How hot does the i9 run vs i7? (15" MBP 2018)

[Synchro3]

This article claims i9 is still slower than i7: https://www.notebookcheck.net/Apple-MacBook-Pro-15-Core-i9-slower-than-Core-i7.396971.0.html

The cooling in the MacBook Pro can dissipate a maximum of 50 watts, while the Core i9, however, would need for full power a cooling of 70 watts.

 

[Thysanoptera]

This article claims i9 is still slower than i7: https://www.notebookcheck.net/Apple-MacBook-Pro-15-Core-i9-slower-than-Core-i7.396971.0.html

The cooling in the MacBook Pro can dissipate a maximum of 50 watts, while the Core i9, however, would need for full power a cooling of 70 watts.

That is misleading somewhat, here on forum we see a good 10-15% difference between exactly the same models, and recent reports from forum members show i9 with Vega running at around 1100 in Cinebench on first pass. It's not that i9 is slower, they virtually identical in multicore loads and both can pull over 80W, i9 can boost momentarily to higher clock and this, all other variables being equal, 'should' result in higher scores because Apple allows the CPU to violate 100C temp limit for a short period of time. This result is because of patented, innovative 'brick trowel and a thermal paste bucket' method of applying thermal compound, which judging by the initial reports of the i9/Vega combo seemed to be retired, but apparently not necessarily so. You buy MacBook nowodays you taking part in a lottery, multi level lottery.

 

[leman]

This article claims i9 is still slower than i7: https://www.notebookcheck.net/Apple-MacBook-Pro-15-Core-i9-slower-than-Core-i7.396971.0.html

All this an unfortunate consequence of Intel's marketing. For all intents and purposes, the i7 and i9 are the same chips*, just configured to run different clocks. In the past, Intel used to put a more strict upper clock limit onto its "cheaper" CPUs. For example, the Skylake CPUs were configured by Intel so that they would maximally reach a multi-core clock of 500Mhz higher over their base clock, even if the CPU itself (sans limiter) would be technically perfectly capable of higher clocks. Now, with Coffee Lake we see the i7 run (multicore) 700Mhz or even higher over its base clock, which is a clear change from before. Why did they do it? Probably because otherwise the performance improvement of 6-core Coffee Lake over the 4 core Skylake would be very small and there wouldn't be enough incentive for users to upgrade.

It is very likely that the 3.2-2.3 clock (multicore sustained operation) is the upper limit of what most of these CPU dies can sustain under the 45Watt TDP. The only way to get more multi-core performance out of the i9 is to run it above its official spec. In other words, you need to run the i9 as if it's a desktop CPU. Which is again, quite awkward, since Intel sells it as a 45Watt part. Frankly, Intel put laptop makers in a very uncomfortable position here: offer it, and get criticised for lacklustre performance, or skip it, and get criticised for not offering it. All this could be fixed simply by selling the i9 as a 60W part (which it really should be) for large enthusiast laptops.

What does all this mean for the i7 vs. i9 debate in the MBP context? Well, that i9 upgrade is even more situational than before. If you rely on sustained multi-core operation (video encoding, raytracing etc.), then i9 probably won't do anything for you. If your workflows are more bursty and utilise mixed single-thread and multi-thread operation, the i9 will most likely have a slight edge over the i7. As it stands, I wouldn't recommend to get i9. I did get it since it does help out a bit with my work, and money wasn't an issue anyway.

*Yes, there is a difference in enabled cache size, but it's not really relevant for making my point.

 

[Queen6]

All this an unfortunate consequence of Intel's marketing. For all intents and purposes, the i7 and i9 are the same chips*, just configured to run different clocks. In the past, Intel used to put a more strict upper clock limit onto its "cheaper" CPUs. For example, the Skylake CPUs were configured by Intel so that they would maximally reach a multi-core clock of 500Mhz higher over their base clock, even if the CPU itself (sans limiter) would be technically perfectly capable of higher clocks. Now, with Coffee Lake we see the i7 run (multicore) 700Mhz or even higher over its base clock, which is a clear change from before. Why did they do it? Probably because otherwise the performance improvement of 6-core Coffee Lake over the 4 core Skylake would be very small and there wouldn't be enough incentive for users to upgrade.

It is very likely that the 3.2-2.3 clock (multicore sustained operation) is the upper limit of what most of these CPU dies can sustain under the 45Watt TDP. The only way to get more multi-core performance out of the i9 is to run it above its official spec. In other words, you need to run the i9 as if it's a desktop CPU. Which is again, quite awkward, since Intel sells it as a 45Watt part. Frankly, Intel put laptop makers in a very uncomfortable position here: offer it, and get criticised for lacklustre performance, or skip it, and get criticised for not offering it. All this could be fixed simply by selling the i9 as a 60W part (which it really should be) for large enthusiast laptops.

What does all this mean for the i7 vs. i9 debate in the MBP context? Well, that i9 upgrade is even more situational than before. If you rely on sustained multi-core operation (video encoding, raytracing etc.), then i9 probably won't do anything for you. If your workflows are more bursty and utilise mixed single-thread and multi-thread operation, the i9 will most likely have a slight edge over the i7. As it stands, I wouldn't recommend to get i9. I did get it since it does help out a bit with my work, and money wasn't an issue anyway.

*Yes, there is a difference in enabled cache size, but it's not really relevant for making my point.

Not far off the mark, a good 8750H will pull approximately 60W at 3.9GHZ, at the PL-1 limit of 45W tops 3.6Ghz if undervolted. Without undervolt you'd be looking at around 3.3GHz, assuming cooling is adequate. I don't really see why Intel/OEM's didn't increase the limits, some W10 notebooks are configured for a PL-1 of 55W I know that much and some can unlock the Turbo limits. 45W is just the TDP, if the cooling system can handle it, the CPU will be fine at higher loads. Personally I'd be very careful of fully unlocking the Turbo limits, as even the 8750H can pull as much as 90W and without detailed specifications that may end badly if the notebooks powertrain isn't up to it

i9 works, yet equally doesn't really suit vast majority of notebook. Few I'm aware of that can deliver the full performance of the i9 pull well over 100W for the CPU in isolation and the power supply if not power supply's are probably as heavy as a 15" MBP Personally I wouldn't opt for the i9 in a MPB, nor any thin & light notebook for that matter as ultimately it will be constrained and the real-world benefits barely tangible for the vast majority.

Q-6

 

[vikramjairam]

Had my 2018 Macbook pro for roughly over a year now. Bought mine in December 2018.
I have an external 3440x1440 monitor attached and under normal conditions temps are around 55 deg C to 65 deg C. When I am running multiple virtual machines in VMware Fusion, with CPU Usage at around 50-60-%, GPU Usage at around 50-100% I see temps hit 90-95 deg C when the VMs are starting up (especially windows). My memory usage is at around 25Gb, and . temps do stabilize at around 65-75 deg C.