Were you SHOCKED by the massive power throttling we found in our torture test? Comment below! NEW M1 Ultra Chip T-Shirt ➡ max-tech-store.creator-spring.com/listing/apple-m1-ultra-soc-package Use Promo Code "M1Ultra" for 20% OFF ANY & ALL of our Merch ➡ max-tech-store.creator-spring.com Like these unique Mac Studio Wallpapers? Download them here ➡ bit.ly/2WNc6Qw Best deals on M1 Macs on Amazon ⬇️ Apple Mac Studio with M1 Ultra ➡ geni.us/cgkl Apple 27" Studio Display ➡ geni.us/GeA3te6 M1 MacBook Air ($850 SALE) ➡ geni.us/1mJ41T NEW 16" MacBook Pro 2021 ➡ geni.us/OuBRWv
I think you are missing the simplest explanation to the diminishing returns: poor software optimisation. It seems quite obvious to me that this is the main problem.
Surely it must be a software or bios issue that's not delivering the right power? Their charts show that they thought these thing should be performing better. Perhaps MaxTech should email Apple with the results and ask?
😏Dont buy mac studio Ultra, remember repair cost of this machine is huge, even certain points this machine some time become unrecoverable. build a pc in very less price: i9 , r9 , TR , system + very good GPU In this pc u can get FREEDOM install any os as u want windows, linux Even MacOS, but don't use openCore for hackintosh, use: type one hypervisor/bare-metal. VMware ESXi or proxmox. You get updates of future versions macOS without any break same as real mac. even iMessage will work.
Yes, it really does seem like the cooling was designed to handle higher power than the M1-Ultra CPU + GPU is capable of drawing. Reminds me of the thin 2016 MacBook Pros, which were opposite in that they were clearly designed for cooler 10nm CPUs that Intel was unable to deliver! If Apple is able to update the M1-Ultra's firmware to deliver better performance, I think it would be quite shocking that it wasn't available at release. Loads of people will have taken home the messages from reviews like this one, and might not notice updates once the Ultra is no longer hot news.
@@wisdomyaw03 Part of the problem may be Apple's desire for secrecy about their new products. Apparently the menu bar widget team didn't know about the MacBook Pro's notch before launch.
@@wisdomyaw03 This depends on if the lack of extra performance is due to app or OS optimization. Allegedly the FCP unreleased update blows the socks off of the current version. Are these apps lmiiting themselves and not using all the resources available? It'll be interesting to do a 1 or 2 month later video to see if app updates make a difference.
Max is enough for me but decided to go with the ultra. If I add the options I want, the max price will be close to 3k, so it would be better to spend a little more and go for the Ultra
Yeah, I had the base Studio on order but decided to cancel and wait for the performance reports. Even with the disappointing results in some areas, I'm now thinking the Ultra will have more longevity if Apple can tweak performance and applications are re-written to take advantage of the new ultra chip so the extra cost is likely worth it -- but just the base Ultra.
@@garfieldirwin I think the max will live a long and healthy life. But i get what you mean i also got the 14 inch max for a little bit more headroom later but there it was only 500. in this case you could buy two max for one ultra.
The issue with limiting power may not be a flaw with the chip die. I'm thinking that it may be much simpler, the power supply cannot deliver a sustained level of high wattage output without cooling. When you dissected that MAC Studio I noted that power supply components looked kind of cheesy and there was no direct cooling mechanism/cooling fan for it. It seems to be relying on convection and some movement of air through the case. It probably was a design oversight and when it was discovered too much of the Studio's design was set in "stone". Most likely the engineering crew was so focused on cooling the M1 they overlooked the power supply cooling. The easiest way to "fix" it was to limit power draw via software/firmware. Since the M1 chips are so efficient and powerful, only a deep dissection of it would reveal some of the wonky things going on inside. As a system's engineer with over 40 years experience designing and overseeing complex electronic systems, these goat ropes happen far more often than you think.
Totally agree on the power supply being the culprit. The other item I'm thinking is it could be a thermal design flaw in the interposer connecting the two Max chips, perhaps too much heat if a ton of bandwidth comes in all at once, so they just decided to lock it down so it doesn't get there - ever. But in the end, I like the power supply hypothesis the best.
It's laughable to suggest the most wealthy company on earth would make such a stupid mistake. On the contrary, virtually everything they do is planned out to a ridiculous degree of precision these days. If the power supply is underperforming in these machines, it's underperforming by design.
@@captain_crunk I won’t argue that Apple has a pile of cash. And no doubt they plan the heck out of what they want to do. But realize that if Apple wants to keep their pile of cash, they (the various design/engineering teams, etc.) have to work to a schedule and budget. I don’t pretend to know how Apple organizes their design and development teams, but all of these teams regardless of industry type or product have one thing in common. They are run by people and people make mistakes. When mistakes are made, planned budgets and schedules can balloon quite quickly to fix the problem(s). So, it usually comes down to two options, fix it right or do a cheaper work around that usually compromises some of the original design or performance goals. You might be surprised how often the latter is chosen. But I see a bright side to this power supply hiccup. I just won’t go into it here as this post is going to be too long as it is. I kind of believe that Apple did not discover the issue until they were well into testing and making changes at that point would probably mess up the release date as the design was fairly set in stone and pre-production had already started or about to start. Apple has made some less than stellar design choices in the past, e.g., butterfly keyboard, and probably will in the future. My last point before I get off my soap box is Apple has become a victim and prisoner of its own success. Its shareholders expect them to remain as one of the most (if not the most) valuable companies in the world. Its customers keep wanting newer and better products ever more quickly. The iPhone 13 has only been out a few months and some folks are expecting an iPhone 14 announcement this June. Insanity! The hyperbole of pundits is not helping either. Apple has positioned itself such that delaying the expected (real or otherwise) new products has nearly become a non-starter. Not introducing either a new M1 Mac Mini or as it turned out the Studio, would have enraged many of its customer base. Delaying release for a power supply heating issue was a non-starter for public relations, marketing and production reasons.
@@lazerbeam3928 I live and work in Silicon Valley. I have several friends who work at Apple. Everything they do is planned and accounted for, including contingency plans in the event their arrogance (in my opinion) is a bit over the top. Obviously they are not 100% perfect, but in the context of what we're talking about here specifically, it's just silly to suggest they didn't check if the power supply worked properly. Maybe I'm not fully understanding what you're saying, but the most basic things like appropriately powering a device are no longer the type of mistakes companies like Apple make. To be clear, of course there are manufacturing defects that may arise, but if everything is built to spec the device will draw the amount of power provided by the spec. And Apple knows exactly how much power their devices require, with a truly obscene degree of accuracy. So why is this happening then? Only Apple knows for certain. It's entirely possible that at some point in the future, Apple will release a new firmware update that miraculously "unlocks" more power for these devices. It's not the first time they have deliberately kneecapped performance. Anyway, at this point in time it's difficult to imagine even copycat companies manufacturing knock-off products would make such simple mistakes. The industry as a whole has evolved beyond such things. Yes, people are people, and yes, people do make mistakes. That said, this is no mistake. Whatever reason Apple has for doing this, this is exactly what Apple designed.
I have ordered the M1 Max, 32 GPU, 1TB, 32GB RAM. Basically the base model with extra storage and extra gpu cores because its a relatively cheap upgrade.
I think the disparity in wattage is a symptom and not the cause. The bottleneck is elsewhere and since the GPUs/CPUs aren’t being feed they just chill. The perfect inverse frequency in that last torture test suggests that, the wattage was lower probably because the frequency was lower and not the other way around. The question is “what is keeping the cores from being feed?” and after that “can a macOS update fix that?”.
To me, it reeks of a memory bandwidth bottleneck. Given a fixed amount of memory bandwidth, adding cores is just going to spread that bandwidth thinner. The gpus look like they're waiting on memory.
The comparison is perfect! Thank you for your effort in informing us. I want to add a comment from a 3D professional perspective. Blender is not an industry-standard tool yet. With that said, Blender benchmarks do not mean much to many 3D professionals. I strongly believe that C4D+Redshift, which has a 14-day trial available, benchmarks would be way more informative for creatives.
