The new iMac is a lot noisier!

[propower]

Thanks EugW All of your posts have been very informative, and I did see your numbers. I guess I'm wondering how many more i7 owners can confirm the same?

I plan on monitoring my temps and will be happy to report back. 95C just seems so high and I just can't help but wonder if you were an outlier or if it's to be expected with heavy load.

As always - totally depends on what you are doing with it. A medium sized audio session runs the CPU ~25% on the i7 and temps are in the 60s to 70s. But then I add in my mirror monitor chain (recording studio environment - 2 mirrored monitors driven by optical thunderbolt daisy chain). Well now the GPU is minimum 35W and I assume it is actually the power supply that adds another 7 to 10degrees of heat in the box and rises the CPU to now 70 to 80degC for this load. But if your use just hits 25 to 50% for a brief moment you won't see sustained high temps. Now if you are video rendering - the CPU will almost certainly be 95 to 100degrees C in very short order and fans 2K to full on. At least that was true for me. I tried setting the fan to 1800 all the time - made pretty much no difference. Other Video editing (4K etc) and many other tasks - you have to find a user that knows.

In my synthetic testing (Yes test) 50% load put the CPU to 95degC and fans went 2K to max on the i7.

Stress test i7 4.2 (at idle ~50degC 1198rpm 10W with Second monitor through TB optical)
1 YES = 15%CPU 27W after 10min Peak~56 to 80degC (Peak moving around from core to core)
2 YES = 26%CPU 38W CPU after 10min Peak~60 - 80degC
3 YES = 38%CPU 58W CPU after 10min Peak~75-91degC 1200rpm
4 YES = 51%CPU 69W CPU …… 95degC ~2K to 2.4K rpm

 

[user1234]

So the i7 has a Tjunction of 100°C, and stabilizes at 95°C under heavy load, and my rMBP i7 has a Tjunction of 105°C and stabilizes around 100°C. It would seem Apple engineered these machines to run at Tjunction-5°C at heavy load.

 

[joema2]

Thanks EugW All of your posts have been very informative, and I did see your numbers. I guess I'm wondering how many more i7 owners can confirm the same?

I plan on monitoring my temps and will be happy to report back. 95C just seems so high and I just can't help but wonder if you were an outlier or if it's to be expected with heavy load.

Max Yuryev tested the 2015 iMac 27 i7 vs the 2017 and found the 2017 ran cooler and quieter under the same load:

This can especially be seen in the temps and fan speeds when running the Unigine Heaven benchmark (see attached). However there are various workloads which stress different elements and there's probably some unit-to-unit variation based on heat sink assembly, thermal compound application, etc.

I'm a professional video editor and next week will have a 2017 iMac 27 with i7 to compare to my 2015 model. I'll post any relevant results, including performance, noise and temps.

 

[Ethosik]

Why would you be surprised a 2016 laptop can beat a 2010 Mac Pro?

I am just surprised that Quick Sync would make that much of a difference, that's all. To cut my export times by 7 hours on average is pretty remarkable for a laptop.

How does the iMac's i5 compare to my 2016 Macbook Pro's i7? I do not have it with me, laptop i7's don't have HT or 4 cores right?

 

[DRuser]

I'm a professional video editor and next week will have a 2017 iMac 27 with i7 to compare to my 2015 model. I'll post any relevant results, including performance, noise and temps.

I'm looking forward to! Which NLE's do you use?

 

[EugW]

Max Yuryev tested the 2015 iMac 27 i7 vs the 2017 and found the 2017 ran cooler and quieter under the same load:

This can especially be seen in the temps and fan speeds when running the Unigine Heaven benchmark (see attached).

The wonders of node shrinks. Two whole nodes to 14 nm in fact.

Also, while I don’t know about video editing performance, for compute performance the 580 is way, way faster than the M395X. Even the “low end” 570 beats or ties the M395X.

 

[joema2]

I'm looking forward to! Which NLE's do you use?

I used Premiere from CS4 to CC but now mostly use FCPX, although I still have a subscription to Premiere CC.

 

[fathergll]

I used Premiere from CS4 to CC but now mostly use FCPX, although I still have a subscription to Premiere CC.

Everything I've read so far the performance increase for FCPX is substantial(I think OpenGL). I think DaVanci Resolve is also benefiting a lot from the new iMac. Adobe not so much.

