As some of you may know, I have had a little windows ivy bridge samsung laptop since december 2012 as a machine to take to mums when i was away from home and needed care for my illness. It is a 2.3ghz i7 quad (http://ark.intel.com/products/64899/Int ... o-3_30-GHz), 8 GB 1600 ram, 480 GB sandisk extreme SSD (aftermarket addon). Windows 7 home 64 bit. Will be adding 16gb ram after the results today.
The machine had always performed better than my 2.2ghz i7 quad sandy bridge "1 year older" macbook pro, but logic was always close (within 10-20%) vs other non logic hosts in windows. Studio one gave at least double the performance of my mac with Studio 1.
To be honest, i was using Live alot on it and had reason rewired, but no RE's installed at all, plus I had had bad experiences with codemeter on windows and didn't want to install the extra runtime ( I was logging into props server with Reason). - more news on codemeter later on.
I have only used the laptop 4 times in those 2.5 years, all those 4 times being the 4 times when I was away for a day or two from home (i.e really in shitloads of pain/unable to take care of myself).
Today, I was looking at it sitting there, unused and unloved, and i thought..WTH.. let's fire this up, get reason and a bunch of RE's installed, do the benchmark, check if the license update errors still happen (i know this is not too common an issue), and test some heavy ass RE's out vs the Mac. What prompted this was that my sister just bought a 2.5ghz Retina quad i7 macbook pro 15", with 16gb ram and 512 gb flash (my goodness it's fast, 4 second total boot from apple logo to functional desktop LMAO). The thing is, even though it performed considerably better in the benchmark song to mine (40 seconds vs 10 seconds!), stacking plugins on single cores didn't even give me 1% better performance. For example, six Softube Tsar's in line on one channel was absolute max before snap crackle pop. same as mine right now, exactly. A laptop that is 4 years older and dated architecture. So the retina only really shined when it's quad cores were allowed to flex their muscles.
The problem was, that on the benchmark test, the retina at 1024 buffer was crackling pretty much after the first 10 seconds all the way to the 40 second cutoff. It also crackled with 6 tsars occasionally (slurred audio) where as my 2.2 doesn't. They both turn to mega rice krispies with 7. LOL.
The total amount of tsars over multiple tracks/cores is 11 for mine. Unfortunately I did not get to do the multi track tsar test in the retina, as my sister has taken it back to do some stuff but she's letting me have it for one more night tonight, where i will test tsar's spread out on all cores also. My 2.2 ghz lappy crackles the *entire* way from the moment you press play to the 10 second time it overloads in the benchmark song test (95% ceiling as specified in test instructions). It really is a stinker in this test.
I did a lot of investigating. There is one thing the retina does that mine doesn't, even without a laptop cooling pad (where as mine has a really good coolermaster massive fan one) - it stays quiet and COOL. I used a benchmark on the retina to max out all cores at 100% (as in all 8 threads, the computer was literally using 800% which is the absolute max, like a prime 95 for mac) - and it stayed under 60 celsius, and fans stayed at no more than 2500 rpm! WTF!
*This * is why it was able to get to 40 seconds in the retina benchmark, over 4 cores. 2.5ghz haswell is not, core for core, much more than the 2.2 sandy bridge. It really isn't. The reason they equal in single core stacking tests is because my 2.2 doesn't overheat at that level, as it's only one core that's heavily loaded, and therefore, it isn't hitting throttling temperature thresholds and well, isn't throttling. Make sense? So retina macbook pro single core performance at 2.5ghz haswell is more or less identical, in Reason, to 2.2ghz older generation single core performance. Remember, this is all about Reason, that's all I have taken into account here. Reason, and of course, RE's. Alot of reason performance problems are due to the mac's inferior audio architecture also, as i have discovered (and suspected), but more of that to come.
Let's do the quick version now then we will proceed to windows.
- Macbook Pro 15", Quad i7 2.2ghz Sandy Bridge, 8GB ram, 750 GB 7200 RPM HDD - 10 seconds in benchmark test (following all guidelines).
Noisy as all hell with fans at 6200 rpm, obviously throttling, crackles and pops from beginning of test.
- Macbook Pro 15" Retina, 2.5ghz haswell, 16gb ram, 512gb pcie flash, 40 seconds in benchmark test (following all guidelines).
