2012 (2.3GHz i7 Quad) vs 2014 (2.6 GHz i7 Dual)

[brandon8u]

Thanks, but I'm seeing it for $749 at Best Buy ...

http://www.bestbuy.com/site/apple-m...548287565&skuId=4840689&st=mac mini&cp=1&lp=2

My bad... I was up late when researching last night. Micro center most definitely has them for $699. But only available in limited stores, and some states don't have a MC. http://www.microcenter.com/product/404239/Mac_mini_MD388LL-A_Desktop_Computer

 

[oneMadRssn]

These arguments are not being made on the basis of efficiency for efficiencies sake. You are providing examples where the additional cores offer tangible benefits. The way I read your streaming example was the dual core would be utilizing the two cores, say, 60% whereas the quad core would be utilizing the four cores, say, 40%. These are just made up numbers to illustrate what I believed you to had been saying. Assuming, for the sake of argument, these numbers are representative of efficiency does it matter to the end user? Not really as the work is being completed as needed either way.

With that said it's unlikely the base 2012 quad core would be 2x faster transcoding video. Why? Because the base 2012 quad core:

• Is, on average, 8% slower (based on Anandtech benchmarks) than the Haswell processor used in the 2014 based on IPC improvements in Haswell.
• Has a base clock speed which is 18% slower than the base clock speed of the 2014 i7. For transcoding all cores will be utilized therefore keeping the clock speed near the base.

I wouldn't be surprised to see the base 2012 quad core to be about 150-160% faster at transcoding.

I'm not against the quad core Mini. I used to own one and found it a great system. And if I had to do it again I'd be looking at it as some of the work I do does scale well with each core (I have an 8 core, 16 thread Z600 system that rips through video transcodes).

No, not quite. During transcoding, depending on settings and how greedy you allow Handbrake to be, it will use 100% of all cores, however many you have. Transcoding is basically decoding, and encoding again one frame at a time. A quad-core will run 2x as fast than an equivalent dual-core because while a dual-core can process 2 frames at a time, the quad-core can process 4 frames at a time. As you said, the haswell is less than 10% faster, per core. So in the time a dual core haswell can do 2 frames, an ivebridge quad-core will do 3.8-3.9 frames. Put another way, if a movie has 12 frames, the quad-core will do frames 1-4 in cycle 1; the dual-core will do frames 1-2 in cycle 1. The QC will do frames 5-8 in cycle 2; the DC will do frames 3-4 in cycle 2. the QC will do frames 8-12 in cycle 3; the DC will do frames 4-6 in cycle 3. So, the QC will finish the job in 3 cycles, the DC will finish the job in 6 cycles. The fact that each cycle on the DC can be done 10% faster isn't an advantage when the QC can do more at once.

 

[Ridley]

Benchmarks are showing better performance from the 2012 model in both singe and multi-core support.

Do you have a link for that benchmark? I do not believe that the i7 Ivybridge @2.3 GHz beats the i5 Haswell @ 2.6 Ghz. In fact I feel like several people have posted 13 inch macbook pro benchmarks (with similar-identical haswell cpu's) that show faster single core performance than the 2012 i7 mac mini.


For example according to primate labs 64 bit single core (higher score is better)

MacBook Pro (13-inch Retina Mid 2014)
Intel Core i5-4308U 2800 MHz (2 cores)
3275

MacBook Pro (13-inch Retina Late 2013)
Intel Core i5-4288U 2600 MHz (2 cores)
3119

Mac mini (Late 2012)
Intel Core i7-3615QM 2300 MHz (4 cores)
3003

http://browser.primatelabs.com/mac-benchmarks

Again this is for single core performance and these chips are essentially identical to what you will get in the 2014 mini. Also I wrote this elsewhere but people underestimate the difference between the HD4000 and Iris 5100. Its huge! And for every argument where people say "well very few things are optimized for the GPU anyway" you can make the same claim for few applications optimized for quad cores.

My dream machine would have been a quad core with iris pro 5200, so I'm not saying this is a perfect update. But I feel like you need to backup your claim with actual benchmarks.

 

[magbarn]

... Also I wrote this elsewhere but people underestimate the difference between the HD4000 and Iris 5100. Its huge! And for every argument where people say "well very few things are optimized for the GPU anyway" you can make the same claim for few applications optimized for quad cores.

My dream machine would have been a quad core with iris pro 5200, so I'm not saying this is a perfect update. But I feel like you need to backup your claim with actual benchmarks.

Please show me the in-game benchmarks where a HD 5000 or HD 5100 kill a HD 4000 at real gaming resolutions (no one's gaming at 720P resolution on their Tbolt or 1080P screens). The HD5000/5100 isn't even close to 50% faster in real life gaming. It's more in the 15-25% range. And no 3DMARK is not a game. Fact is, Handbrake (which I use several times a week) and Lightroom run much faster on my 2012 rMBP quad core vs. 2013 Haswell rMBP dual core Haswell and I was hoping Apple would make a mini with 2014 rMBP 15 performance in a small desktop shell. That obviously didn't happen....

