Windows 10 runs hotter than windows 7?

[sdt]

playng a youtube video is a universal thing which 99% of the computers sold will do at some point in their life. So it is expected that hardware/software codecs/drivers/browsers/OS etc would be well optimised for it and if it not then it will show as greater heat output

 

Plus heat reduces life of a machine exponentially. If a machine runs cooler while doing the same task on XYZ OS then it is more efficient.

 

heat measurement in this test should show all efficiencies and inefficiencies combined in a system.

 

Lets get some test results. I would like to see how win10 is doing vs win7.

Edited May 10, 2015 by sdt

[NoelC]

dudes this is a serious question! its pretty easy to check for someone willing to oblige thanks! NoelC this one is for you if you can take it up, since you are the main win10 tester here with oodles of time in your hands. Thanks!

 

Unfortunately, the new system I have dual booting on 7 / 10 is now online in a kind of critical role, but if I take it down for maintenance I'll check.

 

-Noel

[Tripredacus]

Due to 4k image size, I am using Imgur to put up the pictures. Temperatures noted in description for each. So far only Windows 7 tested.

Chrome with HTML5 is certainly the hottest of the bunch, just starting to play the video at full screen caused all fans to speed way up. Flash based 4k does not have this happen in either IE10 or Chrome. I used a plug-in from the Chrome store in order to force Youtube to show videos in Flash rather than HTML5.

WARNING: Screenshots in this album at at 4k resolution... old timey "56k Warning" be advised.

http://imgur.com/a/sv38a

Edited May 11, 2015 by Tripredacus

[jaclaz]

Well, we actually trust you , no real need of those senselessly large images to provide the actual data:

 

4k 7PRO64 Chrome Flash CPU 49-70c, Video 49-50c
4k 7PRO64 Chrome HTML5 CPU 49-79c, Video 43-49c
4k 7PRO64 IE10 Flash   CPU 44-70c, Video 44-50c

 

 

jaclaz

[Tripredacus]

Oh should it be prefaced with "56k warning" like the good old days?

Anyways, another comparison can be seen with the full screenshots as to how 4k resolution works on Win 7 vs 8.1. Win10 may be a bit... I also wanted to have the full Hwmonitor info available if anyone wanted to see it.

Windows 8.1 test done

4k 8.1PRO64 Chrome Flash CPU 47-57c, Video 47-49c

4k 8.1PRO64 Chrome HTML5 CPU 49-79c, Video 43-47c

4k 8.1PRO64 IE11 Flash CPU 46-63c, Video 45-49c

[NoelC]

The extra heat from Win 10 - assuming we do find it to objectively run hotter - might just be the added load from Microsoft sending all your private data to their cloud servers for, er, later analysis to determine if your Windows experience can be improved.  

 

-Noel

[Tripredacus]

Windows 10 is a pain. The Start Menu and task bar does not work so I won't be able to test with the Spartan browser. Maybe I can do some updates and see what happens later.

4k Windows 10 Build 10074 x64 Chrome Flash CPU 48-66c, Video 47-50c

4k Windows 10 Build 10074 x64 Chrome HTML5 CPU 51-69c, Video 47-49c

4k Windows 10 Build 10074 x64 IE11 Flash CPU 49-58c, Video 46-48c

The only bump I am seeing is HTML5 in all the tests. In most test cases, I would only test using IE, however since it was a Google website I decided to test Chrome as well. It is likely that other browsers that default to the HTML5 mode on Youtube will also see a temperature increase.

Now all 9 tests together.

 

4k Windows 7PRO x64     Chrome Flash CPU 49-70c, Video 49-50c4k Windows 7PRO x64     Chrome HTML5 CPU 49-79c, Video 43-49c4k Windows 7PRO x64     IE10 Flash   CPU 44-70c, Video 44-50c4k Windows 8.1PRO x64   Chrome Flash CPU 47-57c, Video 47-49c4k Windows 8.1PRO x64   Chrome HTML5 CPU 49-79c, Video 43-47c4k Windows 8.1PRO x64   IE11 Flash   CPU 46-63c, Video 45-49c 4k Windows 10 10074 x64 Chrome Flash CPU 48-66c, Video 47-50c4k Windows 10 10074 x64 Chrome HTML5 CPU 51-69c, Video 47-49c4k Windows 10 10074 x64 IE11 Flash   CPU 49-58c, Video 46-48c

Manufacturer has 2 different drivers. One for Windows 7 and one for Windows 8. Windows 10 testing was done with the Windows 8 driver. Also to note that the CPU temp threashold was not met on Windows 10, so no screaming fans.

[jaclaz]

Hmmm.

The reduction in temperature of Windows 8.1 and 10 (CPU) are impressive  while the video GPU seems like the same or too little to be meaningful...

 
jaclaz

Edited May 11, 2015 by jaclaz

[NoelC]

So, given that one person has found Win 10 to run the CPU hotter, while another has found it to run the CPU cooler, is there anything at all we can infer?  I'm thinking it's a bit too course a measurement to be meaningful.  Might need multiple iterations averaged to be more certain there's not just normal variation in the numbers.

 

Thinking on this a bit...  Didn't Microsoft make having an SSE2 instruction set necessary for Win 8 or later?  The use of such instructions can make a HUGE difference in the amount of brute force CPU time spent in graphics processing.