Quiet as a bird with fans in the 2000's, no obvious throttling, temps at low level. Audio pops and crackles from ~ 10 seconds through to the 40 second cutoff.
- Macbook Pro 15", Quad i7 2.2ghz Sandy Bridge, 8GB ram, 750 GB 7200 RPM HDD, 256 buffer on board sound, reason cpu ceiling 80%, 44.1 - single core test using Softube Tsar full. Internet disabled, using ignition key auth - one audio track doing a 4 bar loop, tsar inserted in series, set to dry for clear evaluation purposes, 6 instances. 7th is crackle and/or overload. Fan, noise and temperature low.
- Macbook Pro 15" Retina, 2.5ghz haswell, 16gb ram, 512gb pcie flash, 256 buffer on board sound, reason cpu ceiling 80%, 44.1 - single core test using Softube Tsar full. Internet disabled, using ignition key auth - one audio track doing a 4 bar loop, tsar inserted in series, set to dry for clear evaluation purposes, 6 instances. 7th is crackle and/or overload. Fan, noise and temperature low.
So what we can ascertain from this is that apple have mastered the cooling design in the latest models and allowed them to work at high cpu without high fan speeds/performance loss. Whereas the 2011 model was affectionately known as a jet engine at load. This haswell chip has same tdp of 45 w , as my sandy bridge from memory - although is of course a smaller die, at 22nm. That might have something to do with it too.
In my eyes, if I stay with mac, it is definitely going to be worth getting the retina 2.8 that I was thinking about. It will be a bit better again than the one i tested, the hard drive speed is pure insanity, and it's quiet - oh and that screen - just stunning. I hated the glare screens but apple have managed to make it look almost matte but just super sharp with some sort of special coating. It works. It looks beautiful. It scales well at all forced lower resolutions also, and reason looks twice as good at 1440x900 than it does on my native 1440x900 laptop.
There might be a possibility that el capitan will provide a boost too.. but the problem is, apple's OS boosts never seem to work for high load audio applications. Logic has a special way of dealing with this, as does, more recently, Cubase 8 and a few other hosts, by buffering the playback channels onto a high latency engine and only having the armed/live channels at the true chosen latency. Some of the most experienced folk in building machines and benchmarking them specifically for audio, have confirmed repeatedly that windows ALWAYS throttles the macs in audio benchmarks on similar spec machines. The only way to ever get close is to "cheat' and use logic (logic still performs the best, even though the other "hybrid" daw's are catching up). But I feel a DAW like Reason is actually more at the mercy than most any other host with mac OS, as Normen has explained before, with the way it's live buffering block system works. I don't know the technical details, but I do know in layman's terms there is absolutely no cheating of any kind ever going on - it is always running at the realtime chosen latency.
This is also why S1 performs so poorly on mac. A customer might be totally satisfied and think what a great machine they have but I mean this in comparison to windows. To give an example, my windows laptop gets 64 ambience reverb at highest quality setting at 128 buffer on S1 V2.5 on windows. S1 also has no cheat buffer, unlike many other DAW's. It is always running at the user's chosen latency, i.e realtime. The mac, at only 100mhz less, get's 15 ambience in S1. 15. Logic get's 50 - that's logic's trickery at play. It really doesn't matter in the end, as long as the tracks we are monitoring are at low latency, and all these hybrid hosts do provide that feature. I am all for any methods to make the mac perform better, as long as it's not obtrusive to me and still let's me record vocals/instruments at whatever latency I want. So I say, the more adopting this "trick", the merrier.
Reason buffers differently, and from what I understand would not benefit from this sort of hybrid engine. Indeed on all my machines, I get more or less the same performance at 128 than I do at 1024, in Reason. Even 2048 doesn't really make a noticeable difference, just a few less crackles. But ultimate RE count is the same.
So now we get to my windows laptop - the first thing I will tell you is this - it gets 24 seconds in the benchmark test. So way more than my similar clock, similar architecture macbook. Short of the retina.. but there is ONE HUGE DIFFERENCE - unlike the retina, it does that *entire* 24 seconds without a single pop or crackle. This is using onboard realtek sound with asio 4 all. The retina is crackling from 10 or so seconds onwards to the overload message.
So what is happening with the windows machine, is that it's simply getting to it's 95% cutout ceiling that has been set, and is displaying the overload message. It hasn't actually had any audible glitches. So now it all makes sense.
