MP 7,1 Deciding between MP7,1 vs 16" MBP

[Cellardoor6]

I'm a software engineer currently working on a mid 2015 2.5GHz 4-core 16GB MBP and it has been seriously struggling to keep up. I'd like to upgrade and I'm having trouble deciding between an entry-level MP and a maxed out MBP.

My situation is similar to this but my primary goal is to be able to run Xcode builds, have both iOS and Android simulators running (for a react-native project), package up electron builds, run anywhere from 5-20 nodeJS services (mixture of apis, microservices, and webpack servers), and have 2 4K monitors plugged in. I don't usually do all of this at the same time, but I frequently find myself needing to work on multiple projects at once and wanting to have one thing build in the background while making changes to another. My current MBP just can't do this, and it has caused a lot of frustration just sitting around waiting for one thing to build before moving on. Even web browsing is laggy during builds.

From what I've read, a new maxed-out 16" MBP should have no problem with 2 4K monitors and running Xcode builds, but my main concern is that all the tests and benchmarks out there are really focused on testing a single thing at a time. I need to run many intensive tasks at once. The processor should theoretically show more than double improvement in what I'm seeing now, but I'm worried that I'll eventually find myself in the same situation and it'll be screaming just to keep up.

I feel like the base MP with just an upgrade to the 12 core is more suited to my needs, and will have no problem satisfying my requirement to multitask builds and use multiple monitors. Portability isn't a huge issue for me, and if I go this route and need to work remotely with my current MBP I would just ssh into the MP to run services/edit code and have a local tunnel running to simulate a remote 'local' environment. I love the idea of being able to swap in a more powerful cpu down the road (once they're cheaper), and I realize I would be paying the extra ~3k over the MBP for this ability to expand/modify.

Any thoughts on all this? Cost also isn't a major issue (though I'd like to keep it below 10k) - I can easily justify paying more because programming is my livelihood and I would be using this machine all day, every day.

 

[ajem]

I'm in a similar situation, just placed an order for the Mac Pro. Here are my calculations.

1) I don't need portability: I have a separate laptop that I can ssh into the main desktop.
2) 16" MacBook Pro, as powerful as it is, gets noisy easily. I rocked a 2014 rMBP for many years without issues, but I've been having lots of issues with the newer models and I no longer trust the build quality of apple laptops.

Given 1) and 2), the only choices are the Minis, iMac (pro), and Mac Pro.
3) I don't need / like the display on iMacs. I use an external ultrawide.
4) The current Minis are underpowered for what I do.

All things considered, Mac Pro is the only choice for me if I were to purchase now. I can also wait for a Mini refresh so it'll have 8 cores at least, but given the expandability of Mac Pros & the fact that I'm not too cash constrained, I went with the Mac Pro.

 

[Cellardoor6]

@ajem Thanks for the response - I agree with your thoughts on the other available choices. What configuration did you go for on the MP?

 

[ajem]

@Cellardoor6

I went with 28 core, 1tb, otherwise everything at base.

90% of my currently daily needs can be fulfilled with a 16 core (I run statistical simulations). Occasionally I can benefit from more than 16 cores.

I went with 28 core because
1) Upgrading the CPU voids the warranty; that means no CPU upgrade for me until 3 years later.
2) Buying 16 core now, and then buying a 28 core down the line, doesn't seem to make financial sense. I can't imagine in 3 years the 28 core would drop significantly below 2k, if it's available at all. So the difference in cost is only 2-3k.
3) some of my current workflow benefits from more cores

I think if you only need 12 cores now, then ordering that and upgrading later makes sense.

I will add more ram and ssd later. Will not upgrade the graphics card since 580 is plenty powerful for my needs.

 

[deconstruct60]

I'm a software engineer currently working on a mid 2015 2.5GHz 4-core 16GB MBP and it has been seriously struggling to keep up. I'd like to upgrade and I'm having trouble deciding between an entry-level MP and a maxed out MBP.