Apart from artists involved in motion graphics, C4D isn't a mainstream 3D package. A render with Karma engine on houdini or Arnold render time with Maya or 3DS Max would be more realistic.
A pity the 32-core Max is left out of this test. That one may actually be the best bang for your buck: 50% more GPU for only $200 extra. Would have been worth mentioning at least.
If the utilization is around 100%, it does not matter what the wattage is. More wattage does not mean more utilization, because some instructions are harder on the CPU/GPU than others. You can see this behavior on Intel as well with programs that use AVX instruction sets for example, like Cinebench R20.
The extra memory on the Ultra 128GB is likely only making a difference during the linking phase of the Xcode compile, so that's why the difference is so small. For certain projects which have massive linking phases that extra RAM is crucial, but if you don't have a big link then thread count matters much more, as long as you have enough memory to support it, which even 32GB is.
Question… getting the M1 Max model I wonder if it’s worth it getting 64 gig over the 32 gig or paying extra for the extra GPU cores to 32 cores? How much of a boost would you get.
CONCLUSION: Unless you work for Pixar (24,000+ multicore perfomance), EMI Music (300+ tracks in Logic Pro!) or NASA's Jet Propulsion Lab, for the rest of us mere mortals the M1 Max Mac Studio @ $2,000 is the steal of the century! Maybe upgrade up to 1TB SSD.
3:32 "So Apple is ACTUALLY giving more power to the Ultras to try to maintain that scaling. But the very interesting thing is that Cinebench is ACTUALLY not utilising these chips very well because I tested the wattage in a different app and it was ACTUALLY quite a bit higher." Can't your editor cut out all these "ACTUALLY"s? They're maddening to listen to! Thanks.
If by chance the Ultra GPUs need to work in some sort of sync with the media engines, it kinda makes sense that the yare being throttled. It seems that Apple has designed these to, sort of equally split the gpu load but if only half the media engines are being utilised, the cores are being underclocked because they can't push data to all the engines at the same time, so the gpu gets delegated, but the half the engines are probably getting maxed, and the rest are just idle.
Great video and comparison! But to compare the power between the Mac Studio Max and Ultra don't you think the best way must be both with the same amount of RAM and the same amount of SSD storage? That will be a nice video. Thanks Max!
It's interesting, I've sent a couple Davinci Resolve test projects to someone with a 64-core Ultra and his times and fps playback was exactly double my 32-core M1 Max. Seems like it scales as it should with proper software code?
Is it possible they’re trying to avoid this machine stepping on the Mac Pro’s toes? If they properly fed power to the 64 core 128 gig model, would it trounce the Mac Pro?
Hey man, cool videos. Just one thing regarding the music production / logic performance test. The amount of tracks means very little on its own. You have to tell us at what sample rate, bit depth, and buffer size the session is running. Also, what kind of processing are those tracks doing? what kind of plugins do they have inserted? how many plugins per track? etc etc etc. Without that information the amount of tracks is meaningless.
Yeah, I am guessing he means that many tracks all recording at the same time? I went to the apple store to test out the m1 max Mac Studio, and it could handle 10,000 tracks :)
Regarding “more power for scaling”, keep in mind that there’s probably power needed for the interconnect between the processors as well, rather than just doubling the memory channels/CPU count.
@@ifiwantyoutofeel There's nothing to "solve" there unless the power used for that (which is a necessary loss) is unusually high. Not having perfect scaling isn't unexpected in the slightest. that's how it's always been & probably always will be. getting near to that perfect scaling is pretty impressive though.
I knew the 64 core would not be worth the 1k however I still got the 128gb ram as I plan on using after effects and blender. Hopefully the gap will improve when these machines and software are more optimized. Either way the ultra will be noticeable better then my work horse M1 Mac mini
Hey guys, thanks for this. I just realised, you posted the Affinity Combined scores which won’t show GPU scaling because it’s across CPU+GPU. The GPU raster score is the one which not only shows decent scaling but also the M1 Ultra kicking the RTX3090’s but at 53K vs 21K.
I'm feeling a bit happier about my decision with the 32 core m1 max with 2tb SSD and 64gb ram. Price was $3200. If I wanted the Ultra with 48 cores and similar storage space it would have been an extra $1200. With my basic photo and video editing this machine should more than enough. Worse case, I sell my mac studio for a newer version with better specs next year if this for some reason isn't enough for my task.
Fabulous video - the one I had been waiting for! Thank you for all your hard work on this. Two question though, please: 1) Logic tracks - were these just simple audio files, or did any of them have plug-in effects or instruments on them? 2) Do you think going to 4TB of SSD would be good for audio? I hear that the 4TB SSD has a faster transfer rate than the smaller ones, which might increase track count? Also many virtual instruments and sample libraries require their assets to be on the main drive, rather than an external SSD. I will also get external SSD for Time machine and general project archiving though. Both these questions are based on me getting the 48 core Ultra but keeping the RAM at 64GB. I will be mostly running a lot of processor-hungry plug-in synths on it, but would also like to dabble with Blender and Unreal Engine for non time-critical projects. Any advice greatly appreciated please.
1. These had simple instruments on them. 2. I don't think it'll make any difference whatsoever. Even the base model drives are already incredibly quick. I would keep it to 64GB of RAM for sure. I don't think the extra cost is worth 128GB. The cores just can't ramp up fast enough to make a big difference.
Thanks for this kind of info. It’s why I’m putting my money towards a XDR display to use with my 14” M1 Max MBP instead of another computer (and matching display) that I can live without. One suggestion for the future, although you’ve covered it in the past, it would be nice to see an overall comparison of the entire M1 Pro/Max/Ultra family in one video, or at least have tiles to refer to your past videos where you performed the same tests with, for example, the 14 and 16” M1 Pro/Max MBPs. Thanks again for your excellent content!
So basically for almost everyone, the base model is the one to get! Glad that is what I picked, thus far I'm blown away with the performance and couldn't be happier! That being said the amount of sheer what you guys have done and provided to all of us is amazing. I know there are people who can take advantage of the Ultra and your work is providing them with a LOT of relevant information when configurating their systems! Big thanks for all you guys are doing, hopefully you will hit that 1M subs very soon!
I appreciate all the testing being done, but I think its just a little too early to be testing for what to buy right now. Sounds like most of the apps aren't optimized for the new CPU's and the ones that are obviously show it. I hope to see another round of testing once everything is optimized.
I think a problem here is that you care comparing temps from Intel x86 chips and TSMC ARM chips. I don't know for sure, but these sort of chips may not have as high tolerances as the Intel chips. Since the TSMC chips are designed initially for portable devices, they may not be designed for 100c temps for long periods. Once you get those higher temps, failure rates may increase substantially, decreasing life of the product, and increasing Apple Care requests (which would increase costs for Apple). This is just a guess on my part but might make sense.
Thank you for checking this, the M1 Max Spec vs M1 Ultra price is huge and getting these comparisons helps in our buyung decision. Keep it up and hope you'll get that 1M subs soon
After watching that video I am definitely sure that I made the right decision with my 1 TB base model. Thank you for the great work and this very informative video! 🙏
For a video editor what would you suggest, where I should have to spend a little more if I buy the base model M1 Max Mac Studio : 1. Upgrade RAM to 64gb from 32gb 2. GPU upgrade to 32 core from 24 core 3. SSD upgrade to 1tb
Tough call. If you layer a bunch of effects and titles and can handle having only 512GB SSD (if you use external drives for example) then upgrade to 32 Core. If you don't do a bunch of GPU-heavy effects then upgrade your SSD. RAM would be the last thing I would upgrade.
Another good video. With every previous Apple Mac hardware release I’ve never got it quite right. This time I started with the M1 Max Studio and went to up the RAM and SSD to find it was not much more to go to the base Ultra. The Mac Studio base Ultra is the best Apple purchase I’ve made in more than a decade. Does it perfectly scale? No! But for a cashed up hobbyist photographer like me with a 100 megapixel stills camera (GFX100) and shooting the odd family 8K video on an R5, the base “binned” Ultra is essentially perfect.
