I have two slightly different DELL 670 machines. A machine has one Intel Xeon 3.4GHz 800MHz 2MB L2 Cache CPU. B machine has two Intel Xeon 3.0GHz 800MHz 1MB L2 Cache CPUs. I plan to use machine A and take one CPU from machine B to install in machine A. That means in machine A these two CPUs will be slightly different. Is this OK? thanks,
Depending on your useage pattern, each of the two computers can be faster than the other, casting doubt at the motivation to even try.
And these are 5 year old CPUs - are you sure you want to invest any work in these machines? If a machine of this age runs well, it is best to leave it alone, and let it work. Your chances of the CPU-hack working beneficially are zero (see below), while your chances of breaking one of the servers aren´t.
Now just for fun, lets look at the facts and chances...
From the incomplete information you gave, I guess it is these CPUs:
2x "Potomac" Xeon MP 3.0 SL8EW (C0) 3000 MHz 1024 KiB 667 MT/s 18x Socket 604
1x "Tulsa" Xeon 7140M SL9HA (B0) 3400 MHz 2 × 1024 KiB 800 MT/s 17x Socket 604
- Provided the Tulsa is able to run with a second CPU, which I guess it is.
- They both fit the same socket.
- The Tulsa alone would run in the System that now holds the two Potomacs.
- If the Tulsa is paired with a Potomac, the Tulsa will be throttled to 167 FSB (RAM side), being 17x making it 2.9GHz - actually slower than the Potomac
Another explanation: Even if the Tulsa worked just fine, and even at 3.4 GHz (which it won't), you would gain like 15% of CPU power on half your threats. RAM would be the same speed. Disks would be the same speed. So only CPU-intensive tasks, only if they run on the Tulsa, and you would have to fight your OS to make it so, you might in theory get a boost in performance. Looking at how well most threats scale, I would expect a theoretical gain of 2-3% - less than the measuring precision. But you wont, because of the frequency mismatch.
Therefore, even if it works, and even if it is stable, and even if you get the threats to use the correct core, and if this misconfiguration actually suits your useage, it will be slower.
Apart from that, I doubt it would boot at all (BIOS), but I am not sure of that. I am however sure, it isn´t worth trying except for a challenge.
Mismatched CPU's is not going to work and makes no sense whatsoever anyway. Just see the other answers why.
I'm guessing the single CPU machine is currently somewhat underpowered and the dual-core is over-powered. It's a much better bet to swap ALL cpu's between the systems. Still... It's risky and not for the faint of heart.
- Before you begin make sure both systems have the latest Bios to make sure you give them maximum chance to recognize the other CPU's. Also make sure that the motherboard drivers (Intel chipset drivers) are the very latest.
- Make bloody sure you have good backups as there is a big chance that either, or both, won't boot and/or break the Windows installation.
- After the swap, before booting into Windows, go into the Bios and reset everything to factory defaults. Save and then power-cycle (cold-boot), just a reboot is not enough. Go back into the Bios again and configure as required. Make sure that the dual-cpu system has SMP enabled in Bios. Save and cold-boot again.
- Only then allow the system to attempt to boot Windows. After first boot Windows will probably install new drivers (because of the changes). Let it do it's thing and make bloody sure you reboot every time it asks you to. When all is done reboot once more for good measure.
- On the dual machine you may have to manually update the driver for the "COMPUTER\ACPI PC" from single CPU to multi-CPU PC. (You can tell if this is necessary if task-manager only shows half the CPU-cores.)
If you have just the slightest doubt that you are able to do all this: DON'T EVEN TRY