New MacBook Pro: Higher clock speeds vs. more cores?

[Crunch]

With CPUs going more and more in the direction of more cores vs. higher clock speeds, what is your advice vis-à-vis someone for their next MBP?

Coming from a 2016 MBP and a 2.6GHz quad-core i7-6700, I could go up to a 2017 MBP w/ 2.8GHz i7-7700 and gain 200MHz, or I could upgrade to the latest and the greatest 2018 MBP and get a 2.2GHz i7-8750H and lose 400MHz but gain 2 additional cores.

While I understand that the additional cores are great for some things, isn't it true that the core clock speed is the driver when it comes to general snappiness in everyday tasks? Or would I make up a loss of 400MHz by getting a CPU that's two generations newer? To what extent, if any, would the higher Turbo Boost speeds make up for the loss in core clock speed?

Decisions, decisions! I appreciate your input!

 

[Stephen.R]

What do you do with your laptop? Some tasks/programs are inherently more or less suited to either multi-core processing at lower frequencies or single-core processing at higher frequency. Newer Intel chips will generally try to suit both through Turbo boost (i.e. if a process is using just one core, it can boost frequency to run a single-core process but then other times run multiple cores at lower speeds).

 

[maflynn]

isn't it true that the core clock speed is the driver when it comes to general snappiness in everyday tasks?

Clock speeds have long ceded the thrown as the major factor for CPU performance. advances in CPU design of minimized the impact of a faster clock rate and there are other design features that have improved performance. One of the major features is cores, basically two cpus on one die is better then bumping up the frequency.

 

[pshufd]

Intel's inability to get to 10 nm process has stalled some amount of progress in performance improvements so they've been mainly throwing more cores to improve performance. In the old days, throwing more cores didn't necessarily help because most programs weren't multithreaded to take advantage of parallelism. Today, many programs that can take advantage of parallelism are multithreaded. So more cores often means better performance.

So it may come down to the software you use and your workflow.

 

[Crunch]

I guess these benchmarks just answered my question. Looks like the improvement in single-core performance of the 2018 MacBook Pro over the 2017 model is even greater than in the previous two generations.

 

[leman]

With CPUs going more and more in the direction of more cores vs. higher clock speeds, what is your advice vis-à-vis someone for their next MBP?

You need to look at the range of CPU clocks nowadays, not at its nominal clock. The nominal clock is basically the minimal guaranteed clock that the CPU should run at when operating all the cores at its TDP (disclaimers apply). But the i7-8750H is not a 2.2 Ghz CPU, its an "up to 4.1Ghz" CPU, and it usually able to sustain multi-core operation at 3.0-3.3Ghz. The i7-7700HQ has the maximal clock of 3.8Ghz, which is lower. Also, 6*2.2 > 4*2.8 (multicore operation at nominal frequency)

 

[T'hain Esh Kelch]

While I understand that the additional cores are great for some things, isn't it true that the core clock speed is the driver when it comes to general snappiness in everyday tasks? Or would I make up a loss of 400MHz by getting a CPU that's two generations newer? To what extent, if any, would the higher Turbo Boost speeds make up for the loss in core clock speed?

Decisions, decisions! I appreciate your input!

Everyday tasks can mean a lot of things. That being said, I find my 2011 iMac at home much snappier in general, than my 2013 iMac at work, simply due to the fact that the former has an SSD, and the latter has a mechanical HDD. I wouldn't buy a machine without SSD now, and unless you do stuff which require plenty of CPU time such as rendering, calculations, compiling, etc. then you won't see much of a difference between modern high end CPU's.

 

[Crunch]

Everyday tasks can mean a lot of things. That being said, I find my 2011 iMac at home much snappier in general, than my 2013 iMac at work, simply due to the fact that the former has an SSD, and the latter has a mechanical HDD. I wouldn't buy a machine without SSD now, and unless you do stuff which require plenty of CPU time such as rendering, calculations, compiling, etc. then you won't see much of a difference between modern high end CPU's.

