Performance reduced after EFI update on MacBook Pro Retina

[WPWoodJr]

My experience

I had an interesting experience. Not only did the graphics card get much slower but the SSD did too. Here's my Windows 7 Experience Index under Bootcamp before re-setting the SMC:

Processor Intel(R) Core(TM) i7-3820QM CPU @ 2.70GHz 7.7
Memory (RAM) 16.0 GB 7.8
Graphics NVIDIA GeForce GT 650M 5.5
Gaming graphics 4095 MB Total available graphics memory 6.4
Primary hard disk 222GB Free (308GB Total) 5.9

After re-setting the SMC it is:

Processor Intel(R) Core(TM) i7-3820QM CPU @ 2.70GHz 7.7
Memory (RAM) 16.0 GB 7.8
Graphics NVIDIA GeForce GT 650M 7.3
Gaming graphics 4095 MB Total available graphics memory 7.3
Primary hard disk 222GB Free (308GB Total) 7.9

See how the graphics went up dramatically as did the SSD speed!

What Windows 7 Experience Index are others getting, before and after re-setting SMC?

 

[runebinder]

No Idea.

Although I've been running the Intel Power Gadget app lately and it seems when Idle my 2.6Ghz CPU runs between 1.2-1.5Ghz until I start something that needs it.

Can others here check to see if the same is true for you? The Intel app is OS X compatible so you don't need to have windows to run it.

When running a 4K video in VLC I have seen the CPU get up to 3.5+ Ghz, so I'll keep the monitor open to see if that bug hits me again.

I don't understand what you mean by bug. CPUs down clock themselves when not needed, Intel's way of doing it is called SpeedStep and has been around for years. So on idle 1.2GHz to 1.5GHz sounds correct.

The CPU is perfectly capable of performing basic tasks at this speed, and it helps battery life and reduce heat.

As long as it hits 2.6GHz and it's turbo boost speeds when more demanding apps are running (which going by your post it does), then there's nothing wrong.

BTW, what do 4K videos look like on the rMBP?

 

[bill-p]

BTW, what do 4K videos look like on the rMBP?

Like watching 1080p. Especially for night clips.

We need much better camera sensors.

But if you ask me how rendered 3D contents look like on the rMBP... I can only say... "surreal". Way too detailed. It's like seeing a miniaturized world. I suspect that when we do get better camera sensors that can capture more details, then the same thing will apply.

 

[WPWoodJr]

PassMark benchmark results

Just to be sure that my machine was back to normal after the SMC reset, I compared it's performance on the PassMark benchmark to other rMBPs. It does very well:

 

[Newfiejudd]

I don't understand what you mean by bug. CPUs down clock themselves when not needed, Intel's way of doing it is called SpeedStep and has been around for years. So on idle 1.2GHz to 1.5GHz sounds correct.

The CPU is perfectly capable of performing basic tasks at this speed, and it helps battery life and reduce heat.

As long as it hits 2.6GHz and it's turbo boost speeds when more demanding apps are running (which going by your post it does), then there's nothing wrong.

BTW, what do 4K videos look like on the rMBP?

The bug is very simple, your system down clocks everything. CPU and GPU and does not allow it to run at full potential. This basically locks speedstep into a low cpu/gpu state.

Put it this way your V8 is running on 2 cylinders, is this normal? That is the same thing people are experiencing with the SMC bug. No matter how far you put your foot through the floor, it will only run on 2 cylinders.

 

[runebinder]

The bug is very simple, your system down clocks everything. CPU and GPU and does not allow it to run at full potential. This basically locks speedstep into a low cpu/gpu state.

Put it this way your V8 is running on 2 cylinders, is this normal? That is the same thing people are experiencing with the SMC bug. No matter how far you put your foot through the floor, it will only run on 2 cylinders.

I understand the EFI bug, but what the person I replied to described was normal usage: Low CPU frequency when doing minimal tasks and then it turbo boosting to 3.5GHz when watching 4K video. A CPU under clocking itself in the correct scenarios in not a bug, only if it stays in that state no matter what you are doing.

The person wasn't experiencing a locked under clock hence why I didn't see why they described it as a bug.

 

[N19h7m4r3]

I understand the EFI bug, but what the person I replied to described was normal usage: Low CPU frequency when doing minimal tasks and then it turbo boosting to 3.5GHz when watching 4K video. A CPU under clocking itself in the correct scenarios in not a bug, only if it stays in that state no matter what you are doing.

The person wasn't experiencing a locked under clock hence why I didn't see why they described it as a bug.

I originally did, which is why I got the power gadget to monitor it after an SMC reset.

