Did someone say virtual is slow?
Вставка
- Опубліковано 9 лют 2023
- Let's compare the performance of a real environment and a virtual one (running MacOS and Linux in a Parallels VM).
Grab Parallels for yourself: prf.hn/click/camref:1100libNI (affiliate)
Use coupon code: ZISKIND10
get TG Pro for yourself: www.tunabellysoftware.com/tgp...
👀 Stuff I use: (including course creation and youtube stuff): www.amazon.com/shop/alexziskind
▶️ Where is Parallels' FREE version? - • Where is Parallels' FR...
▶️ Visual Studio 2022 on M1 Max | First Look - • Visual Studio 2022 on ...
▶️ My Parallels 17 first impressions | Windows 11 on a Mac - • My Parallels 17 first ...
▶️ M1 MacBook Air Running Visual Studio and Xamarin for Mac - • M1 MacBook Air Running...
🛠️▶️Windows on Mac Playlist▶️🛠️
• Windows on Mac
- - - - - - - - -
❤️ SUBSCRIBE TO MY UA-cam CHANNEL 📺
Click here to subscribe: www.youtube.com/@azisk?sub_co...
- - - - - - - - -
📱LET'S CONNECT ON SOCIAL MEDIA
ALEX ON TWITTER: / digitalix
#m2max #parallels #windowsonmac - Наука та технологія
Grab Parallels and support my channel: prf.hn/click/camref:1100libNI - Use coupon code: ZISKIND20
This comparison is kinda synthetic in sense that the sample was doing purely computational stuff, where hardware makes virtualization transparent. The real applications do lots of system calls, and that's where virtualization overhead can be noticed, since many syscalls or driver calls have to be intercepted (and emulated) by the hypervisor.
It still depends. Inside server environment, hypervisor can have many more technologies assisting offloading, especially related to networking and storage. Even losing 2% of total performance is negligible, but separation and multi-os environment advantages on single physical machine are worth it.
@@brylozketrzyn Well, I didn't say it's not worth it, I meant such comparison should be taken with care, as not in all scenarios you will get this near-native performance. I myself started working in Windows in parallels, and I'm surprised how fluent it is.
Yeah, and even that works only if you emulate the system that's based on the same architecture
E.g. emulating x64 OS on x64 OS gives almost no speed loss, while trying to emulate ARM on x64, or even x64 on x32
@@narrativeless404 it is not an emulation. You may emulate network, audio or video hardware (but at least with network you got either VMQ or VFs present) but instructions to processor are essentially passed through hypervisor. If you got differenent architecture then it is emulation.
There are different kinds of emulation. You can emulate some processor instruction set, or you can emulate a device or part of OS. E.g. virtual machine emulates hard drive via single file on a real hard drive. Or emulates DirectX 11 API on top of Mac's Metal API. So while you're doing "pure" calculations in your program, you're running the code on "bare metal". But most programs do many other thing, which implicitly involve hypervisor and emulation.
I'm gonna say that the biggest flaw of that test is that it's purely CPU Load, the moment, you need to access HDDs or any other virtual devices, you will generate some overhead to do so, and it will consume a non negligible part of the host resources to do so.
you can passthrough storage devices
@@snowwsquire while that is true, in an heavily mutualized context, and Host (ESX) relying on iSCSI for the storage, you don't really want to do that for a lot of reasons.
@@snowwsquire what does parallels do in this regard?
@@snowwsquire
@@DanielHarrisCodesParallels maps local storage as network drive in my computer by default, to passthough it requires eject from macOS first then connects in Parallels, or manually choose always connects to virtual machines when VM is off
Amazing test Alex!! Never disappointing to watch your videos
I just got my first MacBook and I tried Windows in parallels. AMAZING! It’s mesmerizing how I can use MacOS and windows in time same time on the same machine. Using MacOS on one display, and Windows on another screen(only when I need windows). Also Parallels it’s very nice because of how well it works and integrates with MacOS.
