Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

fate0311

Suspended
Dec 31, 2015
548
451
Apple did a lot of fan boys who did not warrant the i9 a favor.

Now they are all returning them for baseline 2.2’s and saving cash.
 

ksj1

macrumors 6502
Jul 17, 2018
294
535
Yeah no thanks.
I dont remember iphone release that overheats and underperforms. Its a huge company with huge budget not some boutique company. They had enough time to prepare.

ROFL - This is not an iPhone. Even if it were, you know those upgrades you get all the time on your iPhone? Those are firmware updates that add features, fix bugs, and improve performance all the time. This happens on laptop computers, desktops, etc.

WTF - Has no one upgraded the bios on a PC to accept a new CPU? It happens all the time. I don't think there is a thing to see here.

Regardless, I just ran a compile that took ~25 minutes on my mid-2015 i7 2.8. It was ~17 minutes on my new i9. I do this several times a day. I think I'll keep it.
 

agaskew

macrumors 6502
Dec 3, 2009
416
253
ROFL - This is not an iPhone. Even if it were, you know those upgrades you get all the time on your iPhone? Those are firmware updates that add features, fix bugs, and improve performance all the time. This happens on laptop computers, desktops, etc.

WTF - Has no one upgraded the bios on a PC to accept a new CPU? It happens all the time. I don't think there is a thing to see here.

Regardless, I just ran a compile that took ~25 minutes on my mid-2015 i7 2.8. It was ~17 minutes on my new i9. I do this several times a day. I think I'll keep it.

What a pity...if only it didn't throttle so much, the compile might only take 12 minutes. Oh well...
 

Eason85

macrumors 6502
Jan 29, 2017
258
308
Hong Kong
ROFL - This is not an iPhone. Even if it were, you know those upgrades you get all the time on your iPhone? Those are firmware updates that add features, fix bugs, and improve performance all the time. This happens on laptop computers, desktops, etc.

WTF - Has no one upgraded the bios on a PC to accept a new CPU? It happens all the time. I don't think there is a thing to see here.

Regardless, I just ran a compile that took ~25 minutes on my mid-2015 i7 2.8. It was ~17 minutes on my new i9. I do this several times a day. I think I'll keep it.

There are firmware updates that improve performance occasionally. It depends on why the performance is bad. Apple can increase the fan curve and achieve the same result people have already gotten by using 3rd party apps. They won't get any better than what people are getting by disable turbo and turning the fans on max -- scores that are very underwhelming for coffee-lake by any measure (I wouldnt' be surprised if Apple kept the fans off more to improve battery life at the expensive of performance).

There is no firmware update that turns one tiny little heatpipe into the 2 or 3 that every other coffee lake laptop uses. I am a laptop reviewer and an enthusiast who does a lot of testing and tweaking to ensure every laptop I have is running at peak performance, as well. I have owned dozens of laptops and benchmarked them all extensively in the last 2 years. Apple's new MBPs are severely under-engineered for heat dissipation, period.
 
Last edited:

pkouame

macrumors 65816
Jul 7, 2016
1,054
2,319
I have a MacBookPro15,1 (i9 with 32GB) and, amid so many facts and/or speculations (including the Intel Power Gadget utility download link removed), I looked into an alternate test.

Basically, I created a bash script that builds a large project (over 1000 C++ source files) using incrementally from 1 to 12 threads and measuring the build time of each case. Then, I compared the results with a MacBookPro14,3 (2017 i7 3.1GHz), with threads from 1 to 8.

Results are quite interesting:
Z0yFkUr.gif


As expected, the i9 is faster with only with one thread (it seems that the CPU frequency was over 4GHz). But immediately with only 2 treads, the old i7 is faster.

By comparing the two absolute minimum build time (564s and 614s), there is a gain of only 8%.
Interesting. A scenario that hits home with many. A few more details would be useful to make any kind of meaningful conclusion here: actual LOCs not just file count, the compiler being used, the filesystems, the project being built, the power gadget profiles. Assuming you carefully staged an honest test, results are disappointing for some that would be doing this all day long. Could you share your project? I wouldn't mind running that test on my 2017 i7 16GB 512SSD (same as yours).
 

ksj1

macrumors 6502
Jul 17, 2018
294
535
There are firmware updates that improve performance occasionally. It depends on why the performance is bad. Apple can increase the fan curve and achieve the same result people have already gotten by using 3rd party apps. They won't get any better than what people are getting by disable turbo and turning the fans on max -- scores that are very underwhelming for coffee-lake by any measure. There is no firmware update that turns one tiny little heatpipe into the 2 or 3 that every other coffee lake laptop uses. These laptops are severely under-engineered for heat dissipation, period.

