Parallels 20 REVIEW - 128GB Virtual Machine vs M4 Max MacBook Pro 🔥

It states that it supports virtual machine up to 128GB in size, so in this review we take it to the extreme in Windows and macOS emulation.
BUY NOW (Coupon Code: Xcreate)
Parallels Desktop: vtudio.com/a/?...
TEST SYSTEM
MacBook Pro M4 Max 128GB RAM: vtudio.com/a/?...
LG C2 42" Monitor: vtudio.com/a/?...
COMPANION VIDEOS
128GB MacBook Pro; • 128GB vs 64GB M4 Max M...
64GB vs 32GB Ram: • Do You NEED the 64GB o...
M4 Max Developer Review: • M4 MAX MacBook Pro - D...
M1 Max Developer Review: • M1 MAX MacBook Pro - D...
SPECIAL THANKS
Thanks for your support and if you have any suggestions or would like to help us produce more videos, please visit: vtudio.com/a/?...
CREDITS
Vibe Tracks - TBF3
Links to products often include an affiliate tracking code which allow us to earn fees on purchases you make through them.

[@dbtech4562]

This is the only KZbin channel that will do this. Everyone else focuses on Final Cut Pro. I was going to spec out my MacBook Pro to maximum so that I had one device that can do my virtualization, work, gaming, video encoding, and everything else. I ended up just getting a regular M4 Pro in a MBP 16 and got a separate high end Windows laptop with a 17 inch screen and a 4090 and 6 TB of storage. Upgrading my existing Intel iMac to 128 GB dedicated for virtualization. So I spent the same money split across 3 machines instead of one.

[@xcreate]

That's awesome mate, and the great thing about multiple machines is that you can remote in and get them to do different jobs for you without bogging down your main system.

[@andikunar7183]

I like + use Parallels, but it still has it's problems: 1) still no nested hypervisor support for Windows (somewhat crippling Windows - no WSL2, WSLg,... even though it should now work with macOS Sequoia on a M3+) and 2) VMs are much slower than native (e.g. llama.cpp in a Windows or Ubuntu VM has 1/2 the performance than on the host, both running CPU-only. Windows with hyper-v or WSL2 has no such performance penalties).

[@dbtech4562]

@@andikunar7183 I have not experienced this issue. You are running Parallels on ARM and it is running the translation layer for Intel binaries? Or High swap due to undersized memory?

[@andikunar7183]

@@dbtech4562 thanks. I have done my tests with a M2 24GB MacBook Air, MacOS Sequoia, Parallels 20 Pro Edition. Running llama.cpp llama-bench (Q4_0 as in their github discussion #4167 benchmarking) with -ngl 0 -t 4 first on the host, afterwards native-compiled in a 4-cpu 16GB Ubuntu or Windows VM. Always looking, that enoughCPU/memory remains available and not doing other work. The pp512 on the host is ~59, in the VMs it is ~23 (regardless if Windows or Ubuntu). pp is compute-bound and affected much more than tg, which is mostly memory-bandwidth bound. The test was done with the llama.cpp version mentioned in the issue #4167, current versions accelerate CPU-processing by using a new matrix-mult. algorithm. Also if I then run it in wsl1 (with the right memory-setting) in a Windows VM, it has a really low pp512 of ~10. For comparison, running the test on WSL2 on a Snapdragon X has the same performance as in its Windows host.

[@andikunar7183]

@dbtech4562 update: I just re-ran the tests with the latest llama.cpp version of today (it now uses special arm-code for accelerating GEMM/GEMV operations) and the difference is now minimal, host macOS: pp512:~71, ubuntuVM pp512:~62. This difference seems acceptable (OK, with the GPU pp512 is ~192, but it cannot be used in VMs).

[@chocoloco9985]

I need to do a lot of research (no coding or anything fancy like that just word processing if anything) along with a Parallels Windows Virtual machine in coherence mode and I'm using 105gb of ram (always around the 100gb mark)

[@Racermousepr0]

My M2 max is happy with its owner

[@Polenabseits]

0:18 *M4 Max

[@paulyflynn]

Did I miss the announcement 😂

[@futtt_buckerson]

If he was MKBHD he’d probably already have an M4 Ultra 🤣🤣🤣

[@virility]

how does it compare to m1 max? running powerbi and alteryx on my m1 max is kinda laggy at times.