How about doing a comparison on the M1 Max with the 24 core verses the 32 core... is it worth the extra $200... and how about the 32GB verses 64GB is that worth the $400... really enjoy the information but I bet there are others like me that can only afford the M1 Max and want to get the best bang for the buck... thanks
Ordered Mac Studio Ultra 2TB for my Recording studio running Protools. $4,706 and change. Hope this does better than that M1 I had. Love your channel Max..Keep it coming.
It s clear what s the problem. They didnt cool the power supply, that cannot supply sustained power to the gpu. In some way it seem the two teams didnt communicate and focused only on cooling the chips. Not much can be done now
Shocking results, especially video editors who Apple are firmly targeting with this machine. Can this really be a design flaw with the MAX chips....? Now actually makes me glad my Studio Ultra is not due until June and Apple issues an OS update to address this or might just cancel and pick up the Max version.
Are you sure you could not hear the fans at all? I got my Mac Studio (M1 Max) yesterday and I was shocked because of the fan noise. I wanted to replace an iMac 5K from 2017 and the Mac Studio is much much louder. Even standing 3 meters away from my Mac Studio I can clearly hear the fans going in idle. It is even louder than my 15" MacBook Pro 2017 during idle. So when you say it is "completely silent" I can't believe that or my Mac Studio has some kind of malfunction.
I’m thinking you are right - it’s largely a silicon issue. Or, it could be that they are leaving room for the the M2 to show performance gains over the M1….. There does seem to be a software issue, but something else is going on that causes them to put the brakes on performance.
So basically the basic Mac Studio will be more than sufficient for (45mp) photo editing? And an upgrade to the 32-core GPU and 64 GB RAM isn't worth it?
I'm ready to place my order, but the real question is if the software will catch up with the hardware? Is it worth investing the extra $1800 now for the extra memory and GPU cores with the understanding that this machine will last me for 5 years? I breathed new life into my 2017 iMac Pro by adding a very powerful eGPU to it, but that's no longer a possibility with Apple silicon. Thoughts?
I am on day ten of sitting on Apple's website trying to figure out what to order. I'm coming from a 2019 5K iMac with an i9, 580X, and 64GB. No issues with the machine other than lack of external displays. I can order and pick up a base M1 Max Studio today, but no Studio Displays, so I'd have to go with the LG's, or something else. The 512GB might hurt a bit too, but with all the additional USB-C and Thunderbolt ports, externals should cover that. Then there's the MacBook Pro for $3,500; also available for pick up today, which should do the trick with four displays; three Thunderbolt, one HDMI, but that would consume all the Thunderbolt ports, leaving me no room for external SSD. That's not the end-of-the-world as I only need the external displays when I trade so I could disconnect one when I had to connect an external SSD, which hold my FCP libraries. I guess the solution to that would be the HyperDrive 4K, connect the three displays via HDMI/DP, regain three USB-C on the HyperDrive, but give up the fourth display via HDMI?
I think you are wrong that Apple ‘is allowing more power’ for the ultra. It’s simple math - each chip needs about 34w. There are two in the Ultra - which is why it takes 68w ?
Just wanted to thank the Max Tech team for this amazing coverage. SO appreciate that you include Logic as you did for the MacBook Pro. You guys just keep delivering the goods! There’s a reason why you’re moving towards the 1 million, and your fans will celebrate with and for you when you get there! 👍🏾😊
Hi, by standard render time is a stat to refer for performance, but what creative user like us would like to see the smoothness when using it. Example like working with large/heavy layers file in ps, when masking high dpi file with special brush, dealing with multi-layering smart object. Or does it smooth to drag and play when editing in AE, esp when preview time. Thanks,
23:26 You are talking about the frequencies of the GPU, the wattage but where are the temps and the fan speed ? It’s maybe that the chip is heating to much… 24:21 Never mind I really hope Apple gonna fix this
Great video guys!!! Again I think we are seeing why we are getting the Mac mini pro last. I think many video editors might find it the best bang for the buck. So Apple released the Mac Studio first to get most sales before releasing the higher end Mac mini which should be an easy machine to ship. It will be nice to see the Mac mini Pro compared to the base model Mac Studio Max. Overall as impressive as the Ultra looks on paper so far it looks like the most disappointing chip Apple has released to date, not that it is a horrible chip just underwhelming. Hopefully Apple can patch the OS or firmware to push this chip to better performance. The fact that this hasn’t happened already makes me wonder if it will.
3:30 - that's a very impressive linear performance scaling going from 10 to 20 CPU cores. Also doubling the performance only results in a 5% additional power draw in excess of an idealized 1:1 power increase / performance scenario. 22:50 - that's crazy and severely limits the usefulness of the M1 Ultra in both 48 and 64 GPU core configurations.
So could you be a little more clear on what you mean by tracks in the music production test? So what I mean is were those just straight audio tracks, or did they have plugins on them? Also what was the Khz of the session? 44.1k, 48k, 96k etc, and what Sample Rate were you running it at? 64, 128, 256 etc?
I've been using a 2018 Mac Mini i7 32GB RAM with an eGPU. My Mac Studio arrives next month. Anybody interested in a comparison video with benchmarks etc before I sell my Mini?
@@DanielToman I figure a lot of audio/video people are making that jump now that the Studio is out. M1 Mini just didn't have enough ports/only 16GB RAM.
It’ll be interesting to rerun these tests after 6ish months of SW upgrades. Some of the throttling may go away. It’s perhaps not impossible that the interconnect between the two M1 Max dies could be maxing out: GPUs generally operate memory-bound, and perhaps there’s enough GPU memory activity having to cross that interconnect that they kill off all available bandwidth.
Excellent information! I have been watching all of your videos, because my mid 2010 Mac Pro has become asthmatic, and recording with microphones is a problem. I will probably go for the Ultra 48 core 64 gigas, and am sure it will serve well for several years. Thanks for your work!!!
Hello - thank you for your in-depth dives into all things mac. Have you, and I've inadvertently missed it, focused less on graphics and more on audio? I'm designing a recording studio (with a softer focus on video/film editing) and need help understanding how to best finesse my new mac(s) (laptop and desktop) and make the best investment for the longterm. can you help?
I was thinking about upgrading to the Max/Ultra but after this I'm just keeping my M1 mini and when they release M2 Max/Ultra I'll see if these problems have been fixed, dissapointing...
The number of power and ground bumps on the M1 Ultra’s are the same and share the same power grid on chip no matter how many good GPU’s are available so power per GPU result is not surprising if it is a bump or power grid limitation
I think you're misinterpreting a few things when you're looking at the power differences. 1: Apple is not pumping more power in to the CPU, the ultra is literally 2 M1 Max chip joined together with an interconnect. So its just normal full power per chip, plus interconnect and more memory modules. 2: Apple's GPU slide is clearly entire package power vs just a discrete GPU. The angle is that their whole Chip is using less than just the single component of a PC. You are gaining memory bandwidth from the increased number of memory modules, where as the raw capacity is basically meaningless unless the workloads are running over the capacity. Either Apple is prioritizing silence over performance, or the they didn't build much extra headroom in to the power supply. There could also be something to their choice in moving the storage controller off of the storage module. That is one component that typically has a upper limit of 70C, now sharing thermal capacity with the M1 SOC. One of the typical drawbacks to Apple designing their devices to not be opened is the dust collection over time. I can see that being largely mitigated from not running the fans basically at all.
Actually, the power metrics I measured are counting the wattage going to only the CPU cores, not the rest of the die or the rest of the package. Definitely not the memory. Only CPU cores are counted.
@@MaxTechOfficial Okay. The shot of the monitoring software does show a package power of 106watts at one point. In reference to Apple slide, that seems most likely what Apple was referencing as opposed to giving the GPU only 100+ watts.