I've had times now when my CPU functions correctly but the GPU lags heavily in 3d environments and games until I reset the SMC again.

 

[kimovski]

This is something related with switching graphics. About an week ago i forced only using GT650M and the whole week i didnt have any problems with the performance. Until last night ticked "automatic switching..." when he goes to sleep then the problem is back again

 

[kurinzo]

This is something related with switching graphics. About an week ago i forced only using GT650M and the whole week i didnt have any problems with the performance. Until last night ticked "automatic switching..." when he goes to sleep then the problem is back again

Not really.
Im always working with my "thunderbolt display", so my rMBP is always using the discrete card (gt650m). And the bug is appears. I mean throttling and sometimes smc glitch...

 

[brockm]

Not really.
Im always working with my "thunderbolt display", so my rMBP is always using the discrete card (gt650m). And the bug is appears. I mean throttling and sometimes smc glitch...

What's crazy is this bug has been around for like six months now and Apple has not fixed it. So we have to assume the hardware itself is flawed or Apple doesn't give a crap.

 

[anarti]

I run external 24" via thunderbold, using intel power gadget I can see that cpu is around 2GHz-2.7GHz (i7 2.6), never less.

Is this a part of the same bug?

 

[Queen6]

This is something related with switching graphics. About an week ago i forced only using GT650M and the whole week i didnt have any problems with the performance. Until last night ticked "automatic switching..." when he goes to sleep then the problem is back again

Nope, my 2.3 base Retina switches as needed and i have never had any issue, it has to run deeper, hence Apple`s lack of fix for those with systems suffering this issue. Fundamentally the user base with the problem is either too small, or Apple is unsure how to proceed; will a fix create more issue, who knows.

Personally i have deliberately tried to induce throttling on my own 15" Retina to no avail. The same attempts on my Late 2011 15" (2.4) result in almost instantaneous throttling yet my own Retina "to all accounts and purpose" appears to be impervious to throttling are all Mac`s equal possibly not, food for thought indeed...

 

[bill-p]

I run external 24" via thunderbold, using intel power gadget I can see that cpu is around 2GHz-2.7GHz (i7 2.6), never less.

Is this a part of the same bug?

No. This bug forces the CPU, GPU, and every other component to run at their slowest states possible. So if your CPU is not locked to 1.2GHz, you're fine.

----

And for those who are affected by this bug, can you guys keep resetting SMC and count how many times? I can't quite remember how many times, but after that last SMC reset, I never got this bug again.

 

[Valkyre]

No. This bug forces the CPU, GPU, and every other component to run at their slowest states possible. So if your CPU is not locked to 1.2GHz, you're fine.

----

And for those who are affected by this bug, can you guys keep resetting SMC and count how many times? I can't quite remember how many times, but after that last SMC reset, I never got this bug again.

i have reseted countless times already and the bug never went away... it will happen sooner or later again. I am beginning to believe that it happens mostly after waking from sleep. Dunno why but that is the impression i got...

 

[luffytubby]

this is pretty disheartening to hear. I never saw any reviews mention this bug. and there are a ton of reviews out there.

 

[bill-p]

i have reseted countless times already and the bug never went away... it will happen sooner or later again. I am beginning to believe that it happens mostly after waking from sleep. Dunno why but that is the impression i got...

I reset mine many times a day, all 7 days of the week for a straight 3 months before the bug stopped itself.

this is pretty disheartening to hear. I never saw any reviews mention this bug. and there are a ton of reviews out there.

This bug won't appear in a review that took the writer less than a day. Much like IR may not appear to a reviewer who hasn't the machine for at least 3 months.

 

[scenox]

this is pretty disheartening to hear. I never saw any reviews mention this bug. and there are a ton of reviews out there.

most people never had this bug, otherwise boards would have been full with it. if you search in this forum, there was a poll about it. i also never experienced this bug (sept machine), if i do, i would go straight to the apple store to get a new machine. i can't understand people which have been struggling with this bug for 3 months or longer.

 

[frimp]

Just change sleepmode to 25 with "sudo pmset hibernatemode 25". Believe me, the bug is history then. On the other hand, wake-up now takes the additional time to recover from disk instead of RAM.
But this is negligible compared to low fps.
I really hope Apple will fix this. I had no problems before the darn EFI update - so why change hardware?!?

 

[Newfiejudd]

Just change sleepmode to 25 with "sudo pmset hibernatemode 25". Believe me, the bug is history then. On the other hand, wake-up now takes the additional time to recover from disk instead of RAM.
But this is negligible compared to low fps.
I really hope Apple will fix this. I had no problems before the darn EFI update - so why change hardware?!?

