Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.

jay508

macrumors regular
Jul 11, 2020
241
153
Hello, everyone. I come from china. My mother tongue is Chinese. My English is not good.
According to China Chinese Weifeng website, someone has replaced E3 -1230v2 with i5-2400s on version 10.8.2 of the 21.5-inch iMac system in 2011, which can be used normally。
10.8.3 the bootrom efi will be updated when the system starts to install, and the Ivy Bridge series cpu will not light up. That is, if you can still find those that have never upgraded the system version above 10.8.2, you can upgrade and replace the Ivy Bridge cpu series.

I also replaced E3 1230 V2 with i5 2400 in my iMac 2011 27-inch system version 10.13.6. Only two LED lights were on the previous year, and there was no display on the screen. I wonder if bootromefi firmware limits the use of Ivy Bridge series cpu

4324372_QQ图片20121229100040.jpg


I500s replace E3 1230 V2 source

 

Attachments

  • 4324372_QQ图片20121229100040.jpg
    4324372_QQ图片20121229100040.jpg
    84.4 KB · Views: 178
  • Like
Reactions: Petri Krohn

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826
Your Geekbench reports are showing an iMac 11,2 - which is not the 27" 2011 model. This has the ID iMac 12,2?

It would be interesting to get a firmware dump of these old pre 10.8.3 models to analyze the firmware. And yes, currently we all assume that you need to modify the firmware of an iMac 2011 to get Ivy Bridge Support, back.

I assume the firmware upgrade Apple made back in history was related to close some security holes..
 

jay508

macrumors regular
Jul 11, 2020
241
153
Your Geekbench reports are showing an iMac 11,2 - which is not the 27" 2011 model. This has the ID iMac 12,2?

It would be interesting to get a firmware dump of these old pre 10.8.3 models to analyze the firmware. And yes, currently we all assume that you need to modify the firmware of an iMac 2011 to get Ivy Bridge Support, back.

I assume the firmware upgrade Apple made back in history was related to close some security holes..
The 21.5-inch model code of 2011 is 11,2, and the 27-inch model code of 2011 is 12,2. Since the 11,2 models can be the same as the 12,2 models of the platform, it is because the system updates the bootrom efi firmware that it does not support the Ivy Bridge series cpu
 

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826
The 21.5-inch model code of 2011 is 11,2, and the 27-inch model code of 2011 is 12,2. Since the 11,2 models can be the same as the 12,2 models of the platform, it is because the system updates the bootrom efi firmware that it does not support the Ivy Bridge series cpu
No, the 21.5 2011 model has the ID 12,1 not 11,2 ...


The 21,5 mid 2010 has the 11,2 ID and it has a different CPU socket.


You are confusing me :)
 
  • Like
Reactions: roscho

roscho

macrumors member
May 17, 2020
59
39
Switzerland
Hello, everyone. I come from china. My mother tongue is Chinese. My English is not good.
According to China Chinese Weifeng website, someone has replaced E3 -1230v2 with i5-2400s on version 10.8.2 of the 21.5-inch iMac system in 2011, which can be used normally。
10.8.3 the bootrom efi will be updated when the system starts to install, and the Ivy Bridge series cpu will not light up. That is, if you can still find those that have never upgraded the system version above 10.8.2, you can upgrade and replace the Ivy Bridge cpu series.

I also replaced E3 1230 V2 with i5 2400 in my iMac 2011 27-inch system version 10.13.6. Only two LED lights were on the previous year, and there was no display on the screen. I wonder if bootromefi firmware limits the use of Ivy Bridge series cpu

4324372_QQ图片20121229100040.jpg


I500s replace E3 1230 V2 source


Please read the thread carefully, especially the posts of StephN999.

Also the information in wikipedia is not really a miracle


that means that Xeon V2 chips are Ivy bridge. Those chips are BGA compatible but the microcode is not in the NVRAM.

Unfortunately google translate does not work with your cited chinese website. Does it contain any information on microcodes? If not, thise pictures are often fake in my experience and only imposture...

Regards,
roscho

EDIT: apart from that, Ausdauersportler is correct about the imac model numbers...
 
Last edited:
  • Like
Reactions: Ausdauersportler

jay508

macrumors regular
Jul 11, 2020
241
153
No, the 21.5 2011 model has the ID 12,1 not 11,2 ...


The 21,5 mid 2010 has the 11,2 ID and it has a different CPU socket.


