And that's what I want- a better workload machine. You just reassured my confidence in the i7
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
The new iMac is a lot noisier!
- Thread starter Darajavahus
- Start date
- Sort by reaction score
And that's what I want- a better workload machine. You just reassured my confidence in the i7
Fwiw, nobody watches 8K YouTube, and most non-pro and non-gaming activities that most people do will usually keep the fans quiet.
Few things in life are genuine needs. Air, water, food, shelter ... sure, those count.
Maybe a silent computer is just a want, but it feels like a need to me, at least. So much so that my other two computers reside in a separate room, tethered to monitors in my office.
To my ears, even the base iMac is too noisy, though I tolerate it. I bought it for its beautiful monitor, and to play with until a "real" Mac Pro is finally available again (which will go in that separate room).
I think Apple should develop a silent computer again, e.g. a silent version of the iMac. I don't care if it's less powerful, even. Oh, and please throw in an OLED display. Thanks in advance!
MacBook is silent.
I agree. I can hear the base iMac's fan at idle too. Only the MacBook, iPads, iPhones, iPods are truly silent as they do not have any fans. Expecting a computer to be completely silent while it has a fan is wishful thinking. I have 5 separate computers and I can hear the fans on ALL of them while they are idle. I can even hear my 2016 MacBook Pro's fans at idle. I can hear my parent's 2015 iMac fans while idle and it is NOT an i7 or even a 4K/5K version. They got the base model with an SSD instead.
There is a reason why we are still around the upper 3 to lower 4Ghz. These processors get SO HOT. When Sylake came out with a 4Ghz processor stock, people were complaining about the heat it generated. Basically needed a noctua BIG heatsink and fan to keep it cool. A fan is needed when these processors do the work they are designed for. I don't know why people are shocked about this. ALL my systems scream while they are exporting a video. From either FCPX or Adobe Premiere on my Windows PCs. They get SO LOUD but that is what they are designed to do. Even when I get my office to around 67 degrees Fahrenheit
Yes these things get hot, but computers these days are designed to completely turn off if they get too hot in order to prevent damage. Back when I had my first MacBook Pro (I think the 2009 model), It would be constantly at around 110 Celsius. That system was SO HOT, but it ran perfectly fine at those temperatures. That was the only system I monitored the temperatures just because it was so incredibly hot!
I don't think people need to worry about the temperatures that much. These are designed to shut off if they get outside the maximum safe temperature. Having loud fans means the system is doing its job to keep it from overheating. If you do not need the power, slower processors will be more quiet, but do not expect silence especially if you are in the middle of a video export or some heavy task.
[doublepost=1499991275][/doublepost]
Yep, I can confirm that. My 2010 Mac Pro with a GTX 980 is around 5 hours slower to export than my 2016 MacBook Pro.
[doublepost=1499991461][/doublepost]
Not to mention the fact of having two fans to begin with will be louder. Like one of my Windows PCs that has 6 fans on it
[doublepost=1499991703][/doublepost]
the new iMac coming in December is the iMac Pro. It says "Available December" right on the Mac page of Apple. So I would think that means confirmed.
View attachment 708472
That is a new product line. Unless people want to risk early adopters issues and potential problems, I would highly suggest waiting for revision 2 or at least another year after it releases.
My 2012 MacBook Pro 15" with the highest i7 at the time and an SSD is rather silent during normal tasks, makes a little noise with 1080p YouTube, and the fans go full blast watching 4K. It's nothing new.
Also this machine scores only 2,000 or so below the new i5 iMac in multicore Geekbench, so such a lateral performance upgrade like that wouldn't be worth it for me. Must go i7.
Also this machine scores only 2,000 or so below the new i5 iMac in multicore Geekbench, so such a lateral performance upgrade like that wouldn't be worth it for me. Must go i7.
I'm not sure it's such a lateral upgrade if your fans are hitting max with just 4K YouTube playback. Here is Chrome YouTube VP9 4K (software decode) on an i5-7600. CPU usage is around 35% and temps stay in the high 50s Celcius. Fan never budges from minimum.
Judging by this CPU performance though, I suspect the i7-7700K wouldn't rev up the fans high either, at least with this material.
[doublepost=1500006119][/doublepost]Out of interest's sake, I looked at some other 4K VP9 videos in Chrome. One popular one is Costa Rica, and it seems in some parts it might be a bit more CPU intensive than the last one, but not hugely so. However, in a few spots I could get the temps up to the high 60s Celcius with more than 50% CPU usage. Then again, CPU usage would often drop below 35%, with temps in the 50s like the last one. Again, fan never revs up. Locked to ~1200 rpm.