Couldn't agree more!
Is the base model enough or I need 32GB ram for these kind of tasks
@@sayednab I have 32. Got the M2 max base model: 16 inch MacBook Pro, m2 max, 32gb ram, 1TB ssd, space grey.
I think you can find some comparison videos on UA-cam testing Parallels on different configurations
Can you use WSL on that Windows VM?
@@sheevys good question. I will try it. Give me 1-2 hours
Exactly what I wanted to see comparing metal vs virtual. How about comparing UTM(qemu) vs UTM(apple virtual) vs virtual box vs parallels vs vmware.
And comparing to emulation(amd64) as well
VirtualBox doesn't work on ARM
@@HuntingKingYT qemu isn't virtualbox.
@@HuntingKingYT works since 7.0 (Oct 2022) ;)
Another awesome test from Alex!
I love how you much info you give on the real world developer performance we get from silicon. Whilst everyone talks about video editing, you talk about the developer aspect of it. I love your content more than I can put into words.
Agreed. I found him searching for stuff as a dev and now he is on notifications for me!
Shill bot 🤖 activated
@@cedricvillani8502 beep boop
how does one use the apple icon
That was very unexpected. Thanks ❤
Didn’t realise you could run Linux VM’s natively on ARM yet, that’s mega!
I was waiting for Schwarzenegger v2.0 🤩
Article about MacBooks being used as a desktop with lid closed. My intel one being used as a desktop most of the time got the display replaced because of video circuits hanging the machine when connecting a second external display which led me to believe some thermal effects of having the lid closed that long of a time. Any better on m1/2 MacBooks?
Hey Alex,
I've been struggling testing this I find that Compiling code on Windows in a VM (Parallels) is much slower than the Volterra Dev box. I thought it was because I was using a 8GB MacBook Air M1 so just purchased a new M2 Pro 14" with 32 GB and its the same. On a small project I am finding that on average the Volterra Box will compile from the command line (msbuild -t:Rebuild -m) in approx. 1.5seconds but the Mac takes like 3.5-4seconds. So I am seeing for Windows Development workflow using Windows in Parallels twice as slow as the Volterra box infact my old Surface Laptop 3 is faster than the MacBook. Any thoughts have you tested compile times rather than CPU performance? Thanks Sean.
I’m impressed by Linux !
Great test, really entertaining 👍🏻😊
How do Apple Silicon Macs do running IntelX86/AMD64 workloads in a QEMU VM?
This is an issue for running docker containers that have not been recompiled for ARM or Apple Silicon.
Does Parallels have Intel emulation option (perhaps using Rosetta)?
Cool video! Can you do a comparison between Parallels and UTM/QEMU?
That Schwarzenegger machine is really creative idea u built for these laptop tests
Can parallels 19 run big sur vm in m1 macbook running ventura?
Hi Alex, which is the code python do you have used?, for test here :)
how to install mac os on parallels please guide
The only thing that might be different are processes that are io bound. There can be software in the VM implementation between the hardware and the virtual machine.
Would be interesting to benchmarks between different linux distros.
The other day I booted up two clean VMs of Ubuntu and Xubuntu and was surprised to find that regular Ubuntu was actually faster and snappier.
But that´s just a difference in desktop environment. They are both ubuntu. You´d not see much difference if any at all running CPU benchmarks. Ubuntu and it´s derivatives are probably slightly slower than others because of it´s reliance on snaps. Id expect debian to be slightly faster and I´d expect Fedora or arch derivatives to be even faster in most cases because they are more recent. Your comparison there isn´t really a distro comparison and more a gnome vs xfce comparison.
Anything that is rendered (relies on the GPU) you'll see more of a difference speed wise. GPUs don´t generally support virtualization in the same way CPUs do. So a lot of it is gonna be software rendered by your CPU. Some desktop environments may not play nice with that, or having a resizeable "screen" attached.
I keep thinking that sometime you'll have a video where the real Arnold starts the tests for you. Someday.