 

[EugW]

Everything I've read so far the performance increase for FCPX is substantial(I think OpenGL). I think DaVanci Resolve is also benefiting a lot from the new iMac. Adobe not so much.

Link?

 

[DRuser]

I used Premiere from CS4 to CC but now mostly use FCPX, although I still have a subscription to Premiere CC.

Thanks. I've switched from Premiere CC to FCPX with grading in Resolve. Just don't really like the user interface of the Adobe products. Now I'm considering to switch completely to Resolve. Version 14 is just so powerful and with a price drop of the studio version to 299$ it is very tempting.

 

[tozz]

Thanks. I've switched from Premiere CC to FCPX with grading in Resolve. Just don't really like the user interface of the Adobe products. Now I'm considering to switch completely to Resolve. Version 14 is just so powerful and with a price drop of the studio version to 299$ it is very tempting.

How is resolve 14 in terms of performance? FCPX slaps Premiere around when it comes to performance in my experience (better HW acceleration I would guess, and better coders)
[doublepost=1499355461][/doublepost]

I am just surprised that Quick Sync would make that much of a difference, that's all. To cut my export times by 7 hours on average is pretty remarkable for a laptop.

How does the iMac's i5 compare to my 2016 Macbook Pro's i7? I do not have it with me, laptop i7's don't have HT or 4 cores right?

I have the 2016 MBP i7 (it's quad core and has HT), it's just a tad slower than the i5 3.8 in Geekbench multicore (the MBP is just a 2.6ghz part). This is how good HT is when the workload is perfectly parallelized, which is rarely the case in the real world.
Any form of native hardware acceleration is going to give you massive speed increases (or just tax your CPU a lot lower if it's about decoding), just to quote Toms hardware:
"The first demonstration highlighted the combined power draw for both the CPU and GPU during local 4K HEVC video playback. The Skylake system exhibited 50% host processor utilization and 10.2W of power draw, while the Kaby Lake system only required 5% of the CPU and a mere 0.5W of power. The increased efficiency equates to a 20x power reduction and up to 2.6x more battery life."

 

[joema2]

Link?

It's the Max Yuryev video I posted above: https://forums.macrumors.com/threads/the-new-imac-is-a-lot-noisier.2051501/page-21#post-24773764

 

[EugW]

It's the Max Yuryev video I posted above: https://forums.macrumors.com/threads/the-new-imac-is-a-lot-noisier.2051501/page-21#post-24773764

Ah OK, thanks.

I am just surprised that Quick Sync would make that much of a difference, that's all. To cut my export times by 7 hours on average is pretty remarkable for a laptop.

How does the iMac's i5 compare to my 2016 Macbook Pro's i7? I do not have it with me, laptop i7's don't have HT or 4 cores right?

Ah, you're talking about hardware acceleration. As tozz mentioned, it's a ginormous boost. As mentioned before, with a 10-bit 76 Mbps 2160p60 4K HEVC video:

4.2 GHz Core i7-7700K iMac with pure software decode (Sierra): 100% CPU usage, and the video can't play cleanly. Fan at 2700 rpm and loud.
1.2 GHz Core m3-7Y32 MacBook with hardware decode (High Sierra): 25% CPU usage, and the video is buttery smooth. It has no fan.

Both machines are actually capable of hardware decode, but Sierra doesn't have it for HEVC.

 

[DRuser]

How is resolve 14 in terms of performance? FCPX slaps Premiere around when it comes to performance in my experience (better HW acceleration I would guess, and better coders)

I've noticed a massive improvement in terms of playback performance in v14. However I didn't do any scientific tests. Think the video of Max Yuryev posted above shows this in detail.

 

[propower]

In another bizarre twist I have talked myself out of the new iMac today and cancelled my order. In the end there just isn't enough performance parity with my 2013 MP Hex to keep me happy with any i5 iMac. The i7 is the one - but I would have to be OK with the temperature issues (and not insignificant price). Will be trying out some new 4K monitors for my MP after being spoiled to pieces by the 2017 screen but for today - the new iMac is on hold for me... LOL !

 

[trsblader]

In another bizarre twist I have talked myself out of the new iMac today and cancelled my order. In the end there just isn't enough performance parity with my 2013 MP Hex to keep me happy with any i5 iMac.