BTW, if you're wondering about the dropped frames, they seem mainly to occur when I do stuff like go full screen or come out of full screen, or right when I bring up the title bar (like in the picture, to show you the CPU and fan stats). Playback is otherwise smooth.
Video is here:
Last edited:
I was finally able to get my i5-7600 fan to speed up... but just by a little bit. Following a discussion between myself and joema2, some Handbrake encoding was done. I will summarize his post and my post below:
http://www.4ktv.de/testvideos/
And then using Handbrake we transcoded it to h.265 using the Roku 2160p30 preset.
My result for the 2017 iMac 27 i5-7600, no hyperthreading was 12:24
That is very similar to the 2015 iMac 27 i7-6700K with hyperthreading off. That makes sense because that CPU has a multi-core Turbo of 4.0 GHz. The 2017 i5-7600 has a multi-core Turbo of 3.9 GHz.
At max speed with hyperthreading turned on for the i7 models, here are the results in seconds, with reduced encoding times achieved by the i7 models vs the 2017 i5-7600. Obviously, lower is better.
2017 i5-7600: 744 seconds
2015 i7-6700K: 653 seconds (-12%)
2017 i7-7700K: 593 seconds (-20%)
The i5-7600 was maxed out to 397% (!) out of 400% in Handbrake with total CPU usage at 100%. The temps gradually increased and eventually touched 94C a little before the 10 minute mark. There was a gradual mild increase of fan speed to just over 1350 rpm at the 10 minute mark (see screenshot), barely audible. Then the machine eventually settled with temps in the low 90s with the fan speed in the low 1300s at around the 12 minute mark. Max CPU wattage of the i5-7600 was 60 Watts.
Overall, this behaviour is just as some of us predicted. Again, if you want guaranteed (near) silence, go with the i5-7500. If you want almost always (near) silence and a better GPU, go with the i5-7600. And if you want mostly (near) silence and top end GPU, go with the i5-7600K. And if you want the fastest possible machine, get the i7-7700K if you can deal with the noise.
We took the Philips Supershop Demo 1 video from here:
http://www.4ktv.de/testvideos/
And then using Handbrake we transcoded it to h.265 using the Roku 2160p30 preset.
My result for the 2017 iMac 27 i5-7600, no hyperthreading was 12:24
That is very similar to the 2015 iMac 27 i7-6700K with hyperthreading off. That makes sense because that CPU has a multi-core Turbo of 4.0 GHz. The 2017 i5-7600 has a multi-core Turbo of 3.9 GHz.
At max speed with hyperthreading turned on for the i7 models, here are the results in seconds, with reduced encoding times achieved by the i7 models vs the 2017 i5-7600. Obviously, lower is better.
2017 i5-7600: 744 seconds
2015 i7-6700K: 653 seconds (-12%)
2017 i7-7700K: 593 seconds (-20%)
The i5-7600 was maxed out to 397% (!) out of 400% in Handbrake with total CPU usage at 100%. The temps gradually increased and eventually touched 94C a little before the 10 minute mark. There was a gradual mild increase of fan speed to just over 1350 rpm at the 10 minute mark (see screenshot), barely audible. Then the machine eventually settled with temps in the low 90s with the fan speed in the low 1300s at around the 12 minute mark. Max CPU wattage of the i5-7600 was 60 Watts.
Overall, this behaviour is just as some of us predicted. Again, if you want guaranteed (near) silence, go with the i5-7500. If you want almost always (near) silence and a better GPU, go with the i5-7600. And if you want mostly (near) silence and top end GPU, go with the i5-7600K. And if you want the fastest possible machine, get the i7-7700K if you can deal with the noise.
Last edited:
Thanks for the good work on these tests. My only comment is the performance differential and noise is highly software-dependent. On FCPX H264 transcoding, my 2017 iMac i7-7700K is 2x (that's 200%) faster than the 2015 imac i7-6700K. I can't really tell a major difference in noise between those two.
So there's no question that if you're running Prime95, Linpack or doing a software-only transcode, the i5 will be a lot cooler and quieter. But the exact H265 test you ran above could radically change if and when Handbrake supports Quick Sync on Kaby Lake. In that case the CPU utilization would likely drop (since dedicated fixed-function hardware is doing much of the encoding), temps would go down on both i5 and i7 and performance would greatly increase.
We don't know when Handbrake will support that but High Sierra and likely all Apple apps will support this within a few months. So anybody streaming H265 or VP9 video using Safari would see greatly improved performance and lower temperatures -- on both i5 and i7, provided those are 2017 Kaby Lake models. Chrome and FireFox will get similar improvements whenever they support Quick Sync.