 

[MadCar]

Do you have a link for that benchmark? I do not believe that the i7 Ivybridge @2.3 GHz beats the i5 Haswell @ 2.6 Ghz. In fact I feel like several people have posted 13 inch macbook pro benchmarks (with similar-identical haswell cpu's) that show faster single core performance than the 2012 i7 mac mini.


For example according to primate labs 64 bit single core (higher score is better)

MacBook Pro (13-inch Retina Mid 2014)
Intel Core i5-4308U 2800 MHz (2 cores)
3275

MacBook Pro (13-inch Retina Late 2013)
Intel Core i5-4288U 2600 MHz (2 cores)
3119

Mac mini (Late 2012)
Intel Core i7-3615QM 2300 MHz (4 cores)
3003

http://browser.primatelabs.com/mac-benchmarks

I think you are missing the point on many levels not least the cost difference.

The 2012 mid 2.3 i7 Mac Mini has a 64bit multi core benchmark of 11696. and in the UK is £569 from Apple.

The 2014 mid Mac Mini has a 64Bit multi core benchmark of 7188 (expected) so the fastest new mini is 40% slower than the fastest new model if you need multi core performance.

But lets look at cost.

The new 2014 Mac Mini with the top processor, same HD size and 8Gb of memory (slightly ahem) faster GPU is £799.

So thats currently 30% more expensive for a machine that is 40% slower if you need to handbrake, run lots of VMs etc. and need the multi cores.

Which is why after waiting I've picked up a Mid 2012 mini 2.3.

 

[Larry-K]

Slightly off-topic, but is there a list available of which Mac apps actually can use all four cores?

Modo, 8 threads of render goodness.

 

[SoCalReviews]

Which is why after waiting I've picked up a Mid 2012 mini 2.3.

You will enjoy the 2012 2.3 i7 Mini. I have four of them for home and business. I chose the quad core for improved Windows VM performance running Parallels Desktop. I run Win XP, 7, 8, 8.1 and Win 10 Tech Preview on my 2012 i7 Minis. I upgraded all Minis with 16GB and using PD10 I like how I can easily assign resources (number of cores, amount system RAM, Video RAM etc.) for each VM.

 

[fa8362]

people underestimate the difference between the HD4000 and Iris 5100. Its huge!

It's barely perceptible, and certainly not a good reason to pay more for an obviously slower machine. It's just an Apple money grab exploitation of consumers who know nothing about computers.

 

[brdeveloper]

It's barely perceptible, and certainly not a good reason to pay more for an obviously slower machine. It's just an Apple money grab exploitation of consumers who know nothing about computers.

I'm still happy with my Mini 2010. Hopefully the Broadwell Mini will bring 4-cores back.

 

[SoCalReviews]

I'm still happy with my Mini 2010. Hopefully the Broadwell Mini will bring 4-cores back.

Now that they control the pricing structure with the soldered RAM upgrades they can charge a premium for upper end models with SSDs just like they do now with all their other Mac models. I wouldn't be surprised to see the future upper end quad core Minis return with maxed out specs starting in the $2K USD range.

 

[Cherish]

Better GPU, additional Tbolt port, PCI SSD, I say that is a clear win. Not a huge win, but a clear win.

The GPU Intel Iris is the same for the mid & high end model. Both have 2 Tbolt port.

Edit: Just checked both model uses the same ssd too so all your info is wrong.

 

[scottsjack]

My dream machine would have been a quad core with iris pro 5200, so I'm not saying this is a perfect update.

Yes. I would have stepped up to the plate and overspent for that one.

 

[keeftone]

Bought a 2012 2.3 i7 quad Mini today as well. For my needs, I don't think I'll have any regrets.

 

[dandingo]

My bad... I was up late when researching last night. Micro center most definitely has them for $699. But only available in limited stores, and some states don't have a MC. http://www.microcenter.com/product/404239/Mac_mini_MD388LL-A_Desktop_Computer

Awesome, thanks for the info. I'm still debating if I should go 2012 or 2014. Decisions, decisions.