This is too funny. I must say for the record, I don't use my computers to run benchmarks. I use them to get work done. My new one gets it done much faster.

BTW- Are you a software developer? I am. Do you know what microcode is? Do you understand what firmware is and what can be done with it? Versus the OS the computer is running?

CPU optimization (or are you all denying that the i9 has different internal timing than an i7) is not trivial. It can take many iterations to get it right. I go back to the heat they got at WWDC and this being an early release. As others have shown, this can be improved upon with Volta, so why you believe the people in control of the hardware and software can't do much better is beyond me.

Or are you a thermal engineer with extensive industry experience and I'm just misreading you (and everyone else on this board)?

Jeez...
[doublepost=1532246416][/doublepost]
Interesting. A scenario that hits home with many. A few more details would be useful to make any kind of meaningful conclusion here: actual LOCs not just file count, the compiler being used, the filesystems, the project being built, the power gadget profiles. Assuming you carefully staged an honest test, results are disappointing for some that would be doing this all day long. Could you share your project? I wouldn't mind running that test on my 2017 i7 16GB 512SSD (same as yours).

I'm confused, 4 threads seem to be ignored here. What was the overall completion time for the job?
 

Eason85

macrumors 6502
Jan 29, 2017
258
308
Hong Kong
This is too funny. I must say for the record, I don't use my computers to run benchmarks. I use them to get work done. My new one gets it done much faster.

Completely irrelevant red herring to the discussion: Apple's implementation of coffee lake is strangled compared to every other OEM's.

BTW- Are you a software developer? I am. Do you know what microcode is? Do you understand what firmware is and what can be done with it? Versus the OS the computer is running?
No, yes, yes, and yes. Apple is on a level playing field with every other OEM with the microcode of coffee lake.

CPU optimization (or are you all denying that the i9 has different internal timing than an i7) is not trivial. It can take many iterations to get it right. I go back to the heat they got at WWDC and this being an early release.
Why would Apple, one of the largest companies in the world, get it so wrong compared to every other OEM? I don't even care about the i9 -- it's a joke CPU. Looking at the pathetic performance of the i7-8750H in the MBP 15 and even the quad-core CPUs in the MBP 13 is more than enough, even ignoring the debacle of the i9.

As others have shown, this can be improved upon with Volta, so why you believe the people in control of the hardware and software can't do much better is beyond me.

You don't seem to understand why it can be improved upon with Volta, so I am not surprised you think Apple can create some magic that breaks the laws of physics. Lots of people without understanding of laptop engineering and a cult-like trust in Apple indeed do seem to think they can. The cooling system of the 2017 mbp was pushed to the absolute limit already. Apple threw in a CPU that generates far more heat and consumes much more power and didn't change a ****ing thing with the cooling solution -- in fact, people are saying that the fan profile is even more minimal on the 2018 than 2017. There is no magic bullet from what they've done. The only thing they could do that users haven't already done themselves is unlock the VID (which Apple locked after 2015 for no reason) and allow users to undervolt their machines themselves to the degree each system allows.

Or are you a thermal engineer with extensive industry experience and I'm just misreading you (and everyone else on this board)?

Thermal engineer or not, I have extensive industry experience with the tweaking and modification of hardware and software to overcome poor engineering. I am extremely familiar with coffee lake and wrote up the very first analyses and benchmarks of it from release.
 
Last edited:

pkouame

macrumors 65816
Jul 7, 2016
1,054
2,319
This is too funny. I must say for the record, I don't use my computers to run benchmarks. I use them to get work done. My new one gets it done much faster.

BTW- Are you a software developer? I am. Do you know what microcode is? Do you understand what firmware is and what can be done with it? Versus the OS the computer is running?

CPU optimization (or are you all denying that the i9 has different internal timing than an i7) is not trivial. It can take many iterations to get it right. I go back to the heat they got at WWDC and this being an early release. As others have shown, this can be improved upon with Volta, so why you believe the people in control of the hardware and software can't do much better is beyond me.

Or are you a thermal engineer with extensive industry experience and I'm just misreading you (and everyone else on this board)?

Jeez...
[doublepost=1532246416][/doublepost]

I'm confused, 4 threads seem to be ignored here. What was the overall completion time for the job?
Yes confusing. The test results suggest the i9 took as long to complete the "job" with 8 threads as it did with 12.
 

