Would LOVE some After Effects and Premiere testing. As others have noted, export times are great indicators of overall power, but doing some testing in a timeline/sequence with background tasks or lots of effects to process would be a great real-world use test. Loved the vid last cycle when you threw those Premiere comparisons in...
the Xcode graph at 4:53 is confusing.. its shows 1:09 vs 0:67, which is a tad confusing, because 0:67 is really 1:07 which is basically identical to 1:09. Or did they mean to write 109 vs 67 (without the colons)? Because, if so, that would be a significant difference.
I know it might sound weird but how about a comparison between m2 mac mini vs m2 ultra mac studio. Both in the base version. Maybe with the addition of performance to cost ratio. Would love to see the video.
Engineer here. There is a good reason why scaling is never linear, and that the Ultra only doubled performance on heavy load. Basically, it comes down to the size and number of the data blocks passed to cores when using multiple cores and what size makes up a dividable core task. I worked for a company call Inspex for years as a semi-conductor engineer. Basically, it looks like the block size passed to a core in Apple Silicon is pretty large and runs out of blocks to serve to cores pretty quickly. Seems like they made it efficient for the number of max cores and not the ultra cores. Not a bad thing. It just means that all the blocks are pretty much parsed out before it needs the extra cores unless there are A LOT of blocks that need to be processed. There is actually processing lag if you optimize a chip to break data into smaller blocks to pass to more cores because A. You don't really know how much time a black takes and you may end up having some cores wait too long for a single demanding block to finish, or B. There is data assembly lag.. meaning, the time it takes to reassemble the blocks of 12 cores is meaningfully less than the time it take to reassemble the blocks sent to 24 cores. More block to more cores may actually be slower than less blocks to fewer cores for smaller black counts. I am thinking they looked at the average potential block size for tasks apple customers will likely have... Also, the weakness of the way the chips are attached means you cannot change block processing parameters. It's hardware fixed in Silicon. When you glue 2 chips together, you get what you get. There is a controller that manages traffic to and from a core. Differences in this controller faced the same issues on Intel and AMD, who use slightly different hard coded core allocation for different reasons making one better for large block processing and others for smaller block processing. Hope that makes sense. I tried not to get too technical.
The analogy I like to use in cases like this is the challenge of keeping a V12 engine fueled. As the engine needs more fuel for higher RPMs, you need to increase the fuel pressure to ensure that all the cylinders get a full charge. If the system is unable to pump enough fuel from the tank, bubbles form in the system, or fuel pressure is variable, then you're going to see a loss of power. Possibly even problems like sputtering where the engine will run *worse* at higher RPMs. It's the same problem with multi-core issues. If you can't keep a steady stream of blocks to be processed, the CPU will sputter and performance will fall off a cliff.
I'm NOT an engineer :) but that the Ultra wasn't arithmetically faster than its smaller brother didn't surprise me either. There are always overhead and bottlenecks that get in the way of a perfectly proportional increase in performance. I remember in the 90s, when multiple CPUs became a thing, that performance was usually ~50% faster with the extra chip, never 100% faster.
Ok dudes. The Figma Ads are starting to become really obvious throughout all of your videos…. I mean at least switch out the tag line. Let it be known. Figma mention: Paid Ad.
Something I would be interested to see is to run all of the benchmarks side to side or progressively and find where does each processor starts to lag. To expect double the performance is not necessarily realistic nor a good thing. I would expect the ultra to excel at parallel tasks running at the same time (because of the increase in core count) without dropping too much performance (probably because of power limitations). I would recommend something like the studio with the ultra in case you need to open several apps at the same time and work on heavy things in parallel.
Strange that you got the same results over the two machines for Davinci Resolve Noise Reduciton, Tyler Stalman clearly showcased that the M2 Max struggled to keep up with live / 24FPS playback, and the M2 Ultra was the only machine that was able to do this. Even citing the M1 Ultra not being able to keep up. I think with a grade, noise reduction, and any other node elements, the Max would struggle big time
I’m still on M1 Pro and I make iPhone apps with Xcode. Having no issues at all. I say don’t upgrade until you noticeably start to feel a slow down or need the horsepower. Don’t upgrade just to do it.
Great video but while export specs for FCPX etc are great , I already know they're fast. What really interests me as an editor on all three of the major systems (FCPX, Resolve & Premiere) is how well do these actually handle editing... esp with effects like CC, text, plugins etc. Does it hang up or does it play smooth like butter. How does it handle codecs like XAVC-I etc? Do I still need to make proxies if I'm going to be working with multiple plug ins/effects etc. At this point, a fast export is a given and isn't as important as smooth reliable editing. Any thoughts/info? Thanks for all the info!!!
yeah I am confused by that too. I wonder if it meant 109 seconds vs 67 seconds? That would be a significant difference. But if they meant 0:67 as in 1:07, then that's not a meaningful difference at all.
I've never heard you say that playing back 8K footage from the RED camera was an issue on any Mac silicone base system previously - but you stated it was *not* an issue here. I know for a fact that 8K canon footage chokes my MacBook Pro M1 system (16 inch). I would really like to know how these systems handle that footage.
This specific case is color graded 8K Red in a full 8K timeline at full-quality playback. The M2 Ultra also plays back 8K Canon RAW with a lut as well!
@@MaxTechOfficial Would love to see a YT Short showing that 👀 My guess is more ppl are dealing with R5 8k footage than RED, given that it's a prosumer camera.
It would be interesting to throw in a decent spec i9/4090 PC in to the mix. Ultra for Logic sounds great but, man, the cost for the performance is difficult to justify.
Is the binned vs unbinned M2 Ultra video out yet? I've been desparately waiting to watch it as I need to decide ASAP. Could you please tell me at least when to expect it be out, sir? That would be awesome, thanks 🙏🙏🙏🙏🙏🎉❤
Great Video as always. But I don’t quite agree with your final verdict . Have a look in timeline performance with user cache or smart cache on in DaVinci Resolve. Many clips with effects and color grading are constantly rendered in background. This is significantly faster with ultra. Timeline performance is more important than export times.
I would also like to add that you compare 512 gb vs 1 tb. Every professional would at least buy 1 tb. If I compare the configurations with prices in Germany you have 3089 Euros vs 4800 Euros. Yea it is more expensive but far away from being double the price. This is click bait title.