The windows machine is set to high performance mode. Using throttlestop utility (excellent btw), i monitored real time core speed and it was always at maximum turbo for that chip. The fans were on all the time, but the temps were always fine. Unlike my macbook, which starts out silent but when under any real load becomes unbearable, the windows laptop does it differently. It is never ever quiet but as a payoff always runs it's clock speed at maximum (of course only if high performance profile is chosen). This is inline with what a daw builder told me, you will never get a quiet laptop in windows like mac, especially the retina ones. You sacrifice noise for performance. Unless you go desktop.
I find laptop noise a little obnoxious to tell you the truth, as it's right in one's face, and the fans do annoy me. But the fans of my macbook at full speed annoy me even more, and when using a medium-busy audio project of any kind, they are *always* at full speed. It's only super quiet when browsing or watching movies. This will be moot with the retina if i upgrade, as that is a super silent machine.. kudos to apple for that. But is it worth it to sacrifice THAT much performance?
You see, this is the dilemma - the mac OS just does not deal with high load/low latency well - at all. Never has never will. It is rice krispie city. I am literally sitting in front of this windows machine with it maxed out at 128 buffer and it is not making ONE SINGLE POP. So even though the ultimate amount of RE's is the same before "overload" message, it has a much much higher usable range. How much higher?
In the single core Tsar RE test - this windows laptop, at 64.... yes 64 buffer... gets 8 instances with zero crackles or pops, inline on one track. This is with the cpu ceiling at 80% like the mac, to be fair. That's 8 tsars at 64 buffer vs 6 at 256 buffer. Get the picture?
At 128 buffer and 90% ceiling, it GETS 10! 40% more! But the big thing once again is it does it without any audio quality degradation.
I still need to test multitrack total tsars over 4 cores/multiple tracks, but am tired and really wanted to get this message out. So the remaining tests will be to do this on the windows laptop and the retina, so i cam compare them with the dismal 11 of my current macbook, and do a few more cpu hungry RE count tests.
And the short version is this: Core for Core, an almost 3 year old windows laptop has minimum 25% better RE performance, at lower buffer (64 or 128), than a current retina mac, at higher buffer (256), in Reason 8.3. And even though it can't beat the retina in the benchmark song test for sheer seconds, it plays it till it can no longer play it, without any sound degradation, where as the retina is rice krispie city. So in effect, in any scenario, an outdated, older, lower clocked (2300 mhz vs 2500 mhz), lower ram, lower speed HDD (450 mb/s vs 2000 mb/s), windows laptop has more usable real world power, when using Reason 8.3, than a current generation retina macbook pro. That's a thousand dollar laptop vs over 3.5K. (aussie prices).
One has to decide, at some point, how loyal they are to apple, especially us Reason users, when we are losing so much performance. I can only imagine how a current gen windows laptop would perform. It would be astonishing. Let alone a desktop. Oh for anyone's that's interested in DPC latency, the windows laptop is in the 50's the whole time with wireless net disabled, and not over 250 (i.e. still green) with wireless enabled (atheros brand, intel would be even better).
Now comes the bad part. Using windows for such a long period (comparatively), reminded me just how much I hate the OS. Nothing makes sense, and it's all so complicated. By the way, codemeter screws your system restore. You have to uninstall codemeter before you go to a restore point or it will fail.. And if you do what I did, and put your reason license to the hard drive in this case, and don't put it back in your account before your system restore, you lose it. For real. You have to "release" the computer with propellerhead and you can only do that one time, being at their mercy for any future mistakes. I was very lucky. I did not want to waste my one "extra license" on a system restore. So i undid the restore, and the license reappeared, then i moved it back to my account cloud, then re uninstalled codemeter, then re did the restore. It was to do with a synaptics driver problem that i fudged up so i had to absolutely get back to that restore point.
I dunno how current gen are, but there's another thing. The windows laptop's trackpad is well, complete shit compared to the liquid gold mac's.
Are they any better these days?
And all this registry bullshit I had to deal with in just one day of use. I removed bing which i had installed for some stupid reason and it didn't want to go out of my msconfig.. ccclean to the rescue.. but why should we have to?
And no quick look with spacebar.. no snapper/audio finder or alternative - heck, i couldn't even see the reason start up log in screen properly and was confused till I realised i had to turn off aero.. and just like mac, reason doesn't take advantage of any graphics acceleration - it just stays on the intel HD4000 rather than switch to the radeon 7670.