 

[reco2011]

No, not quite. During transcoding, depending on settings and how greedy you allow Handbrake to be, it will use 100% of all cores, however many you have. Transcoding is basically decoding, and encoding again one frame at a time. A quad-core will run 2x as fast than an equivalent dual-core because while a dual-core can process 2 frames at a time, the quad-core can process 4 frames at a time. As you said, the haswell is less than 10% faster, per core. So in the time a dual core haswell can do 2 frames, an ivebridge quad-core will do 3.8-3.9 frames. Put another way, if a movie has 12 frames, the quad-core will do frames 1-4 in cycle 1; the dual-core will do frames 1-2 in cycle 1. The QC will do frames 5-8 in cycle 2; the DC will do frames 3-4 in cycle 2. the QC will do frames 8-12 in cycle 3; the DC will do frames 4-6 in cycle 3. So, the QC will finish the job in 3 cycles, the DC will finish the job in 6 cycles. The fact that each cycle on the DC can be done 10% faster isn't an advantage when the QC can do more at once.

My point is they're not equivalent. Haswell is, on average, 8% faster than Ivy Bridge...which you've already acknowledged. However you appear to have overlooked the clock speed difference. Not only is Haswell 8% faster clock for clock but the base 2014 i7 Mini is clocked 21% faster than the base 2012 quad core Mini. One can also bump up the speed of the 2014 Mini to a base clock of 3.0GHz...giving it 30% higher clocking than the 2012. I'm ignoring Turbo boost as that tends to come into play only when you have idle cores.

Edit: See benchmarks in post #30 which appear to support my claim the 2012 quad core is roughly 150-160% faster than the 2014 dual core (11696/7188 = 1.63).

 

[CausticPuppy]

One can also bump up the speed of the 2014 Mini to a base clock of 3.0GHz...giving it 30% higher clocking than the 2012. I'm ignoring Turbo boost as that tends to come into play only when you have idle cores.

You can't ignore turbo boost for single-threaded CPU performance, because that's exactly where turbo boost benefits.

The 2014 3.0GHz i7 maxes out at 3.5GHz for single-core performance. It can turbo up to 3.3GHz with both cores.

The 2012 2.6GHz i7 (quad) maxes out at 3.6GHz for single-core performance. It can turbo up to 3.5GHz with two cores fully utilized, and 3.4GHz with all 4 cores. The turbo frequency for the older Mac using all 4 cores is still higher than the turbo frequency of the 2014 version using both cores.

That's why single-cpu performance for both chips is pretty much equivalent, but the older model has vastly better multi-core performance.

The Mac Mini has pretty good thermal management, so these chips are rarely running at the base frequency; they will always run at their maximum turbo frequency until they get too hot and need to step down. Comparing base clocks is rather useless since that represents a worst-case scenario in a warm environment.


Reference for determining turbo frequencies for a given cpu with a given number of cores operating:

http://en.wikipedia.org/wiki/List_of_Intel_Core_i7_microprocessors

 

[reco2011]

You can't ignore turbo boost for single-threaded CPU performance, because that's exactly where turbo boost benefits.

The discussion is not about single thread performance and therefore the reason I am excluding Turbo Boost clock speeds.

 

[CausticPuppy]

The discussion is not about single thread performance and therefore the reason I am excluding Turbo Boost clock speeds.

I updated my previous post. Turbo-boost also applies to multicore performance.

It's intellectually dishonest to dismiss Turbo Boost clock speeds since those are the clock speeds that actually matter for performance; otherwise, the base model 2014 Mac Mini would only be half as fast as the top spec model, which is clearly not the case, despite the top spec model having more than twice the base CPU speed. (3.0 vs 1.4).

 

[reco2011]

I updated my previous post. Turbo-boost also applies to multicore performance.

It's intellectually dishonest to dismiss Turbo Boost clock speeds since those are the clock speeds that actually matter for performance; otherwise, the base model 2014 Mac Mini would only be half as fast as the top spec model, which is clearly not the case, despite the top spec model having more than twice the base CPU speed. (3.0 vs 1.4).

There's nothing dishonest about it. The concept of Turbo Boost is to address the problem of few core processors having higher clock speeds than higher core count processors. By implementing Turbo Boost a higher core count processor can achieve similar single thread performance to a lower core count processor. Thus one does not have the conundrum of choosing a higher clocked, lower core processor over a lower clocked, higher core count processor. Turbo Boost helps narrow that gap.

The limitation of Turbo Boost is when all cores are being fully utilized the power / heat constraints are realized and therefore the CPU has to be clocked down. Thus when performing highly threaded work and maximizing core use the clock speed is base (otherwise base would be higher).

I've seen this in real life as I have a 2010 2.8GHz quad core Mac Pro and a late 2012 2.7GHz 15" rMBP. The base clock speeds are within 100MHz of one another. However the rMBP Turbo Boost frequency is 3.7GHz whereas the Mac Pros maximum Turbo Boost is 3.0GHz. The rMBP has a much high Turbo Boost frequency. The Geekbench scores of the rMBP are considerably better. Each is a quad core CPU with Hyperthreading technology.