I mean, 8 displays is a good start, but WTF with the 5? Who uses only 5 displays these days, does Apple think its 1990 or what? Im holding off until the 18 displays connection.
Would be interested in seeing the unbinned M2 Max vs the binned M2 Ultra. Does having 38 GPU cores ever beat having 30+30 GPU cores? How close is it in some of these tests?
Out of curiosity, would love to see these two machines against the M1 Max Mac Studio and the M1 Max Ultra Mac Studio. Thanks again, always find these interesting. Blessings
I'd love to see a poll of your viewers to see how many are video editors vs designers vs software engineers. I'm glad you added Xcode Benchmark back in but I wonder whether you'd find that the % of software engineers would make it worth showing more programming tests.
Nice comparison, thanks. However, it really does depend on what you use the Studio for. I've found that Fusion in DaVinci Resolve is still sluggish on the base M2 Max. Export times aren't an issue. But real-time editing and compositing was underwhelming. Even rendering compositions in place still happens at below 10fps...just like my M1 Mini! If you want super smooth compositing the base M2 Max doesn't cut it...even with proxies etc. enabled.
@@NirHason No, I didn't try an M2 Ultra....double the price. I've since settled on an M2 Max with 64GB RAM. The extra memory has helped a lot. Basically, Resolve seems to soak up as much RAM as you can give it, at least in my experience.
10 месяцев назад
I hope the benchmarks and thermals was done BEFORE tearing it apart. Just to get a feeling on the thermal paste that come from stock. That said. I thank the fact that you showed the processors size difference.
Regarding your resolve noise reduction test. How about making a 60fps timeline, or even 120. Just to see where the hardware chokes out and not both being 24 frames. It would be more clear for those editing 30 or even 60 fps content.
Great comparison video, would be great if you can include on your tests Topaz Video AI and compare if makes any difference the number of neural cores on different setups. Cheers!
Anyone who expects perfect scaling, or anything near that, simply doesn’t understand physics or the laws of computing. The fact that the graphics are nearly perfect is incredible. Stick to tear downs, please. Don’t give people advice. Your criteria are based on no professional workflows or models, and barely possess an understanding of computing.
Love your channel. To answer your question; It is if you need it, it isn't if you don't. That said I'd add the 1TB SSD upgrade to the Max. I'd also add, for those on the fence, the next M3 Max will outperform or be close to the M2 Ultra. In the words of Kurt Vonnegut, "So it goes." Disclosure, owner of several Apple Dealerships.
CPU performance doesn't scale linearly for a single workload. It's odd that MaxTech would expect them to. GPU scores should scale in a much more linear way though.
I do get that the you used the 2 barebone versions of the Products... but to compare it "Apples to Apples" it would have made sense to have an upgraded version as well of the Pro chip where it matches the Ram and the SSD to see what diffrence those 2 components make and what is the diffrence in the CPU itself. Like this there are too many variables
On CPU multicore performance, you need something is embarrassingly parallel workloads. These things are expected to be use by a professional so use software like them and they would likely have software to ingest a lot of those cores. My field uses such multicore usage. I'm not a radio astronomer guy but you can use AIPS or CASA to determine the multicore workload. And also, again, you need to use the Moana scene for GPU testing. That scene is large and heavy enough to see if there is any difference.
Fantastic Video! I'm coming here as a newly full time music & content creator looking to take his business to the next level **woot woot** and this video helped A TON!!!!! Even though I am far from a tech geek, I found that I was able to follow along with a lot of the topics discussed here. Especially the section on Logic, I don't see me or my clients budget being able to afford a 300+ track composition lmao! Also the comparison for video was very easily digestible in terms of seeing why the extra money wasn't worth it! So thank you SO MUCH for a great video
I might get a M3 Ultra when it's available. 😄 When you upgrade the storage and RAM on the Max to match the Ultra, the price does get closer and to the point where I should rather bite the hard bullet and get the stock Ultra.
M3 Ultra will probably arrive on fall 2024 as earliest. If your current Mac is still good enough then it's worth waiting for. But that's a so good question: go for M2 today or wait for M3? If it's urgent with a new Mac then you got no choice... just buy something that's available. The Mac is honestly only a tool. But a very nice one :)
Yeah then it's no need for M2. M1 Max will stay strong for a few more years. If I were you I would wait for that M3 Max/Ultra or M4 Max/Ultra. Depending on the money situation and the needs for more performance :)
@@cjeelde My previous "serious" machine was the trash can and it lasted 7 years. It was the longest serving Mac I have ever used. Got a spec-bumped M1 mini as a stopgap because the trash can started to have hardware issues. When the Mac Studio came out I got it immediately as it's pretty much the exact type of machine I was looking for. Don't need PCI slots but need a machine fast enough and letting me choose my own monitor(s).
I would really like it if you have a virtual machine or two running while you do some of these tests. A lot of developers have a virtual machine for a server or iOS simulator running while they develop. Would be great to know if those affect the performance between the max and ultra.
Thanks for this. I got the Max and was wondering how the Ultra would compare for the work that I do (Adobe PS, AE and PR) . This put my mind at rest that I made the right call and don't need the Ultra.
Mac Studio M2 Ultra host internal 370W power supply, the maximum power of this machine is 295W, when the full-speed operation is only 75W more, it can supply power for 6 Thunderbolt 4 and 2 USB3. It also needs 90W, and the frequency needs to be reduced to accommodate this situation? The price is double, but many machine hardware does not have double
Just for giggles, I decided to do the spedometer test on my iPhone 14 pro and got a score of 393, not too far off either of those machines. Not sure what the a16 chip has over some other PC chips, but it's interesting that my phone out preformed a solid AMD Threadripper with 64 gig of ram.
I recently bought M2 Max studio 2023 , I used it for editing in Final Cut Pro and is working well for me. The cpu is definitely a good buy from my review. I added external memory too.
I don't get the ssd speeds on the M2 Max version.... My base M1 Max MacBook Pro gets 6-7mbs read write on its 512gb ssd... Why 2 years later do they go backwards?
I got the most load on my M2 Ultra with running two conversions in Topaz Video AI and one Handbrake conversion at the same time. Then you hear the fans running (not loud but really blowing).