M.Rizk

macrumors 6502a
Original poster
Apr 20, 2015
785
613
There are firmware updates that improve performance occasionally. It depends on why the performance is bad. Apple can increase the fan curve and achieve the same result people have already gotten by using 3rd party apps. They won't get any better than what people are getting by disable turbo and turning the fans on max -- scores that are very underwhelming for coffee-lake by any measure (I wouldnt' be surprised if Apple kept the fans off more to improve battery life at the expensive of performance).

There is no firmware update that turns one tiny little heatpipe into the 2 or 3 that every other coffee lake laptop uses. I am a laptop reviewer and an enthusiast who does a lot of testing and tweaking to ensure every laptop I have is running at peak performance, as well. I have owned dozens of laptops and benchmarked them all extensively in the last 2 years. Apple's new MBPs are severely under-engineered for heat dissipation, period.

They can actually issue an update that offers better performance than what we can achieve right now.

1. Volta is limited to 45W. The 2.6 i7 and i9 require a little higher power limit for better performance. Apple has no limitations on their end, they can choose any value.

2. We are limiting the power limit as soon as the OS boots. Apple won’t do this. They can enforce power limits based on time intervals and thermal status as other manufacturers. Other manufacturers for example let the CPU pull full power for the first 28 seconds then enforce an initial power limit then a secondary power limit in a sustained load.

When an i9 user here tried letting full turbo for the first 15 seconds then enforced a power limit fast the Cinebench score improved from around 850~950 to 950~1000. (Again, i9 requires more than 45W so there is even more room left for improvement)

3. When a 45W power limit was enforced on the 2.2 GHz i7 the temp reached around 80c, and the fans were still quiet as Mac only goes full fans when at 100c. There is still room for higher performance (Max is 100c)

4. We are not even sure if that’s the only sort of performance optimization that can be done. We have so little when it comes to access to the deep layer of the Mac. There is little we can do.
 

pkouame

macrumors 65816
Jul 7, 2016
1,054
2,319
They can actually issue an update that offers better performance than what we can achieve right now.

1. Volta is limited to 45W. The 2.6 i7 and i9 require a little higher power limit for better performance. Apple has no limitations on their end, they can choose any value.

2. We are limiting the power limit as soon as the OS boots. Apple won’t do this. They can enforce power limits based on time intervals and thermal status as other manufacturers. Other manufacturers for example let the CPU pull full power for the first 28 seconds then enforce an initial power limit then a secondary power limit in a sustained load.

When an i9 user here tried letting full turbo for the first 15 seconds then enforced a power limit fast the Cinebench score improved from around 850~950 to 950~1000. (Again, i9 requires more than 45W so there is even more room left for improvement)

3. When a 45W power limit was enforced on the 2.2 GHz i7 the temp reached around 80c, and the fans were still quiet as Mac only goes full fans when at 100c. There is still room for higher performance (Max is 100c)

4. We are not even sure if that’s the only sort of performance optimization that can be done. We have so little when it comes to access to the deep layer of the Mac. There is little we can do.
Any bets on when Apple will deliver this performance update everyone is talking about? Let alone issue a statement about all of this?
 

ksj1

macrumors 6502
Jul 17, 2018
294
535
You don't seem to understand why it can be improved upon with Volta, so I am not surprised you think Apple can create some magic that breaks the laws of physics. Lots of people without understanding of laptop engineering and a cult-like trust in Apple indeed do seem to think they can.

Thermal engineer or not, I have extensive industry experience with the tweaking and modification of hardware and software to overcome poor engineering. I am extremely familiar with coffee lake and wrote up the very first analyses and benchmarks of it from release.

I worked at Microsoft for 14 years working with this stuff. And yes, I now use a mac pretty much exclusively.

Yes, Apple can do all kinds of stuff with the cpu to optimize it. Do you really think you just plug some random plug compatible cpu into a computer with a proprietary OS and it just works optimally without anyone doing anything?

Or do you think that cpu timing doesn't matter if the OS doesn't really quite work with what they want there yet, but it's good enough to release?

As I said, you can all have your benchmark wars and complaints. I'll be getting work done.
 

M.Rizk

macrumors 6502a
Original poster
Apr 20, 2015
785
613
Any bets on when Apple will deliver this performance update everyone is talking about? Let alone issue a statement about all of this?

They won’t. It really won’t surpise me if they ignore our complaints and won’t be a first.

As sad as this sounds, I really wish they surprise us for once.
 

Eason85

macrumors 6502
Jan 29, 2017
258
308
Hong Kong
They can actually issue an update that offers better performance than what we can achieve right now.

1. Volta is limited to 45W. The 2.6 i7 and i9 require a little higher power limit for better performance. Apple has no limitations on their end, they can choose any value.