My situation is similar to this but my primary goal is to be able to run Xcode builds, have both iOS and Android simulators running (for a react-native project), package up electron builds, run anywhere from 5-20 nodeJS services (mixture of apis, microservices, and webpack servers), and have 2 4K monitors plugged in. I don't usually do all of this at the same time, but I frequently find myself needing to work on multiple projects at once and wanting to have one thing build in the background while making changes to another. My current MBP just can't do this, and it has caused a lot of frustration just sitting around waiting for one thing to build before moving on. Even web browsing is laggy during builds.

Should check to see what the primary wall(s) running into with multiple workloads. If also largely running out of RAM then more newer cores isn't a solution panacea.

From what I've read, a new maxed-out 16" MBP should have no problem with 2 4K monitors and running Xcode builds, but my main concern is that all the tests and benchmarks out there are really focused on testing a single thing at a time.

Neither would an iMac Pro. If open to buying an additional display to couple to the two 4K monitors. There is nothing here that is a hard and/or volume requirement for PCI-e cards. That is one reason why the MBP is so highly viable. With the Mac Pro a huge chunk of what paying for is empty slots inside the chassis. If not going to use those then not a big bang for the buck there.

[ The iMac Pro has problems in that Apple is lollygagging on getting an upgrade out. It doesn't have a "newest" shine to it. But for these software development workloads it would work more than reasonable well. And drive those 4K displays actually better than the nominal Mac Pro GPU. The 5K display could be used to the side for documentation if just want to focus on two twin displays most of the time. But you can buy 14 cores on a iMac Pro for what you probably are looking at buying 12 cores in a Mac Pro for. If it is simply just basic core count you want that is a more cost effective path. If later Apple adjusts the iMac Pro with a core count upgrade for the current price points ... even more so.
If you can pretty easily predict your future RAM consumption in advance for a reasonable service life then can buy the iMac Pro with amount necessary. It is what would be doing with the MBP also. ]

I need to run many intensive tasks at once. The processor should theoretically show more than double improvement in what I'm seeing now, but I'm worried that I'll eventually find myself in the same situation and it'll be screaming just to keep up.

If it turns out need more affordable RAM than more cores then Mac Pro has substantive advantages. Just about any cores in a modern Mac is better than the 2015 you are rolling with now.

I feel like the base MP with just an upgrade to the 12 core is more suited to my needs, and will have no problem satisfying my requirement to multitask builds and use multiple monitors.

If primary store all your workload in a single macOS instance volume will probably also want to upgrade to 1TB storage. ( since operating with a MBP as primary system then that is very likely your nominal set up. )

Any thoughts on all this? Cost also isn't a major issue (though I'd like to keep it below 10k) - I can easily justify paying more because programming is my livelihood and I would be using this machine all day, every day.

For parallel workloads a jump from 4 to 12 cores is already a 3x increase in core count. That would push out the likelihood were ever going to get to a core count problem more than several years. (i.e., far more likely will never upgrade ) Paying $3K 'extra' for a possibility of solely a CPU upgrade 3-4 years down the road is likely to turn into a future fixation on the sunk costs you have thrown at the CPU upgrade path. ( sank $3K into this upgrade possibility so should sink more money into it to make that "pay off". )

3-4 years from now that $3K will probably by a system at least as good at this Mac Pro in the large parallel workload space. Intel probably would have cleaned up their act and AMD is already on that path.

Running iOS and Android simulators doesn't require bleeding edge GPU cards. if they did that would give more weight to the Mac Pro.