@@delcanvideoproductions4895 yes it scales much better than the M1 Ultra. It does not reach factor 2 but at least 1.8 but this depends on the scale and algorithm. Best are 2x or 4x scaling Artemis and Proteus scale well.
I'm at that decision trying to replace a 2018 iMac Pro which has always been terrible video machine for me. I edit in Premiere, nothing super crazy. This video is very helpful.
Amazing, I spent hours searching for this exact information. I was already leaning toward the M2 Max anyway, but I almost always buy top of the line, because I don't buy often. I really want an M4, but I need something now (my 2017 imac pro is showing its age and had an odd failure the other day). Now I just need to decide if bumping up the memory, GPU, and hard drive on the M2 max will be getting so close to the base M2 ultra model that I might as well go to the ultra for the price difference.
The Xcode times were strange. Is it fair to say that the 0:67 timing is the same as saying 1:07 ? That would show the times for both systems to be very close, correct? Thanks
I have doubts about the Logic Pro benchmark, are you sure you ran both machines on a 1024 buffer as the instructions suggest? It seems crazy to see such a difference and also that my basic M1 Pro mbp does more tracks that M2 max Mac Studio. Could you please check? Thanks
Hey Max, always appreciate your reviews! Did you test Canon RAW 8K? That’s what we shoot. It’d be great to find out how these two Macs handle it. Not having to create proxies would be amazing. Thanks 🙏
I love your videos, but could you please update the video export section of your videos. At the speed you’re getting now for example 4K five minute video exporting in less than a minute. Unless for whatever reason, the next generation of tip chips takes longer than that this does not matter anymore to most people. I’d much rather be interested in what a 10 minute or average eight minute or 15 minute 4K video takes to export. On top of that, would I really want to see is HEVC 10 bit HLG or HDRPQ export times. Those are still taking much longer than an hour to export on M1 Pro for an eight minute video.
I know you can’t test everything but it would be great to test a really heavy photoshop illustration and really heavy indesign file with a lot of spreads. I can chew through a tone of memory using indesign and photoshop in tandem.
You can’t expect linear increase especially with multi-core, most it depends on the software, just the fact some software achieve a 80-90% increase on the Ultra is incredible
Hey Max, could you tell me how you were able to get the BMW test for blender? I've looked _everywhere_ to download this to test my Mac, and I can't find it! Great video!
Thank you for clearing my doubts. I appreciate very good content. For software development (backend, iOS, Docker) and video editing cheaper M2 Max with 2TB SSD disc and upgraded RAM up to 64GB will be good enough.
Thanks Max for another brilliant video on the new M2 Mac Studio. Would love to see some comparisons between the previous gen M1 Ultra Mac Studio with M2 Max Mac Studio. From Apple's charts on the performance comparison they show that M1 Ultra outperforms M2 Max in some areas, and some people may be eyeing M1 Ultra Mac Studio in the refurbished/2nd-hand markets
Whether something takes 4 minutes to export or 2 minutes to export just doesn't really matter. Unless you just care about specs, and not on actual workflow and time 'waiting' for your computer.
Thanks as always for such excellent content! I was really surprised about idle temps compared between the Ultra and the Max! I fully expected that the copper heat sink on the Ultra would be a bigger help. Instead, it seems like it's actually NECESSARY just to keep temps in check, due to how much more energy the Ultra uses and heat that it generates, than the Max. The only thing for me that never gets addressed by any of these reviews - on ANY channel - is how various Macs compare with respect to flight simulation, especially a flight sim setup using multiple monitors, which, because of the unique, dynamic content on each screen, adds lots of extra demands on the computer. For pretty much all of personal computation history, Apple has lagged behind PCs with respect to the graphics needs of flight sim. Consequently, no serious flight simmer uses Mac. Up until recently, that was not a problem because there was no serious entry from Apple that was worth consideration. The change came with Apple's ARM chips, at which time, rather than tackling performance by just throwing more and more horsepower at the problem (more powerful chips, meaning more bulk, heat, & energy consumption) Apple finally approached the problem from a different angle... efficiency. Suddenly, Macs became - to paraphrase an "On the Waterfront" line - "a contenduh!" However, here's the problem. For all of history, we've had PC experts and Mac experts, based on the computational skills they cared about. For that reason, Flight Sim experts are quite knowledgeable on PCs as well as the comparisons between PC performance levels and how that translates into Flight Sim skills. Meanwhile, Mac experts know a ton about relative Mac performance levels, especially with respect to graphics skills when used for photo and video editing. However, NOBODY (seemingly) who is a Mac expert, knows anything about Flight Sim needs because Mac doesn't have the history of performing well with flight sim programs. These 2 paradigms have diverged so totally over the years, that I'm despairing of finding a flight sim expert who knows enough and is interested enough in flight sim application, to do a test like the one you did in this video, BUT... with specific emphasis on its application to flight sim work. (Yes, "work". Sure, there are flight simmers who do it more as a gaming activity, but there are also plenty of flight simmers who use it very seriously to augment their flight training for their pilot's license and additional certifications.) For folks lucky enough to be able to afford to spend $2k to $4k on a PC dedicated to flight sim purposes, plus the $2k to $4k on additional dedicated monitors for a flight sim station, AND have the extra space for such a flight sim station, and then doubling that cost for a "regular" personal/work setup of a Mac-based work station... NO PROBLEM !! Just buy 2 completely separate systems, one PC-based with 3+ monitors and another Mac-based 3+ monitor work station for personal/work purposes. Of course there's another option. Throw out all my Apple products, iPhone, iPad MacBook, etc, and convert EVERYTHING to PC, including all my calendar entries, address book entries, etc (need I go on?) so that I can have just a single, PC-based work station. Ugh!! For the rest of us poor sods that don't have the budget nor the space for 2 totally different work stations and don't want to do a full-scale conversion from an Apple ecosystem over to a PC/Android ecosystem... but ALSO care about quality flight sim... there seems to be no light at the end of the tunnel as to finding some who is both a Mac expert AND a flight sim expert. Any thoughts? (My sincere apologies for the long comment! I know, I know. My problem cannot and SHOULD not be your problem nor the problem of any of your subscribers. Just not sure where to turn. If you made it this far... THANK YOU!!!
Software has to be optimized. If so, you get up to 2x speed on the Ultra. Which justifies the price. Which is double. Plus, longevity. You can technically have it twice as long.