Do regular Windows apps run properly on the VM or only apps for Windows on ARM work?
For all CPUs I've used they have done so gracefully but in my case it has mostly been AMD and Intel
How much RAM does the same program uses between both OS? In mac experience every time I use my Debian 11 WorkStation instead of my MBP I think Linux uses less RAM, then when I have to work with a Java BackEnd , 3 Node JS FrontEnt at the same time I have to use the Linux WS.
For some reason Mac uses almost 50% more RAM
Throughput performance is the same or better in a VM than in the host, but latency performance is much better in the host. That includes system calls, CPU scheduling, interrupts.
Can run in I pad and i phone ?
I dont understand anything is it save to use parallel on mac m2 base model for music production
what is your DAW and effects chain? If it has ARM compatible versions then you’re good. But many DAWs already have Mac versions (except Cakewalk I believe)
Hey, would you be able to tell me whether the M1/M2 Max+'s can actually run 3 external displays in extend mode? They're supposed to but I don't quite trust it.
Hi Alex, I redid your same test on my hardware and I got controversial results, what do you think?
From what you can see, the Parallels hypervisor is more efficient than the Apple one. And the Mac Arm, in vm, doesn't seem to be as fast as bare metal native execution. Surely there is to take into account that the operating systems are different.
Native macOS x64 (iMac 19.1 i9 9900K): 33 seconds
Ubuntu 22.10 VM with Apple Hypervisor: 45 seconds
Ubuntu 22.10 VM with Parallels Hypervisor: 29 seconds
Native macOS ARM (Macbook Pro 14, base model M1 Pro): 22 seconds
Ubuntu 22.10 VMs: 29 seconds
I enjoy learning but have no idea what this means. Linux uses the same hardware as the host, so how can it be faster than the host?
I wonder if macOS doesn't allow you to choose # of cores because its actually just some hybrid container using the existing macos image. Perhaps with an overlay FS
Containerization only works for operating systems that share the same kernel. Parallels may very well be doing that for MacOS, I'm not familiar enough with containerization in that space to say for sure. But I guarantee it wasn't doing that for the Ubuntu test, which had higher performance. MacOS does not use the Linux Kernel. It is UNIX like, but it's kernel is unique and proprietary.
Hello Alex, this is a very nice test. Furthermore, you expressed that you use VMs a lot. Can you tell me if you see any bottlenecks while using Windows 11? I will try to explain the reason for me writing this. Lately I have tried the new M2 Pro in MBP16. I though I want to see how faster/slower is my Windows Software running on Apple CPUs. I have tested some Codesys, Siemens, Adobe products. My baseline was MBP16 i9. Sad truth is all of the compilation, computing done in VM was 2x slower. And it is hard to blame DLLs or Rosetta? BTW. Geekbench in a VM was better then i9. What thoughts do you have on this?
Virtualisation -
If its a cpu run/bench, then it should not equate to a large difference.
But drivers, GPU, Hardware APIs or systems will differ and can differ a lot. So when we say Virtual vs host, its not just a simple CPU test. (At least in the mobile, desktop workloads!)
Alex, is your Ubuntu VM an x86 or Arm version? I would like to find out if I can run an install of Windows 10 (x86) or do I have to run the Arm version of Win11? I work with an x86 Windows application that I would love to move onto my MBP… its the last real holdout in my migration from Windows… There’s always one…🤨
It’s all ARM
I'm a Mac and Windows sysadmin. You would have to run the ARM version of Windows using parallels, and from there can execute x86 programs. That being said, there are some programs that simply do not work, including ones which rely on external hardware components and occasionally VPN utilities. There's also a performance hit, which is not typically noticeable using smaller legacy programs, but can be a major issue when running much more intensive applications like ArcGIS and Solidworks.
@@whenhen i consider any x86 software that has to run on Windows for ARM a non-starter. Even when running on windows for ARM natively on ARM hardware like i described in my Volterra tests.
