Cinebenched 2.2 & 2.6Ghz, lid open & Clamshell

[airlied]

I just ran some Cinebench R15 tests of 2.2Ghz & 2.6Ghz 2018 MBP respectively.

Here are some details.

The purposes of the test is to try to verity:

1) Some claim that CPU performance drops slightly on clamshell mode. The test is to see if this is true.

2) And if you care about these two CPUs' theoretical performances this might give you some info (I know there are already plenty of them, but more is always better).

Two 2018 MBPs were being used:

15inch 2.2Ghz/16GB RAM/Radeon pro 560X/512GB
15inch 2.6Ghz/16GB RAM/Radeon pro 560X/512GB

they are identical except CPUs.

On each laptop, I tested 8 times:
4 times with lid open, AC power on, internal display only;

and 4 times on clamshell mode, 4K monitor connected via Blackmagic eGPU (I did this to prove they are indeed on clamshell mode, you can check the screenshots. With lid open 'GFX Board' are AMD Radeon Pro 560X and with clamshell they are 580).

They are recovered from the same 10.14.0 backup (I know it's beta but they are on the same beta), then I did a reinstall again, then waited a full 12 hours without changing anything.

All other applications are closed (all of them), except Cinebench R15 itself and Intel Power Gadget (same version).

Both rebooted then waited 15 minutes before test.

Before every test, I waited long enough to make sure clock and temp drops to normal. However I can't guarantee since even a mouse movement could trigger a slight clock boost.

I know benchmark doesn't mean real world performance, but it also doesn't mean it's useless. Don't argue this please.

Results:

Each CPU indeed performs slightly worse on clamshell mode.

Screenshots:

>2.2Ghz, Lid open:

>2.2Ghz, Clamshell:

>2,6Ghz, Lid open:

>2.6Ghz, Clamshell:

BTW my wallpaper changes every 30 minutes so don't mind...

 

[shavou]

Very good analysis , thanks for sharing.

However, I think you are not presenting this comparison in the best way. Your test shows that the MBP performs slightly worse when connected to an external monitor and not when in clamshell mode (which is not surprising). In other words, to be fully comprehensive, you should run the same tests under the following scenarios:

- MBP only
- MBP connected to an external monitor with the lid open
- MBP connected to an external monitor in clamshell mode

Only then we'll be able to see whether the clamshell mode really affects performance or not...

 

[airlied]

Very good analysis , thanks for sharing.

However, I think you are not presenting this comparison in the best way. Your test shows that the MBP performs slightly worse when connected to an external monitor and not when in clamshell mode (which is not surprising). In other words, to be fully comprehensive, you should run the same tests under the following scenarios:

- MBP only
- MBP connected to an external monitor with the lid open
- MBP connected to an external monitor in clamshell mode

Only then we'll then be able to see whether the clamshell mode really affects performance...

Thanks!! I'll do the test you said when I got time then refresh the post.

 

[keysofanxiety]

Clamshell will always have a slight hit in performance. Why? Because it's connected to an external monitor and that's utilising the graphics card.

As Shav said, you should test it with lid open & connected to an external monitor, then compare that to clamshell & connected to an external monitor. That would be the best test to show if clamshell mode makes a difference. IMO there wouldn't be any real difference between the two.

 

[leman]

Did you check whether there were any other processes taking significant CPU time? Looking at the variance (I used student's t test) in your scores, the non-clamshell 2.2. is faster than the non-clamshell 2.6, there is no performance difference between non-clamshell and clamshell operation of 2.6 and finally, in the clamshell mode, no difference between 2.2 and 2.6.

All in all: inconclusive. The only think I'd take from it is that for Cinebench it doesn't seem to matter which CPU you have. I'd like to see the same done on more straightforward numeric code (raytracing could be insuring a lot of cache misses).

 

[kre62]

B&H running the 2.2ghz model at 2,099 with no tax - quite a steal considering its the top performing model in the group.

 

[roland.g]

B&H running the 2.2ghz model at 2,099 with no tax - quite a steal considering its the top performing model in the group.

Only issue is that B&H unlike Apple doesn't accept returns of opened computers. You get 30 days but if you send it back, you take a 15% hit if they allow the return. Apple will charge more but allow you to return or exchange to a different machine within 14 days.

 

[kre62]

Only issue is that B&H unlike Apple doesn't accept returns of opened computers. You get 30 days but if you send it back, you take a 15% hit if they allow the return. Apple will charge more but allow you to return or exchange to a different machine within 14 days.

A chance I’m willing to take

 

[maflynn]

A chance I’m willing to take

If you're sure you know what you want, I agree, its a no brainer to buy it from them. I think people tend to over-state the short coming of returning the computer. If the consumer such as myself or you, are completely sure this is what you want. Then yes save the money.

 

[roland.g]

If you're sure you know what you want, I agree, its a no brainer to buy it from them. I think people tend to over-state the short coming of returning the computer. If the consumer such as myself or you, are completely sure this is what you want. Then yes save the money.

At this point with B&H offering $150 off through today on the machine I am looking there is some decent savings. I’m more inclined to pick one up from Apple in late Sept or Oct and trial it for 2 weeks, determine if it is what I really want, wait to see what comes out in the fall (do we get a 14” MacBook) and then buy during Black Friday sales.

 

[Queen6]

I just ran some Cinebench R15 tests of 2.2Ghz & 2.6Ghz 2018 MBP respectively.

Here are some details.

The purposes of the test is to try to verity:

1) Some claim that CPU performance drops slightly on clamshell mode. The test is to see if this is true.

2) And if you care about these two CPUs' theoretical performances this might give you some info (I know there are already plenty of them, but more is always better).

Two 2018 MBPs were being used:

15inch 2.2Ghz/16GB RAM/Radeon pro 560X/512GB
15inch 2.6Ghz/16GB RAM/Radeon pro 560X/512GB

they are identical except CPUs.

On each laptop, I tested 8 times:
4 times with lid open, AC power on, internal display only;

and 4 times on clamshell mode, 4K monitor connected via Blackmagic eGPU (I did this to prove they are indeed on clamshell mode, you can check the screenshots. With lid open 'GFX Board' are AMD Radeon Pro 560X and with clamshell they are 580).

They are recovered from the same 10.14.0 backup (I know it's beta but they are on the same beta), then I did a reinstall again, then waited a full 12 hours without changing anything.

All other applications are closed (all of them), except Cinebench R15 itself and Intel Power Gadget (same version).

Both rebooted then waited 15 minutes before test.

Before every test, I waited long enough to make sure clock and temp drops to normal. However I can't guarantee since even a mouse movement could trigger a slight clock boost.

I know benchmark doesn't mean real world performance, but it also doesn't mean it's useless. Don't argue this please.

Results:

Each CPU indeed performs slightly worse on clamshell mode.
View attachment 778707


Screenshots:

>2.2Ghz, Lid open:
View attachment 778701
>2.2Ghz, Clamshell:
View attachment 778702

>2,6Ghz, Lid open:
View attachment 778703
>2.6Ghz, Clamshell:
View attachment 778705

BTW my wallpaper changes every 30 minutes so don't mind...

Interesting and informative, although a longer test such as Corona Renderer may be more revealing.

Q-6