Spend the 2000 dollars, and in two to three years, when you upgrade it spending another 2 grand, that will outperform the one you spent 4 grand on just a couple years earlier.
Max, you have a xCode performance "thing", it is time to have a ML performance "thing", maybe an "Apple Core ML demo run between the different Mac Studio/MacPro models??
we arrived too early, leave youtube some time to process the video at higher quality... i bet in 5 minutes if you come back to the video it will be in full quality
I have to stick with reality and stay with the Mac Studio M2 Max with 1 TB. I need to use the extra $1000 for the 14 inch iPad Pro which for me would be a more useful combination for what I do. So Mac Studio M2 Max 1 TB for $2400+iPad Pro 14 inch for about $2000 makes more sense than spending $4000 for the Mac Studio M2 Ultra. Thing is, what about when the Mac Studio M3 Max come out. Probably in 2025 so can't wait that long. So, now just need Apple to get the iPad Pro 14 inch out. Right now I multiply the M2 benchmarks by 1.2 to see what the M3 will be.
I wonder if you could get a sense of how energy efficient a Mac is by the thickness of its power cable. I was thinking the magsafe ones compared to the one pictured in this video.
Right now, im in a point of my life of opening my eyes once again, and perceiving clearly the true nature of others, and destroying those bridges that no longer serves me and burning those fake or draining connections in my life.
The fact that doubling the CPU or GPU cores doesn't double the performance when having demanding tasks is a big no for me and it kills apple's marketing
Yes that would be really helpful. Is the M2 Ultra going to speed up Topaz Video AI processing significantly ? Or does this depend on the AI model chosen and the processing settings - do some Topaz AI models use multi core / graphics more than others ?
thanks for the comparison... although lol, "double performance" is way too high a bar to set! 42% is a high boost. Plus, there are many more tiers of the Ultra - up to 192 GB, and 76 GPU's. I'm sure it gets there. But... yea double the money doesn't mean double the performance, that's kinda weird to expect, it just doesn't happen, we already know how it takes like 10 years to get a computer to be twice as fast.
When looking at SSD performance there's more than just that "read & write". There's IOPS and 4K random read/write for example and we never see those numbers. These "read & write" speeds from BlackMagic can be compared with normal classic HDDs that can reach maybe 50-150 MB/s. So 2-3 GB/s or more for a SSD is really really fast! We need those great speeds for raw footage and big MP4 files. I'm curious about the future for SSDs. That 3D Xpoint from Intel and Micron got announced in 2015 but in 2021 and 2022 both companies discontinued it. I remember that announcement speech in 2015 and it was so promising but now 8 years later only a few early adopters gave it a chance. Apple never used it for the Macs. Apple probably tested it for evaluation, but we can never know. The BIG jump in performance was switching from HDD to SSD. Now it's incremental performance boosts. Now keep in mind that a DVD movie backuped for a DVD-R is up to 4,7 GB. These better SSDs can read and write that in less than 1 second! But if you copy that from internal fast SSD to an external SSD that will take longer because of the speed for the external SSD. 20 years ago with those hard drives it would probably take an entire minute to copy that between two fast HDDs.
Wow Moore's Law: 6:15 2x the speed for less than 1/4 the price -- only what 8 years later? 2/(1/4)/(1/4) = 16x performance/$ in 8yrs = 2x performance/$/year.
Would LOVE some After Effects and Premiere testing. As others have noted, export times are great indicators of overall power, but doing some testing in a timeline/sequence with background tasks or lots of effects to process would be a great real-world use test. Loved the vid last cycle when you threw those Premiere comparisons in...
Thanks guys. Just purchased the M2 Max Mac Studio. This channel gave me the best info between it, the Mac mini pro, and the Ultra
We're glad to hear and very happy that we helped you out! Thank you for your membership!
the Xcode graph at 4:53 is confusing.. its shows 1:09 vs 0:67, which is a tad confusing, because 0:67 is really 1:07 which is basically identical to 1:09. Or did they mean to write 109 vs 67 (without the colons)? Because, if so, that would be a significant difference.
I caught that too. Really confusing.
Ignoring the 0:67 which is clearly wrong, the proportion of the bars is 1.7 to 1, which yields 41 seconds for the Ultra.
I know it might sound weird but how about a comparison between m2 mac mini vs m2 ultra mac studio. Both in the base version. Maybe with the addition of performance to cost ratio. Would love to see the video.
Engineer here. There is a good reason why scaling is never linear, and that the Ultra only doubled performance on heavy load. Basically, it comes down to the size and number of the data blocks passed to cores when using multiple cores and what size makes up a dividable core task. I worked for a company call Inspex for years as a semi-conductor engineer. Basically, it looks like the block size passed to a core in Apple Silicon is pretty large and runs out of blocks to serve to cores pretty quickly. Seems like they made it efficient for the number of max cores and not the ultra cores. Not a bad thing. It just means that all the blocks are pretty much parsed out before it needs the extra cores unless there are A LOT of blocks that need to be processed. There is actually processing lag if you optimize a chip to break data into smaller blocks to pass to more cores because A. You don't really know how much time a black takes and you may end up having some cores wait too long for a single demanding block to finish, or B. There is data assembly lag.. meaning, the time it takes to reassemble the blocks of 12 cores is meaningfully less than the time it take to reassemble the blocks sent to 24 cores. More block to more cores may actually be slower than less blocks to fewer cores for smaller black counts. I am thinking they looked at the average potential block size for tasks apple customers will likely have... Also, the weakness of the way the chips are attached means you cannot change block processing parameters. It's hardware fixed in Silicon. When you glue 2 chips together, you get what you get. There is a controller that manages traffic to and from a core. Differences in this controller faced the same issues on Intel and AMD, who use slightly different hard coded core allocation for different reasons making one better for large block processing and others for smaller block processing. Hope that makes sense. I tried not to get too technical.
Was it not linear on graphics?
The analogy I like to use in cases like this is the challenge of keeping a V12 engine fueled. As the engine needs more fuel for higher RPMs, you need to increase the fuel pressure to ensure that all the cylinders get a full charge. If the system is unable to pump enough fuel from the tank, bubbles form in the system, or fuel pressure is variable, then you're going to see a loss of power. Possibly even problems like sputtering where the engine will run *worse* at higher RPMs.