I can now say without a shadow of a doubt, that if you are worried about Reason performance, your going to get the most bang for your buck with windows, by a landslide (especially when we factor in money). There is simply no comparison. And you'll be able to work at a lower latency consistently without crackles and pops (imagine if reason gets fully hyperthread optimised also in future - wowee).
Unless you are a hardcore logic or DP user, I can't see a compelling reason to buy a mac, unless you just love using the OS. Which is where I am at. What do I do? Do i just force myself to get used to windows? After all, I used it at home (macs were in the studio), for all my personal composition from 1997-2008. From Logic 3 through 5, to cubase sx3 through 4. And i never even thought of macs back then. I didn't care. They actually gave me the shits in the studio (G3 450 - hoovers) and I had my quiet custom built windows desktop with pulsar 2 card and pentium 3 1000 LOL. I LOVED it. And afterwards, especially my athlon 2.6ghz (single core days) - man, are they memories or what.
Or do i just face reality, that my current laptop SUCKS for reason performance and that i need to buy a retina to have semi decent performance, or just keep this little windows gem of a laptop, and buy a custom made, kick ass, quiet, windows 6 core desktop to use in my room and have the laptop if I am about and about - and let the mac finish off my remaining logic projects? But if I choose a retina, even though my parents are getting me the next laptop as a gift, how can I feel right taking 4500 off them (that's how much a full specced 2.8 is here) when for 2.5K i can have a hexacore windows beast with 32gb ram, and 4 ssd? And keep the perfectly good windows laptop i already have for mobile duties? And that this old windows laptop is beating a brand new retina mac in reason performance tests? Does it really make any sense to buy a mac for reason at all? I can imagine logic in the retina would be amazing, but I haven't even bothered trying it as all my composition is continually is transitioning to reason. No point even discussing a mac pro, it is 11K with screen for the way i would want it, and that's only with 1TB of storage, not counting extra external drives I would need.
There are so many pros and cons - on site applecare VS custom made desktop = different warranties and manufacturers for different parts. VS osx sleek and ergonomic interface to windows explorer disaster..VS no clutter to registry BS? VS no real virus worry to having to very seriously get a top antivirus suite? VS losing so many features like core audio being able to use as many apps open simultaneously as you want, at different latencies with the same 2 outputs, vs asio host hogging? VS having integrated audio preview to not? VS the amazing spotlight to terrible windows slow search? VS integrated midi app to app routing to having to find the right 3rd party utility (think reason controlling external vst synths here)? VS knowing for a fact that reason mac never crashes to reading about some crashes from members here in windows? VS buying a turnkey machine to having to do heaps of research and pick every part correctly? The list goes on.. I can tell you this much, IF the mac performed just "on par" with windows, i would never ever consider using a windows machine ever again. But as my daw professional building friends have been telling me for years, windows simply performs much better for audio, especially at load.
So to summarise, and the real important bit: Windows thrashes OSX in reason/RE/low latency performance. There is no comparison in that one area. I felt it necessary to get that info out there as I have read too many "retina mac not giving me great reason performance" posts over the years. Oh and FWIW, the license did update quicker on windows, but failed 2 out of 4 times. LOL.
------------------------------------------------IF TL;DR - conclusion below---------------------------------------
Just to make this a bit easier to digest and not torture you to read the lot, let's make it simple. Let's list the results of the machines first that are very close in terms of specs. 2 laptops, one 2.2 ghz (mac), one 2.3 ghz (windows). Only sandy and ivy bridge separating them. The mac gets 10 seconds in the benchmark test with crackles the entire way. The windows box gets 28 seconds with not a single pop or crackle. The windows box can then do 24 seconds at 256 buffer which is a latency I did not even attempt on the mac, as it scored so poorly at 1024 buffer SO, the windows laptop has 2.8x the usable power of the more expensive mac, which cost me $2500 vs $1000.
That really says it ALL, doesn't it? I mean really, the performance difference is frightening. Ridiculous. Amazing. All of the above.
For the icing on the cake, the retina 2.5ghz haswell mac with supercharged everything inside, got 40 seconds at 1024 buffer - and only 10 seconds of that was clear audio - from 11seconds through 40 it was all snap crackle pop. THE END.
Share and download patches!
Join ReasonTalk on Slack!