iMac Retina 5K disappointing: Heat and fan noise!

[Mindinversion]

The entire point is: you come barging in, posting a link to some random website listing some specs from GPU 1 and GPU 2 and expect us to believe you.

Random website:

http://support.apple.com/en-us/HT201918

which shows load wattage on the i5 290 @ 176, and the i7 295 @ 288. Apple makes it a PITA to track down CPU part numbers, but a quick check on intel's site shows i5 and i7 with non-turbo clock values both having being rated at 84w.

Yea, because the manufacturer's website is wrong too

I do amend my original statement however. The difference in wattage between the R290x and R295x appears to be somewhere between 100-110w.

as for:

Also, it helps tremendously if you use a Watt meter on an actual iMac 5K with that M295x like I did. The amount of Watt mine is doing is not even near the CPU Max in the Apple article when I'm stressing both the CPU and GPU. On average mine is at 80~82W (i5, 16GB of mem, 512GB ssd, M295x, another ssd via TB and some USB peripherals). When stressed this rises to 250~260W with only a few spikes to 280W

So at this point you're telling us that your word, without any pictures or documented evidence, without even citing what CPU AND GPU stress tests you're running. . . . is more accurate than the MANUFACTURER'S OFFICIALLY DOCUMENTED SPECIFICATIONS?

As for your wattage not coming close to the max from the specifications page, I'm willing to bet you're not running a combo of stress testing software such asPrime95 +3dMark.

You run em' both simultaneously, THEN see what you're pulling wattage wise. Don't forget pics btw.

[I'd actually really like to see the results of that TBH]

 

[AlecZ]

Sounds a lot like the original (2012) 15" rMBP. As you can imagine, it's even worse using a laptop that runs like that. I think they fixed it in later versions. Hopefully, they fix the retina iMac.

 

[leman]

If I am allowed to butt in on your very interesting discussion about the GPU power consumption

The power consumption of the i7 CPU under heavy load could be over 40W compared to the i5 one (based on this with some extrapolation). That would put the TDP difference of the GPUs at around 70W. Subtracting the GPU, a sensible max power load of the i5 CPU (with display and system components) should be somewhere around 120W, maybe more. That would give us around 60W TDP for the M290X and 130W TDP for M295X. The later is in-line with the GPUs like the 680MX that was previously used in the 27" iMac and also what Notebookcheck quotes for the card (125W TDP). And the M290X is likely to be underclocked, otherwise the Apple power consumption specs do not make much sense. Of course, Apple could be doing something weird with their thermal management again. I'd also love to see some properly done full load measurements.

 

[dyn]

Yea, because the manufacturer's website is wrong too

The only thing that is wrong is your interpretation; the website and I are correct. Apple only displays the maximum load for the cpu aka processor. The Gpuboss website does not discuss the cpu at all, therefore this link is useless. A cpu is NOT a gpu or vice versa!

I do amend my original statement however. The difference in wattage between the R290x and R295x appears to be somewhere between 100-110w.

Yet you keep failing to provide any proper source for that information whilst requiring others to provide evidence that your source is incorrect.

So at this point you're telling us that your word, without any pictures or documented evidence, without even citing what CPU AND GPU stress tests you're running. . . . is more accurate than the MANUFACTURER'S OFFICIALLY DOCUMENTED SPECIFICATIONS?

Nope since I am measuring the entire power usage of the iMac whereas Apple is only measuring the CPU.

As for your wattage not coming close to the max from the specifications page, I'm willing to bet you're not running a combo of stress testing software such asPrime95 +3dMark.

I am using the builtin UNIX tools for that as well as running those exact same tools from within a few vm's. Also simulating a normal workload like running a few 4K videos from YouTube in both Safari (which uses the HTML5 player) and Opera (which uses Flash).

Since you can't even differentiate a CPU from a GPU any further discussion will be pointless. I'm happy to discuss things further with you once you are able to differentiate those two very different components.

 

[SmallDane]

So are you set on the non-retina iMac as your replacement?

I don't game all the time, but when I do I have been using my 2011 iMac which also runs hot and "not quiet". My thinking is to go with a Mac Pro for my next desktop.