It's the same problem with multi-core issues. If you can't keep a steady stream of blocks to be processed, the CPU will sputter and performance will fall off a cliff.
HaHa Very funny. I bet Orange Works Inc. of Silicon Valley just luvs comments like this smashing their @@thewiirocks Apple Babies into AppleSauce! 😎😎
Thanks for enlightening us. Your comment serves a lot to make a final decision on purchasing.
I'm NOT an engineer :) but that the Ultra wasn't arithmetically faster than its smaller brother didn't surprise me either. There are always overhead and bottlenecks that get in the way of a perfectly proportional increase in performance. I remember in the 90s, when multiple CPUs became a thing, that performance was usually ~50% faster with the extra chip, never 100% faster.
Ok dudes. The Figma Ads are starting to become really obvious throughout all of your videos…. I mean at least switch out the tag line. Let it be known. Figma mention: Paid Ad.
Correct
Thank you for being the first person to test logic pro, been waiting a while!
Please do Music Production benchmarks on all of these!
Something I would be interested to see is to run all of the benchmarks side to side or progressively and find where does each processor starts to lag.
To expect double the performance is not necessarily realistic nor a good thing. I would expect the ultra to excel at parallel tasks running at the same time (because of the increase in core count) without dropping too much performance (probably because of power limitations). I would recommend something like the studio with the ultra in case you need to open several apps at the same time and work on heavy things in parallel.
Strange that you got the same results over the two machines for Davinci Resolve Noise Reduciton, Tyler Stalman clearly showcased that the M2 Max struggled to keep up with live / 24FPS playback, and the M2 Ultra was the only machine that was able to do this. Even citing the M1 Ultra not being able to keep up.
I think with a grade, noise reduction, and any other node elements, the Max would struggle big time
I’m still on M1 Pro and I make iPhone apps with Xcode. Having no issues at all. I say don’t upgrade until you noticeably start to feel a slow down or need the horsepower. Don’t upgrade just to do it.
I own Apple stock so do not listen to this person. BUY BUY BUY
Great video but while export specs for FCPX etc are great , I already know they're fast. What really interests me as an editor on all three of the major systems (FCPX, Resolve & Premiere) is how well do these actually handle editing... esp with effects like CC, text, plugins etc. Does it hang up or does it play smooth like butter. How does it handle codecs like XAVC-I etc? Do I still need to make proxies if I'm going to be working with multiple plug ins/effects etc. At this point, a fast export is a given and isn't as important as smooth reliable editing. Any thoughts/info? Thanks for all the info!!!
In the Xcode benchmark the values are probably wrong. 0.67 and 1.09 are just 2 seconds of difference.
I was wondering about the Xcode times also. I'm guessing it should be 1:07 and 1:09 minutes.
@@a4e69636b idk, if so the difference would be meaningless
Guess the Xcode benchmark should be 1:09 and 0:47? (Not 0:67)
This is pretty much the only one I am interested in, so got a bit confused 😊
yeah I am confused by that too. I wonder if it meant 109 seconds vs 67 seconds? That would be a significant difference. But if they meant 0:67 as in 1:07, then that's not a meaningful difference at all.
I've never heard you say that playing back 8K footage from the RED camera was an issue on any Mac silicone base system previously - but you stated it was *not* an issue here. I know for a fact that 8K canon footage chokes my MacBook Pro M1 system (16 inch). I would really like to know how these systems handle that footage.
Most of the things he says are biased towards apple, which makes comparing difficult
@@ahyaan2552 Max stands accused of some faint praise and quite a lot of whinging....not exactly biased towards Apple
RED's codec is very different than the awful, power hungry codec Canon uses for their 8K files. R3D files run very smooth comparatively.
This specific case is color graded 8K Red in a full 8K timeline at full-quality playback. The M2 Ultra also plays back 8K Canon RAW with a lut as well!
@@MaxTechOfficial Would love to see a YT Short showing that 👀 My guess is more ppl are dealing with R5 8k footage than RED, given that it's a prosumer camera.
It would be interesting to throw in a decent spec i9/4090 PC in to the mix. Ultra for Logic sounds great but, man, the cost for the performance is difficult to justify.
A Mac Studio could easily fit in your backpack, and take it anywhere as you go. Good luck doing that with your RTX 4090.
@@rigolettofranc It's a desktop machine. For use in a desktop situation. ie it "should" be about performance 1st. Mobility is for laptops.
Would love to see an m2 max vs m1 max base review please, Max and Vadim!
Is the binned vs unbinned M2 Ultra video out yet? I've been desparately waiting to watch it as I need to decide ASAP. Could you please tell me at least when to expect it be out, sir? That would be awesome, thanks 🙏🙏🙏🙏🙏🎉❤
Yes that is the comparison I'm also looking forward to too!
Great Video as always. But I don’t quite agree with your final verdict . Have a look in timeline performance with user cache or smart cache on in DaVinci Resolve. Many clips with effects and color grading are constantly rendered in background. This is significantly faster with ultra. Timeline performance is more important than export times.
Is that so with after effects, too?
@@ekarron4646I used to use it a few years ago and it didn’t, but it might of changed since it’s been like 4 years since I last used it
I would also like to add that you compare 512 gb vs 1 tb. Every professional would at least buy 1 tb. If I compare the configurations with prices in Germany you have 3089 Euros vs 4800 Euros. Yea it is more expensive but far away from being double the price. This is click bait title.
I mean, 8 displays is a good start, but WTF with the 5? Who uses only 5 displays these days, does Apple think its 1990 or what? Im holding off until the 18 displays connection.
Would be interested in seeing the unbinned M2 Max vs the binned M2 Ultra. Does having 38 GPU cores ever beat having 30+30 GPU cores? How close is it in some of these tests?
I'd like to see a test where they compare all four of the M2 Studio versions
@@PerNormanSpygg NEVER - we have to spend the next 6 months going over video by video until the content is dry
Out of curiosity, would love to see these two machines against the M1 Max Mac Studio and the M1 Max Ultra Mac Studio.
Thanks again, always find these interesting. Blessings
I'd love to see a poll of your viewers to see how many are video editors vs designers vs software engineers. I'm glad you added Xcode Benchmark back in but I wonder whether you'd find that the % of software engineers would make it worth showing more programming tests.
