Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Ynk

macrumors member
Original poster
Jan 19, 2013
69
41
Hey all - I have both my old 2015 m3 and my new 2016 i7 with me now, so if anyone wants me to run any specific (free) benchmarks or checks between the two just let me know I would be happy to provide any results which might help.
 
Hey all - I have both my old 2015 m3 and my new 2016 i7 with me now, so if anyone wants me to run any specific (free) benchmarks or checks between the two just let me know I would be happy to provide any results which might help.
I'd like to know the temperature of the palm rests and the underside corners near the hinge after one of these tests. Could you keep it on your lap?
 
My CPU load test of choice is the distributed.net RC5 project (https://www.distributed.net/Download_clients)

My 2015 m3 achieves 14.3M keys/second steady state with 2.0GHz and 82 degree CPU temperature
My 2017 i7 achieves 24.4M keys/second steady state with a 2.5 - 2.6 GHz throttle and 94-97 degree CPU temperature

The palm rests are not warm at all on either machine, the 2015 is only slightly warm on the bottom - the 2017 is definitely a lot warmer but I wouldn't describe it as hot just warm.
[doublepost=1497441038][/doublepost]After 10 minutes
 

Attachments

  • Screen Shot 2017-06-14 at 9.49.03 pm.png
    Screen Shot 2017-06-14 at 9.49.03 pm.png
    126.9 KB · Views: 288
The palm rests are not warm at all on either machine, the 2015 is only slightly warm on the bottom - the 2017 is definitely a lot warmer but I wouldn't describe it as hot just warm.
[doublepost=1497441038][/doublepost]After 10 minutes
Thank you! Good to know it doesn't get that warm.
 
Hey all - I have both my old 2015 m3 and my new 2016 i7 with me now, so if anyone wants me to run any specific (free) benchmarks or checks between the two just let me know I would be happy to provide any results which might help.

Do you have 512GB on each? Can you run the SSD speed test? Also what about the Geekbench benchmarks? The numbers online have been quite inconsistent. Thanks. You must be excited? :D
 
Geekbench 4 2017: 4180/8171
Geekbench 4 2015: 2787/4885

Geekbench 4 2017 OpenCL: 19216
Geekbench 4 2015 OpenCL: 11629

Disk speed: [Note different SSD capacity and I have filevault enabled on both devices]
2015 (256gb) 437/433
2017 (512gb) 1247/1324
 
  • Like
Reactions: effineji
Geekbench 4 2017: 4180/8171
Geekbench 4 2015: 2787/4885

Geekbench 4 2017 OpenCL: 19216
Geekbench 4 2015 OpenCL: 11629

Disk speed: [Note different SSD capacity and I have filevault enabled on both devices]
2015 (256gb) 437/433
2017 (512gb) 1247/1324
Very impressive, especially for multicore. How's your overall impression?
 
When I bought the 2015 I knew it was going to be a big compromise, the CPU is quite limited and the GPU performance generally poor even just for basic tasks/web browsing. Moving to the 2017 I'm really happy, it finally has enough grunt for my daily needs and the general responsiveness (especially with 20+ Safari tabs open) is so much better. Will be keeping the 2017 for sure and selling off my 2015 :)
 
When I bought the 2015 I knew it was going to be a big compromise, the CPU is quite limited and the GPU performance generally poor even just for basic tasks/web browsing. Moving to the 2017 I'm really happy, it finally has enough grunt for my daily needs and the general responsiveness (especially with 20+ Safari tabs open) is so much better. Will be keeping the 2017 for sure and selling off my 2015 :)

What are your thoughts on the 2nd generation keyboard?
 
It's much better, keys are quite firm and stable where as the 2015 keys feel very loose in comparison. I would say it's a huge improvement compared to the last version.
 