Does virtualbox have opengl support for Linux guest ?
this test can be right even when i'm using an intel model ?
*last 16'' with intel*
I'm curious about the disparity between the Linux and macOS Python interpreters. Is it due to the how well the interpreters have been optimized on each platform or the underlying system itself? Remember, the virtual Linux running on top of macOS has all the macOS resources running on the CPU too.
So what's up with that?
With this and your other Windows 11 video you sold me Parallels Desktop haha
What a great video, keep them coming, absolutely had a blast watching it while drinking my coffee.
Don't forget to drop a like folks, Alex is the best.
Glad you enjoyed it
Those Ubuntu and Windows are x86 or arm?
Thank you for awesome tests that give us some insights.
Could you try running a test for Hackintosh (MacOS virtualized on Windows Laptop host) this will be compared to all you already had tests in this video (MacOS host, MacOS virtualized on MacOS) the goal here is to have a comparison between running a MacOS natively on a MacOS machine, and running it on a Windows machine
Is there a service from Parallels where anyone can get Ubuntu on Apple ARM over the cloud?
That would scare AWS maybe
I'm impressed that Linux inside of Parallels runs basically on a native level. I've only been using VMWare so far. Will definitely try Parallels.
By the way, this is what ChatGTP has to say in regards to why Python runs faster on Linux than on Mac.
Optimized compilers: Linux systems typically use more optimized compilers, such as GCC, compared to MacOS, which can result in improved performance for CPU-bound Python programs.
Native support for POSIX: Linux is based on the POSIX standard, which is well-supported in Python. This means that Python programs on Linux can take advantage of the underlying POSIX support, resulting in improved performance.
Which is wrong, because MacOS is actual POSIX Unix, whereas Linux is merely compatible with it.
@@katrinabryce It is true that macOS has full POSIX support and is fully certified for it, and it is also true that Linux is not officially certified for POSIX. However macOS uses the much older BSD implementation of POSIX utilities, which is heavily outdated in terms of performance. Linux uses the much more modern GNU implementation of POSIX utilities since kernel 2.6. The confusion starts with POSIX certification. Since POSIX is a group of standards, it requires an official certification by the IEEE, which neither the Linux foundation nor any Linux distribution ever cared about. Not even Google cared about having Android being certified. Why? Because the POSIX certification simply checks for compatibility and not for performance. Linux since kernel 2.6 is fully POSIX compatible, and has a better and more modern implementation of POSIX utilities than macOS. Apple however decided to get macOS officially certified, even though they're using the older BSD implementation.
Edit: I just read that since 2017 macOS is technically no longer POSIX compatible, due to an increase in missing features that have been added to the POSIX in the 2017 standard. MacOS has been POSIX certified in 2007 and still holds a valid certification, though it no longer complies with the 2017 POSIX standards.
Would love to see how a docker VM tests out. Another great video!
Thank you!
Curious what results would have been for a test running on a Linux as a host os on mac. Would it beat Linux guest results!?
There's not really support for running MacOS on non apple hardware, and not doing so I believe is in the MacOS EULA. Apple does all it can to prevent people from doing this, both with legal protections, as well as software security. There are people who have hacked MacOS onto other hardware, and there's documentation out there for doing it yourself, but in my experience it's very difficult, there's some aspect of the experience missing (graphics or audio drivers are common culprits), and there's often a performance hit as well, as the operating system is highly optimized for specific hardware configurations.
@@shishsquared it's not what I meant. In video test was performed in Linux os running in vm on mac os host on apple hardware. And I asked about running the test on om linux os host on apple hardware.
I’m a long-time user of Parallels and use it to run Windows 10 and Ubuntu for work projects. Before I buy an Apple Silicon Mac, I’d like to know how Windows 10 performs under Rosetta2 in virtualization.
As far as I’m aware, it doesn’t. You install an ARM build of Windows, and use WOW (Microsoft’s equivalent of Rosetta) to run Intel binaries.
