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.
MacBook Pro 13", i7 (2012) VS i5 (2014)
- Thread starter rmger
- Start date
- Sort by reaction score
2014 all the way
Better wifi, better graphics, faster architecture in the processor, thunderbolt 2.
If you want to save some cash the lower clocked 2013 model on the refurb store is the same in every way except the processor speed.
Better wifi, better graphics, faster architecture in the processor, thunderbolt 2.
If you want to save some cash the lower clocked 2013 model on the refurb store is the same in every way except the processor speed.
No. Only fundamental difference is a bit more cache on the i7, for a lot more money. They're both dual core hyperthreaded CPUs based on a similar architecture. The improvements in efficiency would outpace any deficit in clock speed. More importantly I'd take the newer i5 just for the Iris graphics, and all the other benefits that Samuelsan2001 already mentioned.
Geek Bench scores for the processors:
MacBook Pro (13-inch Retina Mid 2014)
Intel Core i5-4308U 2800 MHz (2 cores)
6874
MacBook Pro (13-inch Mid 2012)
Intel Core i7-3520M 2900 MHz (2 cores)
6741
The 2014 model is about 2% faster than the 2012 model, something you aren't likely to notice in everyday use. The choice comes down to price, what you need in terms of connectivity, and optical drive or not.
MacBook Pro (13-inch Retina Mid 2014)
Intel Core i5-4308U 2800 MHz (2 cores)
6874
MacBook Pro (13-inch Mid 2012)
Intel Core i7-3520M 2900 MHz (2 cores)
6741
The 2014 model is about 2% faster than the 2012 model, something you aren't likely to notice in everyday use. The choice comes down to price, what you need in terms of connectivity, and optical drive or not.
For what activities a normal consumer expects would benchmark speeds be relevant, if I might ask?
Well OP asked, and the benchmarks give some context however limited it might be...
I didn't see a reference to whether the OP was comparing retina machines or non-retina machines, but the CPU in the 2012 model should be the same in both models as they were updated together.
The benchmark shows that the expected CPU performance will be more or less identical. I agree benchmarks are more or less pointless (subjective comparison at best for real world use), as the rest of the system has a huge effect on perceived performance (SSD vs. HDD, GPU model, etc.). Without more context I would call it a draw, and if the OP needs to run the CPU full bore for long periods, I doubt there will be much of a discernible difference.
Frankly, depending on the software used, a dual core machine with a higher clock speed could outperform a quad core machine with a slower clock speed, if the software is not designed for multi cores and multithreading. Much of the software that I use in my analyses is written this way. Making the analysis support multiple cores and threads seems like a good idea for future plans, but it makes things more complex to hand it off to the next student that may or may not have programming experience.
To me the difference between the machines are amenities, wireless ac vs n, TB2 vs TB, etc. It's not as though many ISPs are allowing users to download at rates that are going to saturate wireless n, and likely won't be for some time. Personally TB peripherals just don't fit my budget.
The benchmark shows that the expected CPU performance will be more or less identical. I agree benchmarks are more or less pointless (subjective comparison at best for real world use), as the rest of the system has a huge effect on perceived performance (SSD vs. HDD, GPU model, etc.). Without more context I would call it a draw, and if the OP needs to run the CPU full bore for long periods, I doubt there will be much of a discernible difference.
Frankly, depending on the software used, a dual core machine with a higher clock speed could outperform a quad core machine with a slower clock speed, if the software is not designed for multi cores and multithreading. Much of the software that I use in my analyses is written this way. Making the analysis support multiple cores and threads seems like a good idea for future plans, but it makes things more complex to hand it off to the next student that may or may not have programming experience.
To me the difference between the machines are amenities, wireless ac vs n, TB2 vs TB, etc. It's not as though many ISPs are allowing users to download at rates that are going to saturate wireless n, and likely won't be for some time. Personally TB peripherals just don't fit my budget.
This may be my own ignorance or lack of understanding, and for that I apologize, but how does hyperthreading help a piece of code written with a single core, single thread design? I thought that at best that code written in this way could only load a single CPU and crunch numbers until it's finished. I'm always excited to learn and see where my misconceptions lie.
The i5 also has hyperthreading.
Both machines mentioned in this thread have 4 virtual cores and two actual ones.
For 99% of users this is irrelevant anyway and so are benchmarks.
The OP hasn't told anyone why he would need a cpu with a slightly bigger lcache. I doubt he knows what this even means and I suspect he thinks cpu speed equals a snappier computing experience in general.
I apologize for interrupting the geektalk about hyperthreading
Hyper threading creates two virtual cores out of one physical core, it's a way of more efficiently running multithreaded code on a single core, Intel used to claim it was worth about 25-30% for code that was compiled with HT set. Apologies for the computer science geekiness, but as I recall it they duplicate the control and general registers for a single execution engine. If a bit of code gets stalled on vCPU1, e.g. waiting for data to be fetched then vCPU2 can be busy processing another thread, assuming the branch prediction works and the data is available in cache RAM.
In general use you won't notice much difference on a multicore CPU. The one place I've seen the benefits plain and clear are when processing video.
I like the computer science geekiness, it helps me learn new things. I'm a physicist with some intro computer science classes under my belt years ago. Since that time, processors have evolved into multiple cores, and now virtual cores, and I've fallen behind the times a bit.
That was more or less my understanding of hyperthreading, and it's great when multitasking, and with well constructed code. It really seems to allow the machine to get the most out of the CPU, however I don't really think it does much for the type of code I run.
That was more or less my understanding of hyperthreading, and it's great when multitasking, and with well constructed code. It really seems to allow the machine to get the most out of the CPU, however I don't really think it does much for the type of code I run.