i5/575 vs. i5/570 [27 inch 5K]

[mcomp112]

Hi all. I've ordered the i5/575 configuration because of the slightly better GPU. But, I'm concerned about the CPU that comes with this mid-tier configuration, which is clocked higher than the base-tier CPU. I fear that this will likely to lead to higher temperatures.

Is this something to seriously consider? It would be nice to have more GPU horsepower for Adobe CC but I definitely, definitely prefer a quieter machine.

Curious to hear what you guys think.

Thanks in advance!

 

[Falcon80]

Should be fine. Even it is any hotter, I believe it will be pretty unnoticeable. You should only be concerned if you are considering the i7 version.

 

[MacDevil7334]

You only really need to worry if you get the top end i5 or the i7, since these are 91 W chips. The i5 7600 chip in the model you ordered is a 65 W chip and tests have shown that it runs much cooler (around 20 degrees under full load). I think the mid-tier model is the sweet spot for performance and temperatures, which is why I ordered this configuration myself for Lightroom/Photoshop work. Just waiting patiently for July 5 to get here...

 

[Moriarty]

It will be fine. My i5 7600K / Radeon 580 model gets to ~1800 rpm playing games, which is still rather quiet. Other than games, there's nothing I can do to make it move off the idle fan speed (1200 rpm). My usage of Photoshop, Lightroom, and coding isn't even close to being intensive enough to lift the fan off idle.

With the slightly slower CPU and GPU you'll probably never hear the fan lift off idle, no matter how much you push the machine.

@MacDevil7334: Intel's TDP ratings are unreliable. The i5 7600K uses less power than the i7. Under CPU load, the i5 7600K can't easily make the iMac spin its fans loudly, but the i7 7700K can.

 

[MacDevil7334]

@MacDevil7334: Intel's TDP ratings are unreliable. The i5 7600K uses less power than the i7. Under CPU load, the i5 7600K can't easily make the iMac spin its fans loudly, but the i7 7700K can.

Fair point. The temperature differential was coming from the test Toms Hardware did on these chips. Under full load, the 7600K got up to almost 90 C while the 7600 topped out at 70 C. While you might not be running at full load very often in real world useage, that's a lot of additional heat for a 0.3 GHz gain in clock speed. I definitely agree that either i5 will be better than the i7 from a thermal standpoint though.

http://www.tomshardware.com/reviews/intel-kaby-lake-core-i7-7700k-i7-7700-i5-7600k-i5-7600,4870-11.html

 

[Falcon80]

It will be fine. My i5 7600K / Radeon 580 model gets to ~1800 rpm playing games, which is still rather quiet. Other than games, there's nothing I can do to make it move off the idle fan speed (1200 rpm). My usage of Photoshop, Lightroom, and coding isn't even close to being intensive enough to lift the fan off idle.

With the slightly slower CPU and GPU you'll probably never hear the fan lift off idle, no matter how much you push the machine.

@MacDevil7334: Intel's TDP ratings are unreliable. The i5 7600K uses less power than the i7. Under CPU load, the i5 7600K can't easily make the iMac spin its fans loudly, but the i7 7700K can.

May I know what kind of coding you do on this machine? Running Xcode?

 

[Moriarty]

May I know what kind of coding you do on this machine? Running Xcode?

LaTeX, MATLAB, and FORTRAN.

Fully loading the CPU to 100% with an FPU-intensive workload, I see from iStat Menus that the CPU draws about 71 W, temp 93 degrees celsius, and the fan will spin up to around 1500 rpm. Still very quiet.

Almost all workloads are less intensive (e.g. Handbrake takes 55 W), and won't come close to spinning the fans – including code compilation.

Notably, the GPU cools better than the CPU. For example, ~75% GPU load (80 W) + 2.5 threads of CPU load (42 W) = high temps but almost idle fan (~1250 rpm).

You need to push the CPU + GPU at the same time – and push them really hard – to make it loud.

 

[mcomp112]

So, there should be no difference in terms of temperatures?

