Intel iMac CPU Swappable?

[apple4life]

Not interesting.... not

This is great news and will drive another thorn in the side of the dark side...

...Apples will have so much more to offer in terms of one up-manship that is so prevalent in the modification of "dull brown little boxes"... the idea of a mac as an email machine for grandma or the novice will begin to change as technophiles begin to take interest and will lead to providing a growth in products and services created by the technically savvy for the growing user base of macs.... who can now one-up one another.... ahh, ego is the downfall of many but it provides such a large customer base...😀

 

[Bassman59]

clock multiplier

This is not true. The internal clock frequency is generated by a multiplier of FSB frequency. The 2 Ghz chips are simply flashed with a different multiplier value then the 1.83 Ghz chip. Notice that they all are using the same FSB speed. You will only have to change the fimware to modify the FSB frequency.

Ummm, I think that's what I said ("may have to hack the firmware").

One more time: if you just swap out Chip A for Chip B, and not change anything else, then Chip B will run at the same speed as Chip A, even though it is capable of running faster.

You will have to tell the motherboard's clock-generation circuitry to run at a higher frequency if you wish to take advantage of Chip B's higher speed rating. I don't know whether this control is in the firmware or hardware jumpers on the motherboard or what.

 

[jhu]

What I usually do with my home-built systems (and sometimes the Dells and HPs) is to:

o By the system early with a lower to mid-range CPU (when the chips first come out, the top speeds are quite expensive)

o At the end-of-life for the chip line, buy the fastest one available (which is probably cheaper than you paid for the mid-range one a year earlier)

I did that with

o 266 MHz PII (66 MHz bus)
-> 533 MHz Celeron (fastest chip that supported the 66MHz bus)

o 1.5 GHz P4 (400 MHz RDRAM)
-> 2.6 GHz P4 (fastest chip to support 400 MHz bus)

you somehow failed with your p2 since the fastest chip that officially supported 66 mhz was the 766 mhz celeron. although you may have still been able to put a 1.4 ghz celeron in there, but it would be running at 924mhz instead.

 

[katieville]

i dont get it

 

[neonart]

Ummm, I think that's what I said ("may have to hack the firmware").

One more time: if you just swap out Chip A for Chip B, and not change anything else, then Chip B will run at the same speed as Chip A, even though it is capable of running faster.

You will have to tell the motherboard's clock-generation circuitry to run at a higher frequency if you wish to take advantage of Chip B's higher speed rating. I don't know whether this control is in the firmware or hardware jumpers on the motherboard or what.

Ah, no. If you have a 1.83Ghz iMac and you put a 2.0Ghz chip your iMac will run at 2.0Ghz. That's it. It will not automatically clock down the faster chip just to tick you off.

Only on certain upgrades, where you are increasing the FSB or the MB is not set to recognize a faster speed will you need to modify anything.

I have done this many times on different platforms. P4 1.8Ghz-2.24Ghz, P3 667Mhz-1.2Ghz(oc'd 1Ghz), Athlon 1.47Ghz to 1.66Ghz, G4 400Mhz to Dual 500 then Dual 800 then Dual 933. (Don't ask, expensive experiment). iMac G3 333Mhz to 466Mhz. Not once did I enter anything in the bios or open firmware to have the faster processor recognized.

I truly do not understand what you mean.

 

[bleachthru]

It is about time!

I am/have been running my trusty Pismo for a long time. The thing that I love so much about its design is that it is so upgradable; swappable processor, all internals are easy to access, hot swappable drives, etc. One of the main reasons I haven't upgraded in past years, aside form cost, is that I can still upgrade my current laptop and it is six years old, and going strong. If the MacBookPro is bringing back this sort of upgradability I might be inclined to switch over...Once the MBP's can beat the spead of my Pismo's slot-loading 8x DL SuperDrive 🙂

 

[EricNau]

I think someone posted pictures comparing the BACK of the iMac G5 internals to the FRONT of the Core Duo internals (or vice versa), leading to false reports that the Core Duo was vastly "less-organized-looking" inside. But in fact if you compare the SAME side of both machines, then the Core Duo Macs and recent G5 Macs look very similar inside.

Now, the ease of access of the FIRST (thicker) G5 iMacs was indeed great, and it was sad to see that go last year. They even had indicator lights to help users self-diagnose problems. I guess other priorities pushed that strategy aside.



As long as it doesn't make my MacBook thicker... yes please!

I thought the only way to get into the intel iMac was through the front (compared to previous versions which opened through the back).

I'm not sure, that whole thread was full of different guesses.

