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

[kbk75]

So I was doing a few simple tests between the various computers I have, just to see how they compare. One of the things I did was run Geekbench 4's 64bit CPU benchmark, since it is cross-platform, so it could give me some idea of how my 2016 tMBP's i7-6920HQ compared to, say, my i7-2600k in my Windows 10 desktop. I was quite surprised that the MBP actually benchmarked a higher GB4 score than my desktop that's running at 4.6GHz, but I digress.

While I ran the GB4 benchmark I had Intel Power Gadget open so I could see how the CPU speed spiked with each test. I was surprised to see that, at no point, did the CPU go over 3.45 GHz during any of the tests, single or multi-threaded. Now I know synthetic benchmarks aren't truly representative of real-world app performance, but I am now curious to know how high everyone else's CPU goes while running GB4. I guess what I'm wondering is, was the $$ I spent on upgrading the CPU to the 2.9 GHz completely wasted, if it doesn't even throttle-up to the 3.8 GHz it's supposedly rated at, for single-thread performance? Even the base 2.6 GHz model is rated for 3.5 GHz when boosting.

For those that are curious the computer had just been started, ambient temperature was 22 degrees C (around 72 F) and the test really doesn't run long enough for the CPU to start thermally throttling.

Please chip-in with your CPU's peak speed while running GB4, together with your computer's info and GB scores. You can get your CPU model number using Mac CPUID.

Mine hit 3.45 GHz, it's an i7-6920HQ (2.9 - 3.8 GHz) in a 2016 tMBP, scores were 4505 single threaded and 13785 multi threaded.

TL;DR Post your Mac's peak CPU speed while running Geekbench 4 along with your CPU model and computer info, along with your Geekbench 4 scores so we can see how the various processor options compare.

 

[ZapNZs]

2014 15-inch with 2.5 GHz i7-4870HQ (turbo up to 3.7 GHz)
4225 Single-Core / 13746 Multi-Core
OS is 10.11.6
And definitely not long enough to show the impact of thermal throttling, although it did reach 204F for a few seconds

 

[zarathu]

2014 15-inch with 2.5 GHz i7-4870HQ (turbo up to 3.7 GHz)
4225 Single-Core / 13746 Multi-Core
OS is 10.11.6
And definitely not long enough to show the impact of thermal throttling, although it did reach 204F for a few seconds

If you look at the already posted scores, my late 2013 MBP IG is faster in the 4 cores than the new one, and only a measly 300 points slower in the single core(which is 4.5% faster), a number that it would be impossible to notice.

Of course if your a processing large RAW video files you could tell, or using high power games, but that's because it has a DG and mine only has the IG.

Bottom line: Unless you have to have the latest and bestest, or you have the need to be driving two monitors one of which is a 4K, or some serious graphic task, you'll never know.

 

[kbk75]

2014 15-inch with 2.5 GHz i7-4870HQ (turbo up to 3.7 GHz)
4225 Single-Core / 13746 Multi-Core
OS is 10.11.6
And definitely not long enough to show the impact of thermal throttling, although it did reach 204F for a few seconds

Thanks for posting your scores. Any chance you ran Intel Power Gadget to check what speed your CPU actually got up to? That's what I'm really trying to find out because mine is supposed to get up to 3.8 GHz but never goes above 3.5 GHz, which is what the $300 cheaper base CPU is rated for.

 

[RoboWarriorSr]

If I recall correctly, the turbo speed listed is only when one core is activated, the speed decreases roughly 100 MHz for each additional core used.

 

[kbk75]

If I recall correctly, the turbo speed listed is only when one core is activated, the speed decreases roughly 100 MHz for each additional core used.

Yeah, that's true, so it should hit 3.8 GHz when the single core tests are running, right?

By comparison my 2012 rMBP's i7-3720QM (2.6-3.6 GHz) went up to 3.41 GHz running Geekbench 4. Scores were 3624 Single threaded and 11449 Multi threaded.

I've also run a terminal script to load up just one core and I can confirm that my 2016 tMBP will not go beyond 3.45 GHz no matter what. This is 350 MHz lower than it's rated for. The 2012, by comparison, gets to within 200 MHz of its rated speed.