2. We are limiting the power limit as soon as the OS boots. Apple won’t do this. They can enforce power limits based on time intervals and thermal status as other manufacturers. Other manufacturers for example let the CPU pull full power for the first 28 seconds then enforce an initial power limit then a secondary power limit in a sustained load.

When an i9 user here tried letting full turbo for the first 15 seconds then enforced a power limit fast the Cinebench score improved from around 850~950 to 950~1000. (Again, i9 requires more than 45W so there is even more room left for improvement)

3. When a 45W power limit was enforced on the 2.2 GHz i7 the temp reached around 80c, and the fans were still quiet as Mac only goes full fans when at 100c. There is still room for higher performance (Max is 100c)

4. We are not even sure if that’s the only sort of performance optimization that can be done. We have so little when it comes to access to the deep layer of the Mac. There is little we can do.

I don't have a macbook anymore and am not familiar with the limitations of this volta software, but I am familiar with XTU and throttlestop to set PL1, PL2, and different multiplier locks based on temperatures, as well as undervolting on regular notebooks. What I mean is, you might be able to slightly refine the level of control you have now with this volta software, but I've already thinking about solutions for this issue assuming such limits don't exist -- as they don't with XTU and TS.

You guys are seeing very high temps (80c) at quite low clocks (2.2 ghz). I just ran a quick test on my XPS 15 9570 (also one of the thinnest & lightest and thus most thermally-constrained laptops with the i7-8750H), running a 12-thread stress test at 2.2 GHz. After 1 minute, temperatures on all cores held at 52C and the fans did not come on. Again, the MBP 15's behavior is telling me that the cooling solution is just woefully underdesigned, and there is nothing that can be done to get these chips to hit their potential. And these are brand new macs, with no dust in the fans, no dry thermal grease, etc.
 
  • Like
Reactions: cfdlab

M.Rizk

macrumors 6502a
Original poster
Apr 20, 2015
785
613
I don't have a macbook anymore and am not familiar with the limitations of this volta software, but I am familiar with XTU and throttlestop to set PL1, PL2, and different multiplier locks based on temperatures, as well as undervolting on regular notebooks. What I mean is, you might be able to slightly refine the level of control you have now with this volta software, but I've already thinking about solutions for this issue assuming such limits don't exist -- as they don't with XTU and TS.

You guys are seeing very high temps (80c) at quite low clocks (2.2 ghz). I just ran a quick test on my XPS 15 9570 (also one of the thinnest & lightest and thus most thermally-constrained laptops with the i7-8750H), running a 12-thread stress test at 2.2 GHz. After 1 minute, temperatures on all cores held at 52C and the fans did not come on. Again, the MBP 15's behavior is telling me that the cooling solution is just woefully underdesigned, and there is nothing that can be done to get these chips to hit their potential. And these are brand new macs, with no dust in the fans, no dry thermal grease, etc.

1. 80c wasn’t at 2.2. I just meant it was 80c running the 2.2 GHz i7, it was actually running at 2.9~3.1 GHz.

2. I am not defending Apple. I never will. They did a terrible job leaving it a hot mess as is and they did an even worse job with the cooling system compared to other brands.
 
  • Like
Reactions: Eason85

pkouame

macrumors 65816
Jul 7, 2016
1,054
2,319
They won’t. It really won’t surpise me if they ignore our complaints and won’t be a first.

As sad as this sounds, I really wish they surprise us for once.
Hear hear...if keyboard gate is any indication, they will follow sop and ride this out mute. Under the covers Adobe will be pressured to tune Premiere or else. Wouldn't surprise me if Mojave will silently roll in some performance fixes too. That way, they claim victory with the new OS. How long did the iPhone 6 throttling snafu with battery fix take to (sort of) get resolved?
 
  • Like
Reactions: M.Rizk

M.Rizk

macrumors 6502a
Original poster
Apr 20, 2015
785
613
There is now a method to undervolt the Mac too, any volunteers?

https://www.reddit.com/r/apple/comm...18_performance_volta/?st=JJWKRJTQ&sh=362b6c5e

Warning: Undervolting is not as easy as enforcing a power limit. If you put in wrong values the system might freeze and in some cases you might need to wipe the system and re-install Mac OS.

(Volta has undervolting features but it doesn’t seem to work with the 8th gen CPUs)
 

ksj1

macrumors 6502
Jul 17, 2018
294
535
Claiming you are better educated than someone else never makes a strong argument.

hmm, I agree, but you would have to show how I claimed to be better educated. I didn't say I was better educated, and lets just say that there are many, many more out there with far more impressive degrees than I.

