Intel iMac CPU Transplant Success

[topgunn]

I had an acquaintance send me an engineering sample of a 2.16GHz Merom T7400 CPU. Within an hour of receiving the processor, I had my 20" iMac disassembled, ready to replace the stock T2500.

Disassembling the iMac was not fun at all. Usually I enjoy disassembling my Macs but this was one exception. It is much more difficult to access than any other Mac I have owned which includes pretty much every line over the last 5 years. First the front bezel has to come off which involves removing a few torx screws and then using a credit card bent at a 30 degree angle to release a latch in the upper corners that you access through the vent. Once the cover is off, you have to peel back gobs of EMI tape and the EMI shield. Next remove the LCD. Now you have access to the logic board. Proceed to remove 13 different wired connections to the logic board, many of which are tiny and frail and quite the little boogers to get out. Remove the logic board (of course they had to put the CPU on the back side) and the rather large heatsink/heatpipe contraption which is shared between the GPU and CPU. Finally, you can remove the old CPU and replace it with your new 64-bit capable, 4MB L2 cache having Merom.

When it came time to reassemble everything, I was torn between rushing through it to finally hear that reassuring startup "bong" and taking my time to make sure everything was correctly put back together. I only missed one thing and that was reconnecting the Airport antenna but it was easily accessible under the EMI shield. So with everything put back together with only the front bezel still off, I start up the machine. Oh, that beautiful bong. It is music to my ears.

Since it was so late, I just ran a quick Xbench test to see the difference between the T2500 and T7400. The results for the T2500 can be found here and the T7400 can be found here. As you can see, the CPU test was 45% faster with the new CPU but this is just one test and Xbench is generally very limited. If any of you have any tests that you would like me to run, post them here and I try and run them in a timely fashion (assuming I have the software, of course).

I did have grand intentions of taking photos along the way but that fell by the wayside once I got into it. The pictures would have been largely unremarkable anyway as there are a number of pictures of Intel iMac guts already posted on the internet. One thing of note, the earliest pictures to be released of a disassembled Intel iMac on KODAWARISAN were completely different from mine, especially the layout of the cooling system. I thought it was a bit unusual.

Lastly, as a preemptive strike to all of you who have to ask why, I respond with "Because it's there."

[MIDI_EVIL]

Lies!

There is absolutely nothing to prove that you did this.

We need more than your word. I can't believe you didn't take photos.

Lies.

Rich.

[grabberslasher]

I was thinking of upgrading my 17" baseline iMac to the top of the range Merom when they come down in price, but after reading all that has to be done to disassemble I'm having second thoughts.

We'll see - I'd like to hear how gaming performance is improved in Quake 4 and Doom 3 at max settings and resolution. Call of Duty 2 if you can get hold of it would be a nice benchmark.

[Chasealicious]

how about a screen shot of the "about this mac" window and apple system profiler?

[emotion]

So does the machine support the power saving features of the Merom properly. There was a test of a similarly upgraded pc laptop posted on here the other day. The power consumption was worse becaus the mainboard chipset didn't fully support the new chip.

Any power consumption figures?

[mkrishnan]

I guess it's too late for you to have done an XBench of your system running 10.4.7 before the chip swap?

[Sutekidane]

I wonder how well the cooling can keep up with the newer processor.

[emotion]

Originally Posted by Sutekidane

I wonder how well the cooling can keep up with the newer processor.

It should actually be cooler, but only if the supporting chipset works correctly.

[topgunn]

Originally Posted by fatsoforgotso

Lies!

There is absolutely nothing to prove that you did this.

I didn't do this for you. I just thought I would share my experience.

Originally Posted by grabberslasher

We'll see - I'd like to hear how gaming performance is improved in Quake 4 and Doom 3 at max settings and resolution. Call of Duty 2 if you can get hold of it would be a nice benchmark.

The only games I have on my iMac are Half-life 2 (in Windows of course) and RTS games. If you would like me to benchmark Half-Life 2, let me know which settings and I will do a lost coast demo.

Originally Posted by Chasealicious

how about a screen shot of the "about this mac" window and apple system profiler?

I will do that when I get home from work although it doesn't really tell you much. I can also take a screen shot of CPU-Z in Windows.

Originally Posted by emotion

So does the machine support the power saving features of the Merom properly. There was a test of a similarly upgraded pc laptop posted on here the other day. The power consumption was worse becaus the mainboard chipset didn't fully support the new chip.