(Granted, I also use my computer for photo/video editing and software development.)

After being an iMac user since 2005 and I have gone through 3 of them, I just get the feeling deep down this machine was not made for me.

My first iMacs felt underpowered, and my later ones felt they were stressed. Maybe in about 5-10 years I'll come back to it.

Your experience sounds very much like mine. I've had 3 Macs since my first in 2007: first an iMac, then a Mac Pro, then my current 2011 iMac. It has been a mixed blessing. As you can probably tell from this thread, my tolerance of noise is low. All of my Macs have been too noisy for me. Or rather they became noisy as they aged. Though the Retina iMac is extreme in that it was noisy from the start which is why I am returning it.

I am not entirely set on a non-Retina iMac yet as replacement. Like you I am also considering a Mac Pro. It seems like low noise is one of the most agreed upon features of the Mac Pro. But the price is just insane. In addition it may actually not even give me the performance I would like.

Another option is a Hackintosh. This is becoming a more and more attractive solution considering how underwhelming Apple's hardware is in my opinion. I'm just not sure I can accept all the compatibility issues with the Hackintosh.

 

[Vapor matt]

Your experience sounds very much like mine. I've had 3 Macs since my first in 2007: first an iMac, then a Mac Pro, then my current 2011 iMac. It has been a mixed blessing. As you can probably tell from this thread, my tolerance of noise is low. All of my Macs have been too noisy for me. Or rather they became noisy as they aged. Though the Retina iMac is extreme in that it was noisy from the start which is why I am returning it.

I am not entirely set on a non-Retina iMac yet as replacement. Like you I am also considering a Mac Pro. It seems like low noise is one of the most agreed upon features of the Mac Pro. But the price is just insane. In addition it may actually not even give me the performance I would like.

Another option is a Hackintosh. This is becoming a more and more attractive solution considering how underwhelming Apple's hardware is in my opinion. I'm just not sure I can accept all the compatibility issues with the Hackintosh.

I have a maxed out 2013 iMac 27" and the same retina iMac maxed out next to it. the 2013 maxed out is silent and you will like that part of it. but i have to say my retina iMac is not noisy either to me, i can live with a little fan noise.

so far i am very pleased with the maxed out retina, as i was expecting a jumbo jet on take off after what some have said. so its a keeper for me! sorry to hear yours was to noisy for you.

 

[gmfeier]

I've had mine with the M295X for 10 days now...

...and it has exceeded my expectations. Only a whisper of fan noise with my most demanding apps. Definitely glad I got it.

 

[hjalte]

...and it has exceeded my expectations. Only a whisper of fan noise with my most demanding apps. Definitely glad I got it.

Awesome.
Is it the i5 or the i7 CPU?

 

[dyn]

And also importantly, what do you use it for?

 

[joema2]

...The difference in wattage between the R290x and R295x appears to be somewhere between 100-110w....

You cannot conclude that. The Apple spec is for two different retina iMacs with different hard drives, different memory, and different CPUs. Their number for "CPU Max" is *total* system power consumption including all components, efficiency loss from the power supply itself, etc. IOW how much power is pulled from the wall plug. How do we know that? It's stated specifically at the bottom of that same page: http://support.apple.com/en-us/HT201918

So the difference in power consumption between the two iMacs is the product of many changed components -- not just the GPU. Therefore you cannot subtract the numbers and attribute the difference to the GPU alone.

This is just common sense. IF the *difference* in wattage between M290 and M295X was 100-110w, then the M290 would have to be running on almost zero power. Otherwise the M295X would require a gigantic cooler. Therefore the difference can't be that great.

The M290 is probably very roughly in the 100w range, and the M295X is probably modestly above that. Whether it's 110w or 120w, I don't think anybody knows. However it's physically impossible that it is 250w as posted on the GPUBoss site, or anything close to that number.

 

[gmfeier]

Awesome.
Is it the i5 or the i7 CPU?

i7.

----------

And also importantly, what do you use it for?