Anybody know if the OS is limiting the processors from going flat-out or if it's a firmware thing or what?

As an aside, my Windows 10 desktop's i7-2600K hit its overclocked 4.6 GHz while running Geekbench 4 and scored 4337 Single threaded and 13721 Multi threaded.

 

[ZapNZs]

Thanks for posting your scores. Any chance you ran Intel Power Gadget to check what speed your CPU actually got up to? That's what I'm really trying to find out because mine is supposed to get up to 3.8 GHz but never goes above 3.5 GHz, which is what the $300 cheaper base CPU is rated for.

Mine peaks at about 3.5-ish GHz with the Geekbench 4 test. However, with some Apps I use, it will go a bit north of 3.65-ish (and generate a massive amount of heat.)
[doublepost=1486751705][/doublepost]

If you look at the already posted scores, my late 2013 MBP IG is faster in the 4 cores than the new one, and only a measly 300 points slower in the single core(which is 4.5% faster), a number that it would be impossible to notice.

Of course if your a processing large RAW video files you could tell, or using high power games, but that's because it has a DG and mine only has the IG.

Bottom line: Unless you have to have the latest and bestest, or you have the need to be driving two monitors one of which is a 4K, or some serious graphic task, you'll never know.

From a pure processor perspective, I don't think we've seen a huge change in the 15-inch for the past half-decade, at least compared to what had been the growth for several decades prior with laptops. Compared to things like iGPUs and SSDs, the performance growth of quad-core laptop CPUs has been anything but eye-opening!

However, I am interested to see more about how the 2016 can better sustain a higher workload. Heat has always been an enemy of the MBP and a substantial reduction of it could mean a lot for sustained tasks. But quantifying that can't be easy to do given it is so User-dependent.

 

[MrGuder]

Is this in the App Store, how do I download geek bench 4 I'd like to test my 2015 and 2016.

 

[kbk75]

Mine peaks at about 3.5-ish GHz with the Geekbench 4 test. However, with some Apps I use, it will go a bit north of 3.65-ish (and generate a massive amount of heat.)
[doublepost=1486751705][/doublepost]

From a pure processor perspective, I don't think we've seen a huge change in the 15-inch for the past half-decade, at least compared to what had been the growth for several decades prior with laptops. Compared to things like iGPUs and SSDs, the performance growth of quad-core laptop CPUs has been anything but eye-opening!

However, I am interested to see more about how the 2016 can better sustain a higher workload. Heat has always been an enemy of the MBP and a substantial reduction of it could mean a lot for sustained tasks. But quantifying that can't be easy to do given it is so User-dependent.

3.65 is 150 MHz higher than my 2016's 2.9 i7 wants to go! This is disappointing for users like me who have paid for the processor upgrade. And yes, this CPU runs a LOT cooler than the one in my 2012 rMBP while idling or running under light loads, however it thermal throttles within 10 minutes of running anything intensive such as Handbrake. In their pursuit of ever thinner and lighter machines, Apple sacrifices the system's ability to disperse heat under heavy loads. I think they have also tuned the fan curve too sedately; by the time the fans start spinning faster, the CPU is already running at 80 degrees C or higher.

Most of all I'm wondering if they've gimped the fastest processors so that they don't generate extra heat, need the fans to spin faster, and eat-up the already reduced battery life on these new systems.

Is this in the App Store, how do I download geek bench 4 I'd like to test my 2015 and 2016.

www.geekbench.com

 

[Sanpete]

it thermal throttles within 10 minutes of running anything intensive such as Handbrake. In their pursuit of ever thinner and lighter machines, Apple sacrifices the system's ability to disperse heat under heavy loads. I think they have also tuned the fan curve too sedately; by the time the fans start spinning faster, the CPU is already running at 80 degrees C or higher.

Could you describe the throttling you've observed? Reviews have found this MBP to be less prone to heat issues including throttling than previous models, so the notion that ability to disperse heat was sacrificed doesn't fit well with the improvement. Of course a large machine could reduce heat issues even more, but there's a balance that more than ever favors effective heat management.