Now… run a virtual machine in a virtual machine and so on until it becomes unusable
Which app using for tempreture?
forgot to put my link in the description. here you go: www.tunabellysoftware.com/tgpro/index.php?fpr=alex
could you possibly start a series where you show how to set up your mac for development for many different languages? I know you already have one or two general videos setting up things like Node and homebrew and vscode, but It would be cool if you did episodes about how to install rust and the best IDE for rust. Or even how to install python and how to set up an IDE/environment like Anaconda. I would love it if you also showed how to do these things on a Windows computer as well. Given that I am still learning to code, I use a mac at home, but if I go to work, I need to use Windows and would love to know how to do these things on Windows as well. But I mostly care about Mac.
I think this is a good idea too!
Already have one on node setup and another on conda setup. i’ll consider doing more of these. thanks
Well this is easily explainable if you know virtualization. It's magic.
If you have a good ram and good hardisk space it runs well for example in VMware. One thing I found out if the disk space is low it slows down but when you connect an external harddisk the speed improves significantly but the downside is that sometimes it might be difficult to enable 3d acceleration graphics.
VMware fusion was a good option before Apple Silicon. I've been waiting for over two years for them to put out an Apple Silicon version, and they haven't put out a usable solution yet.
Ah reminds of running Pc anywhere with Windows server/SQL server on a slow usb disc drive. Beat the hell out of the costly HP hardware servers.
@@AZiskthat’s just not true and you know it, I don’t think you understand the difference between virtualization and containerization
it's nice to virtual machine efficiency has improved. It used to be VM automatically took a 30% overhead.
Hasn’t been this way for years I believe. Even back when I had a Ryzen 3950x, performance on a VM was only 3-5% slower. So basically almost margin of error. I think in the GPU space the performance hit is a bit deeper if you’re using GPU passthrough, but it’s still not very high.
@@livedreamsg this just shows my age :) Back when vmware first started, that was a real issue. Even today, not all virtual machine platforms are equal. For example Citrix remote environments have lots of over head. In the consulting space where Citrix is common, lots of people like myself have to suffer the horrible latency. I've lost count of how many times my team had to call service desk to troubleshoot slow Citrix.
@@woolfel Ah, yeah. I have mostly used virt manager. It’s extremely fast.
Unfortunately for me my jest tests run unbelievably slow on my maxed out parallels
You didnt show version of python in ubuntu. Maybe it is newer and thats why faster?
3.9.12
Can you do an emulation test with UTM? So testing x86 emulation OSes like Windows XP/Windows 7/Windows 10 on ARM? Or even PowerPC on ARM? Would be fun
It will be a pain. I tried to run the corporate VM image with Windows 7 x86 and login into the system took like 20 minutes. Very high CPU usage and the system is so slow that running anything inside is pointless.
Virtualization is not limited to CPU. It include storage and I/O devices. Linux is used mostly for database and web server or docker container, so realistic virtualization benchmark should include such area to get practical comparison.
You need to find a jingle for the Shwarze..
whats a good jingle?
good video
Linux VM faster than MacOS baremetal is within my expectation 😂😂
In general my VMs are faster than the same OS on raw HW, due to the disk IO. I run everything on the AMD64 instruction set. As Host OS I use a minimal install of Ubuntu 24.04 LTS on OpenZFS 2.2.2 and the hypervisor is Virtualbox. The trick is, that the L1ARC cache of OpenZFS in my Host OS is 4 GB and all caches and storage are lz4 compressed. Almost instantly the VM disk IO runs from the L1ARC cache of the host. It is like running the VM from a RAM disk.
Both system VM and raw HW are installed using the default ext4 filesystem. The VM gets ~99% of the disk IO from the L1ARC cache of the host (99% is my measured hit rate of the L1ARC cache for the disk IO). The raw HW (ext4) has to read the data from the SSD, but the default IO caching is based on LRU and uses less space, while the L1ARC is based on a dynamic optimization between LRU and MFU and it has smart mechanisms to optimize expected future sequential reads. Ext4 is uncompressed and OpenZFS is by default lz4 compressed and that almost halves the number of records you have to read.