 

[reemas]

The cost of such a minor upgrade ($480 for a 2GHz chip) doesn't make sense at this time but does offer more of a upgrade path than previously available on the iMac.

MORE??? Since when have we been able to upgrade iMacs? 😀

 

[MrCrowbar]

I like the way you uprade RAM in the newer iMacs. Just unscrew the plate at the bottom, change the RAM modules, screw on the plate, done. It would be awesomely nice if apple let us change the CPU in the same manner. 😱

There might be a problem with the heatpipes that are usually glued to the processor. Solutions for that are either to enable the heat pipes to easily go off the CPU or to ship the CPU with glued on HeatPipes. Latter would require a bigger piece of case to unscrew.
The PowerMacs though would be easier to handle. I imagine pulling out these silver blocks (with the big G5 logo actually on it) and popping in a faster CPU-Module (silver Block) with the cooling system inside. This way you could really upgrade and have water cooling (bigger silver Block and pipes coming out) if required. I'd like to see a CPU-Block exchange program for PowerMacs. Kind of going into an apple store and get the blocks exchanged. 😀

But honestly, Macs are made for people that don't want to learn about computers, so I think the consumer line won't be upgradable in terms of CPU. I've studied Computer Science, still I love Macs because I don't spend half of the time keeping the mashine alive.

 

[mrichmon]

I thought the only way to get into the intel iMac was through the front (compared to previous versions which opened through the back).

I'm not sure, that whole thread was full of different guesses.

Revision A and B of the G5 iMacs opened through the back and were fairly easy to gain access to the common upgradable components: hard drive and optical drive.

Revision C of the G5 iMac opened through the front making access to the hard drive and optical drive more complex.

Revision A of the Intel Core Duo also opens through the front in a similar manner to revision C of the G5 iMac.

 

[mrichmon]

There might be a problem with the heatpipes that are usually glued to the processor. Solutions for that are either to enable the heat pipes to easily go of the CPU or to ship the CPU with glued on HeatPipes.

The heat pipes are attached to the CPU using thermal paste. This paste is sticky, but does not set like glue does. It is common to clean off this paste and apply new paste when changing cpu heat sinks or doing major disassembly of computers.

 

[neonart]

MORE??? Since when have we been able to upgrade iMacs? 😀

The very first 233(Rev a-b),266(c),and 333(d)Mhz models have the CPU on a daughter card like the Black powerbooks. They're upgradeable to faster G3's and G4's.

 

[Fiveos22]

I heard that the Yonah will be pin-compatible with the Merom - this could aso be huge then. 😱

The reason I am waiting for the next revision is to get ahold of a Merom chip. Even if they are pin compatible (yonah and merom) won't the initial lack of fast interconnect technology be revised by the time Merom is released. I don't think that Intel is so gutsy as to rely on antiquated bus architecture for an entire year. I bet (hope) they will have some CSI implementation when Merom is released, thus making a chip-swap between yonah and merom much less beneficial. Personally I am hoping that the intel northbridge will disappear similar to AMD's at which point I will buy a new power book.

Either that or I am wrong.

Core Solo and Core Duo both use the 945 chipset from intel. When Merom is introduced this summer, it will initially be released on the 945 chipset as well. The replacement chipset supporting 1033 MHz FSB won't be released until next year. (Note: I'm posting from memory, it might only be 800 MHz on the mobile core, I know the desktop revision will support 1033 and 1333 MHz)

Deimo seems to think the latter.

 

[MrCrowbar]

The heat pipes are attached to the CPU using thermal paste. This paste is sticky, but does not set like glue does. It is common to clean off this paste and apply new paste when changing cpu heat sinks or doing major disassembly of computers.

Well, I once tried to recycle the cooling of an Athlon (attached with silver paste) and it came with a bit of the processor case. So I think you can call it some kind of glue. I had to use brute force (wrench, hammer...) to get the thing off. The old CPU was pretty worthless anyway (when I had the new CPU, I overclocked he old one to see when it would fry 🙄 . The 800MHz Athlon died after app. 5 seconds on 1600 MHz!) so it didn't matter too much.

I'd really like to upgrade CPUs the way you upgrade RAM on Macs.

 

[reyesmac]

This is really encouraging. It might be tough to upgrade future iMacs though, I think they will make the case so thin that it will be hard to put a hotter chip in them. But it is good to see them test the waters with this and the monitor spanning capability.

 

[splintah]

upgradeable cpu in the macbook pro would be so great

also interchangeable graphics cards would be nice
with pciExpress that shouldnt be a problem🙄

 

[danielwsmithee]

Clocks

Ummm, I think that's what I said ("may have to hack the firmware").

