Page 1 of 2

2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Sat Aug 16, 2014 6:48 pm
by JJ Deviation
Why does a new mid-2014 2.5Mhz i7 quad core machine perform worse than (or equal to) 2.66Ghz i7 (dual core) from 2010?
What is making my background CPU meter read higher on newest MacBook.

I bought a new machine thinking I might be able to get improvements in latency. But that’s not happening even though it’s a true quad core. Overall performance about the same or worse a big chunk of money later. For some reason I was led to believe the new architectures even in 2011 were an improvement. 2014?

The Symptoms:
Same project, same OS, same version:
CPU meter starting by hovering at 18% on 2014 i7; when on the old machine it starts at 6%.
Both machines running Mavericks and Ableton 8.
The new machine glitches sooner than the old one with the same load (when I start playing). 128 sample buffer (and still at 10% with 256 sample buffer)

Guesses as to the problem:
1) old hard drive vs SSD? (background drive access not designed efficiently for SSD?)
2) old Nvidia graphics chips vs. simple new Intel iris graphics is using CPU clocks? (Would getting the Nvidia option have helped?)
3) old 2.66Ghz (dual-core i7) vs 2.5Ghz (quad-core i7) Is it a clock speed problem?
4) old Firewire vs Firewire adaptor from Thunderbolt (Using MOTU828. Using internal DACs on either machine makes it worse as expected.)
5) Background process eating up the CPU that I can't see.

Since then I upgraded to Live version 9 thinking that may help. No improvement

I have 20 tracks of Kontakt and other Native Instruments synths BUT turning all of those “OFF”, which usually did the trick before, doesn’t improve things this time. (For instance turning OFF software synths would make my background CPU meter change by 5% each time.) Mutli-core support is on.

Maybe it’s not bad enough to go back to the old machine, but I think I just wasted my money.

Return?
If I try and return this very new machine for something else, I want to know what’s going to help me. (I also do circuit board layout and could use the bigger 15.4” screen for that when I’m on the road.) I have another few days before the 14-day period ends.

Any suggestions?

Why do I need best latency performance?
I’m an electronic percussionist almost exclusively working live shows with a lot of expressive MIDI input capabilities and I need both very ow latency and the power of some processing. Right now I’m still hold onto an old 90’s E-mu sampler for some percussion sounds partly because it still responds faster. (And that feels ridiculous.)

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Sat Aug 16, 2014 8:23 pm
by steko

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Sun Aug 17, 2014 7:46 pm
by andreX
I'd be pretty sure it's gotta be some kind of background process (spotlight indexing on a new machine?) or anything like that.
I'm having a 2013 retinaMBP and a 2010 MBP and the retina is WAY faster, in all respects...

cheers,
a

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Mon Aug 18, 2014 4:34 am
by joeyfivecents
Turn off Spotlight

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Tue Aug 19, 2014 12:55 am
by cmreal04
I'm not trying to hijack this thread, but didn't want to start a whole new topic.

If you buy a referb from apple with an original release date of 2012, does it still come installed with maverick? I'm really trying to avoid that OS for the time being.

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Tue Aug 19, 2014 6:11 am
by JJ Deviation
I am finding Kontakt to be much of the problem. (I though I checked this.) I don't know why it behaves differently on a new computer with the same OS, but on the new computer, every instance of Kontakt adds 2% to the CPU load. As I turned off all the Kontakt instances I turned 30% background CPU on the meter to 10%. That didn't seem to be the case before. In fact Kontakt added very little to the load before. Hmmm. Reaktor requires 6% CPU, but Absynth adds nothing now. I don't know if it's the architecture of having an SSD drive, but changing from DFD mode to Sampler mode on Kontakt did not make a difference. Maybe it's time to check the Native Instruments forums.

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Tue Aug 19, 2014 3:30 pm
by steko
cmreal04 wrote:If you buy a referb from apple with an original release date of 2012, does it still come installed with maverick? I'm really trying to avoid that OS for the time being.
I dunno. But don't install a version of Mac OS X earlier than what originally came with your Mac.

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Tue Aug 19, 2014 3:33 pm
by steko
JJ Deviation wrote:I am finding Kontakt to be much of the problem. (I though I checked this.) I don't know why it behaves differently on a new computer with the same OS, but on the new computer, every instance of Kontakt adds 2% to the CPU load. As I turned off all the Kontakt instances I turned 30% background CPU on the meter to 10%. That didn't seem to be the case before. In fact Kontakt added very little to the load before. Hmmm. Reaktor requires 6% CPU, but Absynth adds nothing now. I don't know if it's the architecture of having an SSD drive, but changing from DFD mode to Sampler mode on Kontakt did not make a difference. Maybe it's time to check the Native Instruments forums.
Do you see any performance difference between Live 32bit & Live 64bit versions?

IIRC forumite Machinesworking saw a huge performance gain with Live 64bit version vs 32bit using NI stuff like Massive.

(with Live's instruments & effects only, I see zero difference in performance between Live 32 & 64bit when doing a maxing out test...)

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Wed Aug 20, 2014 2:18 am
by AustinThumper
how many Rams U got?

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Wed Aug 20, 2014 2:44 am
by H20nly
yeah, what it be with the Ramz?

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Wed Aug 20, 2014 3:26 am
by rhythmhead
Try turning off App Nap. I had a few performance issues my self with that.

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Wed Aug 20, 2014 5:15 am
by Machinesworking
Naming just the CPU and that it's a Macbook Doesn't tell us much. Your 2010 machine could be a top of the line 15" and the 2014 could be a 13".
There's a huge difference in performance between these two models. Things like system buss, i7 type, and RAM can affect performance.

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Wed Aug 20, 2014 6:56 am
by JJ Deviation
I haven't tried the 64-bit version. But if that might help I will try it. (I'm trying to think of what might not be happy in 64-bit mode.) I did mention I have a new 2014 15" with an i7 is you read my post, specifically . . . it's a 2.5GHz i7 16GB RAM 1TB SSD. (Intel Iris pro graphics.) Running Thunderbolt to Firewire adaptor.

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook pro

Posted: Wed Aug 20, 2014 8:04 am
by Machinesworking
Well one thing, a single track is going to respond in Live to a single CPU, that's how Live works.
So a 2.66 could very well outperform a 2.5 in this case. Though loading samples isn't a CPU issue.
The fact that Kontakt is the PITA is interesting. Have you checked performance issues with Lives drums?
I would bet Lives drums respond better CPU wise.


Another thing, if you're triggering pads and using Kontakt set Kontakt to load larger chunks of the samples.
the SSD should outperform the regular hard drive, there's definitely something fishy here?

Re: 2014 15" MacBook pro underperforms to 2010 15" MacBook -go64

Posted: Sun Oct 05, 2014 5:43 pm
by JJ Deviation
I've seen another post about this identical problem. I've now been running the 64-bit version and that seems to help significantly--at least with the CPU idling high problem.