Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
For those doing the 4k tests, are you playing it through a 4k monitor or just selecting 4k and watching it on your laptop's non-4k screen?
 
So if I'm understanding you right, MacOS doesn't support the codec that Google is using on Youtube to support 4K video, forcing users using MacOS to brute force the playback using hardware acceleration?

Yes.

To play VP9 on macOS your machine needs to use non-hardware-acceleration based software. The CPU in any machine more recent than Skylake has specialist hardware for decoding VP9 (and h.265 as well), but apple is not exposing it, because they want the industry to use h.265. To be fair h.265 is more efficient, but it is patent encumbered.

In theory vp9 is less patent encumbered, but is controlled by Google. It is less efficient.

Apple paid the license fees to use h.265. Google don't want to.

Upshot of all this: Macs won't play vp9 content very well. Youtube doesn't transcode uploaded 4k videos to h.265. End users lose because they're either consuming more bandwidth for VP9 videos and(if on macOS)/or more CPU consumption, worse battery life, etc.
[automerge]1586669563[/automerge]
For those doing the 4k tests, are you playing it through a 4k monitor or just selecting 4k and watching it on your laptop's non-4k screen?

I have a 4k monitor.
[automerge]1586669651[/automerge]
Either way the consumer loses in this case, but I think it does take some of the heat (pardon the pun) out of the completely broken thermal design hyperbole.

Yeah. It's like taking marathon runner and tying his legs together then making him do the marathon. And complaining he's off pace. Or (for those who remember the days before GPUs) running a 3d game in software rendering. The hardware acceleration features make a massive, massive difference.

But, for VP9 video on macOS, they are basically disabled. Safari just doesn't play VP9 at all. Chrome does it in software.
 
  • Like
Reactions: HopelesslyConfused
I've always had weird behaviour with Chrome when watching YouTube. I have a 13" 2016 base nTB MBP.

Most of the time here's my temps when watching YouTube in Chrome with about 10 other tabs open:

1080p - 44°C fan off
2k - 49°C fan off
4k (on my MBP's 2K screen) - 69°C, fan 1200rpm

This is all fine, but I've noticed when I've left my MBP on for a day or two, or opened 5+ YouTube tabs then playing one 1080p video will cause temps of 90°C and force the fan to run quite high. Even when I close all the other tabs it gets this hot. But if I restart Chrome, then it returns back to normal, about 45°C no fan playing 1080p video.

Anyone have any theories to explain this?
 
I've always had weird behaviour with Chrome when watching YouTube. I have a 13" 2016 base nTB MBP.

Most of the time here's my temps when watching YouTube in Chrome with about 10 other tabs open:

1080p - 44°C fan off
2k - 49°C fan off
4k (on my MBP's 2K screen) - 69°C, fan 1200rpm

This is all fine, but I've noticed when I've left my MBP on for a day or two, or opened 5+ YouTube tabs then playing one 1080p video will cause temps of 90°C and force the fan to run quite high. Even when I close all the other tabs it gets this hot. But if I restart Chrome, then it returns back to normal, about 45°C no fan playing 1080p video.

Anyone have any theories to explain this?
Hardware have a hard limit on how many rendering context. my guess is multiple codec is active so it falls back to software rendering pages.
 
Last edited:
Hardware have a hard limit on how many rendering context. my guess is multiple codec is active so it falls back to software rendering pages.

Makes sense except when it starts exhibiting this hot behaviour, I can close all the other tabs and the one 1080p video will still make temps go up to 1080p and make it stutter a bit too.
 
Hello from Barcelona !!! I hope you are all in good health. I could not bear the temptation to write what I have been testing on my 2019 Air bought a week ago.
The first thing I did was change the thermal paste for an Artic MX4 and I managed to lower a couple of degrees in some applications but the video of the snake from 4K at 60fps is jerky ... yesterday I removed the back cover and put a Pro case protector and also ...
Finally I could see that there are foams on the lid that create a tunnel through which the wind must pass ... because I took and lined the entire part of the lid with metal tape and thus also made contact with the cpu and dissipated by the tape the heat. It almost reaches the top of the fan turbine. Well close everything and test, first put the Macs Fan Control to create the 8,000 rpm to the fan and then I put the famous video of the snake, because nothing went up to 100º and did not go down from there, instead at 1440P I had managed to lower the temperature somewhat . It is new and in guarantee and I do not want to touch it anymore but I think that a tube of the turbine above the heatsink would reduce the temp of the cpu.
 

