hefty annual energy charge, at least in regions where electricity costs are significant. So one more decision to make is whether the added security is worth the cost, i.e. annual recurring $200-300 in electricity bill on top of initial device cost. Home users may want to look into milder forms of network hardening. That said, a great video providing insights into professional network protection. There are lots of valid use cases. Thank you for sharing.
Operating system as a whole package will use both for various components. For example Linux is just kernel, but GNU/Linux is operating system and has userspace and kernel space tools.
I love all the outro throwbacks to my childhood. Growing up in Saskatoon with only channel 4. 8, and 11 *oh and the fresh channel, but that doesn't count*
I used to be an AP computer science teacher. The students absolutely were given information about the Therac-25 (though with the naming of the machine/company removed to prevent unintended liability). They learned of other critical software failures that had far reaching consequences and the importance of software quality. That was a long while ago (my teaching) but I hope that those students learned that not just from a software basis, but also for any profession they entered.
Didn't Stuxnet exploit 2 or 3 zero day vulnerability's? Whats the chance zero days are actively being written into software TO BE sold to Nation States?
It's the damned 2 mode architecture, god/not god. This problem was fixed in other architectures long, long ago. Get Intel to fix the frigging level ring transitions. But since you MS guys abandoned that for speed and the hardware guys figured why work on it if the largest code base didn't use it. Things got worse in true multi-core systems. Another great video.
How does this provide a secure connection when it has to connect to an open and unsecured public wifi network that doesn't even use a portal. I presume that a VPN is required in that case? I'm very much interested in this device because of your video.
So what we can and cannot do in Windows (or Linux) User Mode, is dependant on Kernel Mode? And if a user could access Kernel Mode, then User Mode functions could be compromised?
Back in the 1970's I worked with Honeywell 200 series operating systems. These systems had a primitive method of storage protection which disallowed the user mode code from executing instructions that could alter the registers involved in implementing the storage protect feature. Studying the operating system code I realized the designers of the OS had included a system call whereby a user could ask the supervisor to turn off the protection feature and return with full authority to see all memory and do whatever they wanted. Fortunately the internet did not exist and most users were unaware of this feature.
Although Apple Mac OS was not mentioned, Apple Mac OS also is very similar to Microsoft Windows and Linux in that it has kernel mode and user mode and the operating system protects itself from misbehaving programs by suddenly quitting the application. Also we are a lot better off than the days of MSDOS. Under MSDOS the CPU operated in real mode and any code that was running had full access to all the memory and hardware. So any program could lock up the machine where a hardware reset was needed. Often by turning the machine off and back on, if there was not a reset button.
Another great video. Thanks! At 4:38, I have a question about the architectural diagram: - In a retail install of Windows 10, the HyperV component is not installed by default. In that case, is HyperV still running as hypervisor, or is the NT kernel running on bare metal? - In this diagram, is it correct to say that both Windows 10 the Windows Subsystem for Linux (WSL2) each run as VM's on HyperV? TIA
Thank you for this video. Was wondering throughout, now with the higher bar set for hardware in order to run Windows 11, how much has this improved baseline safety and security of ones computer. I get the impression that these new standards, while they do impact the consumer and leave older systems behind, ultimately this was a necessary step considering all of the "new and improved" ways of compromising ones computer. Perhaps even a bit long overdue? Was wondering what your thoughts might be.
