What speed do your CPUs get up to in Geekbench 4?

[Mindinversion]

however it thermal throttles within 10 minutes of running anything intensive such as Handbrake.

Something's got to be wrong there.... I just finished a 31 small file [110Mb-ish ea] .mkv - .mp4 set, and it never went over 53c with fan just under 6k RPM [measured via SMC Fan Control with default profile]

I mean, I suppose it's possible SMC isn't reading accurately, but I'm a former Razer Blade man. . . I know what 95c feels like.. and while the machine is moderately warm on the bottom and above the touchpad, at no time does it feel "hot" like. . say.. the 2015 Blade.

I also noticed that with SMC fan control the fans spin up a lot more in handbrake than they do when playing casual games like Heroes Of The Storm, and under HoTS the temp stays in the 61-70c temp range [but still well within tJunct.]

Also

MacBook Pro (15-inch Late 2016)
Single-Core Score Multi-Core Score
4452 13641
Geekbench 4.0.4 Tryout for Mac OS X x86 (64-bit)

 

[ZapNZs]

kbk75,

This is running a Windows 10 Pro x64 VM with "4" 2+2 cores to the physical and "4" 2+2 to the virtual while I was crunching some numbers (so mainly this is CPU heat as the GPU wasn't doing a tremendous lot given it was a VM and the UI was stripped of all of the visualizations that normally eat up resources.) I did not expect it to hold quite at 3.35-3.5, but the temps are about maxed at this point, and both fans were at 6200 RPM. If I was doing work that caused significant GPU heat (or if the battery wasn't already charged and hence I was getting some charging heat), I assume I would then see more of a scale back with the CPU in order to reduce overall heat...?

(If I put this on my lap, I probably wouldn't have needed that vasectomy...)

I never upgraded to Sierra because El Capitan has been such a flawless OS for my needs that I've just stuck with it (plus I hate the babbling lady.) It makes me curious though if going to Sierra would affect performance one way or another.

Do you know if the new MBP does much in the way of GPU offloading? If so, could this affect the numbers you are seeing if certain tasks are being pushed over to the GPU?...and if the GPU is in fact being used to augment certain CPU tasks, how would this factor into benchmarks? When I run the GB benchmark, my 750M GPU sits near idle, even though it is set to discrete graphics. Could this be different with the newer GPUs?

 

[kbk75]

kbk75,

This is running a Windows 10 Pro x64 VM with "4" 2+2 cores to the physical and "4" 2+2 to the virtual while I was crunching some numbers (so mainly this is CPU heat as the GPU wasn't doing a tremendous lot given it was a VM and the UI was stripped of all of the visualizations that normally eat up resources.) I did not expect it to hold quite at 3.35-3.5, but the temps are about maxed at this point, and both fans were at 6200 RPM. If I was doing work that caused significant GPU heat (or if the battery wasn't already charged and hence I was getting some charging heat), I assume I would then see more of a scale back with the CPU in order to reduce overall heat...?


View attachment 688030
(If I put this on my lap, I probably wouldn't have needed that vasectomy...)

I never upgraded to Sierra because El Capitan has been such a flawless OS for my needs that I've just stuck with it (plus I hate the babbling lady.) It makes me curious though if going to Sierra would affect performance one way or another.

Do you know if the new MBP does much in the way of GPU offloading? If so, could this affect the numbers you are seeing if certain tasks are being pushed over to the GPU?...and if the GPU is in fact being used to augment certain CPU tasks, how would this factor into benchmarks? When I run the GB benchmark, my 750M GPU sits near idle, even though it is set to discrete graphics. Could this be different with the newer GPUs?

Those graphs actually look pretty good to me, Zap. My 2016 will run 3 GHz for well over two hours with all cores maxed while running Handbrake. At no point did it switch from the integrated 530 graphics to the Radeon 460 I have, I watched closely to see if it would utilize the dGPU, it didn't. Perhaps that's unsurprising, given there was no real computational work taking place. Also the GPU temps didn't go up much, only the CPU did.