P.S.

and need to work remotely with my current MBP I would just ssh into the MP to run services/edit code and have a local tunnel running to simulate a remote 'local' environment.

support clock is ticking on your 2015 MBP.

https://www.macrumors.com/2020/04/01/apple-vintage-2013-2014-macbook-air-pro-models/


Technically the MBP 15" was replaced/discounted by the 2016 MBP 15" (its vintage clock started runing in 2016). If the 2014 MBP 15" just fell of the edge. The 2015 is not that far behind (not necessary has to be next year but it is coming in next couple of years.). If only doing remote ssh for the laptop then may be sliding back down to 13-14" Mac laptop model and therefore a lower budget outlay than a MBP 16". But if looking to keep the same screen size later it is a bit of a toss-up to on whether to get a MBP 16" now and live with it for 2 years and then revisit the double digit core desktop solutions in 2 years when Apple may have iterated past the current Mini , iMac Pro , Mac Pro line up .

I suspect the next iteration of the Mac Pro to be incrementally more affordable. Even if kept Intel the competition between AMD and Intel will be higher and the base costs for the CPU won't be quite as high. For folks who need something right now for those workload levels it can easily be a buy. But buying 'future' core horsepower. That is highly likely going to be cheaper in the future, so that is nominally better a future buy. In short, if going to buy two systems over next 3-4 years then MBP is something that have and is coming 'due'.

 

[Cellardoor6]

@deconstruct60 I appreciate the feedback - You present some great points here, especially about the sunk cost. In reality, there's a good chance that I wouldn't end up swapping the cpu. But there's just something so attractive about the ability to do so on a mac...

Here's a quick snapshot of what I'm looking like with an Xcode build running + iOS simulator + a few nodeJS services running:

So the cpu is definitely straining. The ram eventually maxes out over time, but it's usually after many successive rebuilds, or having additional apps open.

I do see the value in getting an iMac Pro vs MP when looking purely at bang for buck, and in reality it probably does makes the most sense for me. But if I'm getting a desktop I don't want it to be that tightly coupled with the monitor - I'm just not a fan of the setup - as trifling as that may sound.

 

[deconstruct60]

....
Here's a quick snapshot of what I'm looking like with an Xcode build running + iOS simulator + a few nodeJS services running:

View attachment 903603


So the cpu is definitely straining. The ram eventually maxes out over time, but it's usually after many successive rebuilds, or having additional apps open.

If tossed that workload at a MBP 16" 8 core , 64GB RAM , 1-2TB SSD , 5300M system you'd likely would have lots of head room even if crank you nodeJS sevries up toward your high end number. with the extra RAM could have a Android simulator paused back there also.

That current memory pressure isn't bad but if doubled up the working set data volume it would get bigger. But there isn't much there that screams something for Mac Pro triple digit GB DRAM capacities. The MBP 16" can go 4x on your current set up. There is decently high compressed RAM there so the faster raw RAM in the MBP 16" will help present substantive improvements too. If the subset of RAM wasn't compressed you'd be pretty close to very high memory pressure. ( you're taking an small incremental speed hit to get that compression savings. It is worth it for your combination of 16GB and disk I/O bandwidth caps , but you are closer than the "green light" there makes it look. )


SMT/Hyperthreading is working pretty good there. So the more+faster RAM and faster storage I/O should scale quite well up by 4 real cores (and 8 'activity monitor' cores/threads). Handling double the workload you are now should work at least reasonably well if not much better. Also enough periodic dips in user load there ( coinciding "longer black bars" in CPU graph) , that better Turbo (clock shifting ) is going to get traction also. (better than both the current MBP and the new Mac Pro) . So if mainly training to get "more stuff to run along side this baseline " there is more than substantive traction the MBP 16" can bring.

I suspect the Mac Pro 12 core would just be more head room in many cases. Could tweak compile process to go wider when it could, but if build times are decent now if just drop "extra stuff" how much can bring times lower by going wider depends upon the code. Long sequential dependencies on builds are limited. Or perhaps the Android simulator is a bigger hog.

But if I'm getting a desktop I don't want it to be that tightly coupled with the monitor - I'm just not a fan of the setup - as trifling as that may sound.

perhaps working with the current MBP 15" with the lid closed, but it pragmatically is what with has be working decently well with now. That's mainly why added it as an option. If spent years using 3 ... going forward with 3 isn't much different. The iMac Pro one can be "ignored" too if have been running with the lid closed.

