I messed up my math and corrected it I guess while you all were replying. My apologies. Need to get some sleep.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M3 Chip Generation - Discussion Megathread
- Thread starter scottrichardson
- Start date
- Sort by reaction score
I think the P-AMX is new while the E-AMX seems to be the same.
AMX is a lot of logic. So with the new unit having much higher transitor density it should be a lot more capable.
M1/M2: From the perspective of one core the L2 is partitioned into several zones with different latency. We will have to wait for an in depth analysis how the new configuration behaves.
Which video encoding codec normally work in or need to be able to work in "any codec a client throws at me"?
If working almost exclusively in the video codec that are supported in both the Pro and Max then the CPU/GPU count differential isn't as much of a differentiator. So for example ProRes/ProRes RAW , H.265 , H.264 then they are more equal. Also if current 2018 QuickSync was doing a very good en/deocode job with your video then either one of these is likely better.
However, if it is RedRAW or something Apple doesn't cover in hardware then more CPU/GPU cores help with the overhead. So Max would have an edge. Editing anything and everything (past and future formats) the specialized fixed function hardware may not match.
If working with a high number of input streams of ProRes RAW the Max has more ProRes en/decoders than the Pro. Editing what is largely a single camera feed input and/or snippet B camera clips, there isn't as more uplift for a Max.
Similar with delivering to output formats that Apple doesn't directly support in fixed function hardware. (although those may be less time critical tasks. )
The M3 Pro and take the 'extra' money and apply it to more RAM , I suspect is a significant step up from a 2018 laptop. (or base model iMac). Pretty good chance you'll want more memory than you have now. ( some folks spin a notion that M-series needs less memory that doesn't ring true. ) So not necessarily any cheaper than the entry Max options.
I think things have gotten to a point that most of the Resolve effects are GPU based. There is an upswing in stuff leaning on the NPU somewhat. There the Max has more leverage on the GPU side and on doing a blended, concurrent mix of both NPU+GPU effects.
I think I’m normally in h.264 or h.265 but this is the part I worry about. Like am I screwing myself over with my visual effects experience if I go with Pro?
Need a P core to run a compiler ? Especially if the source code compiling is latent on storage drive before process assigned to the core starts.
End stage linkage of a set of binaries gets back to single threaded.
On my M1 Pro cpu utilization appears to max out all 10 cores in brief bursts as I compile large projects… yes it doesn’t stay maxed out, it is bursty, but having fewer p cores and more e cores would only improve performance if there were a lot more e cores. Simply trading them out 1 for 1 wouldn’t be to my benefit.
I am a Japanese person who is not sleep deprived because I live on the other side of the earth.
Floor plans for connecting two or four M3 Max are being discussed, but the Mac Pro and Mac Studio, which are shipped in small numbers, are not designed with the expensive N3B, and the N3E is a cheap one, Is there any possibility of designing with Extreme/Ultra?
By the way, the 14-inch M3 Pro 11Core 2TB that I want costs $2,599, but since the yen is weak in Japan right now, it costs the equivalent of $3,755, which is very expensive!
Floor plans for connecting two or four M3 Max are being discussed, but the Mac Pro and Mac Studio, which are shipped in small numbers, are not designed with the expensive N3B, and the N3E is a cheap one, Is there any possibility of designing with Extreme/Ultra?
By the way, the 14-inch M3 Pro 11Core 2TB that I want costs $2,599, but since the yen is weak in Japan right now, it costs the equivalent of $3,755, which is very expensive!
Last edited:
For the specific effects you heavily rely on there may be documentation that highlights whether they are CPU or GPU based. I wouldn't put it into the 'screwing myself' category. It should work better than what you have now ( 2018 MBP 15" ranged from Radeon Pro 500 series. Pro GPU puts those in the rear view mirror on compute tasks with decently optimized, native code.
Back in M1 era some folks with lots of effects layered on a clip would point to the M1 Max as being a better option than the Pro. The M3 Pro has crept up into the range that the old M1 Max used to benchmark at. ( better hardware and now better optimized code as well. )
If the numbers work on getting reasonable return on investment to cover a Max then it is more flexible. You get 'more' but pay a lot more also. The M3 Pro 14" 36GB is 2,799 the M3 entry Max with same memory is $3,199 $400 isn't a big gap ( 14%). For the same 36GB of RAM the Max version gives you more bandwidth for GPU compute (along with the more cores). if not at the 'redline' of your budget limit it is probably worth the additional $400.