You are confusing me :)
I mean, the i5 2400s of 21.5 Sandy Bridge is upgraded to the e3 1230v2 of Ivy Bridge, which is definitely a 2011 model, but it is different from the 27-inch model. Sandy Bridge is the platform to upgrade Ivy Bridge, which is the same in theory.
The model should be MC309
The details are here

4324351_nEO_IMG_DSCN0441.jpg
QQ截图20200730184211.jpg
 

Attachments

  • 4324372_QQ图片20121229100040.jpg
    4324372_QQ图片20121229100040.jpg
    84.4 KB · Views: 155
  • 4324351_nEO_IMG_DSCN0441.jpg
    4324351_nEO_IMG_DSCN0441.jpg
    159.9 KB · Views: 175
  • QQ截图20200730184211.jpg
    QQ截图20200730184211.jpg
    48 KB · Views: 174

jay508

macrumors regular
Jul 11, 2020
241
153
It is true that 10.8.3 system upgrade started to change the motherboard bootrom efi firmware, but 10.8.2 can support Ivy Bridge. It is useless as long as you have installed any version of degraded system above 10.8.3.
QQ截图20200730192132.jpg

 

Attachments

  • QQ截图20200730192132.jpg
    QQ截图20200730192132.jpg
    43.7 KB · Views: 144

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826
If I will find an older 2011 iMac containing an older MacOS than 10.8.3 version I will get the firmware off it using a Ch341A clip. Probably somebody can work out then if there is another firmware part allowing the installation of Ivy Bridge.

Does anybody know of an online firmware database for such 2011 iMacs?
 

StephN999

macrumors 6502
Apr 12, 2020
288
228
Cergy, France
Bonsoir,

I tried to read everything after translation and what I read was that nobody has ever run an Ivy Bridge on a 2011 iMac, and then the GeekBench with an iMac 11,2 bothers me a lot (https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=iMac11,2+ivy)... :confused:

No one has run an Ivy Bridge on a 2011 iMac, until proven otherwise. 😓

@Ausdauersportler there's no database of intel machine firmware (I don't know about older Macs) simply because each firmware contains the machine's info, unique serial number among other things. ;)

Otherwise for those who can with a 21.5" I chewed through the work, such a test is not going to kill your machine, unless you're not as rough as I was. 😟

I stand by what I said, I would test my modification which seems right to me, but for the moment not the financial means, that's all. 😅

I am a little tired of question after question, it would be necessary to make an effort of reading and comprehension, if you are not a tinkerer in the soul go your way. :mad:

I remind you that I am not good in English and that despite translations it is never simple. 😓

Sorry for my anger, but it's going around in circles here! 🥺

Edit: I made my sentences and expressions explode well by the translations, sorry if there is something incomprehensible. :eek:
 
Last edited:

Pacific1972

macrumors regular
May 2, 2020
128
104
Germany
Bonsoir,

I tried to read everything after translation and what I read was that nobody has ever run an Ivy Bridge on a 2011 iMac, and then the GeekBench with an iMac 11,2 bothers me a lot (https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=iMac11,2+ivy)... :confused:

No one has run an Ivy Bridge on a 2011 iMac, until proven otherwise. 😓

@Ausdauersportler there's no database of intel machine firmware (I don't know about older Macs) simply because each firmware contains the machine's info, unique serial number among other things. ;)

Otherwise for those who can with a 21.5" I chewed through the work, such a test is not going to kill your machine, unless you're not as rough as I was. 😟

I stand by what I said, I would test my modification which seems right to me, but for the moment not the financial means, that's all. 😅

I am a little tired of question after question, it would be necessary to make an effort of reading and comprehension, if you are not a tinkerer in the soul go your way. :mad:

I remind you that I am not good in English and that despite translations it is never simple. 😓

Sorry for my anger, but it's going around in circles here! 🥺

Edit: I made my sentences and expressions explode well by the translations, sorry if there is something incomprehensible. :eek:



Sorry, I don't understand your anger yet, perhaps this can help ?

:apple: :apple: :apple: this guy should analyse correct to find the right CPU :apple: :apple: :apple:
 

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826

Sorry, I don't understand your anger yet, perhaps this can help ?

:apple: :apple: :apple: this guy should analyse correct to find the right CPU :apple: :apple: :apple:
The Xeon E3-1240 is a Sandy Bridge CPU...nothing really new except that you sacrifice sleep with changing to this CPU. Where is the gain?

