Mac Studio Server Rack! CI/CD & App Dev
Вставка
- Опубліковано 28 сер 2024
- Major app development companies around the world use Mac Mini's for continuous integration and continuous delivery of iOS Apps. How is the new Mac Studio with M1 Ultra going to impact the many racks of Mac Mini's we currently use everyday? Let's compare physical footprint and dollars and see which device really is worth scaling with long term.
I knew there was a reason for maintain the cross-section footprint and expanding only in height. Now it makes sense! Cool vid!
This channel is so underrated.
By the way I’ve just graduated and working as sde. I work with react, spring even worked with android to some extent.
Could you please go into more in-depth of what exactly you code. Like no need for confidential stuff but like the scale of project, how many requests it handles, maybe lines of code etc. because I use 16gig machine and I feel it’s okay for small work I do.
I am blown by when you say 5mins to build on i9 and stuff. So more interested. No one shows like the kind of apps this big on UA-cam.
Big fan of you. Please go more into software also.
Thank you so much for your kind words! I will absolutely go deeper into the software and general workflow that I do everyday.
This guy is lying bro when he said his build won't run properly on a 16 GB machine.
How can apple do this but not provide ecc memory or a way to add ecc memory? I keep hearing how EVERYONE needs ecc for home lab or servers, but then I also see how pro creatives are running everything from video editing to machine learning on Macs headless without any ecc. So I have to ask, having all these minis running as servers are there any stories of a Mac crashing or a loss of data because of a mac not having ecc?
Interesting view. Really hoping Apple embraces web assembly and we can move toward a create once use by many.
you really spoke to me like I was 5yro explaining how the Studio compared to the Mini is a little taller and a lot more powerful.
Great vid! btw the lights in the background are flickering because the shutter speed on your camera is too fast. A longer shutter speed will let in light for more of the duty cycle of the LEDs and smooth out, if not totally get rid of, the flicker. Hope that helps.
Great video ! You already talked about the project you are working on ? Very curious to know !
Since you went into costs, I would have liked to see a comparison of the cost of doing this in-house / on-prem server farm vs 3rd party, cloud based CI/CD platform like Circle CI. If it's really a business decision, this should be a big factor as buying your own hardware can be very costly, even if you amortize it over time.
Love the gilfoyle reference.
Ventured in a different world I had no idea existed. Enjoyed the content.
Great video. A newbie question please. Why do you need 20 build agents? Do you work with a team or?
Great video. I would also be interested to know how much power do 4 Studios consume vs 12 Minis and the associated cooling costs.
This is the kind of channel that should be featured when Apple decides to unveil the new Mac Pro at WWDC. Excellent explanation of how you would use your Mac minis vs Mac Studios in a dev environment.
Wow what an improvement from its the LUT, well done.. I know your still working on the sound side of things. Great video.. and love to have a chat with you off line as I work in development from a product side of things..
Sounds fun! My channel page has a way to email me.
Nicely done, good analysis.
I was thinking the same, but the price is prohibited for me. Too expensive. I was expecting these machines to be like Mac mini pros, with a price relative to the current Mac mins but more power.
Could you provide more detail on how you knew 24GB of RAM was needed to compile the app you mentioned? Great vid
Curious to see how much RAM mine takes
Thank you. Great content.
This video repeats itself at 5:29 and 6:13
I went out of development but I really love your video. So I gave a sub
The Mac studio rack mounts are on back order, did you buy them all 😂
I only bought a dozen... 😄
Cool Workflow !
Thank you!
I feel like you could fit even more build agents if your virtualization solution allowed for some memory to be shared.
How about heat output -cooling it-in your server room? These are going to be toasters with loud fan noise :)
@d3vtec re. your easter egg about the strobing led lights: bc the electronic rolling shutter in the camera interferes with the led's light strobing from pwm ....
nice video and very relevant too!
Not involved in development at all - yet still quite interesting!
just wish they’d bring back XServe
They don’t even like making Mac Pros lol. Why should they bring Xserver back for?