If you have hour long 8K Videos use 2 or 3 base model Mac Studios…Apple should excus for the dimishing returns and sell more base models and optimize the wattage usage of the topmodel
At this point, I'm guessing they designed this thing for future iterations of Apple Silicon chips. If you remember; one thing Phil Schiller said about the 2013 Mac Pro was that they backed themselves into a thermal corner. It looks like they have plenty of room to work with now.
Both variants cost the same. Which model would you recommend for video editing? 1. Mac Studio 2022 M1 Max/32/2 TB 10C CPU 32C GPU 2. Mac Studio 2022 M1 Max/64/1 TB 10C CPU 32C GPU SSD over RAM upgrade? or would you spend more for this variant? 3. Mac Studio 2022 M1 Max/64/2 TB 10C CPU 32C GPU
Your graphic is showing the $5,799 version and not the $8,000 version? I'm assuming you are testing the $5,799 one. I love these kind of videos that show the actual performance and not just thinking you get better performance!
IDK all the specific workings of those GPU benchmarks, but could it be you’re “drag racing pickup trucks” rather than seeing how much they can tow? Maybe these benchmarks are too low res to show a significant difference. I’ve seen some crazy difference with GPU’s with more RAM when pushing tons of pixels (LED, etc. video walls), yet very little FPS gains at low res. I suspect if you pushed a “canvas” of 4x 5K displays, for example, the 64c would get better fps than the 24c & 48c pushing the same number of pixels. While their 0-60 could be similar, the “pickup truck” with a twin turbo V12 diesel will be able to tow more weight, and maintain speed, up a hill vs a NA V8.
What do you think these top-end studios with maxed out configs are really going to be for? Will we see optimizations dealing with this extra hardware? All I can think of besides niche cases is servers.
It almost feels like some driver updates are missing. Hopefully Apple can still improve this. It'll be worth rechecking it in a few months. I have a 5950X for software development (I mostly develop on Windows, for Windows and Linux).
I wish you included the 16” mbp w/32 core in this test series to see if the sweet spot for the lower cost studio pros is actually the 32 core rather than the 24 core.
I am beginning to get suspicious about software optimization. I do recall that, when I was using my 2009 Mac Pro Cheese Grater in 2009, there was a lot of software that only saw a single core. In about six months, there was software that could take advantage of a four-core Intel CPU. It wasn’t until almost 2012 that software really started taking advantage of 8 cores in the two processors. You are seeing a seriously diminishing return when you move to the Apple chip with double the number of CPUs and GPUs of the Max. I would like to “blame” software optimization as not all software will spawn enough threads to take advantage of as many cores as are available in the Ultra (remember, the OS is what will send threads to different cores) and the operating system, itself may not be sufficiently optimized for the Ultra, making threads wait even though there are CPU cores waiting. We did see this when Apple released the dual hex-core Intel systems in 2012 and, I think that the Trashcan Mac Pro also did not scale up properly. So, the fact that you are forced to use Apple’s current operating system and also the current software may be the real limitation in the Ultra system.
Am I the only one that find the Mac Studio way too loud? My 48-Core Ultra has two forms of noise, the ground noise (too loud in a quiet room) AND a high frequency chirr noise. Sadly sending this back to apple and staying with my silent MacBook Pro Max :(
"photo editing" is only a large panorama in PS? I know you can only show so much on these videos and I appreciate the content very much but I use photoshop with hundreds of layers and many smart objects and tons of blends and effects etc. I am so worried to buy this machine - every time I upgrade to a new mac there's never a noticeable improvement in PS. For years PS has been slow on new macs because they don;t use the multicore as much and require tons of RAM. Love to see a test with a massively layered, 300 dpi, large canvas photoshop file and see if it's any better than say the m1 imac, current mac pro intel or even an older Xeon 10 core iMac Pro from 2017 (which is what I have).
Is there a logic test results table to compare with other Macs available somewhere? Results for an M1 Mac Mini with 16GB, an i7 2018 Mac Mini with 64GB and a 5,1 series 3.46Ghz 12-core Mac Pro would be very helpful in planning what Mac is worth getting. Particularly as it's £800 more for an M1 Max Studio with 32GB and 512GB SSD than an M1 Mac Mini with 16GB, a 512GB SSD and the ethernet upgraded to 10xGigabit.
To the MaxTech guys, is it possible those GFX Bench and similar GPU tests to do some more extreme resolutions? For example, instead of 1440p to go 2880p on the three models? It’s just to take out of the way the possibility of too fast frame rates affecting it. It is my understanding that when a frame finishes a lot of stuff has to be done (in classic architectures at least), like maybe it needs to sync the cores to finish it, pack, collect, copy, transfer, flush buffers, etc and these things are usually not that thread friendly as it is ‘a whole page’ if you will. So maybe, if the M1 Max is pushed hard to run at say 30fps, maybe the ‘frames per core’ might be different. Just curious about it.
M1 Ultra 48 cores. The other two models are not worth it because the M1 Max is easily outperformed by a MacBook Pro and the advantages of the Ultra 64 are not worth the extra 1000 bucks.
@@aram.v I am thinking about going 48c/128gb/2TB. I have it in my cart, ready to be purchased. I just gotta make one more payment to my credit card, so I will have enough limit to plug the cost of my Mac Studio. I wish you enjoy yours and get the most of it. Congratulations aramv!!! 👍🏻👍🏻
The question that remains unanswered is why has apple restructure the performance of the Ultra machine (and arguably all the lowest model as well to some extent as it performs little better than the 16 inch MBP) ? Is it that OSX needs to be upgraded to take advantage on the duo core SoC chip (along with metal and obviously non-apple software) and we can (eventually) expect much better performance or is there some hidden limitation(s) in the physical hardware derived from doubling to SoC to two Max chips ?
Nice video! I either missed it or it wasn't mentioned. What buffer size was used on the logic pro test?? Very interested to know. Thank you for the vid
Were you SHOCKED by the massive power throttling we found in our torture test? Comment below!
NEW M1 Ultra Chip T-Shirt ➡ max-tech-store.creator-spring.com/listing/apple-m1-ultra-soc-package
Use Promo Code "M1Ultra" for 20% OFF ANY & ALL of our Merch ➡ max-tech-store.creator-spring.com
Like these unique Mac Studio Wallpapers? Download them here ➡ bit.ly/2WNc6Qw
Best deals on M1 Macs on Amazon ⬇️
Apple Mac Studio with M1 Ultra ➡ geni.us/cgkl
Apple 27" Studio Display ➡ geni.us/GeA3te6
M1 MacBook Air ($850 SALE) ➡ geni.us/1mJ41T
NEW 16" MacBook Pro 2021 ➡ geni.us/OuBRWv
Yep!.
I think you are missing the simplest explanation to the diminishing returns: poor software optimisation. It seems quite obvious to me that this is the main problem.
Surely it must be a software or bios issue that's not delivering the right power? Their charts show that they thought these thing should be performing better. Perhaps MaxTech should email Apple with the results and ask?
Can you do a chrome tabs or multiple virtual os test to check the three different ram specs?
😏Dont buy mac studio Ultra, remember repair cost of this machine is huge, even certain points this machine some time become unrecoverable.
build a pc in very less price: i9 , r9 , TR , system + very good GPU
In this pc u can get FREEDOM install any os as u want windows, linux
Even MacOS, but don't use openCore for hackintosh, use: type one hypervisor/bare-metal. VMware ESXi or proxmox. You get updates of future versions macOS without any break same as real mac. even iMessage will work.
Max, I need to nominate you and Vadim for a public service award for this video!!! Thank you guys!
Thank you so much for the support! :)
Totally agree. The clip is jam packed with technical info. Truly a buyer’s guide.
I am wondering if Apple will push out some software updates to tune the performance. Very thorough and informative video guys!
Thanks for the reply and checking out the video! :)
Yes, it really does seem like the cooling was designed to handle higher power than the M1-Ultra CPU + GPU is capable of drawing. Reminds me of the thin 2016 MacBook Pros, which were opposite in that they were clearly designed for cooler 10nm CPUs that Intel was unable to deliver!