@StephN999 has the plan to put an Ivy Bridge CPU in the iMac 2011 - which has not been done so far. And he is losing temper because all people start to share irrelevant content on this thread :)
 
  • Like
Reactions: StephN999

StephN999

macrumors 6502
Apr 12, 2020
288
228
Cergy, France
Sorry, I don't understand your anger yet, perhaps this can help ?
Bonjour,

Because "ça se barre en couilles" as they say in France.
as they say @Ausdauersportler, it's irrelevant, and above all we ask the same question 20 times (sorry for the rudeness). 😅 😅 😅

It would be necessary to ask a moderator to put the first message in writing history to make a mini tutorial of what has been done (Xeon Sandy bridge) and what will be done (Ivy bridge) without but then without guarantee that it works!!! ;)

As a scientist would say it's necessary to prove that it doesn't work, to validate or not a theory. 😊

my regards,
Bonne journée.;)
 
  • Like
Reactions: roscho

roscho

macrumors member
May 17, 2020
59
39
Switzerland
Bonjour,

Because "ça se barre en couilles" as they say in France.
as they say @Ausdauersportler, it's irrelevant, and above all we ask the same question 20 times (sorry for the rudeness). 😅 😅 😅

It would be necessary to ask a moderator to put the first message in writing history to make a mini tutorial of what has been done (Xeon Sandy bridge) and what will be done (Ivy bridge) without but then without guarantee that it works!!! ;)

As a scientist would say it's necessary to prove that it doesn't work, to validate or not a theory. 😊

my regards,
Bonne journée.;)


As we are back on topic, I wonder if it is possible to inject CPU Microcodes via OpenCore / Clover / DSDT / SSDT - has anyone with more hackintosh experience insight in this?

Another option would be implementing LibreBoot / CoreBoot as OpenCore seems to be a fork of the TianoCore payload. SeaBIOS payload seems to support BSD too...

Some work on Apple HW has already been done:

Flashing coreboot on MacBooks without external programmer by using IFD hack

Libreboot Port to MBP 2,1


Best,

roscho
 
Last edited:

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826
It would be necessary to ask a moderator to put the first message in writing history to make a mini tutorial of what has been done (Xeon Sandy bridge) and what will be done (Ivy bridge) without but then without guarantee that it works!!!
So, you have your wiki at the first post. Go ahead and create a documentation there...
 

StephN999

macrumors 6502
Apr 12, 2020
288
228
Cergy, France
Bonjour,

@roscho, I don't think so unfortunately.😓

@Ausdauersportler, Thank you! but I wouldn't do it directly because of my low level of English, I would submit it to anyone for correction when I write it unless someone else takes care of it first. 😊;)

Bests regards,

Stéphane. 😉
 

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826
Bonjour,

@roscho, I don't think so unfortunately.😓

@Ausdauersportler, Thank you! but I wouldn't do it directly because of my low level of English, I would submit it to anyone for correction when I write it unless someone else takes care of it first. 😊;)

Bests regards,

Stéphane. 😉
Start with it and let others do the language adjustments, later. We do it in the same way on the GPU thread, once in a while a native speaker passes by and silently we got the corrections...the matter expert has to deliver the content.
 
  • Like
Reactions: StephN999

StephN999

macrumors 6502
Apr 12, 2020
288
228
Cergy, France
All right, I'll try it. o_O

I'm thinking about starting next week. 😊

I'll be cleaning up my posts by the way.
I would also put forward the videos of the members who made them.
 
  • Like
Reactions: roscho

Ed Hartouni

macrumors member
Jul 18, 2020
53
56
recently swapped the i5-2400S with a i7-2600S on my mid2011 21.5" iMac running Mavericks 10.9.5

The system performed on GeekBench as expected for the two cpus, but the "real life" performance was a disappointment. In the four Mac systems I have access to, early 2008 iMac with 2 duo, the mid 2011 with the i5 and i7, and a much newer MacBookPro with an i7, the cpu performance basically followed the thermal load profiles. The early 2008 was twice as slow, but the three other systems basically had identical "total time to complete" when I ran multiple applications simultaneously. The application was a Geant4 Monte Carlo (example B1) for 1,000,000 events.

I started with 1 process, and then simultaneously started 2, 3, etc processes. Naive expectation is that the total time to complete (the sum of the time for all processes) would scale linearly with the number of total threads. Interestingly, it does not.

computer performance.jpg


the dashed lines are the linear expectation drawn from the 1 process result and (0,0).