Most of all I'm wondering if they've gimped the fastest processors so that they don't generate extra heat, need the fans to spin faster, and eat-up the already reduced battery life on these new systems.

Similarly, the battery life of the 15" is significantly improved over the 2015, the opposite of already reduced. Maybe you refer to the battery size? Things that would spin up the fans are generally not things that people expect good battery life for anyway.

 

[kbk75]

I ran a 3.5GB Handbrake mp4 to H265 conversion this afternoon. When the job began my CPU was turbo-ing up to 3.5 GHz and Handbrake showed that the job would take around an hour and 47 mins to complete (which was faster than my 4.6 GHz i7-2600k Windows 10 PC was going to do the job, by around 5-7 mins).

Ten minutes later the CPU had stepped down to 3.2 GHz and the job showed 2 hours required to complete, later still it throttled down further to around 3 GHz. Eventually it took around 2 and a half hours for the job to complete, while the desktop finished the job in an hour and 53 mins because it chugged along at 4.6 GHz for the entire job.

Edit: Yes, I meant reduced in-size battery which, in turn, leads to poorer battery life than it might have been, had they gone with the larger, shaped batteries they originally wanted to use.

 

[Sanpete]

I ran a 3.5GB Handbrake mp4 to H265 conversion this afternoon. When the job began my CPU was turbo-ing up to 3.5 GHz and Handbrake showed that the job would take around an hour and 47 mins to complete (which was faster than my 4.6 GHz i7-2600k Windows 10 PC was going to do the job, by around 5-7 mins).

Ten minutes later the CPU had stepped down to 3.2 GHz and the job showed 2 hours required to complete, later still it throttled down further to around 3 GHz. Eventually it took around 2 and a half hours for the job to complete, while the desktop finished the job in an hour and 53 mins because it chugged along at 4.6 GHz for the entire job.

I see. Not very severe throttling to merely reduce the turbo boost. Compare that to any other MBP, and the vast majority of other laptops, and it will probably look pretty good. How was the fan speed and temperature during the reduced turbo, if you happened to notice?

 

[kbk75]

No, it managed to at least keep up the speeds at 100 MHz turbo. The fans were running flat out and temps would go as high as 80 degrees C or so, before falling around 10-12 degrees with each drop in speed.

 

[Sanpete]

No, it managed to at least keep up the speeds at 100 MHz turbo. The fans were running flat out and temps would go as high as 80 degrees C or so, before falling around 10-12 degrees with each drop in speed.

I gather it isn't what you'd hope for, but that's actually rather encouraging considering what I've seen from previous models!

 

[kbk75]

To be honest, I wasn't even disappointed with the turbo not being sustained, I always knew that would be the case. What's really eating at me is that this thing just will not go over 3.5 GHz, no matter what. Why the heck did I pay $300 extra to get the same performance as the base CPU?

That's why I was actually hoping to get some input from other 2016 15" owners while they run GB4, just to see if it was just my unit or symptomatic of the lot.

 

[Sanpete]

To be honest, I wasn't even disappointed with the turbo not being sustained, I always knew that would be the case. What's really eating at me is that this thing just will not go over 3.5 GHz, no matter what. Why the heck did I pay $300 extra to get the same performance as the base CPU?

If you're concerned about whether the performance is worth paying for, just compare the performance against the base CPU, rather than the turbo speed against the claimed max turbo. My understanding is that the performance is significantly improved, for whatever reasons.

That's not to say your investigation isn't interesting in a general way. Would be interesting to know why it isn't turboing higher. A post above suggests it may be related to the task it's doing.

 

[kbk75]

That's just it, the base CPU is supposedly capable of turbo-ing to 3.5GHz. This one is meant to go to 3.8, which is barely 10% faster, but it's something I've paid to have. If all I'm getting is what the base CPU can also do, I've received zero value for my purchase.

Yes, the max turbo speed will only happen in a single threaded application. As I mentioned in one of my earlier replies, I've even run a script in Terminal to fully load up just one core to see if the system would go higher than 3.5GHz, but no joy. :-/

 

[Sanpete]

What I mean is that the performance of the 2.9 in accomplishing tasks is significantly better than that of the base CPU, whether due to extra turbo or something else. That seems to be the value to focus on, not the max turbo you observe.