I am still eyeing the M1 Max. And how many people really need 8 displays.
MacBook Pro M2 Max vs Mac Studio M2 Max comparison and heat throttling analysis, please!
Coming from a 5 year old PC-my Ultra is a badass.
After a few projects it will be paid off, then time for the next one. It’s a tool for business
Watts and Temperature- higher is better? Don't think so :)
Nice comparison, thanks. However, it really does depend on what you use the Studio for. I've found that Fusion in DaVinci Resolve is still sluggish on the base M2 Max. Export times aren't an issue. But real-time editing and compositing was underwhelming. Even rendering compositions in place still happens at below 10fps...just like my M1 Mini! If you want super smooth compositing the base M2 Max doesn't cut it...even with proxies etc. enabled.
Did you try that on an M2 Ultra?(:
@@NirHason No, I didn't try an M2 Ultra....double the price. I've since settled on an M2 Max with 64GB RAM. The extra memory has helped a lot. Basically, Resolve seems to soak up as much RAM as you can give it, at least in my experience.
I hope the benchmarks and thermals was done BEFORE tearing it apart. Just to get a feeling on the thermal paste that come from stock. That said. I thank the fact that you showed the processors size difference.
Regarding your resolve noise reduction test. How about making a 60fps timeline, or even 120. Just to see where the hardware chokes out and not both being 24 frames. It would be more clear for those editing 30 or even 60 fps content.
So is the Xcode benchmark difference 2 seconds? Because the chart had 0:67 and 1:09 which is confusing
Great comparison video, would be great if you can include on your tests Topaz Video AI and compare if makes any difference the number of neural cores on different setups. Cheers!
Anyone who expects perfect scaling, or anything near that, simply doesn’t understand physics or the laws of computing. The fact that the graphics are nearly perfect is incredible. Stick to tear downs, please. Don’t give people advice. Your criteria are based on no professional workflows or models, and barely possess an understanding of computing.
Love your channel. To answer your question; It is if you need it, it isn't if you don't. That said I'd add the 1TB SSD upgrade to the Max. I'd also add, for those on the fence, the next M3 Max will outperform or be close to the M2 Ultra. In the words of Kurt Vonnegut, "So it goes." Disclosure, owner of several Apple Dealerships.
My old MacBook Pro with the M1 Pro chip has a similarly fast SSD as the M2 Ultra. I'm glad to have it.
CPU performance doesn't scale linearly for a single workload. It's odd that MaxTech would expect them to. GPU scores should scale in a much more linear way though.
I do get that the you used the 2 barebone versions of the Products... but to compare it "Apples to Apples" it would have made sense to have an upgraded version as well of the Pro chip where it matches the Ram and the SSD to see what diffrence those 2 components make and what is the diffrence in the CPU itself. Like this there are too many variables
On CPU multicore performance, you need something is embarrassingly parallel workloads. These things are expected to be use by a professional so use software like them and they would likely have software to ingest a lot of those cores. My field uses such multicore usage. I'm not a radio astronomer guy but you can use AIPS or CASA to determine the multicore workload.
And also, again, you need to use the Moana scene for GPU testing. That scene is large and heavy enough to see if there is any difference.
wanna see M2 Ultra Vs PC with RTX 4080 / Core i9 / 64gb ram. )
xcode benchmark slide is weird.. 0min 67 sec ??? doesn't add up.
Insane to be able to keep these processor at such low temperature, with idle fans 😮
Fantastic Video! I'm coming here as a newly full time music & content creator looking to take his business to the next level **woot woot** and this video helped A TON!!!!! Even though I am far from a tech geek, I found that I was able to follow along with a lot of the topics discussed here. Especially the section on Logic, I don't see me or my clients budget being able to afford a 300+ track composition lmao! Also the comparison for video was very easily digestible in terms of seeing why the extra money wasn't worth it!
So thank you SO MUCH for a great video
I might get a M3 Ultra when it's available. 😄
When you upgrade the storage and RAM on the Max to match the Ultra, the price does get closer and to the point where I should rather bite the hard bullet and get the stock Ultra.
M3 Ultra will probably arrive on fall 2024 as earliest. If your current Mac is still good enough then it's worth waiting for. But that's a so good question: go for M2 today or wait for M3? If it's urgent with a new Mac then you got no choice... just buy something that's available. The Mac is honestly only a tool. But a very nice one :)
@@cjeelde I have the stock M1 Max, I am not in rush to have another computer.
Yeah then it's no need for M2. M1 Max will stay strong for a few more years. If I were you I would wait for that M3 Max/Ultra or M4 Max/Ultra. Depending on the money situation and the needs for more performance :)
@@cjeelde My previous "serious" machine was the trash can and it lasted 7 years. It was the longest serving Mac I have ever used. Got a spec-bumped M1 mini as a stopgap because the trash can started to have hardware issues. When the Mac Studio came out I got it immediately as it's pretty much the exact type of machine I was looking for. Don't need PCI slots but need a machine fast enough and letting me choose my own monitor(s).
I would have liked the 2200$ vs 4000$ max studios, to see if it’s worth 1800$ more for ultra
I would really like it if you have a virtual machine or two running while you do some of these tests. A lot of developers have a virtual machine for a server or iOS simulator running while they develop. Would be great to know if those affect the performance between the max and ultra.
How much power do I neeeeeeed? ALL of it! Bring forth the M3 Ultra!
Thanks for this. I got the Max and was wondering how the Ultra would compare for the work that I do (Adobe PS, AE and PR) . This put my mind at rest that I made the right call and don't need the Ultra.
If you use Ae, can you give us your comments? I'm thinking of buying a Max Studio but I haven't decided yet, I mainly use Ae
Mac Studio M2 Ultra host internal 370W power supply, the maximum power of this machine is 295W, when the full-speed operation is only 75W more, it can supply power for 6 Thunderbolt 4 and 2 USB3. It also needs 90W, and the frequency needs to be reduced to accommodate this situation? The price is double, but many machine hardware does not have double
Just for giggles, I decided to do the spedometer test on my iPhone 14 pro and got a score of 393, not too far off either of those machines. Not sure what the a16 chip has over some other PC chips, but it's interesting that my phone out preformed a solid AMD Threadripper with 64 gig of ram.