What is interesting is that the iMac and the MacBookPro follow the same curves, linear for the one and two processes, but then start to climb up in time. I speculate that the Apple cooling "philosophy" is at work here.

For this application there is no gain going from the i5 to the i7.

So while it is possible to upgrade the hardware, you might want to look at what is limiting your computing, and if possible, benchmark you apps on the cpu you want to upgrade to.

I understand that it is quaint to use a computer to compute... but that's just me.
 
Last edited:

Ausdauersportler

macrumors 603
Nov 25, 2019
5,007
5,826
recently swapped the i5-2400S with a i7-2600S on my mid2011 21.5" iMac running Mavericks 10.9.5

The system performed on GeekBench as expected for the two cpus, but the "real life" performance was a disappointment. In the four Mac systems I have access to, early 2008 iMac with 2 duo, the mid 2011 with the i5 and i7, and a much newer MacBookPro with an i7, the cpu performance basically followed the thermal load profiles. The early 2008 was twice as slow, but the three other systems basically had identical "total time to complete" when I ran multiple applications simultaneously. The application was a Geant4 Monte Carlo (example B1) for 1,000,000 events.

I started with 1 process, and then simultaneously started 2, 3, etc processes. Naive expectation is that the total time to complete (the sum of the time for all processes) would scale linearly with the number of total threads. Interestingly, it does not.

View attachment 943065

the dashed lines is the linear expectation.

What is interesting is that the iMac and the MacBookPro follow the same curves, linear for the one and two processes, but then start to climb up in time. I speculate that the Apple cooling "philosophy" is at work here.

For this application there is no gain going from the i5 to the i7.

So while it is possible to upgrade the hardware, you might want to look at what is limiting your computing, and if possible, benchmark you apps on the cpu you want to upgrade to.

I understand that it is quaint to use a computer to compute... but that's just me.
This is what I am talking about for a long time now.

The Sandy Bridge is/was a great CPU step forward, but the HT implementation was not the best. Effectively you have in most cases exactly the four cores of the CPU to do the compute work and see no or only minor time savings. And if you enter ark.intel.com and compare the max. boost frequency the difference become irrelevant. All the CPU came out of the same fab and where labeled due to market demand or in rare cases quality tests.

It might be interesting to compare the same compute tasks with disabled or enabled HT and watch the power budget of the CPU at the same time, it might be more power efficient to use HT. But I do not have data points. There has been a longer discussion about that in 2011 when the CPU was brand new and was used for the famous Seti@Home project my millions od private users...

What you measured here can show - of course - another different bottle neck of the system doing simulations: At one point in time data has to be read from or written to disk, which is a shared resource, as all the different internal communication lines..
 

Ed Hartouni

macrumors member
Jul 18, 2020
53
56
What you measured here can show - of course - another different bottle neck of the system doing simulations: At one point in time data has to be read from or written to disk, which is a shared resource, as all the different internal communication lines..

I was aware that other resources may produce a bottle neck in the computing, that was the reason to include the MacBookPro, which is mid 2017, has an i7-7660U "Kaby Lake" 7th Gen running base at 2.5 GHz.

The memory on the MacBookPro is LPDDR3, while on the 2011 iMac it is DDR2, both systems max out the memory and the process requirements do not indicate that memory resource is limited when running.

The iMac has a SATA bus to an SSD, I'm not sure what bus is on the MacBookPro, but I assume it is faster if only because of the integration and the updated chip set.

The process I'm running is pure number crunching, with some output at the end.

I had noticed this when running some pre-production simulations on the MacBookPro. I had setup a problem with Geant and did successively longer runs to bench the total time. Then launched 4 very long jobs for a simulation. It took twice as long to complete (24 hours later) than I expected. The plot above shows why. I was also at the mercy of Apple's thermal management and, thankfully, didn't destroy the MacBookPro (I generally support it off the table to improve cooling when I run these sorts of jobs).

I am very surprised that the performance of the 2011 iMacs is the same as the 2017 MacBookPro, but that is what you might expect if the thermal management is the same for the two.

The HPC I have access to has cores that are somewhat slower than any of the cores I have for my personal computing (exception, they are faster than the Core 2 Duo, but not a lot), but there are a lot of cores, AND the thermal management is much much better, so they scale linearly for the sorts of production runs I do. Of course, half the power requirement for the computer center is in cooling.