Why are my VM faster than raw HW?
- The difference between CPU performance is neglectable.
- The VM profits from the advanced features of the OpenZFS file system provided by the Host OS for free, features that are largely missing for raw hardware.
- The ext4 cache is lost after a reboot, the L1ARC cache is only lost if the Host OS is rebooted. If the VM (re)boots part of the L1ARC host cache is still available dependent on passed time and system usage. Everything will remain cached, if you select the VM reboot button :)
Now compare real vs virtual DOM performance 😮
Yes
nice
The second half of the video involving M2 benchmarks got a bit confusing for me.
my ryzen 5950X runs this exact same code with same parameters on bare metal arch linux in an average of 24.2 seconds using python 3.10.9, and 25.9 seconds using python 3.9.9. I though I was missing out on M1/M2, but I guess I'm still good
Windows probably works as well/better in Parallels as it does on Microsoft’s ARM hardware.
Nice. But Parallels you buy frequently (whenever there is a new macOS release). Then, it does not really work that well with 8gb ram. Therefore I would not recommend the setup for lower speced machines. And a native windows is usually better aligned to the hardware like graphics or network cards. In other words. Instead of upgrading a Mac to spec it for windows, I would recommend a dedicated windows machine with a native windows install.
That thumb pressing machine - what is it called? - looks awesome haha
it’s the schwarzenegger
I used the migration tools for the first time recently and it was nice, but I had to do a factory reset. There were so many quirks after a few weeks. The configuration and the apps was nice that I was able to continue working like nothing. Did save me a lot of time. I just wouldn't recommend it. It's hit or miss really
Sorry to hear that. Works for me so far 🤞
@@AZisk oh yeah hopefully you don't get issues. Crossing my fingers for you too 😆 🤞🤞🤞
@@pqsk i made a video about the migration assistant. perhaps i’ll publish it.
@@AZisk yeah that would be nice señor
So it's cool to run MAC OS I love it but sometimes it's quicker to run some tasks in a Linux VM. On Windows, I run a Linux VM for flexibility and speed in specific tasks. So I find myself working in VM more than Windows.
virtualization != emulation, I see many confuse the conceps, normally emulation must be slower by design but for pure math calculations it may not see this differnce, especially when code is in cache
exactly
@@AZisk this makes me wonder how emulation performance compares :)
I do not get how the softwares I use based on excel/word macros + Outlook are so fast on Parallels (M2 MB air) compared to my top of the range laptop 12700. Everything is snapier. looking at benchmark it should not be that fast. There is a bit of magic in this M2
could be memory bandwidth... depends on what the scripts do
@@mahonar That s what suppose too. or W11/office on ARM64 is just so much faster
It can be faster cause glibc is that much faster. You also did not stop the virtual machine when runing host.
actually, you missed a big point here.
all of the CPUs (from maybe 10 generations ago) has some features to accelerate virtualization. for example, intel has a VT-x feature.
With this feature, most CPU tasks can be directly run by the CPU, without any host machine intervention. something like mode-bit that is used for kernel-mode and user-mode separation in the OSs.
this benchmark is not a good example to separate the speed difference between virtualization and bare-metal in modern systems.
For this case, something like DB speed test is a better benchmark.
You’re missing out on a key point. All operating systems are multi-core supported. If you have 8GB with 8 mediocre cores, your host machine would outperform your virtual machine by a very significant amount, especially in smoothness and UI fluidity.
Do Xcode build comparisons between M1 pro, M1 Max, and M2 pro, M2 Ultra
It seems that single-core performance is more important.
ok you read comments 😁try installing crossover and steam , test network lag ad fps on mortal kombat , tekken 7, guilty gear on the M2 Max for gaming & development, we all wana know
I don't know much about games on the Mac, but @Andytizer reviews that stuff in detail.