If Apple is able to update the M1-Ultra's firmware to deliver better performance, I think it would be quite shocking that it wasn't available at release. Loads of people will have taken home the messages from reviews like this one, and might not notice updates once the Ultra is no longer hot news.
I was thinking the same thing…
@@wisdomyaw03 Part of the problem may be Apple's desire for secrecy about their new products. Apparently the menu bar widget team didn't know about the MacBook Pro's notch before launch.
@@wisdomyaw03 This depends on if the lack of extra performance is due to app or OS optimization. Allegedly the FCP unreleased update blows the socks off of the current version. Are these apps lmiiting themselves and not using all the resources available? It'll be interesting to do a 1 or 2 month later video to see if app updates make a difference.
i changed my order and went ofr the 48core with 64memory. I cancelled my 128memory one. Saved myself £2K. THANK GOD. THANKS GUYS
Max is enough for me but decided to go with the ultra. If I add the options I want, the max price will be close to 3k, so it would be better to spend a little more and go for the Ultra
Ultra is is w waste of money max is more then enough
Yeah, I had the base Studio on order but decided to cancel and wait for the performance reports. Even with the disappointing results in some areas, I'm now thinking the Ultra will have more longevity if Apple can tweak performance and applications are re-written to take advantage of the new ultra chip so the extra cost is likely worth it -- but just the base Ultra.
@@garfieldirwin I think the max will live a long and healthy life. But i get what you mean i also got the 14 inch max for a little bit more headroom later but there it was only 500. in this case you could buy two max for one ultra.
The issue with limiting power may not be a flaw with the chip die. I'm thinking that it may be much simpler, the power supply cannot deliver a sustained level of high wattage output without cooling. When you dissected that MAC Studio I noted that power supply components looked kind of cheesy and there was no direct cooling mechanism/cooling fan for it. It seems to be relying on convection and some movement of air through the case. It probably was a design oversight and when it was discovered too much of the Studio's design was set in "stone". Most likely the engineering crew was so focused on cooling the M1 they overlooked the power supply cooling. The easiest way to "fix" it was to limit power draw via software/firmware. Since the M1 chips are so efficient and powerful, only a deep dissection of it would reveal some of the wonky things going on inside. As a system's engineer with over 40 years experience designing and overseeing complex electronic systems, these goat ropes happen far more often than you think.
Totally agree on the power supply being the culprit. The other item I'm thinking is it could be a thermal design flaw in the interposer connecting the two Max chips, perhaps too much heat if a ton of bandwidth comes in all at once, so they just decided to lock it down so it doesn't get there - ever. But in the end, I like the power supply hypothesis the best.
It's laughable to suggest the most wealthy company on earth would make such a stupid mistake. On the contrary, virtually everything they do is planned out to a ridiculous degree of precision these days. If the power supply is underperforming in these machines, it's underperforming by design.
@@captain_crunk I won’t argue that Apple has a pile of cash. And no doubt they plan the heck out of what they want to do. But realize that if Apple wants to keep their pile of cash, they (the various design/engineering teams, etc.) have to work to a schedule and budget. I don’t pretend to know how Apple organizes their design and development teams, but all of these teams regardless of industry type or product have one thing in common. They are run by people and people make mistakes. When mistakes are made, planned budgets and schedules can balloon quite quickly to fix the problem(s). So, it usually comes down to two options, fix it right or do a cheaper work around that usually compromises some of the original design or performance goals. You might be surprised how often the latter is chosen. But I see a bright side to this power supply hiccup. I just won’t go into it here as this post is going to be too long as it is. I kind of believe that Apple did not discover the issue until they were well into testing and making changes at that point would probably mess up the release date as the design was fairly set in stone and pre-production had already started or about to start. Apple has made some less than stellar design choices in the past, e.g., butterfly keyboard, and probably will in the future. My last point before I get off my soap box is Apple has become a victim and prisoner of its own success. Its shareholders expect them to remain as one of the most (if not the most) valuable companies in the world. Its customers keep wanting newer and better products ever more quickly. The iPhone 13 has only been out a few months and some folks are expecting an iPhone 14 announcement this June. Insanity! The hyperbole of pundits is not helping either. Apple has positioned itself such that delaying the expected (real or otherwise) new products has nearly become a non-starter. Not introducing either a new M1 Mac Mini or as it turned out the Studio, would have enraged many of its customer base. Delaying release for a power supply heating issue was a non-starter for public relations, marketing and production reasons.
@@lazerbeam3928 I live and work in Silicon Valley. I have several friends who work at Apple. Everything they do is planned and accounted for, including contingency plans in the event their arrogance (in my opinion) is a bit over the top. Obviously they are not 100% perfect, but in the context of what we're talking about here specifically, it's just silly to suggest they didn't check if the power supply worked properly. Maybe I'm not fully understanding what you're saying, but the most basic things like appropriately powering a device are no longer the type of mistakes companies like Apple make. To be clear, of course there are manufacturing defects that may arise, but if everything is built to spec the device will draw the amount of power provided by the spec. And Apple knows exactly how much power their devices require, with a truly obscene degree of accuracy. So why is this happening then? Only Apple knows for certain. It's entirely possible that at some point in the future, Apple will release a new firmware update that miraculously "unlocks" more power for these devices. It's not the first time they have deliberately kneecapped performance.
Anyway, at this point in time it's difficult to imagine even copycat companies manufacturing knock-off products would make such simple mistakes. The industry as a whole has evolved beyond such things. Yes, people are people, and yes, people do make mistakes. That said, this is no mistake. Whatever reason Apple has for doing this, this is exactly what Apple designed.
I have ordered the M1 Max, 32 GPU, 1TB, 32GB RAM. Basically the base model with extra storage and extra gpu cores because its a relatively cheap upgrade.
It’s the best option because you get more performance for 300$ not 1000$ with the ultra
Is 32gb of ram good for video editing?
@@northparkfilms3045 yes
I think the disparity in wattage is a symptom and not the cause. The bottleneck is elsewhere and since the GPUs/CPUs aren’t being feed they just chill. The perfect inverse frequency in that last torture test suggests that, the wattage was lower probably because the frequency was lower and not the other way around. The question is “what is keeping the cores from being feed?” and after that “can a macOS update fix that?”.
Thanks Fernando
To me, it reeks of a memory bandwidth bottleneck. Given a fixed amount of memory bandwidth, adding cores is just going to spread that bandwidth thinner.
The gpus look like they're waiting on memory.
Would be great to have the 32c M1 Max as well, as that is only 200 bucks more than the 24c version, and likely an much better value upgrade
That’s the one I bought. I’m setting it up now actually.
f you have the money upgrade. If not, the 24 core is enough
@@mattspalace i think i will get that one. how is it?
@@opaque_totc9275 Seems to run fine for me. 👍
I can’t wait to see your video on how the Mac Studio does with video editing.
The comparison is perfect! Thank you for your effort in informing us. I want to add a comment from a 3D professional perspective. Blender is not an industry-standard tool yet. With that said, Blender benchmarks do not mean much to many 3D professionals. I strongly believe that C4D+Redshift, which has a 14-day trial available, benchmarks would be way more informative for creatives.
Apart from artists involved in motion graphics, C4D isn't a mainstream 3D package. A render with Karma engine on houdini or Arnold render time with Maya or 3DS Max would be more realistic.
Thanks for this great comparison but I still need one more comparison between M1 max 32 vs 64 because I decided to buy Max not ultra
The model I shouldn’t have bought was the model I bought. Perfect. 😂
What model did you buy?
Same
A pity the 32-core Max is left out of this test. That one may actually be the best bang for your buck: 50% more GPU for only $200 extra. Would have been worth mentioning at least.
If the utilization is around 100%, it does not matter what the wattage is. More wattage does not mean more utilization, because some instructions are harder on the CPU/GPU than others. You can see this behavior on Intel as well with programs that use AVX instruction sets for example, like Cinebench R20.
The extra memory on the Ultra 128GB is likely only making a difference during the linking phase of the Xcode compile, so that's why the difference is so small. For certain projects which have massive linking phases that extra RAM is crucial, but if you don't have a big link then thread count matters much more, as long as you have enough memory to support it, which even 32GB is.