I recently bought M2 Max studio 2023 , I used it for editing in Final Cut Pro and is working well for me. The cpu is definitely a good buy from my review. I added external memory too.
I don't get the ssd speeds on the M2 Max version.... My base M1 Max MacBook Pro gets 6-7mbs read write on its 512gb ssd... Why 2 years later do they go backwards?
Apple now uses 256gb nand chips
Instead of the 128gb ones nand chips for the m1 family
probably costs
Very complete review, guys! You should make a puget benchmark for After Effects.
I got the most load on my M2 Ultra with running two conversions in Topaz Video AI and one Handbrake conversion at the same time. Then you hear the fans running (not loud but really blowing).
Is the M2 Ultra much faster than M2 Max for Topaz Video AI processing ?
@@delcanvideoproductions4895 yes it scales much better than the M1 Ultra. It does not reach factor 2 but at least 1.8 but this depends on the scale and algorithm. Best are 2x or 4x scaling Artemis and Proteus scale well.
@@stargator4945 Thanks.
I'm at that decision trying to replace a 2018 iMac Pro which has always been terrible video machine for me. I edit in Premiere, nothing super crazy. This video is very helpful.
Is there a binned vs unbinned test coming for the M2 Ultra and the M2 Max?
@Maxtech where's the video on the logic Pro benchmark with the Mac Studio M2 ultra?
Question :
Can you link these together, like have a Max connected to an Ultra.
Amazing, I spent hours searching for this exact information. I was already leaning toward the M2 Max anyway, but I almost always buy top of the line, because I don't buy often. I really want an M4, but I need something now (my 2017 imac pro is showing its age and had an odd failure the other day). Now I just need to decide if bumping up the memory, GPU, and hard drive on the M2 max will be getting so close to the base M2 ultra model that I might as well go to the ultra for the price difference.
The Xcode times were strange. Is it fair to say that the 0:67 timing is the same as saying 1:07 ? That would show the times for both systems to be very close, correct? Thanks
Finally! I've been waiting for this video for a week!
at 4:53 it says 0:67 and i think you meant 0:57 😁😁😁😁😁😁😁😁😁
I have doubts about the Logic Pro benchmark, are you sure you ran both machines on a 1024 buffer as the instructions suggest? It seems crazy to see such a difference and also that my basic M1 Pro mbp does more tracks that M2 max Mac Studio. Could you please check? Thanks
How many seconds you have in a minute at 4:54?
Hey Max, always appreciate your reviews! Did you test Canon RAW 8K? That’s what we shoot. It’d be great to find out how these two Macs handle it. Not having to create proxies would be amazing. Thanks 🙏
I love your videos, but could you please update the video export section of your videos. At the speed you’re getting now for example 4K five minute video exporting in less than a minute. Unless for whatever reason, the next generation of tip chips takes longer than that this does not matter anymore to most people. I’d much rather be interested in what a 10 minute or average eight minute or 15 minute 4K video takes to export. On top of that, would I really want to see is HEVC 10 bit HLG or HDRPQ export times. Those are still taking much longer than an hour to export on M1 Pro for an eight minute video.
I know you can’t test everything but it would be great to test a really heavy photoshop illustration and really heavy indesign file with a lot of spreads. I can chew through a tone of memory using indesign and photoshop in tandem.
You can’t expect linear increase especially with multi-core, most it depends on the software, just the fact some software achieve a 80-90% increase on the Ultra is incredible
Best reviewer I’ve seen on RUclips, and you saved me some money. Thank you!
can you do a comparison between M1 ULTRA vs M2 Max
Hey Max, could you tell me how you were able to get the BMW test for blender? I've looked _everywhere_ to download this to test my Mac, and I can't find it!
Great video!
Thank you for clearing my doubts. I appreciate very good content. For software development (backend, iOS, Docker) and video editing cheaper M2 Max with 2TB SSD disc and upgraded RAM up to 64GB will be good enough.
Is it worth it for professional photo editing? I understand the export is faster but what about the editing itself with Lightroom?
I made the same decision when I purchased my M1 Studio Max a year ago March, the performance difference just wasn't worth the extra $2K.
Thanks Max for another brilliant video on the new M2 Mac Studio. Would love to see some comparisons between the previous gen M1 Ultra Mac Studio with M2 Max Mac Studio. From Apple's charts on the performance comparison they show that M1 Ultra outperforms M2 Max in some areas, and some people may be eyeing M1 Ultra Mac Studio in the refurbished/2nd-hand markets
I’d love that comparison, too. I’m
exactly at that decision point and don’t know what would make more sense
I see how you snuck that “ nice “ in there 4:26
What tests are Neural Engine bound? It would be interesting to see how the NE performance scales.
Whether something takes 4 minutes to export or 2 minutes to export just doesn't really matter. Unless you just care about specs, and not on actual workflow and time 'waiting' for your computer.
M2Max 12CPU,38GPU,1TB,32GB Enough for editing 8K not very complicated?
I'd like to see M2 Ultra vs M1 Ultra direct comparison (unless you've done it already and I missed it).
Thanks as always for such excellent content! I was really surprised about idle temps compared between the Ultra and the Max! I fully expected that the copper heat sink on the Ultra would be a bigger help. Instead, it seems like it's actually NECESSARY just to keep temps in check, due to how much more energy the Ultra uses and heat that it generates, than the Max.
The only thing for me that never gets addressed by any of these reviews - on ANY channel - is how various Macs compare with respect to flight simulation, especially a flight sim setup using multiple monitors, which, because of the unique, dynamic content on each screen, adds lots of extra demands on the computer.
For pretty much all of personal computation history, Apple has lagged behind PCs with respect to the graphics needs of flight sim. Consequently, no serious flight simmer uses Mac. Up until recently, that was not a problem because there was no serious entry from Apple that was worth consideration.
The change came with Apple's ARM chips, at which time, rather than tackling performance by just throwing more and more horsepower at the problem (more powerful chips, meaning more bulk, heat, & energy consumption) Apple finally approached the problem from a different angle... efficiency.
Suddenly, Macs became - to paraphrase an "On the Waterfront" line - "a contenduh!" However, here's the problem.
