2018 i9 MacBook Pro throttling discussion [merged]

[Elektrofone]

Could someone who has a 2018 i7 model with 32GB of RAM run this After Effects Benchmark Test? I'd like to compare a real world test to my 2018 i9. Also, if you could run it from idle and then also run it after the render is done.

mediafire.com/file/qgoacxne6567wxn/AE_Test_2016.zip

I got 6 minutes 49 seconds both times I ran the test back to back

 

[PeterJP]

After watching the AppleInsider video & reading some comments in the i7 throttling thread, I've noticed something. No matter which of the three processors you have, the system always settles at 2.2-2.6GHz. If I remember correctly, all of these processors are the exact same silicon. But during testing, the ones that perform better get a fancier name (i9 or i7 2.6GHz), whereas the ones that fail at higher speeds quicker get the lower names. If that is the case, then it's logical that they all settle at the same speed. When the limitation is the cooling system and Apple's use of the fans, then all of them ramp up to their turbo speeds, then they need to ramp down because of thermal constraints and end up in the same thermally limited zone.
If you're doing stress tests, then that's the behaviour you should see. In real life, though, many users will have bursts of activity mixed with calm periods. In those situations, the higher spec'ed processors could give an overall benefit.
Still disappointing, though, this whole situation. I'm convinced that it's fall-out of Intel's failure to deliver lower TDP processors because of their 10nm problems. If Apple banked on that aggressively when designing their 2016 ultra thin chassis, then they're the victim now. Tough luck for us, and tough luck for Apple because they get beat up in forums and in reviews and they are stuck with Intel without a quick way out.
Do note that I am describing the situation with the current MBP chassis. I have heard enough whining about how Apple should not have gone so thin. Fine. I understand many people think that here. But it's not the situation at the moment so it's not relevant.

 

[zshane1125]

^^ I was about to post something similar.

I just did Cinebench test on my 2016 13" nTB i7 2.4ghz. Super consistent temp, clock speed, and voltage....

 

[M.Rizk]

One interesting point to add (and not sure if anyone has made this before), has anyone tried measuring the i9 with Turbo Boost switched off?

Background: I ultimately got annoyed by the fans on my MBPs always spinning up on even simple tasks based on the CPU's aggressive behaviour with ramping up Clock Speed on tasks that sometimes barely have any benefit from it.
I came across a bunch of tools that allow you to disable/enable Turbo Boost manually:
http://tbswitcher.rugarciap.com

I had done some simple benchmarks with Handbrake converting video at the time and the difference between TB on and off was roughly 5-10fps when converting 1080p.

I know this isn't a solution to a much bigger underlying issue, but I'm curious as to see how this affects the overall performance with aggressive throttling in place.

https://www.reddit.com/r/apple/comm...rkaround_with_the_i9/?st=JJSL0JYG&sh=2eb4c2c9

 

[Naimfan]

Interesting that the new 6 core MBPs score so high on Geekbench but so poorly on Cinebench. The 2.6 I had scored effectively the same as my 3.33 GHz MP 12 core, but is annihilated by it in Cinebench (965 v. 1632). And the MP has the advantage of not burning my fingers when I use the keyboard.

All the more frustrating because Apple seems to have corrected the keyboard failing due to heat. Irony, there!

 

[zshane1125]

Here are my results, i9, 32/512/560X, prime95 ~5 min torture test.

p95 was never an issue for me....not sure why that dude turned off Turbo, when I did that, it only decreased my performance and the throttle was the same on Cinebench.
[doublepost=1532007601][/doublepost]

https://www.reddit.com/r/apple/comm...rkaround_with_the_i9/?st=JJSL0JYG&sh=2eb4c2c9

I don't think it was an issue on P95...

Try Cinebench, same shxt will happen.

 

[Elektrofone]

Does anyone have an i7 to compare some After Effects benchmarks to?

 

[afir93]

Let me guys tell you something about Intel CPUs. Intel offers desktop, low-power CPUs somewhat similar to laptop ones.

There is 6C/12T Core i7-8700T. It is supposed to have 35W TDP.

It doesn't. Under load, like Cinebench the CPU draws AVERAGE of 82W of power, exceeding the TDP target over twice.

If you will power gate the CPU to 35W, you will not exceed 2.4 GHz in core clocks. And core i7-8700T has 500MHz lower stock speed than Core i9 from MBP, and 800 MHz lower Turbo Boost clock speed.

You cannot cheat physics. Those CPUs will throttle like a pigs. Of course Apple is not helping them, with their cooling solutions.

The only way you can slightly help this is by, increasing the fan speeds under load, and repasting the thermal paste on the laptops.

There might not be any other option.

I'm curious, how much improvement could we realistically expect by putting the MBP on a dedicated cooling pad? I don't have much experience with them so I'm genuinely curious if that would or could make a noticeable difference or if it would be negligible. I imagine a lot of heat is distributed downwards to the surface the MBP is placed upon, so if this surface would thoroughly and constantly dissipate the heat, could it at least somewhat help? Or would it be a pointless endeavor?

 

[ronincse]

Welp now that there are a lot more tests, including from typically very pro-apple sites, I will happily say it is in fact a problem. I still say the original D2D video was a bad test even if it turned out to be correct in the end .

Honestly if I had gotten an i9 I would strongly consider exchanging it for an i7 model.