Web surfing, general purpose computing, including business applications like Excel and FileMaker Pro and some gaming, like Civilization V. I'm sort of retired, but have a small business and I spend a lot of time on the Mac because I can. ;-)

 

[leman]

The only thing that is wrong is your interpretation; the website and I are correct. Apple only displays the maximum load for the cpu aka processor. The Gpuboss website does not discuss the cpu at all, therefore this link is useless. A cpu is NOT a gpu or vice versa!

You are wrong. Those numbers are the total system power consumption. A Haswell CPU would blow up before it can push 280 Watt. Apple's wording is confusing.

----------

You cannot conclude that. The Apple spec is for two different retina iMacs with different hard drives, different memory, and different CPUs. Their number for "CPU Max" is *total* system power consumption including all components, efficiency loss from the power supply itself, etc. IOW how much power is pulled from the wall plug. How do we know that? It's stated specifically at the bottom of that same page: http://support.apple.com/en-us/HT201918

As I have written above, the CPU + other components would account for max. 40-50 Watt difference. Furthermore, I don't see how M290X can push 100W and still be consistent with Apple's specs. The i5 CPU + system will push significantly more then 76W under load. So either the M290X TDP is lower or Apple's numbers are not correct.

I definitely agree with you that the M295X can't be a 250W part. As you say, something in the ballpark of 125W is much more realistic.

 

[dyn]

Web surfing, general purpose computing, including business applications like Excel and FileMaker Pro and some gaming, like Civilization V. I'm sort of retired, but have a small business and I spend a lot of time on the Mac because I can. ;-)

Ah, the iMac 5k with either the 290 or 295 will be silent with those tasks.

You are wrong. Those numbers are the total system power consumption. A Haswell CPU would blow up before it can push 280 Watt. Apple's wording is confusing.

Ha, finally 1 person who's awake! I also believe that Apple actually means the entire iMac and not the CPU (lot's of companies do this, for example there are no computercase holders, only CPU-holders). With Mactracker you can quickly find out which i5 and i7 CPU is being used in the iMac 5K. They have a TDP somewhere around 84W. Even though TDP is something entirely different than actual CPU usage (in Watts) it does give you a good idea of what to expect (dissipating less heat than a CPU creates is rather idiotic). This leaves less than 200W for the other components (display, wifi/bluetooth, ssd, hdd, GPU) which is lower than the aforementioned 250W for the GPU.

 

[leman]

I also believe that Apple actually means the entire iMac and not the CPU (lot's of companies do this, for example there are no computercase holders, only CPU-holders).

Ok, my mistake then, I have interpreted your post differently

This leaves less than 200W for the other components (display, wifi/bluetooth, ssd, hdd, GPU) which is lower than the aforementioned 250W for the GPU.

Exactly. See also my posts above where I try to give an estimate based on what we know about those CPUs.

 

[JGRE]

Thanks for all the replies. I tried to make a video of the problem. But I'm not sure how useful it is since you really have to be in the room to experience how loud the fan noise is.

YouTube: video

The sensors shown in he menu bar are as follows:

1. Fan speed
2. GPU Die
3. CPU Core 1
4. CPU Core 2
5. CPU Core 3
6. CPU Core 4

The machine is doing nothing other than showing this video stream. I should note that this is with Chrome using its own Flash plugin. Using the Flash plugin from Adobe (either by disabling the built-in one in Chrome or by using Safari) does improve performance somewhat. It is still pretty poor performance though.

I am still leaning mostly towards returning it. The only thing is that I don't know what to replace it with. I tend to want the biggest and fastest when buying new. But some of you have suggested that this may be a bad idea if I also want low noise. It just seems quite counter-intuitive getting a slower model in order to reduce noise. Wouldn't a slower GPU be pressured more and thus produce even more heat?

Sound like a jet engine, this is not normal.

 

[joema2]

....the Retina iMac is extreme in that it was noisy from the start which is why I am returning it....

My best guess is something was unique about your iMac, since playing that video caused much higher CPU activity than on my 2013 iMac 27 with i7. I can tell that by watching the video you made. I was using Chrome and "best" quality. Obviously if CPU is very high, fan speed will be high. This has nothing to do with the GPU -- your CPU cores were high. My CPU activity on each core was about 15%. Yours was about 80%.