If you need more than 36GB then M3 Pro isn't really an option (that is where it stops).
N3E is only 'cheap' relative to N3B. It is still more expensive than N5/N4.
The whole notion of forcing the monolithic Mn Max die into playing the two/four game is to save money on the design reusing an mostly completely paid for part from another product. Going to from N3B to N3E is expensive. It only makes sense if have significantly high volume to sell.
The "small numbers" problem can't really be solved by going to a different, cheaper fab process. It is still there because would need new layout , new dies , and all the other overhead. Overhead is what is hard to 'pay for' (amortize) with just a relatively small number of dies sold. (low system unit numbers.)
I’m between the
$3239 (edu)
MBP 16” M3 Pro 12CPU/18GPU 36GB Mem with 2TB SSD
Or
$3559 (edu)
MBP 16” M3 Pro 14CPU/30GPU 36GB Mem with 2TB SSD
I use effects like film emulation and noise reduction and it drags my current laptop down to 2 fps even with a 1080p proxy. I would LOVE to get mostly-smooth playback with effects on native 4K files.
My videos tend to be less than 30 min and two or three layers of clips in some places
For 1080p there is pretty good chance wouldn't need a proxy for either one of those.
[ this was a M1 in context of new M2 . Not really pitching discount M2's here. More so that the M1 Max quite capable and M3 Pro is in the same range. (and M3 Max is 'gravy on top'. Probably has longer useful service lifetime over incrementally increasing workload. ) ]
'or three' is another nod in direction of the Max.
We still don't know it's flat. The calculations you and @altaic did don't take into account scaling issues. I would expect the M3 generation to be at least a little better than the M2 for that, though we'll need to see what the benchmarks say.
Even if it benches flat, due to Amdahl's law I'd expect compilation to still be better on the M3.
You're confusing chips and laptops.
The M3 Pro is just fine. Not what I would have done, but you can make a good argument for segmenting things the way they are.
The MB Pro with M3 Pro is a different issue. MP performance on it (likely either flat as you suggest, or up just a little, compared to the previous gen) will not be a huge step up. But it will come with a little more memory and a slightly better price (I think, I don't recall for sure). With the better screen, slightly better battery performance (for iso-weight workloads, due to N3), slightly more RAM, and notably better SC performance, it's not a bad step up, just a bit less than I would have liked. (It has other minor improvements too, like AV1 hw decode, etc.)
I'm not sure what you're trying to say, but everything that was announced is on N3B. There's no particular reason to think the Studio/Pro will be on N3E, except... hm, I'll post about that separately.
Is it? My vague recollection is that N3B is the first time that you don't save money on transistors, but I don't think the silicon is more expensive. (That may not be true for wire-heavy stuff, like a chiplet entirely for static RAM.) I think N3E is a little cheaper (unless you go for the highest-performance transistors; it gives you a wide variety of choices).
This really isn't the thread for that sort of question. But that said, the smart thing to do is organize yourself so you can test something as soon as you buy it. Then buy the Pro, and see if it handles what you want well enough. If not, you have two weeks to return it (at least - if you return it to buy a Max they'll probably let you slide more than two weeks) and buy the low-end Max instead.
Of course, if you're looking to buy something for work, I would expect work demands to grow over time. If you want to buy something that lasts a good long while, I think it'd likely be worth getting a Max. And if you need to save, think about your plans for external storage (you're going to need a bunch) and consider if you can comfortably move down to 1TB internal storage.
no.
you have no idea what you're talking about.
none whatsoever.
there have been many, 10,20,30x comprehensive reviews on YouTube where people run exhaustive tasks to try and get the m1 and m2 chips to throttle under load. we are talking about people running 10 YouTube videos simultaneously while scrolling through 4x 40MP images in adobe. the gist being: it is near impossible for a normal user to be able to discern the difference between m1 and m2. you are far more likely to be vram and ram bound.
laughable to even use 3d rendering as an argument. what subset of users purchasing these products do you think consider 3d rendering capabilities? at this point, you dont need anyone to give you advice if thats the case.
sweet spot: now I know you're bs. the sweet spot is the highest specced CPU that will cost 3-4k?