ZapNZs reports that his matches your experience for turbo in GB4, but that it exceeds 3.5 in other applications.

 

[kbk75]

Actually, it won't be any better than the performance of the base 2.6 CPU, if that CPU manages to turbo to its rated 3.5GHz. These chips all scale dynamically based on how many cores are being stressed, so if the base CPU manages to hit its rated 3.5GHz under single-threaded loads, then it is every bit as fast as my 2.9 CPU, which should be able to hit 3.8, but doesn't seem to want to go over 3.45 or so under single-threaded loads.

As we speak I'm converting another mp4 to H265, this time with turbo disabled, just to see if the processor will thermal throttle down from it's 2.9 base speed or not. So far, it's managing to hold its 2.9 clock, which isn't surprising, given that it managed to sustain over 3 GHz earlier, with turbo enabled. CPU temps at the moment are 93 degrees C and fans are spinning at 3800 on the left side and 3500 on the right, at the moment.

Edit: Zap's machine is a 2.5GHz older CPU that is managing to sustain 3.5GHz. This means there will be virtually no difference in performance between his system and mine.

 

[Sanpete]

Actually, it won't be any better than the performance of the base 2.6 CPU, if that CPU manages to turbo to its rated 3.5GHz.

But is is better according to a variety of measurements, which is what matters. More than max turbo speed is involved in performance. You should just compare performance directly instead of turbo, if performance is your concern.

As we speak I'm converting another mp4 to H265, this time with turbo disabled, just to see if the processor will thermal throttle down from it's 2.9 base speed or not. So far, it's managing to hold its 2.9 clock, which isn't surprising, given that it managed to sustain over 3 GHz earlier, with turbo enabled. CPU temps at the moment are 93 degrees C and fans are spinning at 3800 on the left side and 3500 on the right, at the moment.

Looks good.

Edit: Zap's machine is a 2.5GHz older CPU that is managing to sustain 3.5GHz. This means there will be virtually no difference in performance between his system and mine.

What I've been trying to say is that it may matter what application is used.

 

[Jefe's MacAir]

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

 

[ZapNZs]

3.65 is 150 MHz higher than my 2016's 2.9 i7 wants to go! This is disappointing for users like me who have paid for the processor upgrade. And yes, this CPU runs a LOT cooler than the one in my 2012 rMBP while idling or running under light loads, however it thermal throttles within 10 minutes of running anything intensive such as Handbrake. In their pursuit of ever thinner and lighter machines, Apple sacrifices the system's ability to disperse heat under heavy loads. I think they have also tuned the fan curve too sedately; by the time the fans start spinning faster, the CPU is already running at 80 degrees C or higher.

Most of all I'm wondering if they've gimped the fastest processors so that they don't generate extra heat, need the fans to spin faster, and eat-up the already reduced battery life on these new systems.



www.geekbench.com

They might have modified settings to reduce heat production or to attempt to improve battery - IIRC on the MBP 2011s with dGPUs they released an update designed to reduce the amount of heat the GPU made, and this had some performance impact. I've also seen some scores on this website that have found differences in Geekbench scores between different Sierra builds.

How quickly does heat build up on your 2016? It builds very rapidly on my 2014. For example, where as yours can hold 3 GHz, I highly doubt mine could for an extended period, and I assume that can yield in major improvements. I'm going to watch this monitor closely when doing some of my more intensive work and play around some VMs that I'll only dedicate one core to...

 

[IngerMan]

2016 13" TB Base 2.9 Model. Gadet showed 3.1 Max, Rated 3.3 Turbo

 

[Sanpete]

They might have modified settings to reduce heat production or to attempt to improve battery - IIRC on the MBP 2011s with dGPUs they released an update designed to reduce the amount of heat the GPU made, and this had some performance impact. I've also seen some scores on this website that have found differences in Geekbench scores between different Sierra builds.

Good point.

 

[Synchro3]

5550 single core
17973 multi core

Turbo 4.4 GHz

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