I am about >< close to doing the same and for very similar reasons. I was holding off for the BTS promo and for more reviews of the i7. I really wanted to upgrade my desktop for the ssd, usb3/TB3, and i7 with hyper threading as the hyperthreading really could have helped out I believe. My laptop has usb 3 and an ssd, so just TB3 and a very minor processor upgrade wasn't enough to justify the cost so I'll just deal with kicking the charging cable every once in a while and go spend 3 grand elsewhere (my car is really begging for the next round of mods. Pretty sure I heard it whimper and I know it give me sad puppy eyes the other day).

 

[EugW]

In another bizarre twist I have talked myself out of the new iMac today and cancelled my order. In the end there just isn't enough performance parity with my 2013 MP Hex to keep me happy with any i5 iMac. The i7 is the one - but I would have to be OK with the temperature issues (and not insignificant price). Will be trying out some new 4K monitors for my MP after being spoiled to pieces by the 2017 screen but for today - the new iMac is on hold for me... LOL !

Honestly, given your desired use and desired workload, as I suggested before I think your best solution MIGHT be to buy a 6-core i5 iMac in 2018, and attach those 4K monitors to it.

I'm not sure how hot those 6-core chips will be, but they won't have HT, so maybe they may behave closer to something like a 7600K in terms of power spikes under load, with the caveat that they will likely be lower clocked than the current top-of-the-line i5 chips. But because your workload makes great use of multi-threading, a hypothetical 3.5 GHz 6-core 8600K might just fit the bill for you. HT will be reserved for the i7 class chip (6-core / 12-thread), but with a 6-core version of the i5, hyperthreading won't be quite as necessary, as it would still be a huge performance boost over the 2017 i5-7500 you were looking at.

 

[propower]

^^^^ yep I had the same thoughts in my head many times. Maybe even the new MacPro depending on everything about it though! If I could only stand Windows (see where i live! Redmond, WA 25 years) I would have built a killer 7700K machine for less than $1500 by now. I have done the Hack route before and have no desire to re-live that...

 

[mcomp112]

In another bizarre twist I have talked myself out of the new iMac today and cancelled my order. In the end there just isn't enough performance parity with my 2013 MP Hex to keep me happy with any i5 iMac. The i7 is the one - but I would have to be OK with the temperature issues (and not insignificant price). Will be trying out some new 4K monitors for my MP after being spoiled to pieces by the 2017 screen but for today - the new iMac is on hold for me... LOL !

FYI: Once you scale 4K monitors to 1440p scaling, they lose quite a bit of their sharpness. The default scaling is sharp but unusably big.

5K is the way to go for monitors >24 inches.

 

[propower]

FYI: Once you scale 4K monitors to 1440p scaling, they lose quite a bit of their sharpness. The default scaling is sharp but unusably big.

5K is the way to go for monitors >24 inches.

Yes - I actually don't mind settling on 1080P with a 4K screen. 2013 MP not so great at 5K to boot. My U2715H is Native at 1440P so it will be interesting to see the difference and decide which makes more sense. Of course the 5K iMac screen is a spoiler for all else sadly

 

[dspdoc]

Not hearing the "noise" unless totally stressing the processor and heating things up. Otherwise things run cool and quiet.

 

[Hrodriguez77]

Not hearing the "noise" unless totally stressing the processor and heating things up. Otherwise things run cool and quiet.

Is this for the i7/580?

 

[dspdoc]

Is this for the i7/580?

Yes.

 

[mcomp112]

Yes - I actually don't mind settling on 1080P with a 4K screen. 2013 MP not so great at 5K to boot. My U2715H is Native at 1440P so it will be interesting to see the difference and decide which makes more sense. Of course the 5K iMac screen is a spoiler for all else sadly

Oh ok. Yeah, for sure some people actually prefer the 1080p scaling since it makes it easier to use from a greater distance than the usual 15-18 inches.

I've actually tested native 1440p and 4K scaled to 1440p and the 4K monitor, even with scaling, is actually a decent jump up in sharpness. Given how reasonably priced 4K monitors are these days (I got mine for $450), I think it's worth it.

5K is drool-worthy though and I really wish we had more (reasonably priced) 5K options in the market.

 

[killhippie]

Not hearing the "noise" unless totally stressing the processor and heating things up. Otherwise things run cool and quiet.

Same here, and yesterday the ambient temperature in my room was 28C and it didn't crank up the fans at all. Very happy with my i7 580 so far.