Would you review book 3 ultra, because im thinking to buy one for my engineering college
Faster in Linux because its scheduler is MILES ahead of *BSD (or Mach) from which MacOs is derived from and has not really evolved since 2001... much like Windows scheduler that didn't (except when AMD Zen came out and then 12th gen Intel big.Little) get any real optimization since 2003. That's also why most servers around the world powering the economy run Linux and not Windows/*BSDs.
im using VirtualBox. My host has everything in the world. i have all the RAM and memory in the world. no matter how much i adjust the setting i can make the vm run at a normal speed. i just want to have linux VM just like everyone else and have it run like normal.
is there any community that exists that i can got to. i will pay money! anything! i just need help to fix this issue!
then people say ubuntu a slow distro :) (ya, i now it's mostly snaps...)
These machines running Linux on bare metal must feel so good
Comparing Python programs running on two different platforms can be problematic. Differences between Python versions and compiler types (Clang vs. GCC) and compiler versions come into play. Also, Python 3.11 is easily 25% faster than Python 3.10, for example, and the default Python on your Mac might be older. Try comparing C or C++ programs using the same version of Clang on Linux and on MacOS. Or maybe Rust or Zig. Or Go.
i dont wanna be nitpicky... but i have to be :D
you should close all the other apps on your host (as well as in the VM) that are all taking 0.5 to 30% (chrome...) cpu.
because while the VM and also the host still will have a full CPU core due to macos cpu scheduler being rather robust... the bottlenecks that can occur here are of rather random nature and can range from memory-subsystem , interrupts to I/O bottlenecks that may mess with the result. agreeably your results were rather close so on first sight this would make you think - oh they perform the same - in some kind of 'expected but at least a bit unexpected' sense so you wont question the result.
but it should be cross-checked to rule out that for example the host system could have been faster.
also! an ADDITIONAL cross comparison only on the host should be made with one time the VM open and one time VM closed to see if it affects the host result.
sadly this cant be tested for the VM case unless you would run a bare-metal hypervisor type of virtual machine
So how does Windows 11 perform?
Of course you can have the same amount of cores and logical cores and FMUs -- whatever -- as in host... In fact you almost always do.
uh oh, no comments 🙃
here you are :)
Wait until he discover that docker for Mac is a virtual machine.
time is here
@@AZisk Hyperkit is an hypervisor that allows Docker for mac to run containers as if it were a real linux environment.
On the other hand wsl2 uses hyper-v hypervisor to run docker on top of windows.
Only docker containers on linux baremetal runs at full speed.
to be honest the m2 does have better single core performance than the m1
so linux is faster than mac?
I appreciate your work, but does this need to be a 9m video if the relevant information presented could be summed up in a table?
this is a long form youtube video, not a short, not a tictoc. so yes.
@@AZisk would it be an option to add the results in the description/comments? I‘d love to be able to go back and quickly get the information I‘m looking for. I usually watch UA-cam kind of passively while I‘m eating so this would help me be able to digest the benchmarks when I‘m ready.
Paralells-Desktop-Subscription sucks alot....over 100bugs per year to use more than 4 cores on your imac....i call this fraud
M2 pro machine learning tests please
Like for the Schwarzenegger!
PC slow? Just virtualise some more ram ez
One day we will have ARM processors everywhere and Linux will be the Boss.. 🙂
What about Windows?
I saw the title and got pumped the video is about Windows or at very least linux (most importantly x86 processor) and then got disappointed that this is about macOS with M1 chip :/
1. the thumbnail literally has mac and macos on it. 2. i have linux in the video and do comparisons with it.
Apple keeps putting Apple craps on iOS, macOS, watchOS release after release. I'd be the happiest man on Earth if I can install back my Leopard (macOS without a lot of Apple craps). Sometimes you think you're upgrading your hardware with a new mac but some of that upgrades is powering the new Apple craps included in the OS.