For all of history, we've had PC experts and Mac experts, based on the computational skills they cared about. For that reason, Flight Sim experts are quite knowledgeable on PCs as well as the comparisons between PC performance levels and how that translates into Flight Sim skills.
Meanwhile, Mac experts know a ton about relative Mac performance levels, especially with respect to graphics skills when used for photo and video editing.
However, NOBODY (seemingly) who is a Mac expert, knows anything about Flight Sim needs because Mac doesn't have the history of performing well with flight sim programs.
These 2 paradigms have diverged so totally over the years, that I'm despairing of finding a flight sim expert who knows enough and is interested enough in flight sim application, to do a test like the one you did in this video, BUT... with specific emphasis on its application to flight sim work. (Yes, "work". Sure, there are flight simmers who do it more as a gaming activity, but there are also plenty of flight simmers who use it very seriously to augment their flight training for their pilot's license and additional certifications.)
For folks lucky enough to be able to afford to spend $2k to $4k on a PC dedicated to flight sim purposes, plus the $2k to $4k on additional dedicated monitors for a flight sim station, AND have the extra space for such a flight sim station, and then doubling that cost for a "regular" personal/work setup of a Mac-based work station... NO PROBLEM !! Just buy 2 completely separate systems, one PC-based with 3+ monitors and another Mac-based 3+ monitor work station for personal/work purposes.
Of course there's another option. Throw out all my Apple products, iPhone, iPad MacBook, etc, and convert EVERYTHING to PC, including all my calendar entries, address book entries, etc (need I go on?) so that I can have just a single, PC-based work station. Ugh!!
For the rest of us poor sods that don't have the budget nor the space for 2 totally different work stations and don't want to do a full-scale conversion from an Apple ecosystem over to a PC/Android ecosystem... but ALSO care about quality flight sim... there seems to be no light at the end of the tunnel as to finding some who is both a Mac expert AND a flight sim expert.
Any thoughts? (My sincere apologies for the long comment! I know, I know. My problem cannot and SHOULD not be your problem nor the problem of any of your subscribers. Just not sure where to turn. If you made it this far... THANK YOU!!!
Software has to be optimized. If so, you get up to 2x speed on the Ultra. Which justifies the price. Which is double.
Plus, longevity. You can technically have it twice as long.
Spend the 2000 dollars, and in two to three years, when you upgrade it spending another 2 grand, that will outperform the one you spent 4 grand on just a couple years earlier.
You just saved me a lot of money! I appreciate you!
Max, you have a xCode performance "thing", it is time to have a ML performance "thing", maybe an "Apple Core ML demo run between the different Mac Studio/MacPro models??
I think you accidentally said M1 instead of M2 around 1:08. Apologies if this has already been pointed out.
why is it only in 360p
we arrived too early, leave youtube some time to process the video at higher quality... i bet in 5 minutes if you come back to the video it will be in full quality
I'll be upgrading from a mid 2012 Pro. I'm thinking that the Mac Max should be just fine. Thanks for the video!
what is gaming performance ?
I have to stick with reality and stay with the Mac Studio M2 Max with 1 TB. I need to use the extra $1000 for the 14 inch iPad Pro which for me would be a more useful combination for what I do. So Mac Studio M2 Max 1 TB for $2400+iPad Pro 14 inch for about $2000 makes more sense than spending $4000 for the Mac Studio M2 Ultra. Thing is, what about when the Mac Studio M3 Max come out. Probably in 2025 so can't wait that long. So, now just need Apple to get the iPad Pro 14 inch out. Right now I multiply the M2 benchmarks by 1.2 to see what the M3 will be.
I wonder if you could get a sense of how energy efficient a Mac is by the thickness of its power cable. I was thinking the magsafe ones compared to the one pictured in this video.
Right now, im in a point of my life of opening my eyes once again, and perceiving clearly the true nature of others, and destroying those bridges that no longer serves me and burning those fake or draining connections in my life.
The fact that doubling the CPU or GPU cores doesn't double the performance when having demanding tasks is a big no for me and it kills apple's marketing
How much difference will they make in Topaz Video AI for 24->120fps Chronos?
Same. I want to see if 192 gigs of ram can be utilized by topaz for upscaling
Yes that would be really helpful. Is the M2 Ultra going to speed up Topaz Video AI processing significantly ? Or does this depend on the AI model chosen and the processing settings - do some Topaz AI models use multi core / graphics more than others ?
thanks for the comparison... although lol, "double performance" is way too high a bar to set! 42% is a high boost. Plus, there are many more tiers of the Ultra - up to 192 GB, and 76 GPU's. I'm sure it gets there. But... yea double the money doesn't mean double the performance, that's kinda weird to expect, it just doesn't happen, we already know how it takes like 10 years to get a computer to be twice as fast.
When looking at SSD performance there's more than just that "read & write". There's IOPS and 4K random read/write for example and we never see those numbers.
These "read & write" speeds from BlackMagic can be compared with normal classic HDDs that can reach maybe 50-150 MB/s. So 2-3 GB/s or more for a SSD is really really fast! We need those great speeds for raw footage and big MP4 files.
I'm curious about the future for SSDs. That 3D Xpoint from Intel and Micron got announced in 2015 but in 2021 and 2022 both companies discontinued it. I remember that announcement speech in 2015 and it was so promising but now 8 years later only a few early adopters gave it a chance. Apple never used it for the Macs. Apple probably tested it for evaluation, but we can never know.
The BIG jump in performance was switching from HDD to SSD. Now it's incremental performance boosts.
Now keep in mind that a DVD movie backuped for a DVD-R is up to 4,7 GB. These better SSDs can read and write that in less than 1 second! But if you copy that from internal fast SSD to an external SSD that will take longer because of the speed for the external SSD. 20 years ago with those hard drives it would probably take an entire minute to copy that between two fast HDDs.
hey thanks for the video , would be great if you could do some machine learning or deep learning tests
What about running virtual machines? What would be the delta between Max & Ultra?
A comparison between M2 Ultra binned vs unbinned would help a lot of us here... Please??!
Do you think the ultra will get longer software support from apple or do you expect them to discontinue support based on chip generation?
Wow Moore's Law: 6:15 2x the speed for less than 1/4 the price -- only what 8 years later? 2/(1/4)/(1/4) = 16x performance/$ in 8yrs = 2x performance/$/year.