Attachments

  • Captura de pantalla 2020-04-11 a las 20.39.49.png
    Captura de pantalla 2020-04-11 a las 20.39.49.png
    2.9 MB · Views: 171
  • Captura de pantalla 2020-04-11 a las 20.42.00.png
    Captura de pantalla 2020-04-11 a las 20.42.00.png
    2.4 MB · Views: 156
  • Captura de pantalla 2020-04-11 a las 20.55.13.png
    Captura de pantalla 2020-04-11 a las 20.55.13.png
    1.7 MB · Views: 133
If anything that is requiring all 4 cores on the i5/i7 for a prolonged period of time means performance drops to the i3 level anyway, why would I buy an i5?
 
  • Like
Reactions: magbarn
If anything that is requiring all 4 cores on the i5/i7 for a prolonged period of time means performance drops to the i3 level anyway, why would I buy an i5?
Because lots of tasks just need 4 cores for a short period of time.
 
  • Like
Reactions: jpn
If anything that is requiring all 4 cores on the i5/i7 for a prolonged period of time means performance drops to the i3 level anyway, why would I buy an i5?
I thought the benchmarks (cinebench) showed the i5 still performed better than the i3 and the previous gen i5. Could be wrong but I thought that was the case
 
  • Like
Reactions: jpn
I thought the benchmarks (cinebench) showed the i5 still performed better than the i3 and the previous gen i5. Could be wrong but I thought that was the case
Cinebench is a good benchmark for "bursty" uses that the MBA is decent at. We need to wait for bare feats to do a test as they do prolonged testing. Even the poster above who is converting audio files would notice a difference as that is a prolonged use of the CPU. But for the average user it's a fine machine. I routinely have to batch process hundreds of megabytes 45 MP RAW DSLR files, in my use case the MBA will likely be heat soaked and slow down to speeds that are probably too close to the speed of my 2017 MB 12"
 
Cinebench is a good benchmark for "bursty" uses that the MBA is decent at. We need to wait for bare feats to do a test as they do prolonged testing.
Cinebench is a test of sustained, multiple cores running at 100%. It's basically a stress test of CPU thermals / TDP. Its not a test of quick, bursty processor use.
 
Hello from Barcelona !!! I hope you are all in good health. I could not bear the temptation to write what I have been testing on my 2019 Air bought a week ago.
The first thing I did was change the thermal paste for an Artic MX4 and I managed to lower a couple of degrees in some applications but the video of the snake from 4K at 60fps is jerky ... yesterday I removed the back cover and put a Pro case protector and also ...
Finally I could see that there are foams on the lid that create a tunnel through which the wind must pass ... because I took and lined the entire part of the lid with metal tape and thus also made contact with the cpu and dissipated by the tape the heat. It almost reaches the top of the fan turbine. Well close everything and test, first put the Macs Fan Control to create the 8,000 rpm to the fan and then I put the famous video of the snake, because nothing went up to 100º and did not go down from there, instead at 1440P I had managed to lower the temperature somewhat . It is new and in guarantee and I do not want to touch it anymore but I think that a tube of the turbine above the heatsink would reduce the temp of the cpu.

That sounds really interesting, thank you for posting! Did you take any photos while you were doing this? I'd be fascinated to see what you came up with.

Hope you're doing well in Barcelona!
 
Cinebench is a test of sustained, multiple cores running at 100%. It's basically a stress test of CPU thermals / TDP. Its not a test of quick, bursty processor use.
One run of Cinebench is measured in seconds not 30 minutes.
 
One run of Cinebench is measured in seconds not 30 minutes.