In video editing, FCPX has used Quick Sync for years but Premiere Pro does not yet support that on Mac, despite Quick Sync being introduced in 2011. This illustrates that High Sierra or the Kaby Lake CPU are themselves not sufficient -- the app itself must be written to use Quick Sync to get those benefits.
The 2017 iMac 27 does not seem any louder or hotter than the 2015 iMac 27, so from that point nothing has changed. What has changed is (1) The recent attention to the i5 vs i7 in CPU-pegged scenarios, and (2) 2017-model i5 performance has improved to the point where a significant class of users may no longer need the i7.
What is being done in your FCPX h.264 comparison? I'm curious. Is some of the speedup GPU related? That is the question I asked in the other thread, which is how much of the FCPX improvements with the 2017 i7 are due to the new 14 nm Radeon Pro 580. The 580 is vastly superior to the 2015 i7's Radeon R9 M395. The reason this is pertinent is you can also get the i5-7600K with the Radeon Pro 580. In fact, in some artificial tests at least, the base model i5's Radeon Pro 570 is as fast as the M395 (if not slightly faster).
It would be good to get a FCPX video editor's review of the 7600K/580 directly compared to the 7700K/580.
Because I am a non-pro and don't update my machines very often, I waited for YEARS for this new HEVC QuickSync support. I really was tempted to buy a new 2016 MacBook because of some killer refurb deals (IBM), and was also a little tempted to buy a 2015 iMac too, but I am glad I waited on both.
Last edited:
The FCPX H264 comparisons included either (1) Importing 4k H264 long GOP material and building ProRes proxies, or (2) Exporting an H264 timeline to 4k H264. This came from a variety of cameras but mainly a Panasonic DVX200 and Sony A7RII.
In the FCPX import and proxy creation case, the 2017 iMac 27 i7 was about 2x faster than the 2015 iMac 27 i7
In the Premiere import and proxy creation case, the 2017 iMac 27 i7 was about 25% faster than the 2015 iMac 27 i7, so they both improved but FCPX improved vastly more.
In the 4k H264 export case, the improvement was less -- 16% for FCPX and 12% for Premiere, however FCPX was 3.7x faster than Premiere, and with far less heat and noise -- on the exact same hardware.
It seems unlikely the improvement is GPU related per se. This is because a GPU by itself cannot significantly accelerate long GOP encode/decode. The core algorithm is sequential: frame 1 must be calculated before frame 2 can begin, etc. Only by doing the innermost part in hardware can it be greatly accelerated, which is what Quick Sync does.
This is obvious from running similar FCPX tests on a 12-core Mac Pro with dual D700 GPUs. Despite the powerful CPU and GPUs, it is 1/2 the speed of a 2015 iMac 27 at exporting H264 or transcoding H264 to ProRes proxy. If a GPU could make this task faster, it would work here but it does not.
Quick Sync is associated with but functionally separate from the GPU. It is entirely separate fixed function logic. Unfortunately due to shared resources (frame buffer, busses, etc) it cannot exist without the on-chip GPU but the work is done by Quick Sync not by the GPU.
I don't know how Apple got such large performance increases out of FCPX on Kaby Lake for the import and proxy transcode task. It's the exact same code. Supposedly the Quick Sync improvements on Kaby Lake were for function (HEVC 10-bit), not performance.
The huge performance, heat and noise difference between apps that support Quick Sync and those that do not are visible when doing these tests in FCPX vs Premiere Pro CC on the exact same 2017 iMac 27 i7:
Import and create ProRes proxies for 10 H264 4k long GOP files, total running length 11 min 43 sec:
FCPX, 2015 iMac 27: 5 min 37 sec
FCPX, 2017 iMac 27: 2 min 40 sec
Premiere 2015 iMac 27: 8 min 6 sec
Premiere, 2017 iMac 27: 6 min 27 sec
Even though Premiere is much slower in the above test, it won't create full-HD ProRes proxies (only 1537 x 790), so this is an artificial advantage. To bypass this we can do a straight export test of a 1 min 51 sec 4k clip to 20 mbps single-pass H264 4k:
FCPX, 2015 iMac 27 i7: 1 min 21 sec; CPU levels: moderate, noise & heat: low
FCPX, 2017 iMac 27 i7: 1 min 8 sec; CPU levels: moderate, noise & heat: low
Premiere, 2015 iMac 27 i7: 4 min 43 sec; CPU levels: high, noise & heat: high
Premiere, 2017 iMac 27 i7: 4 min 11 sec; CPU levels: high, noise & heat: high
Can someone explain what Quick Sync is? Is this something that comes standard on FCPX? Does iMovie have it? And what does it do?
