2014 In-depth Review

[Crosscreek]

Any one wanting benchmarks of 2014 Mac Mini performance can view this website. For english translation view in Google and use translator. This is in French.


http://www.macg.co/tests/2014/11/test-des-mac-mini-fin-2014-85345

As you can see the 2011 still kicks ass in graphics.

 

[Darby67]

Of particular interest is that the 2014 i5 2.8 and 2.6 are the faster than the 2012 QC i7 2.3. Curious

 

[Crosscreek]

Of particular interest is that the 2014 i5 2.8 and 2.6 are the faster than the 2012 QC i7 2.3. Curious

Some of this maybe the fusion drives on the 2014 2.6 and 2.8 with additional 8GB ram as opposed to 4GB and spinners in the 2012s.

 

[Darby67]

Some of this maybe the fusion drives on the 2014 2.6 and 2.8 with additional 8GB ram as opposed to 4GB and spinners in the 2012s.

I forgot to finish my comment and qualify the speed results for Handbrake, the rest is interesting as well but Handbrake is the most interesting to me. Not sure that the fusion drive would really have an dramatic impact although the Apple:Apple 2012 QC i7 results with and without fusion sees a meager jump in speed for HB.

I would also expect a small jump in speed with increase in RAM but HB is mostly a processor intensive program. I'm even flabbergasted how well the 2012 base did in the HB trial.

Overall, I find the results rather curious and would be interested to see this repeated.

 

[magbarn]

I seriously doubt the handbrake tests were done properly. I know there's been a few beta releases of quick sync enabled handbrake versions which AFAIK never came out for OS X that may give an edge for the IRIS over a quad core.

 

[Thunderchicken]

I seriously doubt the handbrake tests were done properly. I know there's been a few beta releases of quick sync enabled handbrake versions which AFAIK never came out for OS X that may give an edge for the IRIS over a quad core.

HandBrake cannot take advantage of Intel Quick Sync under OSX. Because the feature is hard baked in and would open security loop holes. Not to say that HandBrake is or hasn't pushed for access.

 

[belltree]

I'm still happy I picked up my new 2012 Mac Mini quad core i7 2.3Ghz last week. Ordered 16GB modules on the weekend.

 

[Crosscreek]

I'm still happy I picked up my new 2012 Mac Mini quad core i7 2.3Ghz last week. Ordered 16GB modules on the weekend.

I have a 2012 I5 and a 2012 I7 and i have handbreak on both. The I5 does as well as the I7 unless I multitask and then the I7 wins out with extra processor power to do what ever I want with out slowing down.

I would expect the 2014 would be the same and if your multitasking while hand breaking they will slow too.

 

[am2am]

hmm .. so the real life usage proves that new mini are not that bad (assuming there is no error in the test results)

encoding 30min podcast in garageband + script execution in photoshop + handbrake encoding of HD film into iphone format + zipping folder containing 4500 files (1,5GB in total) =

mac mini 2012 (i7 2.3, fusion drive) = 26min 25 s
mac mini 2014 (i5 2.8, fusion drive) = 23min 46 s

faster fusion, 8GB vs 4GB - I know we are not comparing apple-to-apple (ups 🙂), but anyhow seems not true that old model outperforms so much in the typical usage scenarios.

If we add on top better graphics, wifi, power consumption ... personally after long reflection I am going to buy 2014 model.

 

[barkmonster]

hmm .. so the real life usage proves that new mini are not that bad (assuming there is no error in the test results)

encoding 30min podcast in garageband + script execution in photoshop + handbrake encoding of HD film into iphone format + zipping folder containing 4500 files (1,5GB in total) =

mac mini 2012 (i7 2.3, fusion drive) = 26min 25 s
mac mini 2014 (i5 2.8, fusion drive) = 23min 46 s

faster fusion, 8GB vs 4GB - I know we are not comparing apple-to-apple (ups 🙂), but anyhow seems not true that old model outperforms so much in the typical usage scenarios.

If we add on top better graphics, wifi, power consumption ... personally after long reflection I am going to buy 2014 model.

Those benchmarks heavily favour the effect a fusion drive has on the zipping folder test. Run a purely CPU test and it would be more indicative of real performance because the PCIe SSD isn't speeding up disk activity.

Look at it this way. I could boot my Core 2 Duo Mac Mini off it's SSD to show how it boots faster, loads applications faster and likely zips a big folder full of files on the SSD faster than a 2012 i5 with a 5400rpm boot drive but if I ran handbrake on both, the i5 would beat it hands down and if I ran the software I use (which takes FULL advantage of all cores and doesn't benefit from turbo-boost mode), the i5 would trounce the C2D the same way a quad i7 would with a dual i5.

 

[Darby67]

if I ran handbrake on both, the i5 would beat it hands down and if I ran the software I use (which takes FULL advantage of all cores and doesn't benefit from turbo-boost mode), the i5 would trounce the C2D the same way a quad i7 would with a dual i5.