There are several cautions here, one is the industry hype for "improved" cpus, one is the blind interpretation of "standard" benchmarks (what is GeekBench and why does it matter?), and finally what performance do you really want to get out of your desktop?

To the last question, gaming is certainly a legitimate answer, and so it would be interesting to see a comparison of some "real life" performance vs. video performance benchmark numbers. [My main contention is that computing hardware will stop improving once it saturates total human I/O bandwidth, scientific computing benefits from these improvements, but is not driving them].

While not as sexy, it would be really nice if Apple had models with improved thermal management. Sadly, until then the potential of the cpus (and gpus) will not be fully realized.

Maybe this is a direction for 2011 iMac modifications?
 
Last edited:

roscho

macrumors member
May 17, 2020
59
39
Switzerland
I was aware that other resources may produce a bottle neck in the computing, that was the reason to include the MacBookPro, which is mid 2017, has an i7-7660U "Kaby Lake" 7th Gen running base at 2.5 GHz.

The memory on the MacBookPro is LPDDR3, while on the 2011 iMac it is DDR2, both systems max out the memory and the process requirements do not indicate that memory resource is limited when running.

The iMac has a SATA bus to an SSD, I'm not sure what bus is on the MacBookPro, but I assume it is faster if only because of the integration and the updated chip set.

The process I'm running is pure number crunching, with some output at the end.

I had noticed this when running some pre-production simulations on the MacBookPro. I had setup a problem with Geant and did successively longer runs to bench the total time. Then launched 4 very long jobs for a simulation. It took twice as long to complete (24 hours later) than I expected. The plot above shows why. I was also at the mercy of Apple's thermal management and, thankfully, didn't destroy the MacBookPro (I generally support it off the table to improve cooling when I run these sorts of jobs).

I am very surprised that the performance of the 2011 iMacs is the same as the 2017 MacBookPro, but that is what you might expect if the thermal management is the same for the two.

The HPC I have access to has cores that are somewhat slower than any of the cores I have for my personal computing (exception, they are faster than the Core 2 Duo, but not a lot), but there are a lot of cores, AND the thermal management is much much better, so they scale linearly for the sorts of production runs I do. Of course, half the power requirement for the computer center is in cooling.

There are several cautions here, one is the industry hype for "improved" cpus, one is the blind interpretation of "standard" benchmarks (what is GeekBench and why does it matter?), and finally what performance do you really want to get out of your desktop?

To the last question, gaming is certainly a legitimate answer, and so it would be interesting to see a comparison of some "real life" performance vs. video performance benchmark numbers. [My main contention is that computing hardware will stop improving once it saturates total human I/O bandwidth, scientific computing benefits from these improvements, but is not driving them].

While not as sexy, it would be really nice if Apple had models with improved thermal management. Sadly, until then the potential of the cpus (and gpus) will not be fully realized.

Maybe this is a direction for 2011 iMac modifications?
Maybe you just want to try some sort of „mining rig“ computing via multiple GPUs under Linux. My impression is that gpu’s surpassed cpus when it comes to number crunching.

By the way, when I used an ePU solution via an old TB1 enclosure and either a cheap GTX 1060 or a RX 570 it gave me a GeekBench result of about 100‘000. Vega 56 Nano even higher. Do maybe that might even be the way to go if you look for maximum computing performance with a 2011 imac - as long as your App supports GPU usage (what even Adobe SW does)
 

Ed Hartouni

macrumors member
Jul 18, 2020
53
56
Maybe you just want to try some sort of „mining rig“ computing via multiple GPUs under Linux. My impression is that gpu’s surpassed cpus when it comes to number crunching.

By the way, when I used an ePU solution via an old TB1 enclosure and either a cheap GTX 1060 or a RX 570 it gave me a GeekBench result of about 100‘000. Vega 56 Nano even higher. Do maybe that might even be the way to go if you look for maximum computing performance with a 2011 imac - as long as your App supports GPU usage (what even Adobe SW does)

There is development in Geant to incorporate GPU resources to accelerate the computations. AFAIK, this has been complicated by the heterogeneous implementation of GPU code... right now OpenCL, Metal, Cuda, ... so building a Geant distro that can take advantage of the local hardware configuration starts to be overwhelming.

But I was interested in the Geant software development as a "hobby" (my day job doesn't pay me to do that) so I'm upgrading the GPU on the 2011 iMac to an Nvidia Quadro (which has better floating point performance than the GeForce).

I suspect that the ultimate limitation of using the GPU will be thermal management.
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.