See below. There are no diagnostics or performance counters to tell whether an app is using Quick Sync. But IF it's used, certain tasks such as H264 import and export and scrubbing an H264 timeline will be a lot faster and consume less CPU. iMovie 10.1.6 is definitely using Quick Sync for H264 export, because it's as fast as FCPX on H264 export.
It is fundamentally a hardware feature of Intel i-series CPUs which started with Sandy Bridge in 2011. To benefit from this the app or utility must write to the relevant API -- it's not automatic. This is similar to how an app must properly access a GPU to obtain those benefits.
Unfortunately Premiere does not use Quick Sync, at least on Mac. Also unfortunately no Xeon CPU except for a few low-end E3 models with four cores have Quick Sync. So the highest-end Xeon workstations such as Mac Pro or Dell Precision cannot use Quick Sync.
To a degree this affects the highest-end video production less because they typically use all ProRes or DNxHD or RAW data acquisition. Quick Sync does not benefit those, only long GOP codecs like H264, H265 and Google's VP9 and AV1. However there's a large slice of video production that uses H264 acquisition -- even if they immediately transcode to an intermediate codec. It is this group that are negatively affected by any software or hardware that doesn't support Quick Sync.
https://en.wikipedia.org/wiki/Intel_Quick_Sync_Video
https://www.intel.com/content/www/u...uick-sync-video/quick-sync-video-general.html
Thanks for this!
So you need an app for it? I didn't see any mac related suggestions on the intel.com link. I thought it was something already enabled so when I click Export Movie it will be a faster process.
Each app or utility must be written to access Quick Sync separately. You don't need a separate utility to enable it. Typically if an app or utility does this it is automatically enabled, but some may have a config option. For FCPX and iMovie it is automatically enabled on any platform that supports Quick Sync.
However Quick Sync can only be used for H264 (or eventually H265) export. Originally it was single-pass only, not multi-pass encoding, so if you picked "multi-pass", or "high quality" (wording varies with each app), it would not use Quick Sync.
Originally there was a viewpoint that single-pass or hardware-accelerated encoding severely hurt quality but the way Apple implements it in FCPX and iMovie I cannot tell any difference. That is why FCPX defaults to "fast" export.
Ah okay! Now on iMovie '11 when I export a project , it gives me the option of 1080, 720, Large movie, etc. Do I do that same export process to activate Quick Sync? Or is there another way I have to do it? Again I'm using older software so maybe the new iMovie will have different options for the export process.
I don't know the answer, but considering iMovie '11 was released in October 2010 and Quick Sync was not released until Intel's Sandy Bridge CPU in 2011, it's unlikely iMovie '11 uses Quick Sync at all.
I think the latest version of iMovie which does support that is free, so you could try it. It does obviously require a Sandy Bridge or later CPU. The Mac Tracker app in the App Store has all the specific details on the CPU and GPU in each Mac: https://itunes.apple.com/us/app/mactracker/id430255202?mt=12
Thanks Joe! I'm getting my new iMac next week so I'll ask during my in store set up process
Sorry for slow reply, only had time to test this weekend.
I took a folder with 2700 D800 RAWs (36MP) and tried to render 1:1 Previews. The screen shot is showing the process, interestingly despite all cores are being used at bursts, and the total power can fire up to as high as 80W at times, the constant average temperature never got above 100 degrees C, so the clock is relatively stable at the peak 4.2GHz without throttling happening. The fan pretty much behaves as expected, fired up to above 2700RPM seconds after the rendering started.
Now for this test I only left LR doing its thing on its own, which isn't the normal workflow of mine; I would usually be editing some other photos or culling and metadata-ing the imported photos as the preview generation goes on. If I go into develop module during this process I am not sure if the machine won't slow down due to being heated up more.
And I agree that Lightroom is one of the worst optimized modern professional software. I really wish the competition catching up close enough to offset Adobe's semi-monopoly on this front. The last time I have seen something substantial from Adobe was InDesign (circa version 2 or 3) where they needed to compete against the then industry leading Quark.