 

-Noel

Edited May 11, 2015 by NoelC

[NoelC]

Can you suggest a link to an online video that would stress a modern (Haswell) processor?

 

I tried with some HD video I have on hand with Win 7 and got a range of 33C - 37C in Win 7.

 

EDIT:  Found several I'll try from this page:  http://box.houkouonchi.jp/4k/

 

SECOND EDIT:  Hm, I'm having difficulty finding one that'll play with the codecs out of the box on my Win 7 system, and I'd rather not install extra software.

 

-Noel

Edited May 11, 2015 by NoelC

[Tripredacus]

So, given that one person has found Win 10 to run the CPU hotter, while another has found it to run the CPU cooler, is there anything at all we can infer?

I don't think so. As they say... YMMV. Considering different hardware and software used, I think that in order to fully understand the why Windows 10 runs hotter in some instances than others, more tests would need to be done. And then to figure out in what conditions is this the case.

This isn't so say the sdt is wrong in saying that Windows 10 runs hotter on his PC than Windows 7 on his PC... I'd wonder if he can also do the Youtube 4k test and see if the results are the same. Or try other media players besides VLC. If the source of the 4k video is on the HDD or ODD, then maybe their extra use is adding to the general increase of temperature on the system? What about other things that might be running? My tests were done on new installs with minimal software installed and in an air conditioned building. Also perhaps the internal cooling flow of a notebook differs greatly from that of a desktop?

Anyways, it seems only that my test shows there is no flaw within Windows 10 causing an overly large spike in temperature on the CPU when trying to play 4k video at fullscreen.

[sdt]

@tripredacus - seems like windows 10 is improving its html5 experience since it is the best of the 3 OSes as far as temps are concerned. overall though win8.1 with chrome and flash seems to be doing the best.

 

its interesting win10 uses win8.1 drivers yet win8.1 doesnt  match its html5 experience.

 

PS i didnt suggest windows 10 runs hotter or cooler, i just suggested we test it. thanks for posting those figures

 

 

@noel youtube doesnt require codecs for 4k videos. wat about using VLC for houkouonchi videos? how does it compare on windows 10 from your testing?

Edited May 12, 2015 by sdt

[jaclaz]

Thinking on this a bit...  Didn't Microsoft make having an SSE2 instruction set necessary for Win 8 or later?  The use of such instructions can make a HUGE difference in the amount of brute force CPU time spent in graphics processing.

That could be a nice explanation of the data Trip gathered, but that would mean that  on a same machine (with SSE2 as hardware) a "same" Chrome browser Flash engine (as an example) makes a request to the OS and depending on the OS the same "call" is routed/translated to a set of SSE2 instructions in 8.1 and later and is routed/translated to more conventional instructions on Windows 7?

 

It may well be but it sounds "queer" to me.

 

Of course much more experimental data are needed to have anything of real value, I would have expected that more recent/better optimized drivers/sybsystems could have transferred some more work to the GPU relieving the CPU of some calculations, thus making the GPU slightly hotter and the CPU slightly less hot or viceversa, i.e. the "sum" of the heat produced to be more or less the same.

 

jaclaz

[NoelC]

 

That could be a nice explanation of the data Trip gathered, but that would mean that  on a same machine (with SSE2 as hardware) a "same" Chrome browser Flash engine (as an example) makes a request to the OS and depending on the OS the same "call" is routed/translated to a set of SSE2 instructions in 8.1 and later and is routed/translated to more conventional instructions on Windows 7?

 

It may well be but it sounds "queer" to me.

 

It's not implausible, because of a few things...

 

1.  Microsoft did publish that SSE2 is a requirement for Windows 8+.

 

2.  There is a setting in Visual Studio (which I assume is what they compile the Windows sources with) that says, "compile this to use SSE2 instructions".  Thus DLLs that the OS provides, even without re-coding at all, may well use different instructions in newer Windows versions, even though they offer the exact same sets of entry points.  This is one of the reasons using DLLs was supposed to be better than static linking - you could get advantages that weren't available at the time the original software was coded.

 

3.  Since Microsoft made SSE2 a requirement, there's no reason the boys and girls involved with the coding of things like Media Player and whatnot couldn't re-code critical parts of the software to specifically use the intrinsics that generate SSE2 instructions.  I recently went through the same advancement of requirements for my own graphics software, and re-coded routines that do things like resample images.  Large reductions in execution time can be seen.

 

4.  Since you can easily query the processor capapbilites, it IS possible in the application itself that different linkages can be set up depending on machine capabilities.  I do this in my software.  Also, I know Adobe Photoshop does this under the covers, even taking it so far as to move some operations to the GPU when it's possible, all the while keeping the intent that the software operates the same for the user.  Taken to an extreme, an installer could install an entirely different application for an older OS vs. a newer one.

 

The advantages of using the better instruction set can be very significant, making it worth the extra complexity.

 

-Noel

Edited May 12, 2015 by NoelC

[MagicAndre1981]

Windows 8/10 do a maintenance task during idle, which does Defrag, WinSxS compression and NGEN which Win7 doesn't do and this causes extra CPU usage, so the idle temperature is a bit higher.