Without getting more actual CPU speeds from 2016 owners, it's hard to deduce what might be going on with my system. What's clear, though, is that it will easily run at it's rated 2.9GHz base clock for extended periods of time, fully loaded on all cores. Temps will go as high as 93 Celsius and the fans will spin to just shy of 6,000 rpm. The left is always 300-400 rpm faster than the right, too.

However, given that your 2014 unit will not only turbo higher, but also sustain higher turbo speeds than my 2016 will, clearly shows that heat is a factor. What upsets me is the thought that perhaps a base 2.6 i7 might just as easily have gone to and sustained 3 GHz for such tasks, making my upgrade a complete waste of money.

I'd imagine something like 3DMark (on Windows) would give you a good idea of whether or not the system will switch to discrete graphics even for physics tests, not involving any great graphics load per se, but I've not got Windows running on this thing, so I can't confirm.
[doublepost=1486851110][/doublepost]

Something's got to be wrong there.... I just finished a 31 small file [110Mb-ish ea] .mkv - .mp4 set, and it never went over 53c with fan just under 6k RPM [measured via SMC Fan Control with default profile]

I mean, I suppose it's possible SMC isn't reading accurately, but I'm a former Razer Blade man. . . I know what 95c feels like.. and while the machine is moderately warm on the bottom and above the touchpad, at no time does it feel "hot" like. . say.. the 2015 Blade.

I also noticed that with SMC fan control the fans spin up a lot more in handbrake than they do when playing casual games like Heroes Of The Storm, and under HoTS the temp stays in the 61-70c temp range [but still well within tJunct.]

Also

MacBook Pro (15-inch Late 2016)
Single-Core Score Multi-Core Score
4452 13641
Geekbench 4.0.4 Tryout for Mac OS X x86 (64-bit)

See what happens if you're converting a single, larger file to H265. I'd imagine it would get much hotter and fairly quickly, too. Also, while both Intel Power Gadget and iStat menus reported my temps at 93 C, the system didn't 'feel' all that hot to me, which is a good thing, no doubt. My 2012 retina MBP feels a heck of a lot hotter on my lap even at 75 C CPU temps than this 2016 unit feels when the CPU is hitting 93 C. This makes me wonder if, perhaps, your temps are reading incorrectly? They might well be correct, but even if they're not, the new MBPs do a good job of not localising heat to a point where the unit feels too uncomfortable to have on your lap.
[doublepost=1486851587][/doublepost]

I just ran it and
3937 single core
7626 multi core

I've gotten over 4000 and 8000 many times before but not always. Generally closer to what I've posted above. i5 base tb w/ 512 GB. I was getting similar numbers w/ a base i5 tb and 256 GB.

https://browser.geekbench.com/v4/cpu/1830680

Thanks for the scores. Any chance you monitored the CPU speed with Intel Power Gadget?
[doublepost=1486851729][/doublepost]

5550 single core
17973 multi core

Turbo 4.4 GHz

https://browser.geekbench.com/v4/cpu/1788957

Those Kaby 7700Ks are sweet chips. I'm still holding on to my Sandy at 4.6 for the moment, though! It's run 4.6GHz for five years straight! Best CPU I ever had in a desktop. The 2016 MBP's 2.9 would actually outrun my desktop if it were able to sustain a 3.4GHz turbo, which is impressive, given it's 1.2GHz slower.

 

[Sanpete]

What upsets me is the thought that perhaps a base 2.6 i7 might just as easily have gone to and sustained 3 GHz for such tasks, making my upgrade a complete waste of money.

Yours has more L3 cache. There's more to speed than clock. The one you should compare to is the 2.7, which has the same L3 cache. Chances are high that yours is still faster, though.

 

[ZapNZs]

Those graphs actually look pretty good to me, Zap. My 2016 will run 3 GHz for well over two hours with all cores maxed while running Handbrake. At no point did it switch from the integrated 530 graphics to the Radeon 460 I have, I watched closely to see if it would utilize the dGPU, it didn't. Perhaps that's unsurprising, given there was no real computational work taking place. Also the GPU temps didn't go up much, only the CPU did.