That implies a software issue, yet you re-installed OS X and it made no difference, so I can't explain that.

If I were you I'd be tempted to try another retina iMac with M295X and see if the same behavior happens. It's a lot cheaper than a nMP.

You say you're sensitive to noise, but if my iMac was frequently at 2700 rpm, it would bug me also -- it's very loud at that speed. Fortunately it rarely does that.

 

[SmallDane]

If I were you I'd be tempted to try another retina iMac with M295X and see if the same behavior happens. It's a lot cheaper than a nMP.

Thanks for the encouragement but I'm not tempted. Although some people say they have no issues with the retina iMac, it does seem other people agree that it runs quite hot. So I don't think it's a machine for me. To be honest I wasn't overly impressed with the display either. It's nice but not overwhelming. And if the high resolution is the (at least indirect) cause of the heat problems, it's definitely not worth it for me.

The 2013 iMac is now more tempting to me. Though the Mac Pro is so damn nice in many ways. Except the price.

 

[grame]

I just ran that annoyingly tedious video on my 4Ghz i7/290. The fan came on after a few minutes - but it was barely perceptible (and this in a very quiet room). I think it pushed up to maybe 1400rpm (it dropped very sharply once I came out of full-screen to grab the iStats data).

I do audio on mine so I tried some DAW stress tests. The fan will come on when the cpus are really pushed - but them this was the case with my Macbook. Handbrake will set the fan going - but it encodes so fast that it's over before you really get a chance to measure it. The sound of the fan is more bearable than the Macbook - it sounds sort of like a soft electric kettle or immersion heater. I don't play games or watch video on it (that I cast to my TV via Chromecast) so it's not something I expect ever to have to endure.

I tried some audio projects - and tons of plugs still won't push it over 30%. I would maybe consider swapping it for a new Mac Pro Hex... but otherwise it's staying.

 

[Cape Dave]

Thanks for the encouragement but I'm not tempted. Although some people say they have no issues with the retina iMac, it does seem other people agree that it runs quite hot. So I don't think it's a machine for me. To be honest I wasn't overly impressed with the display either. It's nice but not overwhelming. And if the high resolution is the (at least indirect) cause of the heat problems, it's definitely not worth it for me.

The 2013 iMac is now more tempting to me. Though the Mac Pro is so damn nice in many ways. Except the price.

It is sad and ironic that something as simple as proper application of CPU/GPU grease could cause these major differences. Who knows? But it sure sounds to me that there is a huge variation in noise from the same configurations.

I practically had one hand on the box to return mine, but a reformat fixed my particular issues. I was going to get a Pro with a decent 4K monitor.

This way I get to save the extra grand for my NEXT computer, which most likely WILL be a Pro and an Apple 4 or 5K monitor! The pro's take silence and cooling seriously, as do I

 

[fastlanephil]

Thanks for the encouragement but I'm not tempted. Although some people say they have no issues with the retina iMac, it does seem other people agree that it runs quite hot. So I don't think it's a machine for me. To be honest I wasn't overly impressed with the display either. It's nice but not overwhelming. And if the high resolution is the (at least indirect) cause of the heat problems, it's definitely not worth it for me.

The 2013 iMac is now more tempting to me. Though the Mac Pro is so damn nice in many ways. Except the price.

If you're applications use multi-core 64 bit then a refurbished 2009(flashed)/2010 upgraded twelve core might be an option. The 3.46GHz twelve core rates 30,000+ for 64-bit multi-core with Geekbench.

You won't get the Apple warranty so it's a little bit of a gamble that the seller will make good on it if there is an issue but you're getting one for about the price of the base retina iMac. They do have some fan noise but at least you won't have it in your face.

Also there is no possibility of Thunderbolt with these models but you can add USB 3.0 ports. Having a certain amount of tech knowledge might be a good idea also. Having a lot, you could build it yourself and save money. It beats building a hackintosh.