Question… getting the M1 Max model I wonder if it’s worth it getting 64 gig over the 32 gig or paying extra for the extra GPU cores to 32 cores? How much of a boost would you get.
Are those tests all from the devices you guys opened last week? Maybe that's why
That would be hilarious
We didn't touch the highest-end model at all. That's the one experiencing the most diminishing returns.
CONCLUSION: Unless you work for Pixar (24,000+ multicore perfomance), EMI Music (300+ tracks in Logic Pro!) or NASA's Jet Propulsion Lab, for the rest of us mere mortals the M1 Max Mac Studio @ $2,000 is the steal of the century! Maybe upgrade up to 1TB SSD.
3:32 "So Apple is ACTUALLY giving more power to the Ultras to try to maintain that scaling. But the very interesting thing is that Cinebench is ACTUALLY not utilising these chips very well because I tested the wattage in a different app and it was ACTUALLY quite a bit higher."
Can't your editor cut out all these "ACTUALLY"s? They're maddening to listen to! Thanks.
So far there is not a big different between the 64 and 128 GB models.
If by chance the Ultra GPUs need to work in some sort of sync with the media engines, it kinda makes sense that the yare being throttled. It seems that Apple has designed these to, sort of equally split the gpu load but if only half the media engines are being utilised, the cores are being underclocked because they can't push data to all the engines at the same time, so the gpu gets delegated, but the half the engines are probably getting maxed, and the rest are just idle.
Great video and comparison!
But to compare the power between the Mac Studio Max and Ultra don't you think the best way must be both with the same amount of RAM and the same amount of SSD storage? That will be a nice video.
Thanks Max!
So 48 core Ultra seems to be the best of the best. Unless Apple does a firmware update to optimize GPU cores. Unfortunately that's highly unlikely.
i think that is very likely to happen, in coming weeks..
It's interesting, I've sent a couple Davinci Resolve test projects to someone with a 64-core Ultra and his times and fps playback was exactly double my 32-core M1 Max. Seems like it scales as it should with proper software code?
It would be great to get the Logic Songfile so we can gauge how our current machines fare with the tracks you ran on your test machine.
Is it possible they’re trying to avoid this machine stepping on the Mac Pro’s toes? If they properly fed power to the 64 core 128 gig model, would it trounce the Mac Pro?
Hey man, cool videos. Just one thing regarding the music production / logic performance test. The amount of tracks means very little on its own. You have to tell us at what sample rate, bit depth, and buffer size the session is running. Also, what kind of processing are those tracks doing? what kind of plugins do they have inserted? how many plugins per track? etc etc etc.
Without that information the amount of tracks is meaningless.
Yeah, I am guessing he means that many tracks all recording at the same time? I went to the apple store to test out the m1 max Mac Studio, and it could handle 10,000 tracks :)
Regarding “more power for scaling”, keep in mind that there’s probably power needed for the interconnect between the processors as well, rather than just doubling the memory channels/CPU count.
I taught that was solved 🤔
@@ifiwantyoutofeel There's nothing to "solve" there unless the power used for that (which is a necessary loss) is unusually high. Not having perfect scaling isn't unexpected in the slightest. that's how it's always been & probably always will be.
getting near to that perfect scaling is pretty impressive though.
I knew the 64 core would not be worth the 1k however I still got the 128gb ram as I plan on using after effects and blender. Hopefully the gap will improve when these machines and software are more optimized. Either way the ultra will be noticeable better then my work horse M1 Mac mini
Hey guys, thanks for this. I just realised, you posted the Affinity Combined scores which won’t show GPU scaling because it’s across CPU+GPU. The GPU raster score is the one which not only shows decent scaling but also the M1 Ultra kicking the RTX3090’s but at 53K vs 21K.
I'm feeling a bit happier about my decision with the 32 core m1 max with 2tb SSD and 64gb ram. Price was $3200. If I wanted the Ultra with 48 cores and similar storage space it would have been an extra $1200. With my basic photo and video editing this machine should more than enough. Worse case, I sell my mac studio for a newer version with better specs next year if this for some reason isn't enough for my task.
Have you guys done a video on functionality of the Mac studio running applications in Windows through Parallels for how feasible it is?
Fabulous video - the one I had been waiting for! Thank you for all your hard work on this. Two question though, please:
1) Logic tracks - were these just simple audio files, or did any of them have plug-in effects or instruments on them?
2) Do you think going to 4TB of SSD would be good for audio? I hear that the 4TB SSD has a faster transfer rate than the smaller ones, which might increase track count? Also many virtual instruments and sample libraries require their assets to be on the main drive, rather than an external SSD. I will also get external SSD for Time machine and general project archiving though.
Both these questions are based on me getting the 48 core Ultra but keeping the RAM at 64GB. I will be mostly running a lot of processor-hungry plug-in synths on it, but would also like to dabble with Blender and Unreal Engine for non time-critical projects. Any advice greatly appreciated please.
1. These had simple instruments on them.
2. I don't think it'll make any difference whatsoever. Even the base model drives are already incredibly quick.
I would keep it to 64GB of RAM for sure. I don't think the extra cost is worth 128GB. The cores just can't ramp up fast enough to make a big difference.
@@MaxTechOfficial Thanks. I really appreciate the extra info.
I appreciate it too. Saved me 800 Euros! 😎
Thanks for this kind of info. It’s why I’m putting my money towards a XDR display to use with my 14” M1 Max MBP instead of another computer (and matching display) that I can live without. One suggestion for the future, although you’ve covered it in the past, it would be nice to see an overall comparison of the entire M1 Pro/Max/Ultra family in one video, or at least have tiles to refer to your past videos where you performed the same tests with, for example, the 14 and 16” M1 Pro/Max MBPs. Thanks again for your excellent content!
So basically for almost everyone, the base model is the one to get! Glad that is what I picked, thus far I'm blown away with the performance and couldn't be happier!
That being said the amount of sheer what you guys have done and provided to all of us is amazing. I know there are people who can take advantage of the Ultra and your work is providing them with a LOT of relevant information when configurating their systems! Big thanks for all you guys are doing, hopefully you will hit that 1M subs very soon!
I appreciate all the testing being done, but I think its just a little too early to be testing for what to buy right now. Sounds like most of the apps aren't optimized for the new CPU's and the ones that are obviously show it. I hope to see another round of testing once everything is optimized.
I think a problem here is that you care comparing temps from Intel x86 chips and TSMC ARM chips. I don't know for sure, but these sort of chips may not have as high tolerances as the Intel chips.
Since the TSMC chips are designed initially for portable devices, they may not be designed for 100c temps for long periods. Once you get those higher temps, failure rates may increase substantially, decreasing life of the product, and increasing Apple Care requests (which would increase costs for Apple).
This is just a guess on my part but might make sense.
This was the video I've been waiting for. Thanks Max, Vadim and Anjelica!
Thank you for checking this, the M1 Max Spec vs M1 Ultra price is huge and getting these comparisons helps in our buyung decision. Keep it up and hope you'll get that 1M subs soon
After watching that video I am definitely sure that I made the right decision with my 1 TB base model. Thank you for the great work and this very informative video! 🙏
For a video editor what would you suggest, where I should have to spend a little more if I buy the base model M1 Max Mac Studio :
1. Upgrade RAM to 64gb from 32gb
2. GPU upgrade to 32 core from 24 core
3. SSD upgrade to 1tb
Tough call. If you layer a bunch of effects and titles and can handle having only 512GB SSD (if you use external drives for example) then upgrade to 32 Core. If you don't do a bunch of GPU-heavy effects then upgrade your SSD. RAM would be the last thing I would upgrade.
@@MaxTechOfficial Thank you so much for your reply, based on your suggestions I think I should go for GPU upgrade.
SUPER DUPER JOB Max and Vadim
Thank you Erfon!