If the data is correct, the graphic shows the 2014 i5 finishing a Handbrake encode significantly faster than the 2012 QC i7.

 

[scottsjack]

Interesting but would like to see the results of a 2012 2.3 and 2.6 with Apple 256GB SSDs and 8GB Apple RAM against the 2014 i5 and i7 with the 256GB SSD. Only those comparisons would be truly accurate.

 

[am2am]

Those benchmarks heavily favour the effect a fusion drive has on the zipping folder test. Run a purely CPU test and it would be more indicative of real performance because the PCIe SSD isn't speeding up disk activity.

Don't agree. I am not buying CPU, I am buying mac mini to do the job.
Its a combination of CPU, GPU, memory, storage, bus etc..
CPU-to-CPU comparison may be interesting, but I am far more concerned by the final product performance.

On top of this - both minis in the comparison I've quoted were with the fusion drive

 

[Mr. Retrofire]

I seriously doubt the handbrake tests were done properly. I know there's been a few beta releases of quick sync enabled handbrake versions which AFAIK never came out for OS X that may give an edge for the IRIS over a quad core.

The x264 CoDec (used by Handbrake) can use AVX2 instructions on the Haswell (or newer) platform. This increases the encode speed on newer Macs with Haswell processors. See also:
https://mailman.videolan.org/pipermail/x264-devel/attachments/20130423/ffd6bfb6/attachment-0001.pdf

 

[barkmonster]

Don't agree. I am not buying CPU, I am buying mac mini to do the job.
Its a combination of CPU, GPU, memory, storage, bus etc..
CPU-to-CPU comparison may be interesting, but I am far more concerned by the final product performance.

On top of this - both minis in the comparison I've quoted were with the fusion drive

I can't run more plug-ins because of a fusion drive. I can't get higher polyphony from a software synth because of an SSD (the exception being the kind that stream audio like Eastwest Synphonic Orchestra Gold or VE Pro. An SSD can literally replace a RAID array or even a complete system in that case but it would have to be a seperate drive from the boot drive so it still doesn't count).

I can with a faster CPU/more cores and to an extent, with more RAM.

Handbrake must be pourly optimised for multiple CPUs to perform so badly.

You can rationalise your decision to pay through the nose for a dual i5 as a sensible one using any manipulated statistics you like but at the end of the day, you have to be very selective to find any advantage over a faster CPU. That's why I/O comes and Turbo Boost comes into play in any benchmark where the 2.8Ghz i5 is close to the 2.3Ghz i7.

In addition, think of the costs involved too.

The previous 2.3Ghz quad i7 was £649, if you add £160 for a Fusion drive, that's £809. £9 more than the 2.8Ghz model that replaces it in the range but that doesn't take into acount the extra RAM which Apple would happily charge another £80 for so really (with Apple inflated prices for £40 worth of SSD and a cable), the previous 2.3Ghz i7 would be £889 for what you get in RAM and storage with the 2.8Ghz model.

You lose a significant amount in multi-core power and the ability to upgrade the RAM yourself at a later date for no more than £125 using companies like Crucial or Kingston so the REAL cost should take into account the fact the 2014 is a soldered brick making it more like:

£809 for a 2012 2.3Ghz quad i7 with 1Tb Fusion + £125 for 16Gb of Crucial/Kingston RAM at later date.

(you can also sell the existing RAM on eBay if you want).

VS

£959 for a 2014 2.8Ghz dual i5 with 1Tb Fusion and 16Gb of Apple "soldered" RAM with no concession for trading in the 8Gb it already comes with and it all has to be paid at once.

The x264 CoDec (used by Handbrake) can use AVX2 instructions on the Haswell (or newer) platform. This increases the encode speed on newer Macs with Haswell processors. See also:
https://mailman.videolan.org/pipermail/x264-devel/attachments/20130423/ffd6bfb6/attachment-0001.pdf

Making Handbrake an entirely dishonest measure of actual CPU power. Thank you for proving my point!

 

[scottsjack]

Users are going to have to decide which features they want and which they do not care about. For many people the 2014 mini will be just fine if not preferred. Of course if you want a 1TB SSD, max RAM and an i7 paying $2200 for a mini has got to be one of the most stupid things ever. Talk about bending over and grabbing your ankles for Apple (no Tim jokes please 🙂.

In my case FW and optical are important, TB1 works fine for my non-RAID ThunderBay 4. My mini is hardwired to a current gen 3TB Time Capsule so wireless AC would not get used. I already had a 960GB M500 SSD so $839 for a 2.6 quad plus $139 for 16GB of RAM was the obvious choice. I made almost the same choice with a 2.3 quad cMBP over a much more expensive rMBP last fall.

 

[blanka]

For many people the 2014 mini will be just fine if not preferred.

For people who use it as HTPC it is nice to have a 499 option, yet it still is stupid as an Intel Q1900 board with 4x SATA socket and 10 watt load power consumption for 90 bucks is a way better way to go for HTPC.