Edit - To also be clear, I am not a thermal engineer and had no intention to make myself out as one. I ha
 

limo79

macrumors 6502
Jan 9, 2009
298
139
Using Volta or another tool is totally against philosophy of Apple product ready to use after unboxing. Personally I do not like to made own bugfixes because I pay for a good product and respect my private time so it is better to go for a walk with Rex than fixing faults of Apple engineers. Maybe I should to replace all inductors to eliminate coil whine in brand new MacBook Pro 2018 for $6900 to TDK because Apple uses some chinese crap from Shenzhen Yangcy ****d up Parts because they are greedy? Hope that they will fix it in mainboard Rev 02 and they release an official patch and finally product will be good (excepting people that still believe that it can handle 4.8GHz or even more because it is Apple). It would be good if MacRumors chief may send an official question to Apple about these problems constructed gently and clever so Apple may respond safely giving reasonable explanation or just a hope to take all these findings into consideration as customer feedback to improve product. Maybe there are reasons that we do not know why TDP limitation was not added eg. because most of Apple customers use it for Facebook or Netflix according data telemetry and 1% of customers use it for work. Then situation is clear what are the Apple customer expectations. Hope that Mr Schiller is an open person and cool guy (they use animoji at work so they need to be cool) so will be glad to send reasonable engineering explanation after consulting with law department. I assume that communication between Apple and customers are on the same level like design of Apple stores. Is not it?
 
Jul 4, 2015
4,487
2,551
Paris
There will probably be a firmware update to control the TDP.

HOWEVER!

Uncomfortably hot laptops have always existed and are not going away soon. The issue is people using their machines in impractical ways.

Is a mobile solution even ready to handle your needs? Rendering 4K or 8K video streams on a laptop isn't recommended by serious professionals. They usually perform low res edits in the field and then complete the high res renders on workstations or server farms. You have to use the right device type for specific tasks.

Those YouTube critics are being disingenuous and looking for clickbait because you can clearly see in other videos (or even in the background) that they use desktop workstations for the final high quality video rendering.

CG is also another example. Even the best mobile CoffeeLake processor with great cooling will perform 60% slower than a 5Ghz desktop CoffeeLake. If you're serious about CG then you don't use a laptop for final CG renders. Use a workstation for the final output. It's a business expense so pros don't care about the cost. Are we really supposed to believe CG pros are typically rendering on laptops? They're not. So why hammer impractical use cases like that on forums?

Does the laptop get too hot under lesser loads that usually handle fine on quad-core or even dual-core laptops? Such as graphics, achitecture, design, word processing, programming? If so, then there is a problem. If not, then there isn't so much of a problem.
 
Last edited:

cfdlab

macrumors regular
Feb 26, 2008
179
220
Apple is putting a very powerful cpu into a laptop that is not able to cool it. It is absolutely correct to point this out. Why are some people trying to shoot the messenger here instead of questioning apple design ?
 
Jul 4, 2015
4,487
2,551
Paris
YESS!!!! They're holding it wrong!!! They're typing on it wrong!!

Maybe there's a new EULA that requires these machines to be operated in freezers?

Your reply is a bit cray cray and irrelevant to what I posted.

There was a time when you couldn't scrub or render video, even on the most powerful AVID desktop workstation. But you could import TARGA sequences, edit, save out your EDL or project files, and then have the work finished by a film (actual film) editor or rendered on a farm.

Times haven't changed much.

If you're a serious pro you don't do 4K/8K video or CG rendering on laptop. You only do lower quality pre-viz on the laptop. No matter what generation of laptop you use, it will always be hot and you will always be laughed at by other professionals using workstations and server farms. This is not going to change for at least another 10-15 years as video and 3D graphics continue to become more demanding.


Those YouTubers are doing their renders on desktop hardware, not laptops. Tell them to be honest.
 

inksane

macrumors newbie
Mar 27, 2018
5
9
I'm starting to get into web design/ development. I'm getting the base 2018. Would the throttling affect me?

I just bought a new macbook pro 15", meanwhile I've been working as a web dev in a 2013 base model macbook air 13" i5 4GB of ram for 2 years.
Swaps like crazy and runs hot but I can get the work done, the alternative is to use the company provided surface books.

I use safari, chrome, webstorm (4 projects open and running concurrently), slack, citrix, spotify and outlook plus other light weight apps.

So yes, a 2018 15" it's actually stupidly powerful for web dev.
 
  • Like
Reactions: ckeck and CodeJoy
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.