I only use 1080p proxies because anything more is rough on my current MBP. I actually am trying to work in 4K and export in 4K, I just currently can’t edit in 4K.
@sv8 wondered about the Ultra (and possible quad-chip M3) being on N3E. Their reasoning was incorrect, but I noticed something just a little while ago that bears on that.
Looking at the Max's die shot, where is UltraFusion?? Did they just hide it (IIRC they did in some of the early pictures of the M1 Max)? If so it would have to be at the bottom, under the GPUs and SLC/RAM controllers, which is reasonable.
But what if there's another explanation?
They've already changed their pattern (if two generations makes a pattern) - the Pro and Max are clearly quite different from each other, not sharing nearly as much of their design as previous generations. Perhaps the chip that will be the basis of the Ultra is NOT the Max, this generation. Maybe there's no Ultrafusion connector in the M3 Max die shot because it doesn't exist.
This would be pretty surprising - I wouldn't expect them to do a whole new floorplan for chips making up an Ultra, as it would seem to be way too low volume for that. But... They obviously know a lot more than I do, maybe they see a good reason to do this. Maybe they count it as another learning step towards a future full of high-density chiplet interconnects, and therefore worthwhile just for that.
Maybe they need to do more work because they want it to go 4-way as well as 2-way?
Anyway, N3E vs. N3B is a sideshow for this, but: If they have to design M3-class CPU, GPU, and NPU cores for N3E anyway (for the A18, and maybe an M3+?), AND they have to do a separate design for M3 Ultra+ chips, maybe the added cost of redoing the entire M3 Ultra chip on N3E doesn't seem so crazy to them.
If this entire chain of speculation is correct, then we could see N3E M3 Ultras in 2024. Otherwise, it's a near certainty that we'll see M3 Ultras on N3B.
Looking at the Max's die shot, where is UltraFusion?? Did they just hide it (IIRC they did in some of the early pictures of the M1 Max)? If so it would have to be at the bottom, under the GPUs and SLC/RAM controllers, which is reasonable.
But what if there's another explanation?
They've already changed their pattern (if two generations makes a pattern) - the Pro and Max are clearly quite different from each other, not sharing nearly as much of their design as previous generations. Perhaps the chip that will be the basis of the Ultra is NOT the Max, this generation. Maybe there's no Ultrafusion connector in the M3 Max die shot because it doesn't exist.
This would be pretty surprising - I wouldn't expect them to do a whole new floorplan for chips making up an Ultra, as it would seem to be way too low volume for that. But... They obviously know a lot more than I do, maybe they see a good reason to do this. Maybe they count it as another learning step towards a future full of high-density chiplet interconnects, and therefore worthwhile just for that.
Maybe they need to do more work because they want it to go 4-way as well as 2-way?
Anyway, N3E vs. N3B is a sideshow for this, but: If they have to design M3-class CPU, GPU, and NPU cores for N3E anyway (for the A18, and maybe an M3+?), AND they have to do a separate design for M3 Ultra+ chips, maybe the added cost of redoing the entire M3 Ultra chip on N3E doesn't seem so crazy to them.
If this entire chain of speculation is correct, then we could see N3E M3 Ultras in 2024. Otherwise, it's a near certainty that we'll see M3 Ultras on N3B.
I'm not taking a side on the whole argument you're having, but it's pretty funny you saying he has no idea what he's talking about when you entirely botched your claim that the M3 Pro SoC doesn't have TB4 and can't attach multiple external displays. (Pro tip: It does and it can.)
I think you are underestimating the performance of the e-cores. I remember reading something before the M1 Macs came out (after the processor was announced) that e-cores fully pushed to it's limit was around 70% as performant as the performance cores... I calculated a guess at what the multicore geekbench benchmark would be for that computer based on everything that was out at the time and using that 70%.... and that geekbench benchmark was actually very close. I highly doubt the e-core is only 1/4 of a p-core. Also you have to factor in the fact that one p-core operating is going to perform at 100%, but when you get up to many p-cores they are not going to run as it was a single core... at a certain point those performance differences between more e-cores and p-cores will narrow in performance, while the e-core will still be just as efficient as it was for the first one. The equation is much more complicated, which is why you hear Apple having multiple different versions fabbed and tested before deciding.