 

[ajem]

@deconstruct60
Thank you so much for your response, I have learned a lot.

Do you know if there's any downside for having too many cores? I.e., is there any benchmark showing 28 core has poorer performance than 12 or 16 in single- or light-threaded tasks?

 

[Cellardoor6]

@deconstruct60 Thanks again for the insight - you're a wealth of information. @ajem From what I've read and seen here the 28 core seems to outperform the 12/16 in both cases, though in single-threading it's almost negligible. Multi-core is obviously where it really shines. I'm sure @deconstruct60 could probably add more to that though.


So just playing out my scenario long-term, let's say I have these two options:

- MP 16-core, 1TB, 580X ~$8400 (opting for 16 over 12 for more "future-proofing")
- MBP 16" 8-core, 1TB, 5500M ~$3900

In either case, assume this machine would get heavy use 8+ hours a day every day. This MP would definitely be overkill for what I need, but with the 16 core xeon I could bet on it still running strong in 5 years and remain stable and viable for up to 10 years. The MBP should theoretically handle my needs easily, but being that the average lifespan of MBPs is around 5 years I would purchase a new one in 5 years for another ~$4k and by that time (just speculating here) could probably get a 12-core MBP that would run closer to the MP's 16-core.

In either scenario, I'm paying around $8k whether it be up front, or over time. With the MP I'd be betting on the longevity of the xeon, being that it has a much higher power rating and is engineered to be running under constant loads. In the case of MBP I'd be getting a refresh after 5 years for newer technology, and while it would still be behind the MP in performance, it would be plenty for my needs.

So given these assumptions are valid, it then becomes a tossup between portability+refresh and more head room+expandability. Of course there are other factors that come into play, but this is just another interesting perspective I'm throwing around

 

[AidenShaw]

but being that the average lifespan of MBPs is around 5 years

Check that assumption - lately we've been replacing MBPs much sooner.

 

[ajem]

@Cellardoor6 My 2014 rMBP lasted about 4 years before I see issues on battery, DisplayPort connections, etc.

Based on what I see of more recent MBPs, I would anticipate replacing one every 3 years.

 

[th0masp]

In either case, assume this machine would get heavy use 8+ hours a day every day.

I wouldn't trust a laptop with that - not engineered for anything like it IMO. Plus the likelyhood of having to deal with throttling.

 

[deconstruct60]

Do you know if there's any downside for having too many cores?

Too many? One downside is possibly a substantially smaller savings account. Or budget that could have been put on something else that don't have too much of. The 24 and 28 core options for the Mac Pro come with a hefty "access > 1TB" tax on them. If not using more than 1TB and also not using more than 16-18 cores then that is an extra tax that is retuning not much in value.

For approximately price gap between the 12 core and 28 core options you could also buy about six, 6 core Mac Minis ( collectively 36 additional cores instead of just 16 . ). If have a diverse group of distinct process workloads that don't absolutely need to be in the same shared memory space that can be a far more cost effective path. ( Mac Pro has two Ethernet ports. So local mini "farm" could be run off one of those as a hyper local network quite easily. Local "mini" farm could be an 8 core MBP 16" too for that price. Doesn't have to be MP or MBP at that point. Buy both is an option. )

I.e., is there any benchmark showing 28 core has poorer performance than 12 or 16 in single- or light-threaded tasks?

Generally, you loose single threaded performance as go to a higher core count. The Mac Pro is using Xeon W 3200 products though that have a uniform Turbo rate (single threaded to correlate closely to that) [the 8 core is probably kneecapped a bit unnecessarily, but the rest are uniform. ] . Intel arranges for that 'privilege' by highly binned high core count dies that also can clock high. (and go back to the above... a hefty privilege tax applied ).

Not "loosing" single thread performance here, but paying substantially more for here at the higher core counts. ( The Geekbench numbers are close enough that the differences are likely just noise of the configuration of the other stuff and whatever else is running in the background. )