[@MrVaJoy]

It's better to use CrossOver to play games on Mac.

[@MirzaXBT]

think it depends due to cheat software

[@andreaslassak2111]

@@MirzaXBT yea, anti-cheat is problem

[@dbtech4562]

I noticed that CrossOver always struggled with the games I wanted to play. For example, I really like Star Trek Armada (original one). I was able to get it to work on a Windows XP VM on an Intel Mac. It will not run on modern Windows or any type of WINE software.

[@gamesthatmatter9374]

if its a lower end game vmware fusion is always the better way . much less issues especially with playing videos inside a game or using special fonts

[@Giljrg]

when you set the system RAM to 112GB I don't think you left any for the Graphics Memory, games might run a bit better if you configure the amount of VRAM available to Windows.

[@tomdrexel112]

Finally, a real test!!

[@muiravenue7182]

Do you think assigning 64gig to Window MS and 64 gig to Mac OSX you basically have 2 machines in 1. I work mainly on the Mac but want to learn Solid works. And use some CAD I have a Lenovo Thinkstation as well that I use for the Creative Suite. I have to upgrade my Mac Set up I am Going 16 inch Mac Book Pro 128gig ram? Some Clients are PC some are MAC? What do you think? Do it?

[@xcreate]

It's always good to use a dedicated Windows PC when possible, but there's no harm giving it a try to see how it performs for you.

[@DRAM-68]

Great job testing the limits of the machine. Have you tried playing those games with nothing else running in the background?

[@mlbrln]

Hello, thanks for sharing! Could you make a video and compare parallels and macOS with geekbench? I am wondering if I should buy a M4 pro and use parallels or just a windows pc..

[@JethroRose]

by allocating 112 GB you may have starved the GPU for video memory, hence games running like shit.

[@paulwoodward8265]

Agree. As a general rule, the host is should have about as much resource as the guest, to ensure the best experience. In your situation, with a lot of resource, you might get away with more. If you actually did have 110gb in use things would get pretty hairy for the host os. Same for cpu cores. But I’m sure you know what you’re doing here makes no sense! It’s just abuse!

[@xcreate]

They also ran the same at lower allocations too, probably because at that point Windows was only actually using 15% of the 112GB.

[@ywueeee]

try running UTM

[@olorin11]

Blender, v-ray 3D Max tests pls 🙌🏼

[@tc5006]

What monitor are you using?

[@xcreate]

LG C2 42" Monitor: vtudio.com/a/?a=lg+c2+42

[@gdotone1]

Clearly we could use 128 as a base. 256 and 1024 gigabytes with an os that will load the complete app in ram.

[@victormuniz695]

I have a simple MacBook Air M1 2020 (8GB RAM). That means I can only select 8GB RAM for the VM?

[@andrzejkulakowski]

Incredible hardware and revelation!

[@YuriySamorodov]

Hi, any chance to test nested virtualization on M4?

[@nyambe]

Fantastic info, thanks!!! good info for coders

[@sleepyelk5955]

sick machine 😂 thanks for trying this out, I am also still doing some load tests and will to some parallels tests later ... but if you imagine, such performance or such a setup was before the apple silicon era only possible on server like system, thinking of Ryzen Epic and so on, but at a cost of power consumption, heat and crazy fan noise in a big tower or server rig 😂

[@xcreate]

Let me know how you get on with your tests

[@fintech112]

I saw yourolder videos do u have app courses ?? U r a nice teacher let me know i would like to learn from u man

[@bartekmatuszak3694]

I wonder how Revit with Parallels would work on this set, otherwise great video.

[@refixed]

That's wild! Is it possible to run parrallels on multiple monitors like on a windows PC? if so, will it allow switching windows between the monitors with the windows shortcut keys? (windows+shift+right/left, obviously using a windows keyboard). I'm hoping to replace my work desktop and run three monitors dedicated to windows and one to macOS as my workflow would really improve with access to MacOS, and I just much prefer apple quality. Currently watching this on my youtube machine, a base model 2015 mbp with 8gb of ram and the original battery LOL still lasts a couple hours streaming with the brightness turned down.