When I bought the 2015 I knew it was going to be a big compromise, the CPU is quite limited and the GPU performance generally poor even just for basic tasks/web browsing. Moving to the 2017 I'm really happy, it finally has enough grunt for my daily needs and the general responsiveness (especially with 20+ Safari tabs open) is so much better. Will be keeping the 2017 for sure and selling off my 2015 :)

How would one go about selling a 2015 that still has AppleCare? ;)
 
My CPU load test of choice is the distributed.net RC5 project (https://www.distributed.net/Download_clients)

My 2015 m3 achieves 14.3M keys/second steady state with 2.0GHz and 82 degree CPU temperature
My 2017 i7 achieves 24.4M keys/second steady state with a 2.5 - 2.6 GHz throttle and 94-97 degree CPU temperature

The palm rests are not warm at all on either machine, the 2015 is only slightly warm on the bottom - the 2017 is definitely a lot warmer but I wouldn't describe it as hot just warm.
[doublepost=1497441038][/doublepost]After 10 minutes

Are you mean the value of "nodes/s", why my 2015 m3 with speed
[24,840,322 nodes/s] after run 30mins ?
 
Are you mean the value of "nodes/s", why my 2015 m3 with speed
[24,840,322 nodes/s] after run 30mins ?

You're running the OGR workload not RC5. Add the following lines(or replace) in your .ini config file (same directory as the application):

[misc]
project-priority=OGR-NG=0,RC5-72
[triggers]
pause-on-no-mains-power=no
[display]
progress-indicator=rate
 
2015 m3 ... 82 degree CPU temperature
2017 i7 ... 94-97 degree CPU temperature

The palm rests are not warm at all on either machine, the 2015 is only slightly warm on the bottom - the 2017 is definitely a lot warmer but I wouldn't describe it as hot just warm.
[doublepost=1497441038][/doublepost]After 10 minutes

Interesting finding! I wonder if all 2017 models get warmer than the 2015 ones or if it is just the i7. I very much like the 2015 m3 always staying relatively cool on the bottom. I would sure like to get a i5 or i7 rather than a m3 though.
 
Apple has changed the throttling temperature from 95C to 100C on the CPU so we are seeing higher temperatures all around apparently.
 
Hey all - I have both my old 2015 m3 and my new 2016 i7 with me now, so if anyone wants me to run any specific (free) benchmarks or checks between the two just let me know I would be happy to provide any results which might help.
Could you please try to run the Intel Linpack benchmark on both machines? You can download the benchmarks from https://software.intel.com/en-us/articles/intel-mkl-benchmarks-suite
(you need to double click to unpack m_mklb_p_2017.3.019.tgz, and then open Terminal - or again double click - to run the benchmark "runme64" in /m_mklb_p_2017.3.019/benchmarks_2017/mac/mkl/benchmarks/linpack )

I've been running the Feb 2017 benchmark on MBA mid 2013 with a 1.3GHz i5 and MBA Early 2015 with a 2.2Ghz i7, and got the following numbers for the 10,000 size:

---
Performance Summary (GFlops)

Size LDA Align. Average Maximal
10000 10008 4 63.0738 63.1265 (Mid 2013, with Intel(R) Core(TM) i5-4250U CPU @ 1.30GHz)
10000 10008 4 73.5309 73.7697 (Early 2015, with Intel(R) Core(TM) i7-5650U CPU @ 2.20GHz)
 
Results from the 2017 i7:

Performance Summary (GFlops)
Size LDA Align. Average Maximal
15000 15000 4 57.5102 57.5102
14000 14008 4 56.5900 57.1852
13000 13000 4 55.4907 55.5754
12000 12008 4 54.7716 54.9216
11000 11000 4 54.5494 54.6181
10000 10008 4 56.0783 56.1736
8000 8008 4 58.0139 59.0826
6000 6008 4 60.8393 61.5137
1000 1000 4 36.9072 37.4723

Results from the 2015 m3:

Size LDA Align. Average Maximal
15000 15000 4 53.0731 53.0731
14000 14008 4 51.4463 51.9657
13000 13000 4 50.3368 50.5473
12000 12008 4 48.7187 49.9225
11000 11000 4 48.9303 48.9321
10000 10008 4 43.5228 49.7383
8000 8008 4 49.4257 49.4569
6000 6008 4 48.4495 48.6142
1000 1000 4 26.8678 27.8419
 
Last edited:
  • Like
Reactions: wol
When you say 2015 "M3" which Core-M are you referring to? Core-M 1.1, 1.2 or 1.3? I'm assuming the 1.1?

Unless I'm mistaken, there was no M3 option back then - that branding came with the 2016 model - so I'm just wondering as I'm interested in the 2017 i7, currently using a 2015 Core-M 1.1.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.