Last edited:
Using LightRoom ver. 2015.10.1, I just repeated a similar test on 100 Sony A7RII 42-megapixel raw stills, running it on both my 2015 iMac 27 i7 and the 2017 i7. I then repeated the test on the 2017 with hyperthreading disabled. Results:
2015 iMac 27 i7: 4 min 42 sec
- relatively quiet until about 1/2 way through
- temp and fan gradually increased to 95C and 2400 rpm
2017 iMac 27 i7: 4 min 3 sec
- a little quieter than the 2015
- temp and fan speed hit about 90C and 1900 rpm
2017 iMac 27 i7, hyperthreading disabled: 4 min 12 sec
- temp and fan similar to HT enabled
So once again, this indicates there's no major difference in heat or noise between the 2015 and 2017 iMac 27 i7. If anything the 2017 might be a little cooler and quieter.
The OP starting this thread said: "....it's now confirmed it's noisier than before". The iMac 27 i7 is NOT noisier than before. It is noisier than the i5 version while producing higher performance. That also is not new -- the 2015 iMac 27 was also hotter and noisier than the i5 version while producing higher performance. There is nothing new here.
What IS new is the top 2017 i5 performance is significantly higher than the 2015 i5: about 19% faster on multicore GeekBench and the Radeon Pro 580 is much faster than the previous Radeon M395.
This opens up an increased possibility that the 2017 iMac 27 with i5 can stretch to some higher-end workloads than before. EugW and others have made that point well, but due to the incorrect OP, I think lots of people have become confused about the 2017 i7 noise and heat issue. It is no worse than the 2015 i7, and possibly a little better.
On some workloads (like FCPX transcoding) the 2017 i7 is *vastly* faster than the 2015 i7. On other workloads the improvement is more modest. Maybe the 2017 i5 version is similarly faster than the 2015 i5, but I don't have those so can't test them.
Those Lightroom tests gives me much more confidence in my choise of the i7. Can't wait to get mine, only 2 to 3 weeks left...
I tested doing other (intensive) stuff in LR while the 1:1 previews were still running. Working in develop module seems to have slower response, noticeable but slight, which is acceptable. Throwing in a batch export doesn't seem to make a dent (looks like LR split loads between concurrent tasks).
The only way to make my iMac throttle within LR is to have started 3+ panorama stitching, while the 1:1 generation was still running. I managed to drag down clock speed to below 4GHz for very brief duration, since even a 11-image pano would only like LR 2 minutes to finish anyway.
Overall I am quite impressed with the capabilities of this i7 iMac, all the places that showed slowness were seemingly due to Adobe's coding issues, or that I am throwing unrealistic loads at it, the hardware looks to have enough headroom for general photography workflow. The fan noise would kick in above 2400rpm, which can happen in seconds for batch preview generation / export, but if just in develop module for small bursts of adjustments then it would take minutes before the cumulated heat piles up. This can be annoying but nothing unacceptable for a powerful photography workflow.
The only way to make my iMac throttle within LR is to have started 3+ panorama stitching, while the 1:1 generation was still running. I managed to drag down clock speed to below 4GHz for very brief duration, since even a 11-image pano would only like LR 2 minutes to finish anyway.
Overall I am quite impressed with the capabilities of this i7 iMac, all the places that showed slowness were seemingly due to Adobe's coding issues, or that I am throwing unrealistic loads at it, the hardware looks to have enough headroom for general photography workflow. The fan noise would kick in above 2400rpm, which can happen in seconds for batch preview generation / export, but if just in develop module for small bursts of adjustments then it would take minutes before the cumulated heat piles up. This can be annoying but nothing unacceptable for a powerful photography workflow.
So, on my dual screen setup with the Radeon Pro 575, just activating Mission Control a few times will get the GPU up to over 40 Watts. Normal idle power usage is a little under 30 Watts. I wonder how taxing on the GPU this action is. It doesn't seem overly taxing, but I note that some people have complained in the past that Mission Control on older 5K iMacs like the 2014 was stuttery in dual screen setups.
It doesn't seem to be a problem here with the 2017 i5 7600 Radeon Pro 575, at least with a 2560x1440 second screen (which happens to be an old iMac). I don't know how it would be with a 5K external screen though.
It doesn't seem to be a problem here with the 2017 i5 7600 Radeon Pro 575, at least with a 2560x1440 second screen (which happens to be an old iMac). I don't know how it would be with a 5K external screen though.
First, What app did you use to keep track the CPU RAM SSD usage? Or CPU watt power and other thermal degrees? I like how it showed on the top bar menu compared to the other apps that I have to open them to look at the information, which is inconvenient for me.
Second, by turning off the HT in i7 , do you think it will reduce the performance of the i7 compared to the fastest i5? I'm still trying to debate which one I should get for my very first imac.
Thanks in advance