Exactly my system was fine before the EFI update. After that I have been fighting it.

 

[Valkyre]

Just change sleepmode to 25 with "sudo pmset hibernatemode 25". Believe me, the bug is history then. On the other hand, wake-up now takes the additional time to recover from disk instead of RAM.
But this is negligible compared to low fps.
I really hope Apple will fix this. I had no problems before the darn EFI update - so why change hardware?!?

can you please explain what exactly this command does. I really want to get rid of that bug, but I want to know what I am about to do to my computer, what is the trade off. Don't want to mess things up.

Thank you.

 

[frimp]

Hi,

this command changes the default behavior of the sleepmode.
Normally, closing the lid or entering the sleepmode by inactivity uses a RAM-buffered sleepmode. In this case, all peripheral devices are put into low-power state and sleepmode. Power is still applied to the RAM to keep contents. After some hours in this sleepmode, the device goes into deep sleep.
In deep sleepmode, all RAM content is written to disk and power is removed from the machine and all components.

You discover the difference when waking up. A wakeup from deep sleep takes the time to copy the disk image back to the RAM. This takes e.g. about 5-10 seconds on a rMPB with 16 GB. This delay is the initial experience you have when you just see the login screen and have to wait the RAM recovery time, before you can enter the password and continue to work.

The command changes the behaviour to deep sleep only. Therefore, the rMBP will always go to sleep without touching the peripheral low-power settings (in this case: the nVidia settings), so your 650M will continue to work at full speed after wakeup.
The drawback: it will always take the above mentioned time to wake up. This is annnoying, but the only known interim solution by now.

Apple is updating all other MBPs with bug fixes but not the Retina.

 

[Valkyre]

Hi,

this command changes the default behavior of the sleepmode.
Normally, closing the lid or entering the sleepmode by inactivity uses a RAM-buffered sleepmode. In this case, all peripheral devices are put into low-power state and sleepmode. Power is still applied to the RAM to keep contents. After some hours in this sleepmode, the device goes into deep sleep.
In deep sleepmode, all RAM content is written to disk and power is removed from the machine and all components.

You discover the difference when waking up. A wakeup from deep sleep takes the time to copy the disk image back to the RAM. This takes e.g. about 5-10 seconds on a rMPB with 16 GB. This delay is the initial experience you have when you just see the login screen and have to wait the RAM recovery time, before you can enter the password and continue to work.

The command changes the behaviour to deep sleep only. Therefore, the rMBP will always go to sleep without touching the peripheral low-power settings (in this case: the nVidia settings), so your 650M will continue to work at full speed after wakeup.
The drawback: it will always take the above mentioned time to wake up. This is annnoying, but the only known interim solution by now.

Apple is updating all other MBPs with bug fixes but not the Retina.

Thank you very much for your lengthy explanation. I already tested the command and now I am waiting to see the results!

Only one thing I have to ask is, what is the command to revert back to the default sleepmode setting?

It is really ugly and disappointing that apple has updated every single machine out there except the retina which was supposed to be their top tier product... thats just wrong...

 

[frimp]

Hi, the standard setting for the sleepmode is 3 (bit 1 and 2 set):

0000 0001 (bit 0) enables hibernation; causes OS X to write memory state to hibernation image at sleep time. On wake (without bit 1 set) OS X
will resume from the hibernation image. Bit 0 set (without bit 1 set) causes OS X to write memory state and immediately hibernate at sleep
time.

0000 0010 (bit 1), in conjunction with bit 0, causes OS X to maintain system state in memory and leave system power on until battery level
drops below a near empty threshold (This enables quicker wakeup from memory while battery power is available). Upon nearly emptying the bat-
tery, OS X shuts off all system power and hibernates; on wake the system will resume from hibernation image, not from memory.

So "sudo pmset hibernatemode 3" is your friend.

 

[Valkyre]

Hi, the standard setting for the sleepmode is 3 (bit 1 and 2 set):



So "sudo pmset hibernatemode 3" is your friend.

Αwesome! Thank you man!

 

[Valkyre]

i think i might have found the culprit (or at least one of the culprits) responsible for the efi bug...

I noticed that mail application is the one that causes my machine to produce this stupid glitch. Having the mail app open and letting the computer go to sleep will have quite a large possibility to produce this bug.

I think it as something to do with power nap and downloading mail messages while the computer is sleeping.

I have now my computer on for about a week without the mail app open and for the first time i have not encountered the efi bug for such a big period of time. Usually, I encountered the glitch everyday. Now its been 6-7 days without it.

o my guess is mail app has something to do about it.

Just sharing some information with you guys.