You are right. One run can be short enough that some machines can turbo boost through much of a single run, if it starts from idle. For example, the Thinkpad T480 lets the CPU draw 29W for 28 seconds straight. So the first run has the high score; it is the burst. That is why it is informative to see a few runs in a row to see sustained performance.
 
Tomorrow I will take a picture of the metal tape that is placed in the tunnel between the cover and the processor.
 
Last edited:
Tomorrow I will take a picture of the metal tape that is placed in the tunnel between the cover and the processor.
You're only supposed to use like a dot of that stuff, not put it on you're frosting a cake.
 
Last edited by a moderator:
Tomorrow I will take a picture of the metal tape that is placed in the tunnel between the cover and the processor.

I'm not an expert but the one time I've used Artic Silver thermal paste I had to really really make sure it was a thing layer - you might get even better performance if you use a specialty spreading tool (or a credit card if you don't have one) to make sure it is even and very, very thing.

Looking forward to seeing what you've done with the metal tape tomorrow! Thanks for posting, it's all good data
 
Last edited by a moderator:
True, it also seems excessive to me, but look at image 1046 and see how they affected the origin ... with respect to leveling, the heatsink itself expanding the surplus.
By the way today here in Spain there is confinement with lesser services, you too ??? Health !!!
 
Last edited:
Hmmm I think I see what you mean, was the original thermal paste in image 1046 about as thick as what you applied?

In Australia it is much the same, though not as bad as Spain. Most people are isolated, but our quarantine is not as strict as yours yet. All the best for you and good health!
 
Last edited by a moderator:
If anything that is requiring all 4 cores on the i5/i7 for a prolonged period of time means performance drops to the i3 level anyway, why would I buy an i5?

For all the things that aren't 4k video rendering in software.

Like... pretty much everything else you do with the computer, including working with video in native apple applications (that use accelerated h.265).

And no, a quad won't drop down to i3 level even when hot, because you have 2x the cores that will be able to sustain more than 1/2 the clock speed of the i3.

in multi-threaded apps, you are:
  • guaranteed a higher base clock than the i3
  • have 2x the cores that run at that guaranteed higher base clock
  • better GPU
in single thread the i5/i7 have
  • higher boost clock than the i3
  • better GPU
  • more CPU cache (definitely with the i7, not sure with the i5)
Additionally, if you want more than poverty-spec RAM and SSD, the it's generally cheaper to start with the higher spec machine that already has 512GB SSD and go up from there. Which has i5 as standard.
 
Last edited:
  • Like
Reactions: HopelesslyConfused
I show you my inventions ... here you have the metallic adhesive paper inside the tunnel and on top I have placed three € 80% copper coins to press on the adhesive. I have found several heatsinks and none work for me ... Think my Apple is 2 weeks old! I have to be careful in case I want to return it and Apple doesn't notice anything I'm trying ... complicated ...
- - Post merged: Today at 1:40 PM - -

Here you have the test with the fan at 8,000 rpm and the famous 4K video at 60 fps (it is a lot of information since my 2015 Macbook Pro played it in jerks ...) In the last photo it is at 1440p. This is on Chrome and Macs Fas Control and on the battery. Before making the coins it also went up to 100º but there were no peaks, which there are now. It is complex.
 

Attachments

  • IMG_2421 2.jpg
    IMG_2421 2.jpg
    439.2 KB · Views: 181
  • IMG_0372.jpg
    IMG_0372.jpg
    478 KB · Views: 169
  • IMG_9166.jpg
    IMG_9166.jpg
    495.7 KB · Views: 156
Last edited:
For all the things that aren't 4k video rendering in software.


Additionally, if you want more than poverty-spec RAM and SSD, the it's generally cheaper to start with the higher spec machine that already has 512GB SSD and go up from there. Which has i5 as standard.

Sorry but its the same price (1 euro difference) if you start from the I3/8/256 and do the i5+16GB+ 512 upgrades, than if you start from i5/8/512 and do the 16 GB upgrade... Apple is not stupid regarding money, no way to trick them (I tried lol)

Out of that I 100% agree with you, and don't agree with the click-buster-youtubers : performance will still be better with i5 than i3 even in high load
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.