No argument there. One of the first things I used to do on laptops around 2009-2012 was rip any hard drive out of MacBooks and, before then, Thinkpads, and replace them with the early Intel X25-M SSDs. Night and day difference.

I realize that you can have one of the fastest CPUs and GPUs but when paired with a hard drive, any hard drive, performance is a joke when compared to a system with a mid-range processor and ANY SSD.
[doublepost=1557746729][/doublepost]

You need to look at the range of CPU clocks nowadays, not at its nominal clock. The nominal clock is basically the minimal guaranteed clock that the CPU should run at when operating all the cores at its TDP (disclaimers apply). But the i7-8750H is not a 2.2 Ghz CPU, its an "up to 4.1Ghz" CPU, and it usually able to sustain multi-core operation at 3.0-3.3Ghz. The i7-7700HQ has the maximal clock of 3.8Ghz, which is lower. Also, 6*2.2 > 4*2.8 (multicore operation at nominal frequency)

Thanks for your post. It looks as though the base model 2018 MacBook Pro is the obvious choice, as even the highest-end CPU option on the 2017 model is far surpassed in both single and multi-threaded operations by the base model 2018 MacBook Pro.

What's also interesting is that the 2.6GHz 6-core i7 in the 2018 model offers very little benefit to the base 2.2GHz i7, again in both single and multi-threaded ops. The i9 seems to come up with a bigger difference in both, but not nearly enough to justify its high upgrade cost as far as I'm concerned.

 

[leman]

What's also interesting is that the 2.6GHz 6-core i7 in the 2018 model offers very little benefit to the base 2.2GHz i7, again in both single and multi-threaded ops.

The base CPU is always the best bang for your buck. Higher tier CPUs are usually 5-10% faster at best in peak usage (irrelevant for the majority of users users anyway). This has pretty much always been the case. I got myself an i9 config since the little extra boost in single-core performance helps my particular use case. For multithreading operations, there is usually very little difference between any mobile Coffee Lake, since even the cheapest one has a very high clock range.

I would caution you from taking Geekbench too serious though. It's a mix of different tests and it is not always representative of the improvement you will see in practice. It's always better to look for tests that cover your intended use. For me (if I remember correctly) the 2018 i9 is about 20-30% faster than the top-tier 2016 model I used previously, which made the upgrade for me worthwhile.

 

[Zwhaler]

Coming from a 2016 MBP and a 2.6GHz quad-core i7-6700, I could go up to a 2017 MBP w/ 2.8GHz i7-7700 and gain 200MHz, or I could upgrade to the latest and the greatest 2018 MBP and get a 2.2GHz i7-8750H and lose 400MHz but gain 2 additional cores.

Unless the machine is paid for by work, upgrading to a 2017 for a net 200MHz boost would be a waste. If you are interested to know the single core performance capabilities, Geekbench Mac Benchmarks (single core) is a good place to start. That includes single core performance scores for higher core, lower clock speed processors This should answer your question. Higher core count is better for software that uses multiple cores, such as video encoding and distributed computing. For day-to-day tasks, it won't make a meaningful difference to have more cores if they aren't being used. Activity Monitor will highlight how your CPU is being used while software is running.

 

[theorist9]

While I understand that the additional cores are great for some things, isn't it true that the core clock speed is the driver when it comes to general snappiness in everyday tasks?

As other posters have already mentioned, the single biggest improvement in snappiness comes from having the fastest SSD you can get. The nice thing is that Apple's MBP's offer some of the fastest SSDs available for consumers.

But your question was about snappiness and computational speed. When it comes to that, I'd reword your statement somewhat but, broadly speaking, you're correct for most home computing tasks. [The way I'd reword it is to say that it's about single-core computational speed, rather than clock speed, since single-core computational speed is determined by a combination of clock speed and architecture.] The reason is that, even today, most programs are only written to run on a single core (e.g., web browsers, MS Office, etc.; Excel is a partial exception, being partially multi-core). That's because it's hard to parallelize code. Thus, unless you are specifically using programs that can make use of multiple cores (video rendering programs, for instance), more than a few cores won't do you any good---they'll just sit unused.