One more time: if you just swap out Chip A for Chip B, and not change anything else, then Chip B will run at the same speed as Chip A, even though it is capable of running faster.

You will have to tell the motherboard's clock-generation circuitry to run at a higher frequency if you wish to take advantage of Chip B's higher speed rating. I don't know whether this control is in the firmware or hardware jumpers on the motherboard or what.

You just don't get it. For simplicity sake, there is a only one clock generated on the motherboard, that is the FSB clock. That clock is what drives the internal clock in the processor for the Core Duo it is a 667 Mhz clock.

Internally to the processor there is what is called a clock multiplier that takes the 667 Mhz clock and creates a 2.0 Ghz or 1.83 Ghz clock from the 667 Mhz clock. So there is nothing in the firmware that needs to be changed, the clock on the mother board is still operating at the same frequency.

When the Chip is designed there is a register that can be "permanently electrically programmed" that represents the multiplier value. They test the chips at different frequencies determine the highest value of this multiplier for which it will pass the tests then program it to that value. All the chips are built the same they're clock rate is just determined by how well they test.

That is why you see a relationship between FSB frequency and CPU frequency. Examples:

Dual-core 2GHz
667MHz frontside bus per processor = multiplier of 3

Dual-core 1.83Ghz
667MHz frontside bus per processor = multiplier of 2.75

dual-core 2.5GHz PowerPC G5 processors
1.25GHz frontside bus per processor = multiplier of 2

dual-core 2GHz PowerPC G5 processors
1GHz frontside bus per processor = multiplier of 2

If I remember correctly the single G5 Powermacs were
1.8 Ghz G5 Processor
600 Mhz FSB = multiplier of 3

AMD's became very popular because they shipped many of their processors with the ability to overclock by changing this multiplier register. You could go into your firmware and change both the FSB frequency and the internal core frequency (or multiplier value). That is where most of your confusion comes from. Most processors to over-clock them you have to increase your FSB frequency, which means you are also stressing your chipsets and RAM more.

 

[mrichmon]

Well, I once tried to recycle the cooling of an Athlon (attached with silver paste) and it came with a bit of the processor case. So I think you can call it some kind of glue. I had to use brute force (wrench, hammer...) to get the thing off.

Yes, cheap pastes can do that. Using mechanical force is begging to crack/chip/snap/break the chip. If you need to do this in future dripping some high purity isopropyl alcohol into the space between the heatsink and the chip and leaving it for a few minutes to work into the thermal paste is an easy way to separate the heatsink from the chip.

 

[Maxx Power]

You just don't get it. For simplicity sake, there is a only one clock generated on the motherboard, that is the FSB clock. That clock is what drives the internal clock in the processor for the Core Duo it is a 667 Mhz clock.

Internally to the processor there is what is called a clock multiplier that takes the 667 Mhz clock and creates a 2.0 Ghz or 1.83 Ghz clock from the 667 Mhz clock. So there is nothing in the firmware that needs to be changed, the clock on the mother board is still operating at the same frequency.

When the Chip is designed there is a register that can be "permanently electrically programmed" that represents the multiplier value. They test the chips at different frequencies determine the highest value of this multiplier for which it will pass the tests then program it to that value. All the chips are built the same they're clock rate is just determined by how well they test.

That is why you see a relationship between FSB frequency and CPU frequency. Examples:

Dual-core 2GHz
667MHz frontside bus per processor = multiplier of 3

Dual-core 1.83Ghz
667MHz frontside bus per processor = multiplier of 2.75

dual-core 2.5GHz PowerPC G5 processors
1.25GHz frontside bus per processor = multiplier of 2

dual-core 2GHz PowerPC G5 processors
1GHz frontside bus per processor = multiplier of 2

If I remember correctly the single G5 Powermacs were
1.8 Ghz G5 Processor
600 Mhz FSB = multiplier of 3

AMD's became very popular because they shipped many of their processors with the ability to overclock by changing this multiplier register. You could go into your firmware and change both the FSB frequency and the internal core frequency (or multiplier value). That is where most of your confusion comes from. Most processors to over-clock them you have to increase your FSB frequency, which means you are also stressing your chipsets and RAM more.

Actually, since the P-M borrows the P4's quad pumped bus, which Yonah eventually inherited, the actual base speed of the bus is 667/4 = 166.75 mhz, and hence, the multipliers are what you suggested multiplied by 4 for every case. The PLL runs at 166.75, data is transferred four times over 1 clock, the clock itself doesn't run at 166.75. So for the 2 GHz Yonah, the multiplier is 12, base clock is 167 (rounded up).