@@Dave102693 well I didn’t say they would. I just loved it and was a delight to use. Linux racks are just meh on meh.
Do ypu think 128gb ram in an unified memory arm as apple silicon eorth it? I’m thinking about it, i do vfx and very heavy stuff but max tech make a video about how stupid how much ram is in an m1 chips bc its achitecture, what do you think about it?
What are you using for virtualization? The OSX License limits you to 2 virtual instances of OSX concurrently on a single Mac machine. How are you getting 5 independent Jenkins Agents running simultaneously on your setup?
Awesome & Thanks :)
What server software will you use since Apple discontinued it server software as of April 2022?
How did you measure the 24g ram to compile your project? How can I determine how much ram my project requires to compile?
What software do you use to run Apple silicon as a server without MacOS Server for apple silicon ?
My org still has us on Intel MacBooks, and everytime I want a nice 30 min break, I hit build on Xcode and walk away, because the system just chugs.
You need to update! From a security standpoint, many organizations are wiping out the Intel based Mac's because supporting two architectures is difficult.
Can someone explain why is compiling computationally intensive? I thought you just write the code and you have an app?
You may already have found your answer but just in case, look up the difference between script based languages and compiled languages. Compiled languages need a compiler to convert code to machine code in order to run. Scripts don't need to be compiled because an interpreter software has already been compiled and it essentially "interprets" your script code on the fly.
Does app development benefit from GPU or neural engine cores at all? Or is it still just pure CPU grunt?
Gpu is needed for the simulator and emulator, not just pure cpu. Base m1 max is good enough I guess.
Apple goofed by not making it fit in a 2u form factor. They could have designed it to fit.
I guess the extra cooling space is needed
May I have some reference for the app require 24Gb of RAM to build (03:23)?? I would like to know more detail of what kind of system have big build. Thanks
Unfortunately I cannot share that information. Very large Android/iOS apps, supported by 40 or so engineers on both platforms.
@@D3VTec I see, thanks for your answer. Great video 👍🏻
@@D3VTec, what kind of NDA did you sign you cannot say the app name?
Feel a waste to use Mac Studio in this case as Studio shines on the GPU side.
4 m1 ultras clustered together. Meanwhile Me who doesnt even have an m1 chip 😂
I HATE when people put KRKs on their side... thanks for letting me know you know nothing about sound. Also the studios I've worked with (Warner, Universal, Sony, MGM, Legendary, etc...) all use the rack-mounted Mac Pro 7,1 (and multiple of them) for our post-production servers.
That is quite the experience you have! Nice sounding career. I admit I don't know much about music, but I have had these for ten years and they've worked well enough for me. I ask that the next comment you leave online, wherever it is be a little nicer. It's hard putting yourself out there. Have a good day Nic.
I have to think Apple has been using M1 Ultra to train models in their own cloud. Why pay NVidia millions for DGX or A100? Makes more sense to use their own M1 Ultra and save millions in electricity.
If you ever trained models, you’ll know that M1s are 10x slower than a consumer Nvidia GPU. A datacenter grade GPU smokes any M1 chip.
@@stanchan I have several Nvidia video cards. As long as the dataset fits in memory, training on GPU is WAY faster than CPU. But I have benchmarked my M1Max 24 gpu against 2060 6G card. Once the dataset is bigger than 5.5G, 2060 is considerably slower due to IO and latency. For me, the cost of A100 is 12K is too much. On 32G, my M1Max 24 GPU can train createML with COCO dataset in reasonable amount of time. On my 2060 Ryzen 3700x training with 330K images is not reasonable.
I have to think apple has even bigger chips they are deploying in their own datacenters for training models. That fits with apple's philosophy and they have the EE talent.
Can anyone advise if you can run multiple vm's on one M2 Mac studio each with a dedicated line? Would it be possible to run 20 VM's at the same time? Thanks.