I keep reading that clock speeds matter just as much as TDP (which is an arbitrary number, it seems) in determining system temperature and there is some good evidence for it being true. See the example of the 7600K and 7700K. Both are rated at the same TDP (91W) but the latter CPU is significantly warmer and can lead to a noisier system.

 

[EugW]

So, there should be no difference in terms of temperatures?

I keep reading that clock speeds matter just as much as TDP (which is an arbitrary number, it seems) in determining system temperature and there is some good evidence for it being true. See the example of the 7600K and 7700K. Both are rated at the same TDP (91W) but the latter CPU is significantly warmer and can lead to a noisier system.

There is variation amongst CPUs, but testing on the PC side indicates the 7600K is usually cooler than the 7700K, which makes sense because the 7600K has no HyperThreading and is also lower clocked.

 

[mcomp112]

There is variation amongst CPUs, but testing on the PC side indicates the 7600K is usually cooler than the 7700K, which makes sense because the 7600K has no HyperThreading and is also lower clocked.

Yes, that was my point. The i5-7600 should run warmer than the i5-7500 as well, then.

 

[EugW]

Yes, that was my point. The i5-7600 should run warmer than the i5-7500 as well, then.

Maybe, maybe not. It is possible they might be similar in power output. The 7500 is lower clocked, but otherwise it is identical to the 7600.

It is possible a particular 7600 was labelled and sold as such just because it was able to achieve a higher speed at the same voltage and power as another chip that could only achieve 7500 speeds at that voltage.

 

[mcomp112]

Maybe, maybe not. It is possible they might be similar in power output. The 7500 is lower clocked, but otherwise it is identical to the 7600.

It is possible a particular 7600 was labelled and sold as such just because it was able to achieve a higher speed at the same voltage and power as another chip that could only achieve 7500 speeds at that voltage.

I thought it's more likely than not that higher clock speeds = higher temperatures. Are there any real world CPU examples that do not follow this pattern?

Even the general consensus on this forum is that, in terms of heat and noise, the configurations can be ranked as follows (starting from lowest heat and noise):

1. i5/570
2. i5/575
3. i5(3.8)/580
4. i7/575
5. i7/580

 

[EugW]

I thought it's more likely than not that higher clock speeds = higher temperatures. Are there any real world CPU examples that do not follow this pattern?

Yes. Binning can address this.

Even the general consensus on this forum is that, in terms of heat and noise, the configurations can be ranked as follows (starting from lowest heat and noise):

1. i5/570
2. i5/575
3. i5(3.8)/580
4. i7/575
5. i7/580

Not sure about 3 vs 4. Depends on the GPU load I guess.

 

[mcomp112]

Yes. Binning can address this.


Not sure about 3 vs 4. Depends on the GPU load I guess.

I can't imagine there would be very much that separates them during binning. But, when it comes to use in a compact all-in-one enclosure, the difference could be magnified, leading to the i5/570 configuration being cooler and quieter.

 

[Asclepio]

#i5/570masterrace

 

[Quash]

I can't imagine there would be very much that separates them during binning. But, when it comes to use in a compact all-in-one enclosure, the difference could be magnified, leading to the i5/570 configuration being cooler and quieter.

The cpu's in the base and the mid model use the same power. Any difference measured is an i5-7500 and i5-7600 is just sample by sample difference (based on luck). There is a case to be made that the i5-7600 might more power efficient statistically, because they are the best binned samples on that voltage... In real world it's just not a factor.
The 575 should use slightly more power than the 570 under full load. Not under full load again no difference.
So if you want a little more horse power (5-10%) for probably no noticable noise gain get the middle one. If that doesn't matter save 200$ get the base model. It's not like the base model and the max model this generation are in different leagues anyway. Any upgrades are always bad return on investment, only buy if needed.(or you don't care about money)
(Full disclosure I have the middle with SSD one on the way)

 

[Ph.D.]

My advice: Base with SSD beats middle without.