 

[MrCrowbar]

Thanks mrichmon. It was a pretty expensive paste but thanks for the hint. I'll try that next time I will have to do that (if I ever have to 🙂 ). Since that little "accidendent" I never had to do it again because I needed to change motherboard and headsink due to changinf sockets. 😡

Has anyone thought of a mainboard upgrade? 😛 Sort of pulling out the insides of a Mac to replace it with a new board with CPU, graphics and cooling system. LOL. You could even buy the upgrade Kit and plug in cheap peripherals (display, keyb, HD, optical drive, RAM...). Would look kinda freaky, but run OSX natively 😀 . Hope we never see THAT happen.😱

 

[Deimo]

One other factor to keep in mind is that Merom is supposed to consume less power than Yonah, this would mean that it will dissipate less heat, and assuming that it is in fact pin compatible, would make for a great replacement in any system that can have its chip upgraded. One estimate that I saw stated that Merom should have a 30% improvement in performance over Yonah in 32-bit applications.

 

[~Shard~]

One other factor to keep in mind is that Merom is supposed to consume less power than Yonah, this would mean that it will dissipate less heat, and assuming that it is in fact pin compatible, would make for a great replacement in any system that can have its chip upgraded. One estimate that I saw stated that Merom should have a 30% improvement in performance over Yonah in 32-bit applications.

Yeah, Merom is going to be the cat's ass. That's what Conroe and Woodcrest are going to be based off as well form my understanding. It's all about Merom. 😎

 

[danielwsmithee]

Actually, since the P-M borrows the P4's quad pumped bus, which Yonah eventually inherited, the actual base speed of the bus is 667/4 = 166.75 mhz, and hence, the multipliers are what you suggested multiplied by 4 for every case. The PLL runs at 166.75, data is transferred four times over 1 clock, the clock itself doesn't run at 166.75. So for the 2 GHz Yonah, the multiplier is 12, base clock is 167 (rounded up).

Good point. I was actually wondering about that a few minutes after I posted.

 

[christian_k]

I think it is a nice advantage, but don't be to entusiastic about it.
You are talking about 3 years+. In my office I am using a PC that was state of the art when it was bought 3 years ago. It has a P4 with 2.4 Ghz and even
a Geforce 6800 (yes ,I do need that in the office).
But can I install a Yonah on that mainboard? No. Can I install a Pentium-D? No. As far as I know I cannot even install a 3 Ghz P4 (or any 64 Bit P4), because it wohld need faster FSB or other sockets and it is not supported by the BIOS. I could install a 2.6 Ghz but that would not be worth the money. If I wanted a faster CPU that does mean a new mainboard, making my memory (PC 2700) and the Gfx Card (AGP!) obsolete.
I think it is better to sell a computer as long as you get a god price for it and go for a new one. Macs used to retain value well , and even more when they are still under AppleCare. You get a "fresh" machine with all the new stuff and new warranty.
There where times when you could use a 100 Mhz Pentium and a K6 with 550 Mhz on the same mainboard, but during the last years CPU sockets, cooling and power requirements and memory specs have changed rapidly making tbig "jumps" without the price nearly impossible.

I think the time of do it yourself upgrading is over, even on PCs (except for some simple cases like installing more RAM / bigger HD which is possible on all Macs).

Christian

The real advantage to this will be seen in a couple of years. When the first Macintel's get old and their AppleCare warranty is done. Pop in a new processor and you have a new machine!! Now if only the GPU were upgradeable. Seriously this is great news for people who like to keep ahold of their machines. Also, this could have the effect of lowering the prices of new macs. Why upgrade to a whole new machine every two or three years when you can pop in a $200 or so new CPU? Apple may need to lower prices of their machines in order to make it attractive to upgrade. Just my .02$

 

[AidenShaw]

you somehow failed with your p2 since the fastest chip that officially supported 66 mhz was the 766 mhz celeron. although you may have still been able to put a 1.4 ghz celeron in there, but it would be running at 924mhz instead.

Let me expand my post:


-> 533 MHz Celeron (fastest chip in a SEC (Slot I) cartridge that supported the 66MHz bus)


You remember the SEC package, right?

😉

The CPU, cache and other components were mounted on circuit board the was about the size and shape of a tall memory DIMM. The socket on the motherboard was long and narrow, like a DIMM socket. The board was enclosed in a plastic case that contained the mounting tabs and attachments for the heat sink and fan.