M3 Pro Has Fewer Performance Cores, Lower Memory Bandwidth Than Previous-Generation M2 Pro, Indicating A Notable Downgrade
The M3 Pro might be fabricated on the cutting-edge 3nm process, but in specifications, Apple has downgraded the SoC compared to the M2 Pro
wccftech.com
I don't buy Apple cut cores and bandwidth to save money.
I think these chips are not performing as expected and Apple went into damage control trying to figure out what to cut to reduce thermal throttling. The GPU is a huge upgrade and that is going to add a lot of heat and battery draw. The 3nm process is not as efficient as they thought.
Obviously with less P cores and still getting a 15% lift shows that the 3nm process is good but they need to optimize the process.
Next year the M4 will probably have more P cores, more bandwidth and have a better GPU with better efficiency and better performance.
Sometimes a new process or denser process can be better and sometimes it has issues. I think Apple has issues with TSMC first gen 3nm. But this all just a big guess on my part.
Last edited:
I agree the e cores are more powerful than people think. That being said if Apple was going to cut the P cores they should have added an additional two e cores for a 14 core setup. Then we would have had better multi core performance and better battery life as things that can be done using more e cores than using the P cores would save battery with no drop in performance. I think Apple hit a thermal envelope wall with the new GPU and they couldn't add more cores or too many P cores due to thermal constraints due to the new 3nm process on TSMC.
Thank you, confused-user-san !
I couldn't explain it because English was difficult for me, but you spoken for all my guesses for me.
I have a feeling that the N3E will made to the A18 and a cheaper version of the M3 (e.g. M3E) for the MacBook Air.
So I thought there might be a possibility of designing Ultra/Extreme with N3E as well.
I couldn't explain it because English was difficult for me, but you spoken for all my guesses for me.
I have a feeling that the N3E will made to the A18 and a cheaper version of the M3 (e.g. M3E) for the MacBook Air.
So I thought there might be a possibility of designing Ultra/Extreme with N3E as well.
Last edited:
This could explain why Apple reduced the memory bandwidth. So maybe we shouldn't expect worse perfromance.
Chips and Cheese
Apple’s recently announced M3 Pro carries forward M2 Pro’s design goals, but memory bandwidth has been reduced to 150 GB/s. The M3 Pro can be configured with 18 GB or 36 GB of DRAM, compared to 16 or 32 GB on the M2 Pro. That divisible-by-3 number suggests Apple has switched to a 192-bit memory bus. In the discrete GPU world, the RX 6500 XT has just under 150 GB/s of bandwidth and slightly less FP32 throughput than the M2 Pro’s iGPU. M2 Pro’s very high DRAM bandwidth likely wasn’t helping most applications. A narrower, better utilized DRAM bus could save packaging and power costs.
Like Van Gogh and console chips, M2 Pro’s large memory bus aims to feed the GPU. Achieving the same 200 GB/s figure from the CPU side is not possible, but the CPU still benefits from it. 125 GB/s is a lot of bandwidth for eight cores, and is out of reach for desktop platforms.
Chips and Cheese
Apple’s recently announced M3 Pro carries forward M2 Pro’s design goals, but memory bandwidth has been reduced to 150 GB/s. The M3 Pro can be configured with 18 GB or 36 GB of DRAM, compared to 16 or 32 GB on the M2 Pro. That divisible-by-3 number suggests Apple has switched to a 192-bit memory bus. In the discrete GPU world, the RX 6500 XT has just under 150 GB/s of bandwidth and slightly less FP32 throughput than the M2 Pro’s iGPU. M2 Pro’s very high DRAM bandwidth likely wasn’t helping most applications. A narrower, better utilized DRAM bus could save packaging and power costs.
Like Van Gogh and console chips, M2 Pro’s large memory bus aims to feed the GPU. Achieving the same 200 GB/s figure from the CPU side is not possible, but the CPU still benefits from it. 125 GB/s is a lot of bandwidth for eight cores, and is out of reach for desktop platforms.
I purposefully avoided addressing scaling because their gripes were not about a specific task, and TBH I'm not interested in getting into the weeds about arbitrary hypotheticals.