If the extra $200 or so means nothing to you, and you plan on an SSD anyway, get the middle (3.5 GHz) model, mostly for the GPU, which has a higher percentage performance difference than for the CPU.

However, Apple is charging a LOT for these slight upgrades compared to the component cost differences. They want a "good, better, best" differentiation, and the middle one is better in only minor ways. The difference in cost between the two CPU's is maybe $10-15 or so, and similar for the GPU. It's the Apple tax in high gear for these minor changes.

These two models should produce no noticeable real-world differences. Benchmarks and perhaps a few certain long, hard production runs (transcoding, etc.) will still only be a few percent different. Heat will be slightly higher for the middle model, but probably not very noticeably. (No, the base model will produce less heat; the TDP numbers are for worst-case maxed-out temperature tolerance for setting throttling points, and not a measure of actual power used.)

In my case, I went with the base model (with 512GB SSD) because I was more concerned about any increase in noise than I was about a slight performance advantage. I would have been more unhappy if the noise was higher with the middle model than I would have been if the performance was slightly lower with the base model.

The other thing about the base model is that, for many people, saving the $200 or so can help you to upgrade to an all-SSD system, which has large advantages in speed and lower heat, noise and vibration. Do that!

In the end, the base model (with SSD) is a very snappy system whose fans rarely if ever budge from their minimum speed. Games at 1920x1440, with maxed out detail and anti-aliasing, etc., seem to run at full frame rates (locked to 60 Hz in all models) with no throttling and minimal if any increase in fan speed. (The higher-end 90+W chips will definitely run a lot hotter - for some that will be OK.)

These are nicely improved machines in many respects. You can't go wrong with any of them.

 

[EugW]

My advice: Base with SSD beats middle without.

If the extra $200 or so means nothing to you, and you plan on an SSD anyway, get the middle (3.5 GHz) model, mostly for the GPU, which has a higher percentage performance difference than for the CPU.

However, Apple is charging a LOT for these slight upgrades compared to the component cost differences. They want a "good, better, best" differentiation, and the middle one is better in only minor ways. The difference in cost between the two CPU's is maybe $10-15 or so, and similar for the GPU. It's the Apple tax in high gear for these minor changes.

These two models should produce no noticeable real-world differences. Benchmarks and perhaps a few certain long, hard production runs (transcoding, etc.) will still only be a few percent different. Heat will be slightly higher for the middle model, but probably not very noticeably. (No, the base model will produce less heat; the TDP numbers are for worst-case maxed-out temperature tolerance for setting throttling points, and not a measure of actual power used.)

In my case, I went with the base model (with 512GB SSD) because I was more concerned about any increase in noise than I was about a slight performance advantage. I would have been more unhappy if the noise was higher with the middle model than I would have been if the performance was slightly lower with the base model.

The other thing about the base model is that, for many people, saving the $200 or so can help you to upgrade to an all-SSD system, which has large advantages in speed and lower heat, noise and vibration. Do that!

In the end, the base model (with SSD) is a very snappy system whose fans rarely if ever budge from their minimum speed. Games at 1920x1440, with maxed out detail and anti-aliasing, etc., seem to run at full frame rates (locked to 60 Hz in all models) with no throttling and minimal if any increase in fan speed. (The "K" chips will definitely run a lot hotter - for some that will be OK.)

These are nicely improved machines in many respects. You can't go wrong with any of them.

Yep, good point.

I bought the 4.2 GHz i7 with 1 TB SSD, but if I HAD to choose between i7 and Fusion vs base and 1 TB SSD, I'd choose the base and 1 TB SSD for my usage.

 

[whatevs]

These two models should produce no noticeable real-world differences.

just curious, where are you getting benchmarks/reviews of the 570 vs 575?

I know I will choose an 512 SSD model, but 570 vs 575 vs 580 is a big debate for me.
Thanks!

 

[Ph.D.]

just curious, where are you getting benchmarks/reviews of the 570 vs 575?

I know I will choose an 512 SSD model, but 570 vs 575 vs 580 is a big debate for me.
Thanks!

