2014/2015 CPU/GPU Heat Conclusions

[vetruvian]

Hope this video helps some of you. Heat was one of the last issues keeping me from ordering a new iMac. I hope the comparison in the below video is typical of all 2015 iMacs:

Conclusions:
1. 2015 GPU runs about 10-15% cooler on identical load tests
2. 2015 GPU runs about 10-15% cooler on identical load tests
3. Apple has apparently implemented new fans and/or software that ramps up fans quicker

Shout out to Max Yuryev! Excellent video!

 

[MaxYuryev]

Thanks for sharing. Specs are not listed in the video but in video description. For those curious;

4ghz i7
256GB SSD
M395X 4GB
24GB RAM (ups lost my second set of 16GB somehow, waiting for a replacement)

The 2014 model I tested was similar 4GHZ i7 and M295X.

I used both iStat menu and TG Pro check temps/speeds and TG Pro to do manual adjustments.

 

[vetruvian]

Thanks for sharing. Specs are not listed in the video but in video description. For those curious;

4ghz i7
256GB SSD
M395X 4GB
24GB RAM (ups lost my second set of 16GB somehow, waiting for a replacement)

The 2014 model I tested was similar 4GHZ i7 and M295X.

I used both iStat menu and TG Pro check temps/speeds and TG Pro to do manual adjustments.

View attachment 595667

Thanks Max!

 

[Smarky]

Awesome! This is the news we were all waiting for, thank you for posting! Looks like for everyone complaining Apple did at least listen.

 

[Rmafive]

Glad to see a more comprehensive analysis of heating in the updated model. Thanks for testing!

 

[Zarniwoop]

Thanks! This was just the info I've been looking for.

 

[seekersu]

Just curious, since i5 has 40% lower TDP than i7-6700k (65W vs 91W), will iMac with i5 cooler than i7?

 

[cynics]

Just curious, since i5 has 40% lower TDP than i7-6700k (65W vs 91W), will iMac with i5 cooler than i7?

Typically, that's the way it's been in the past. It's also less powerful so that's your trade off.

 

[seekersu]

So if I use Vsync to limit CPU load, will i5 still be cooler than i7 or they will be the same because they seems do the same job??

 

[cynics]

So if I use Vsync to limit CPU load, will i5 still be cooler than i7 or they will be the same because they seems do the same job??

CPU generally doesn't get that hot when gaming anyway (using vsync) so in that regard it won't matter.

Also if you are trying to synthetically give a faster CPU or GPU the same load as a slower CPU or GPU then you might as well get the slower one to begin with.

 

[seekersu]

Thank you for explanation

 

[The Doctor11]

That was the best video I've watched in a long time! This video just makes me even more excited for my October 13th order to get here .

 

[SBruv]

Nice work, Max. Certainly reflects my experience with the same machine.

 

[steve62388]

Thanks for sharing. Specs are not listed in the video but in video description. For those curious;

4ghz i7
256GB SSD
M395X 4GB
24GB RAM (ups lost my second set of 16GB somehow, waiting for a replacement)

The 2014 model I tested was similar 4GHZ i7 and M295X.

I used both iStat menu and TG Pro check temps/speeds and TG Pro to do manual adjustments.

View attachment 595667

That's odd, because Barefeats' results contradict yours. Maybe you should get your heads together to make sure you're using consistent testing methodology.


What about heat and noise? The M395X runs just as hot as the M295X. Both quickly climb from 45C to over 100C when stressed. And at that temperature, the internal fan is running at the 2300RPM max (versus 1200RPM idle). Ambient room temperature was 21C.

A rough measure of sound level using an iPhone app (dB Meter Pro) positioned under the iMac reported a rise from 48dB to 58dB in both cases.
Source: http://barefeats.com/imac5k13.html

 

[vetruvian]

That's odd, because Barefeats' results contradict yours. Maybe you should get your heads together to make sure you're using consistent testing methodology.


What about heat and noise? The M395X runs just as hot as the M295X. Both quickly climb from 45C to over 100C when stressed. And at that temperature, the internal fan is running at the 2300RPM max (versus 1200RPM idle). Ambient room temperature was 21C.