If you guys disappear we know now who to blame💀
Another good video. With every previous Apple Mac hardware release I’ve never got it quite right. This time I started with the M1 Max Studio and went to up the RAM and SSD to find it was not much more to go to the base Ultra. The Mac Studio base Ultra is the best Apple purchase I’ve made in more than a decade. Does it perfectly scale? No! But for a cashed up hobbyist photographer like me with a 100 megapixel stills camera (GFX100) and shooting the odd family 8K video on an R5, the base “binned” Ultra is essentially perfect.
shut up spambot.
Tracks? Audio tracks? MIDI tracks? FXs? Plugins? Sample rate?
How about doing a comparison on the M1 Max with the 24 core verses the 32 core... is it worth the extra $200... and how about the 32GB verses 64GB is that worth the $400... really enjoy the information but I bet there are others like me that can only afford the M1 Max and want to get the best bang for the buck... thanks
Ordered Mac Studio Ultra 2TB for my Recording studio running Protools. $4,706 and change. Hope this does better than that M1 I had. Love your channel Max..Keep it coming.
It s clear what s the problem. They didnt cool the power supply, that cannot supply sustained power to the gpu. In some way it seem the two teams didnt communicate and focused only on cooling the chips. Not much can be done now
I completely agree that the M1 Max is probably the machine that 90% of buyers should get.
Shocking results, especially video editors who Apple are firmly targeting with this machine. Can this really be a design flaw with the MAX chips....? Now actually makes me glad my Studio Ultra is not due until June and Apple issues an OS update to address this or might just cancel and pick up the Max version.
Are you sure you could not hear the fans at all? I got my Mac Studio (M1 Max) yesterday and I was shocked because of the fan noise. I wanted to replace an iMac 5K from 2017 and the Mac Studio is much much louder. Even standing 3 meters away from my Mac Studio I can clearly hear the fans going in idle. It is even louder than my 15" MacBook Pro 2017 during idle. So when you say it is "completely silent" I can't believe that or my Mac Studio has some kind of malfunction.
Even though I won't purchase any mac any time soon, I wait everyday for these videos. I appreciate all the efforts you guys put in these videos.
I’m thinking you are right - it’s largely a silicon issue. Or, it could be that they are leaving room for the the M2 to show performance gains over the M1….. There does seem to be a software issue, but something else is going on that causes them to put the brakes on performance.
Firm believer they are throttling both ultra models this can be seen in the power draw and ultra still needs to be optimized
So basically the basic Mac Studio will be more than sufficient for (45mp) photo editing? And an upgrade to the 32-core GPU and 64 GB RAM isn't worth it?
Apple needs to address why they are limiting the wattage on the Ultras.
I just received the 48 core Ultra. I am now debating on returning it for the Max.
Thank you for the over the top video. Excellent work
I'm ready to place my order, but the real question is if the software will catch up with the hardware? Is it worth investing the extra $1800 now for the extra memory and GPU cores with the understanding that this machine will last me for 5 years? I breathed new life into my 2017 iMac Pro by adding a very powerful eGPU to it, but that's no longer a possibility with Apple silicon. Thoughts?
what you also should consider is... that even the unbinned m1 pro on the macbook has exactly the same CPU performance as the m1max on the studio...
I am on day ten of sitting on Apple's website trying to figure out what to order. I'm coming from a 2019 5K iMac with an i9, 580X, and 64GB. No issues with the machine other than lack of external displays. I can order and pick up a base M1 Max Studio today, but no Studio Displays, so I'd have to go with the LG's, or something else. The 512GB might hurt a bit too, but with all the additional USB-C and Thunderbolt ports, externals should cover that. Then there's the MacBook Pro for $3,500; also available for pick up today, which should do the trick with four displays; three Thunderbolt, one HDMI, but that would consume all the Thunderbolt ports, leaving me no room for external SSD. That's not the end-of-the-world as I only need the external displays when I trade so I could disconnect one when I had to connect an external SSD, which hold my FCP libraries. I guess the solution to that would be the HyperDrive 4K, connect the three displays via HDMI/DP, regain three USB-C on the HyperDrive, but give up the fourth display via HDMI?
I think you are wrong that Apple ‘is allowing more power’ for the ultra. It’s simple math - each chip needs about 34w. There are two in the Ultra - which is why it takes 68w ?
Just wanted to thank the Max Tech team for this amazing coverage. SO appreciate that you include Logic as you did for the MacBook Pro. You guys just keep delivering the goods! There’s a reason why you’re moving towards the 1 million, and your fans will celebrate with and for you when you get there! 👍🏾😊
Hi, by standard render time is a stat to refer for performance, but what creative user like us would like to see the smoothness when using it. Example like working with large/heavy layers file in ps, when masking high dpi file with special brush, dealing with multi-layering smart object. Or does it smooth to drag and play when editing in AE, esp when preview time. Thanks,
23:26
You are talking about the frequencies of the GPU, the wattage but where are the temps and the fan speed ? It’s maybe that the chip is heating to much…
24:21
Never mind
I really hope Apple gonna fix this
Great video guys!!! Again I think we are seeing why we are getting the Mac mini pro last. I think many video editors might find it the best bang for the buck. So Apple released the Mac Studio first to get most sales before releasing the higher end Mac mini which should be an easy machine to ship. It will be nice to see the Mac mini Pro compared to the base model Mac Studio Max.
Overall as impressive as the Ultra looks on paper so far it looks like the most disappointing chip Apple has released to date, not that it is a horrible chip just underwhelming. Hopefully Apple can patch the OS or firmware to push this chip to better performance. The fact that this hasn’t happened already makes me wonder if it will.
Mac Mini M1 16GB vs. Base Mac Studio?
Memory doesn't mean crap when ing multiple piplelines inside of docker / kuber clusters
3:30 - that's a very impressive linear performance scaling going from 10 to 20 CPU cores. Also doubling the performance only results in a 5% additional power draw in excess of an idealized 1:1 power increase / performance scenario.
22:50 - that's crazy and severely limits the usefulness of the M1 Ultra in both 48 and 64 GPU core configurations.
So could you be a little more clear on what you mean by tracks in the music production test? So what I mean is were those just straight audio tracks, or did they have plugins on them? Also what was the Khz of the session? 44.1k, 48k, 96k etc, and what Sample Rate were you running it at? 64, 128, 256 etc?
I've been using a 2018 Mac Mini i7 32GB RAM with an eGPU. My Mac Studio arrives next month. Anybody interested in a comparison video with benchmarks etc before I sell my Mini?
I usually make how to tech videos, but a comparison would be fun to make.
@@DanielToman I figure a lot of audio/video people are making that jump now that the Studio is out. M1 Mini just didn't have enough ports/only 16GB RAM.
Would love to see the gaming performance.
edit: You already made one! awesome
For photo and video editing, this should be an ideal configuration (best bang for the buck): Mac Studio M1 Max/64GB RAM/2 TB 10C CPU 24C.
I wonder what is the difference for 360 video with 48 and 64 GPU.
It’ll be interesting to rerun these tests after 6ish months of SW upgrades. Some of the throttling may go away.
It’s perhaps not impossible that the interconnect between the two M1 Max dies could be maxing out: GPUs generally operate memory-bound, and perhaps there’s enough GPU memory activity having to cross that interconnect that they kill off all available bandwidth.
Excellent information! I have been watching all of your videos, because my mid 2010 Mac Pro has become asthmatic, and recording with microphones is a problem. I will probably go for the Ultra 48 core 64 gigas, and am sure it will serve well for several years. Thanks for your work!!!
Hello - thank you for your in-depth dives into all things mac. Have you, and I've inadvertently missed it, focused less on graphics and more on audio? I'm designing a recording studio (with a softer focus on video/film editing) and need help understanding how to best finesse my new mac(s) (laptop and desktop) and make the best investment for the longterm. can you help?
I was thinking about upgrading to the Max/Ultra but after this I'm just keeping my M1 mini and when they release M2 Max/Ultra I'll see if these problems have been fixed, dissapointing...