I've been looking at one of these as an option because the current desktop options from Apple/Intel don't look that attractive to me.

 

[SmallDane]

This way I get to save the extra grand for my NEXT computer, which most likely WILL be a Pro and an Apple 4 or 5K monitor! The pro's take silence and cooling seriously, as do I

That was my understanding as well. But then I made the mistake of reading about the coil whine that some nMP users seem to be experiencing.

But maybe it's better not to read about everything that can potentially go wrong.

----------

If you're applications use multi-core 64 bit then a refurbished 2009(flashed)/2010 upgraded twelve core might be an option. The 3.46GHz twelve core rates 30,000+ for 64-bit multi-core with Geekbench.

I don't think Apple sells refurbished models here in Denmark unfortunately.

I actually own a 2009 Mac Pro though. But I gave up on it a long time ago after it completely killed no less than 2 video cards and several RAM modules. And the 3rd video card (actually the original) is also unstable. I kid you not. I think it's cursed!

 

[Nosferax]

One thing to consider in regard to noise, is where the iMac is situated. If it's back is close to a wall, the sound level will be more loud then if the back is facing an empty space. The same thing happens with speakers btw due to reverberation.

 

[fastlanephil]

That was my understanding as well. But then I made the mistake of reading about the coil whine that some nMP users seem to be experiencing.

But maybe it's better not to read about everything that can potentially go wrong.

----------



I don't think Apple sells refurbished models here in Denmark unfortunately.

I actually own a 2009 Mac Pro though. But I gave up on it a long time ago after it completely killed no less than 2 video cards and several RAM modules. And the 3rd video card (actually the original) is also unstable. I kid you not. I think it's cursed!

Apple dosen't sell these upgraded Mac Pro's. You'll find them for sale by individuals on eBay and there are a few small businesses that deal in them. I doubt if you could get one shipped to Denmark. If you were pretty techie you could just buy the CPUs and do the conversion yourself. You do need to do a successful ROM firmware flash for the 2009 so the computer thinks it's a 2010.

 

[thundersteele]

I found a review that addresses heat and noise in the 5k iMac. Unfortunately it is in french... http://www.macg.co/tests/2014/10/test-des-imac-retina-27-pouces-fin-2014-85217/page/0/5

Notes from the review:
The i5 newer consumes more than 60W under load, the i7 happily exceeds 85W (just the CPU, not total system consumption).
The i5 + M290 does a handbrake encode without warming up very much (60C with base fan speeds)
The i7 + M295 heats up (CPU to 100C), and fans at 2400 rpm. Article claims that noise and heat is on par with the 2013 model.
Total system power consumption with handbrake: 130W for the base model, 160W for the high end model.

My comments:
It seems clear that the handbrake test doesn't task the GPU fully. Unfortunately the review doesn't report fan speeds or temperature while gaming. One thing that seems relevant though is that while the i5 and i7 nominally have the same TDP, only the i7 actually reaches the max TDP (88W on intels website).


BTW, the review also clearly says that they consider the M290 underpowered for this machine

 

[hjalte]

I found a review that addresses heat and noise in the 5k iMac. Unfortunately it is in french... http://www.macg.co/tests/2014/10/test-des-imac-retina-27-pouces-fin-2014-85217/page/0/5

Notes from the review:
The i5 newer consumes more than 60W under load, the i7 happily exceeds 85W (just the CPU, not total system consumption).
The i5 + M290 does a handbrake encode without warming up very much (60C with base fan speeds)
The i7 + M295 heats up (CPU to 100C), and fans at 2400 rpm. Article claims that noise and heat is on par with the 2013 model.
Total system power consumption with handbrake: 130W for the base model, 160W for the high end model.

My comments:
It seems clear that the handbrake test doesn't task the GPU fully. Unfortunately the review doesn't report fan speeds or temperature while gaming. One thing that seems relevant though is that while the i5 and i7 nominally have the same TDP, only the i7 actually reaches the max TDP (88W on intels website).


BTW, the review also clearly says that they consider the M290 underpowered for this machine

Chrome can translate the page, so it's understandable