A rough measure of sound level using an iPhone app (dB Meter Pro) positioned under the iMac reported a rise from 48dB to 58dB in both cases.
Source: http://barefeats.com/imac5k13.html

I'll go with the guy who spent hours comparing and preparing a video with over 8 minutes of analysis just on heat alone. I love Barefeats comparisons, but they only had three sentences on the subject in the above article.

 

[steve62388]

I'll go with the guy who spent hours comparing and preparing a video with over 8 minutes of analysis just on heat alone. I love Barefeats comparisons, but they only had three sentences on the subject in the above article.

Whatever floats your boat, Barefeats has been around for an incredibly long time. It's worth somebody trying to figure out what the real story is, but it won't be me I don't have one.

 

[MaxYuryev]

Whatever floats your boat, Barefeats has been around for an incredibly long time. It's worth somebody trying to figure out what the real story is, but it won't be me I don't have one.

I'm really not sure about barefeats and what testing methods he/they use but Mine is using Unigine heaven 4 (free version) on extreme preset so anyone can download it and test it. For the CPU Test I'm using wondershare to convert 4K h.265 to ProRes LT, which uses all the cores maxed out at 100%. The video clips being converted are the same exact ones I used for my testing of the late 2013 6 core, D700 MacPro as well as both iMacs.

I guess I can't really "Prove" my work being quality of consistent, but there would be no reason for me to make something up.

 

[steve62388]

I'm really not sure about barefeats and what testing methods he/they use but Mine is using Unigine heaven 4 (free version) on extreme preset so anyone can download it and test it. For the CPU Test I'm using wondershare to convert 4K h.265 to ProRes LT, which uses all the cores maxed out at 100%. The video clips being converted are the same exact ones I used for my testing of the late 2013 6 core, D700 MacPro as well as both iMacs.

I guess I can't really "Prove" my work being quality of consistent, but there would be no reason for me to make something up.

Fair enough. But we at least need more data points.

When last year's model was being tested one of the members here (I can't remember their name) had a M295X which ran incredibly cool. Nobody could really explain it. Was it a problem with their testing, were they making it up, did they have some sort of freak on their hands? Who knows, but in the end as more data came in it became settled that the M295X runs hot.

 

[Aspekt]

I am guessing something is off...

Late 2015 iMac, i7, 16GB RAM, m395x, 1TB SSD.

Only Safari is open.

 

[mattoligy]

I am guessing something is off...

Late 2015 iMac, i7, 16GB RAM, m395x, 1TB SSD.

Only Safari is open.

Wow, wtf...

 

[Travisimo]

I am guessing something is off...

Late 2015 iMac, i7, 16GB RAM, m395x, 1TB SSD.

Only Safari is open.

You sure those are Fahrenheit degrees? iStat defaults to Fahrenheit. I have the i7/395x/512ssd and my CPU is hovering in the 40's Celsius with a light load. Under heavy load (encoding a 1080p video), the CPU cores got up into the 80's.

 

[h9826790]

I am guessing something is off...

Obviously those temperature are in degree F, and all parameters well within normal range.

 

[dyn]

I'll go with the guy who spent hours comparing and preparing a video with over 8 minutes of analysis just on heat alone. I love Barefeats comparisons, but they only had three sentences on the subject in the above article.

I'd rather go with the guy who has his testing methodology straight and is transparant about it. Not some random website or lengthy YouTube video. Or in other words: quantity and quality are still very separate things, they are not the same. Oh and let's not forget what happens in the scientific world and the medical world: try to find scientists/doctors who agree with each other...it is easier to find a needle in a haystack. Those guys have proper testing methodology and yet they contradict each other continuously (is GSM radiation harmless or bad? ). I'm not surprised that Barefeats and the video have opposite experiences.

That said, it is nice to see some extensive testing. That's what we need instead of all the theoretics and whining about specs so thanks! The 10-15% difference is also to be expected from the new architecture. Good to see it meets the expectations.