Without getting more actual CPU speeds from 2016 owners, it's hard to deduce what might be going on with my system. What's clear, though, is that it will easily run at it's rated 2.9GHz base clock for extended periods of time, fully loaded on all cores. Temps will go as high as 93 Celsius and the fans will spin to just shy of 6,000 rpm. The left is always 300-400 rpm faster than the right, too.

However, given that your 2014 unit will not only turbo higher, but also sustain higher turbo speeds than my 2016 will, clearly shows that heat is a factor. What upsets me is the thought that perhaps a base 2.6 i7 might just as easily have gone to and sustained 3 GHz for such tasks, making my upgrade a complete waste of money.

I'd imagine something like 3DMark (on Windows) would give you a good idea of whether or not the system will switch to discrete graphics even for physics tests, not involving any great graphics load per se, but I've not got Windows running on this thing, so I can't confirm.

At least looking at GB averages, there appears to be a pretty significant difference between the 2.6/2.7 and 2.9? I also noticed that my scores are different than the average on GB, especially the multicore score. Assuming that GB results are automatically set to drop any sample where something is clearly outside the theoretical peak for that system as well as drop any sample where the results are so low it indicates a major problem with that specific system's hardware, does that indicate binning variations among individual specimens?

 

[kbk75]

Yours has more L3 cache. There's more to speed than clock. The one you should compare to is the 2.7, which has the same L3 cache. Chances are high that yours is still faster, though.

Fair point

At least looking at GB averages, there appears to be a pretty significant difference between the 2.6/2.7 and 2.9? I also noticed that my scores are different than the average on GB, especially the multicore score. Assuming that GB results are automatically set to drop any sample where something is clearly outside the theoretical peak for that system as well as drop any sample where the results are so low it indicates a major problem with that specific system's hardware, does that indicate binning variations among individual specimens?

View attachment 688170


View attachment 688170

Mine is actually benching well above those indicative scores, on the single thread test anyway, so yeah I'm guessing there is a fair bit of variation, not only between various chips, but also from run-to-run, I've seen my own scores fluctuate a bit too.

 

[IngerMan]

I have always scored higher on GB MAC average listings. From my iMac to my MBP's. I am not sure how they calculate the average but if some users perform the test while the MAC has many open processes or it is indexing from new it will produce a lower score, which you can see on many registered scores for the specific model.

 

[kbk75]

At least looking at GB averages, there appears to be a pretty significant difference between the 2.6/2.7 and 2.9? I also noticed that my scores are different than the average on GB, especially the multicore score. Assuming that GB results are automatically set to drop any sample where something is clearly outside the theoretical peak for that system as well as drop any sample where the results are so low it indicates a major problem with that specific system's hardware, does that indicate binning variations among individual specimens?

View attachment 688170


View attachment 688170

Zap, any chance you have Bootcamp on your system? I'm wondering if these systems will clock higher under Windows than they will under MacOS. Unfortunately I don't have Bootcamp on mine, but it is making me wonder...

 

[ZapNZs]

Zap, any chance you have Bootcamp on your system? I'm wondering if these systems will clock higher under Windows than they will under MacOS. Unfortunately I don't have Bootcamp on mine, but it is making me wonder...

I'm afraid my Windows work is done through VMWare Fusion or VirtualBox, as many of my OS' are installed on external SSDs and I often need access to multiple at a time. I imagine someone on here with a 2.9 uses Bootcamp?

Have you tried running a different benchmark software?

 

[kbk75]

I'm afraid my Windows work is done through VMWare Fusion or VirtualBox, as many of my OS' are installed on external SSDs and I often need access to multiple at a time. I imagine someone on here with a 2.9 uses Bootcamp?

Have you tried running a different benchmark software?

I've tried various benchmark tests and real world tests but absolutely nothing will make this machine go over 3.4 - 3.45GHz, which is its max turbo speed for all 4 cores being loaded. Disappointing that with even single core loads it will not go any faster, though.