I'm not. For the I5 CPU's, the difference is approximately 3.5/3.4. That's an extremely small difference (3%). The GPU (575 vs. 570) has a bigger performance difference. Benchmarks are out there, but as many people recognize, they tend not to correlate well with real-world perceived performance differences anyway, except perhaps under fairly unusual conditions.

The infuriating thing about Apple is the artificial differentiation they impose. You can't pick 3.4 GHz I5 with 580 GPU, etc.

(If it helps, I would have picked the 3.5GHz I5 and 575 if I had to do it over again, but honestly the 3.4/570 is a great machine anyway.)

 

[mcomp112]

I'm not. For the I5 CPU's, the difference is approximately 3.5/3.4. That's an extremely small difference (3%). The GPU (575 vs. 570) has a bigger performance difference. Benchmarks are out there, but as many people recognize, they tend not to correlate well with real-world perceived performance differences anyway, except perhaps under fairly unusual conditions.

The infuriating thing about Apple is the artificial differentiation they impose. You can't pick 3.4 GHz I5 with 580 GPU, etc.

(If it helps, I would have picked the 3.5GHz I5 and 575 if I had to do it over again, but honestly the 3.4/570 is a great machine anyway.)

I would buy a 3.4/580 configuration in a heart beat because I care more for GPU power than CPU speed. But, no way am I getting one of those 91W CPUs for the 580. Too much of a noise/temperature trade off, imo.

Out of curiosity, why would you go for the middle one?

 

[EugW]

I'm not. For the I5 CPU's, the difference is approximately 3.5/3.4. That's an extremely small difference (3%).

Well, it could be 3.9/3.6, which is 8.3%. Still not a huge amount, and probably not noticeable to most people, but almost 3x that 3% number. Geekbench multi-core benchmarks seem to bear that out.

I would buy a 3.4/580 configuration in a heart beat because I care more for GPU power than CPU speed. But, no way am I getting one of those 91W CPUs for the 580. Too much of a noise/temperature trade off, imo.

FWIW, my i7 4.2 GHz iMac is silent for regular day-to-day bursty usage. The fan ramps up when I put a sustained high load on it though, like video encoding. But then again, the i7-7700K can be something like 25-30% faster than the i5-7600K in that task, and as much as 50% faster than the i5-7500, so it's all a trade off.

 

[mcomp112]

FWIW, my i7 4.2 GHz iMac is silent for regular day-to-day bursty usage. It ramps up when I put a sustained high load on it though, like video encoding.

Right, but you also value/need that CPU speed. Getting an i7 is prudent in that situation. For my workflow, though, it's really not as important as GPU; hence, my decision to get a 575 over a base 570.

 

[DisMyMac]

My mac mini is already loud. The chime sounds like an explosion.

FWIW, my i7 4.2 GHz iMac is silent for regular day-to-day bursty usage. The fan ramps up when I put a sustained high load on it though, like video encoding.

I like a little fan noise. It's the 580 itself I fear will burn out or turn the screen YELLOW.

 

[Ph.D.]

Well, it could be 3.9/3.6, which is 8.3%. Still not a huge amount, and probably not noticeable to most people, but almost 3x that 3% number.

Sure, that's true, assuming it stays in boost mode forever. That's only likely with truly excellent cooling, or those would be the stock speeds. I was using the more conservative base speed.
[doublepost=1498589346][/doublepost]

I would buy a 3.4/580 configuration in a heart beat because I care more for GPU power than CPU speed. But, no way am I getting one of those 91W CPUs for the 580. Too much of a noise/temperature trade off, imo.

Out of curiosity, why would you go for the middle one?

Because both base (3.4GHz) and middle machines (3.5GHz) are cooler max 65W units. That's not such a big deal speed-wise, but the 575 is a more noticeable step up over the 570 (in benchmarks, at least). Having learned more since the purchase, I'd have gone with the middle model (or, like you, with a 580 if only that was available). But the base one is more than adequate for my purposes anyway - it's very nice!