The number of power and ground bumps on the M1 Ultra’s are the same and share the same power grid on chip no matter how many good GPU’s are available so power per GPU result is not surprising if it is a bump or power grid limitation
It sounds like a lot of what is being tested is not yet optimized to use the available power along with Apple not allowing for that power to be used.
Do the Ultra machines have the "HIGH PERFORMANCE" checkbox in the OS settings????? Like the M1 Max in the MBP???
Nope.
I think you're misinterpreting a few things when you're looking at the power differences. 1: Apple is not pumping more power in to the CPU, the ultra is literally 2 M1 Max chip joined together with an interconnect. So its just normal full power per chip, plus interconnect and more memory modules. 2: Apple's GPU slide is clearly entire package power vs just a discrete GPU. The angle is that their whole Chip is using less than just the single component of a PC. You are gaining memory bandwidth from the increased number of memory modules, where as the raw capacity is basically meaningless unless the workloads are running over the capacity.
Either Apple is prioritizing silence over performance, or the they didn't build much extra headroom in to the power supply. There could also be something to their choice in moving the storage controller off of the storage module. That is one component that typically has a upper limit of 70C, now sharing thermal capacity with the M1 SOC. One of the typical drawbacks to Apple designing their devices to not be opened is the dust collection over time. I can see that being largely mitigated from not running the fans basically at all.
Actually, the power metrics I measured are counting the wattage going to only the CPU cores, not the rest of the die or the rest of the package. Definitely not the memory. Only CPU cores are counted.
@@MaxTechOfficial Okay. The shot of the monitoring software does show a package power of 106watts at one point. In reference to Apple slide, that seems most likely what Apple was referencing as opposed to giving the GPU only 100+ watts.
If you have hour long 8K Videos use 2 or 3 base model Mac Studios…Apple should excus for the dimishing returns and sell more base models and optimize the wattage usage of the topmodel
At this point, I'm guessing they designed this thing for future iterations of Apple Silicon chips. If you remember; one thing Phil Schiller said about the 2013 Mac Pro was that they backed themselves into a thermal corner. It looks like they have plenty of room to work with now.
Both variants cost the same. Which model would you recommend for video editing?
1. Mac Studio 2022 M1 Max/32/2 TB 10C CPU 32C GPU
2. Mac Studio 2022 M1 Max/64/1 TB 10C CPU 32C GPU
SSD over RAM upgrade?
or would you spend more for this variant?
3. Mac Studio 2022 M1 Max/64/2 TB 10C CPU 32C GPU
Your graphic is showing the $5,799 version and not the $8,000 version? I'm assuming you are testing the $5,799 one. I love these kind of videos that show the actual performance and not just thinking you get better performance!
IDK all the specific workings of those GPU benchmarks, but could it be you’re “drag racing pickup trucks” rather than seeing how much they can tow?
Maybe these benchmarks are too low res to show a significant difference. I’ve seen some crazy difference with GPU’s with more RAM when pushing tons of pixels (LED, etc. video walls), yet very little FPS gains at low res. I suspect if you pushed a “canvas” of 4x 5K displays, for example, the 64c would get better fps than the 24c & 48c pushing the same number of pixels.
While their 0-60 could be similar, the “pickup truck” with a twin turbo V12 diesel will be able to tow more weight, and maintain speed, up a hill vs a NA V8.
I am so disappointed that apple killed the 27 inch iMac. Is there any hope for this to come back in 2022/2023?
What do you think these top-end studios with maxed out configs are really going to be for? Will we see optimizations dealing with this extra hardware? All I can think of besides niche cases is servers.
It almost feels like some driver updates are missing. Hopefully Apple can still improve this. It'll be worth rechecking it in a few months. I have a 5950X for software development (I mostly develop on Windows, for Windows and Linux).
Do update again once the mac studio firmware is out to solve the scaling issue.
I wish you included the 16” mbp w/32 core in this test series to see if the sweet spot for the lower cost studio pros is actually the 32 core rather than the 24 core.
I am beginning to get suspicious about software optimization. I do recall that, when I was using my 2009 Mac Pro Cheese Grater in 2009, there was a lot of software that only saw a single core. In about six months, there was software that could take advantage of a four-core Intel CPU. It wasn’t until almost 2012 that software really started taking advantage of 8 cores in the two processors.
You are seeing a seriously diminishing return when you move to the Apple chip with double the number of CPUs and GPUs of the Max. I would like to “blame” software optimization as not all software will spawn enough threads to take advantage of as many cores as are available in the Ultra (remember, the OS is what will send threads to different cores) and the operating system, itself may not be sufficiently optimized for the Ultra, making threads wait even though there are CPU cores waiting.
We did see this when Apple released the dual hex-core Intel systems in 2012 and, I think that the Trashcan Mac Pro also did not scale up properly. So, the fact that you are forced to use Apple’s current operating system and also the current software may be the real limitation in the Ultra system.
Am I the only one that find the Mac Studio way too loud? My 48-Core Ultra has two forms of noise, the ground noise (too loud in a quiet room) AND a high frequency chirr noise. Sadly sending this back to apple and staying with my silent MacBook Pro Max :(
"photo editing" is only a large panorama in PS? I know you can only show so much on these videos and I appreciate the content very much but I use photoshop with hundreds of layers and many smart objects and tons of blends and effects etc. I am so worried to buy this machine - every time I upgrade to a new mac there's never a noticeable improvement in PS. For years PS has been slow on new macs because they don;t use the multicore as much and require tons of RAM. Love to see a test with a massively layered, 300 dpi, large canvas photoshop file and see if it's any better than say the m1 imac, current mac pro intel or even an older Xeon 10 core iMac Pro from 2017 (which is what I have).
Is there a logic test results table to compare with other Macs available somewhere? Results for an M1 Mac Mini with 16GB, an i7 2018 Mac Mini with 64GB and a 5,1 series 3.46Ghz 12-core Mac Pro would be very helpful in planning what Mac is worth getting. Particularly as it's £800 more for an M1 Max Studio with 32GB and 512GB SSD than an M1 Mac Mini with 16GB, a 512GB SSD and the ethernet upgraded to 10xGigabit.
To the MaxTech guys, is it possible those GFX Bench and similar GPU tests to do some more extreme resolutions? For example, instead of 1440p to go 2880p on the three models?
It’s just to take out of the way the possibility of too fast frame rates affecting it. It is my understanding that when a frame finishes a lot of stuff has to be done (in classic architectures at least), like maybe it needs to sync the cores to finish it, pack, collect, copy, transfer, flush buffers, etc and these things are usually not that thread friendly as it is ‘a whole page’ if you will. So maybe, if the M1 Max is pushed hard to run at say 30fps, maybe the ‘frames per core’ might be different. Just curious about it.
M1 Ultra 48 cores. The other two models are not worth it because the M1 Max is easily outperformed by a MacBook Pro and the advantages of the Ultra 64 are not worth the extra 1000 bucks.
Same. Went with Ultra/48c/128/4TB. You?
@@aram.v I am thinking about going 48c/128gb/2TB. I have it in my cart, ready to be purchased. I just gotta make one more payment to my credit card, so I will have enough limit to plug the cost of my Mac Studio. I wish you enjoy yours and get the most of it. Congratulations aramv!!! 👍🏻👍🏻
@@BestAudiovisualTricks awesome 👏🏼 same to you . Got another 2 months to wait lol
@@aram.v You are very welcome! I share the excitement! I should get mine by September and already cannot wait. Have a wonderful week! 😎👍🏻👍🏻
The question that remains unanswered is why has apple restructure the performance of the Ultra machine (and arguably all the lowest model as well to some extent as it performs little better than the 16 inch MBP) ?
Is it that OSX needs to be upgraded to take advantage on the duo core SoC chip (along with metal and obviously non-apple software) and we can (eventually) expect much better performance or is there some hidden limitation(s) in the physical hardware derived from doubling to SoC to two Max chips ?
Nice video! I either missed it or it wasn't mentioned. What buffer size was used on the logic pro test?? Very interested to know. Thank you for the vid
Apple should have named the Mac Studio the Mac Maxi in honor of the great content being produced by the crew at Max Tech.