Any power consumption figures?

The Speed Step (I am not sure what Intel calls it now) aspect of the CPU works. When the CPU is not being used, the clock drops to 1GHz. I have a kill-a-watt that I can check it out with to see what the system is drawing. I will do that tonight. I do have numbers preconversion to compare it to as well.

Originally Posted by mkrishnan

I guess it's too late for you to have done an XBench of your system running 10.4.7 before the chip swap?

I did not think about running another Xbench test before the switch although in hindsight it would have been nice. I may break the machine down again for pictures since I am more comfortable with it. If I do, I may swap the CPU out again to do another Xbench test but I doubt that will happen.

[bowens]

That's awesome. I would love to have a faster processor in my mini.

[emotion]

FWIW I think this is pretty cool.

[topgunn]

Originally Posted by Sutekidane

I wonder how well the cooling can keep up with the newer processor.

At first, it looked like the CPU was running much warmer. I was idling in the low 40's and load was getting into the mid 60's. This was on the second boot, however. Ceramique, the thermal compound I used, needs to go through a number of thermal cycles before it reaches its max conductivity (something about the crystals aligning). Well, after a few thermal cycles, I am now back to idling in the low 30's and at load in the mid 50's. These are the same results as the Yonah. My thought was since the Meroms have a much larger core (physically) that it should cool better. I don't know if there is any truth to that, though.

[topgunn]

Originally Posted by bowens

That's awesome. I would love to have a faster processor in my mini.

I will be selling the T2500 from my iMac soon if you are interested. I was going to post it in the Marketplace before moving it to eBay.

[risc]

Very nice!

[emotion]

Originally Posted by topgunn

I will be selling the T2500 from my iMac soon if you are interested. I was going to post it in the Marketplace before moving it to eBay.

An ace upgrade particularly for the Core Solo Minis

[cruxed]

Good job! Now I've gotta pass this on to my friend who thinks his "Yonah" iMac is "outdated"

[kgarner]

This really bodes well for the future of my mac purchases. I really hope they keep this ability in all future models as I would love to buy a reasonably powerful machine and then in about three years upgrade the processor to whatever the machine can safely handle to keep it usable for another 3 years or so.

Anyone have any idea if you can upgrade a Core Solo Mini to a Core 2 Duo? Just wondering if there are some MB or ROM restrictions in there.

[JGruber]

Yes you can swap out a Core Solo for a Core Duo... its been done a number of times on Xtremesystems.org

[thestaton]

awesome report! Keep us updated I'm looking at doing this same thing very soon. My iMac is pretty much instant with everything now I could not imagine adding even more power.

To the people asking about power consumption? What does it matter? Its hooked up to your wall.

[JGruber]

Well Power Consumption = more heat being generated = degraded CPU performance over time...

[emotion]

Originally Posted by thestaton

awesome report! Keep us updated I'm looking at doing this same thing very soon. My iMac is pretty much instant with everything now I could not imagine adding even more power.

To the people asking about power consumption? What does it matter? Its hooked up to your wall.

I'm more interesting to see if it is actually properly supported by the existing board.

Anyway, it's always sensible to have these things operating within normal operating temp range (for speedstep reasons etc).

Also where do you think that power in the wall comes from? Nuclear? Coal? Wind? (unlikely over there). Lower power consumption is _always_ good.

[crees!]

So did you update your system between XBench tests? The merom test has 10.4.7 and before has 10.4.4. Was the previous one taken at an earlier date?

[JGruber]

If you notice as well, the 2 different tests are using 2 different harddrives as well

[topgunn]

Originally Posted by crees!

So did you update your system between XBench tests? The merom test has 10.4.7 and before has 10.4.4. Was the previous one taken at an earlier date?

Yes, I generally run and submit Xbench results when I get a new personal system for my reference when I get a new machine down the road. This one was run in February.

[thestaton]

people calm down. he's already stated its running 30c under normal operating conditions & 50c when being put under a load. i'm sorry but the very mild at most power consumption differences is not going to mean anything. if something this small bothers you I'll go hug a tree for you before it gets cut down.

I'm curious to know as well how it mated up to the mobo, and if the stock heat sink was used. can't wait to this mod myself, any idea how much these processors are going to run?