 

[NJRonbo]

Welp now that there are a lot more tests, including from typically very pro-apple sites, I will happily say it is in fact a problem. I still say the original D2D video was a bad test even if it turned out to be correct in the end .

Honestly if I had gotten an i9 I would strongly consider exchanging it for an i7 model.

I have the i9. I just spent a day setting it up.

The last thing I want to do is return it, that is, unless I have to.

Two questions...

What do you guys think will Apple eventually do about this situation? It's obviously starting to garner a lot of traction online.

Will a cooling pad help the situation at all?

 

[Standard]

I have the i9. I just spent a day setting it up.

The last thing I want to do is return it, that is, unless I have to.

Two questions...

What do you guys think will Apple eventually do about this situation? It's obviously starting to garner a lot of traction online.

Will a cooling pad help the situation at all?

I was actually going to make a thread about this. Due to the nature of the chassis, do cooling pads even work effectively on MacBook Pros? If so, can anyone recommend a particular model?

 

[NJRonbo]

...and what is Apple's return policy once an item is opened?

 

[bhodinut]

You have 14 days to return it to Apple. No questions asked. At least in the US.

 

[drvelocity]

...and what is Apple's return policy once an item is opened?

14-days no questions asked.

 

[Mockletoy]

I was actually going to make a thread about this. Due to the nature of the chassis, do cooling pads even work effectively on MacBook Pros? If so, can anyone recommend a particular model?

You could start by making sure the bottom of the MacBook can breathe. Just having it elevated a bit in the back can make a big difference, so that the chassis can radiate heat away more quickly.

 

[NJRonbo]

Is there any benefit to keeping this i9?

...and please serious answers only. I have read all the freezer jokes and it's getting old.

 

[Queen6]

I have the i9. I just spent a day setting it up.

The last thing I want to do is return it, that is, unless I have to.

Two questions...

What do you guys think will Apple eventually do about this situation? It's obviously starting to garner a lot of traction online.

Will a cooling pad help the situation at all?

Apple may issue a firmware update to tune the fan profile.

Cooling pad will have negligible impact, case will feel cooler to touch internals will remain the same with no meaningful cooling.

Q-6

 

[OC40]

Is there any benefit to keeping this i9?

...and please serious answers only. I have read all the freezer jokes and it's getting old.

It’s really up to you and your needs. You have 14 days to return it, assuming that you are in the US.

I opted for the i9 on a whim because it only cost $276 for me. It’s currently sitting unopened on our desk and I have an i7 on order. Unless new information is released before the i7 arrives, I am returning it.

 

[Naimfan]

Is there any benefit to keeping this i9?

...and please serious answers only. I have read all the freezer jokes and it's getting old.

I would like to believe that Apple will sort this out, and that over time the i9 models will throttle less and offer faster performance.

However, for right now, I'd suggest returning it in favor of the 2.2 2018, which *should* run cooler. I returned a 2.6 yesterday and ordered a 2.2, will report back when it arrives.

 

[Elektrofone]

It’s really up to you and your needs. You have 14 days to return it, assuming that you are in the US.

I opted for the i9 on a whim because it only cost $276 for me. It’s currently sitting unopened on our desk and I have an i7 on order. Unless new information is released before the i7 arrives, I am returning it.

Any chance you could run some tests in apps like Adobe After Effects to see the difference between the i7 and i9. I know that the i9 is throttling but if the i9 is faster than the i7 in those tests then it doesn't really matter to me.

 

[aevan]

Is there any benefit to keeping this i9?

...and please serious answers only. I have read all the freezer jokes and it's getting old.

It depends on the workflow, but the i9 is certainly a lot faster for burst tasks, even with all the throttling. If a program is multi-threaded and requires short bursts of CPU power, the benefit will be obvious.

If you're getting it for something that requires sustained performance, like video export or rendering, then the benefits over an i7 may be smaller, but I still believe it will be faster.

 

[Elektrofone]

It depends on the workflow, but the i9 is certainly a lot faster for burst tasks, even with all the throttling. If a program is multi-threaded and requires short bursts of CPU power, the benefit will be obvious.

If you're getting it for something that requires sustained performance, like video export or rendering, then the benefits over an i7 may be smaller, but I still believe it will be faster.

Are there any benchmarks out there comparing the i7 to the i9 in rendering benchmarks? The D2D video shows a brief mention of the 2018 i7 compared to the i9 but I would like to see more detailed info.

 

[aevan]

I will happily say it is in fact a problem

Why would you be happy about that?

 

[rdowty]

It depends on the workflow, but the i9 is certainly a lot faster for burst tasks, even with all the throttling. If a program is multi-threaded and requires short bursts of CPU power, the benefit will be obvious.

If you're getting it for something that requires sustained performance, like video export or rendering, then the benefits over an i7 may be smaller, but I still believe it will be faster.

I think Apple needs to tweak their fan profiles as well and afterward I'd imagine sustaining the base frequency will be possible and the cost of the i9 justified.

 

[OC40]

Any chance you could run some tests in apps like Adobe After Effects to see the difference between the i7 and i9. I know that the i9 is throttling but if the i9 is faster than the i7 in those tests then it doesn't really matter to me.

The i7 isn’t scheduled to be delivered until the end of the month (icry). I am sure that others will have completed similar tests by then.