In real world Handbrake transcodes the Mac Pro handily beats the rMBP. Why? I suspect because the rMBP is running into thermal constraints forcing it to run slower. Does this mean the Mac Mini will fare similarly? I believe so. However the only way to know is to have someone perform Handbrake transcodes on the two systems. I suspect the 2012 quad core will outperform the 2014 i7 by about 150-160%

 

[oneMadRssn]

My point is they're not equivalent. Haswell is, on average, 8% faster than Ivy Bridge...which you've already acknowledged. However you appear to have overlooked the clock speed difference. Not only is Haswell 8% faster clock for clock but the base 2014 i7 Mini is clocked 21% faster than the base 2012 quad core Mini. One can also bump up the speed of the 2014 Mini to a base clock of 3.0GHz...giving it 30% higher clocking than the 2012. I'm ignoring Turbo boost as that tends to come into play only when you have idle cores.

Edit: See benchmarks in post #30 which appear to support my claim the 2012 quad core is roughly 150-160% faster than the 2014 dual core (11696/7188 = 1.63).

I think we're saying the same things back at each other, and splitting hairs. Whether a 2012 quad-core can transcode a movie 1.63x faster or 2x faster than a 2014 dual-core isn't the argument. The point is, when transcoding a blu-ray takes over an hour on high quality settings, whether it's 1.63x or 2x faster, it makes for a significant difference in real world home use.

It's the difference between getting the job done in 60 minutes (2012 QC), or 100-120 minutes (2014 DC). Or it's the difference between getting 12 tv show episodes done overnight, or 6-7.

 

[starfoxx]

Perhaps these tests can help clarify some matters:

http://www.macg.co/materiel/2014/10/premiers-tests-de-trois-mac-mini-2014-85045

Some real-world CPU, GPU and Application performance tests comparing 2014 i5-2.8 GHz, 2014 i5-2.6 GHz, 2014 i5-1.4 GHz, 2012 i7-2.3 GHz, 2011 i5-2.5 GHz, 2011 i5-2.3 GHz and 2010 core2duo - 2.4 GHz

 

[reco2011]

I think we're saying the same things back at each other, and splitting hairs. Whether a 2012 quad-core can transcode a movie 1.63x faster or 2x faster than a 2014 dual-core isn't the argument. The point is, when transcoding a blu-ray takes over an hour on high quality settings, whether it's 1.63x or 2x faster, it makes for a significant difference in real world home use.

It's the difference between getting the job done in 60 minutes (2012 QC), or 100-120 minutes (2014 DC). Or it's the difference between getting 12 tv show episodes done overnight, or 6-7.

We're in agreement the 2012 quad core would be preferred for this particular task. Especially when you factor in cost (as it's less expensive than the base model 2014 i7). I was merely stating, with the 2012 quad core, one is not achieving a 2x increase in transcoding performance when compared to the the 2014 i7 dual core.

 

[Ridley]

I think you are missing the point on many levels not least the cost difference.

The 2012 mid 2.3 i7 Mac Mini has a 64bit multi core benchmark of 11696. and in the UK is £569 from Apple.

The 2014 mid Mac Mini has a 64Bit multi core benchmark of 7188 (expected) so the fastest new mini is 40% slower than the fastest new model if you need multi core performance.

Uhh, i think you are missing the point? The OP was claiming the quad core ivy bridge was faster in BOTH single and multicore performance than the dual core haswell. I even bolded that I question only the single core claim.

 

[phrehdd]

Most of my major apps take advantage of multi-core so for me, the quad 2012 is the way to go. You need to figure out which apps you are going to use will "stress" your system and if those apps exploit multi-core. If yes, you have your answer.

I see the only advantage of the 2014 is the WiFi and possibly TB2 availability. If I had to buy today, I would get the quad 2.6 2012 again for Photoshop, Capture One, and a few multi-media related software. Truth be said, I think unless Apple next year comes out with a quad Mini next year, I'll be leaving the Apple camp as I have zero use for the iMac and the Mac Pro is not budget friendly.

 

[Ridley]

It's barely perceptible, and certainly not a good reason to pay more for an obviously slower machine. It's just an Apple money grab exploitation of consumers who know nothing about computers.

haha barely perceptible? You are crazy man, and you are claiming that I know nothing about computers?

Iris 5100 vs HD 4000
Shader Processing Units 160 vs 16
Actual Shader Performance 314 vs 51

What else do you want to bring up? ROPs? Texture Compute? ALUs? Feature sets? The Iris 5100 is vastly superior in all of these.

http://www.game-debate.com/gpu/inde...-5100-mobile-vs-intel-hd-graphics-4000-mobile

I'm not debating that the update could have been better. I obviously would have shelled out for a quad core iris pro 5200, but come on man. Don't be ignorant! Iris 5100 vs HD 4000 barely perceptible!? Please! A game that is unplayably slow on the HD 4000 will work on the Iris 5100. That is perceivable