[@Snackdance]

I’m not running the current version, but I used to run Parallels on 2011 i7 MBP with 16 GB of RAM with a 3 monitor setup. The Windows experience was complete. You could run it across displays any way you wanted and if it was running full screen, the experience was identical as running with my windows laptop, keyboard shortcuts and all. But I finally changed it so I “lived” in MacOS and ran the Windows apps like they were MacOS apps. It was a game changer. But I think to answer your question, you could run Windows full screen and across monitors, make MacOS nearly disappear. I certainly wouldn’t use it for final Windows software testing, but for the day-to-day workflow, the experience was (nearly) indistinguishable to running a native windows machine.

[@refixed]

@@Snackdance Thanks man, I ended up going with a binned m4 max as I don't need insane power, but want to run the monitors natively. So far, windows is a little buggy compared to on a pc, but that's likely due to all my custom mouse shortcuts interfering between the two systems. I will likely swap to coherence mode once I get some things figured out. cheers

[@Snackdance]

@@refixed real quick, I wonder if your quirks could also have something to do with Windows for ARM? I was using parallels on an Intel Mac, so it was running x86 windows. I’ve never used parallels with an ARM chip Mac, so with Windows ARM, x86 app emulations, etc. maybe there are a few more possible points for problems?

[@refixed]

@@Snackdance Definitely I think you're right. I can't even get the mouse software to run on the ARM windows. Otherwise it's working well.

[@Sequel7]

You stand no chance with 8gb of ram unless u want to run Linux or windows 95 lol Never buy a MacBook Pro with 8 gb of ram

[@victoririas2366]

how do you get the gpu meter without it having a game with metal running ?.

[@xcreate]

It's in Activity Monitor, just hit Command+4

[@pablosepulvedavaldes]

Good video 👍 It would be interesting another one to show advantages for using the M4 Max for development, I mean why M4 Max and not M4 Pro

[@xcreate]

I did a video on development on M4 Max previously you can check out. M4 Max is worth it to me for double the GPU cores and memory upgrade options.

[@josiaphus]

You don’t have the m4 ultra

[@robert3design489]

Do Stable Diffusion 1111 Testing on m4 Max

[@andreaslassak2111]

Use CrossOver for games.

[@asafblasbergvideographer]

M4 Ultra? :)

[@mickecast]

Were your fans like crazy while doing all of that? Thank you

[@xcreate]

Nah, they just revved up slightly when uncompressing (as shown), then calmed down after.

[@nastybadger-tn4kl]

I want to develop app and test but no one willing to test. So i have to vmware or parallel. anyone have experience in both?

[@crippledmind]

Now this is real testingggg🔥🔥😂. Great videoo!!. Also really glad to have the M series mac. Proud owner of m1 macbook air that still is much powerful and power efficient🫶

[@juzan9999]

do I need to purchase a Windows 11 license?

[@xcreate]

You can use it without a license, but it won't let you do certain things like change the wallpaper.

[@MoFields]

Is this the x86 version?

[@dbtech4562]

ARM based Macs will never run x86 virtual hypervisors even with Rosetta. There are 4 different ARM hypervisors that I know of that will work natively with Macs and the OS needs to be ARM but with Windows and Linux they do have their own Rosetta type technology to run x86 binaries.

[@zoezebra4013]

@@dbtech4562 never say never. parallels just released preview version which allows to run x86 vm images on ARMs. but it's super slow apparently.

[@dbtech4562]

@@zoezebra4013 yeah, I saw that. Reminds me of SoftPC.

[@jazzsoft]

I am using VMWare Fusion, which became free a while ago.

[@xcreate]

That's excellent, I love freeware. Will have to give it another try.

[@muiravenue7182]

Now you are Talking

[@giantbear829]

Can you not conceptualise that it is not possible to use all of the parent system memory which is hosting the virtual machine. Logically that is an impossibility, due to parent system needing memory for critical functions and has to stay active.

[@TerkleTon]

So basically avoid 128GB.

[@euphoriantraveller8651]

Why?

[@ashm3697]

try counter strike 2

[@sanchesfree]

+1 :D

[@biggaryhunk]

M4 ultra? for where? LOL😂🤣

[@alessandromanuel728]

Hey what happen with the white shirt? and the hot girl? 😅 just kidding! thanks for the info

[@klaymoon1]

꼼수 안쓰고 당당히 인정하고 책임지는 모습 같은 남자로써 멋있네! 상남자야. 남자던 여자던 탑배우 100명중 98명은 이렇게 대처하지 못했을거라 생각함. 그리고 이정도 사생활이면 연예인이 아닌 일반 사람하고 대비해서도 깨끗하다고 생각함.