Is the Mac Pro WAY overpriced?

[nanofrog]

With or without increasing the core voltage?

Voltages were increased, but not substantially so (~1.38V IIRC, as it took 1.46V to get 4.2GHz).

I dropped it back, as temps were a tad high for my liking (idle = 54C, load = 80C @ 4.2GHz). Before the cooler was lapped though. I've not repeated that high a clock after lapping (Noctua NH-U12P w/1366 mount and push-pull fan config).

 

[Transporteur]

That's really a remarkable voltage increase.

Honestly I doubt that your CPU will be able to run at that speed, with such high temps for a long time.
Ever heard of electromigration?

A little anecdote from my overclocking times. Intel Pentelum 4 650, 3.2GHz clocked at 4.2GHz with a very similar voltage increase like yours, watercooled with a maximum temp of about 55°.
It worked absolutely fine for over half a year, till the said electromigration kicked in. I had to reduce the clock speed almost every week, till a few weeks later the CPU didn't even reach it's stock speed.

Such high clock speeds paired with a voltage increase and no decent cooling (which can reduce electromigration, but of course can't stop it), may be stable for quite a long time, but they significantly reduce the durability of the CPU.

 

[nanofrog]

That's really a remarkable voltage increase.

Honestly I doubt that your CPU will be able to run at that speed, with such high temps for a long time.
Ever heard of electromigration?

Yes, I'm well aware of electromigration. That's why I backed it off. The temps are for 4.2GHz, not 4.12. As the summer came on, I dropped it back to stock, as the ambient was too high (room with A/C hits 85F). Too hot for both the system and user.

Good cooling helps to push clocks higher, but they're much more able to maintain a lower OC, say the 3.8GHz-ish range, as the voltage increase isn't that bad at all. Electromigration is mitigated here from the data available, and the temps are acceptable. It's trying to go over that the voltage increases get more substantial. As it cools down (winter), this is likely where I'll set it. It's a really nice performance boost for the cost of the chip.

Such high clock speeds paired with a voltage increase and no decent cooling (which can reduce electromigration, but of course can't stop it), may be stable for quite a long time, but they significantly reduce the durability of the CPU.

I'm thrilled if I can use a processor for 3 years for it's initial task.

 

[Transporteur]

I'm thrilled if I can use a processor for 3 years for it's initial task.

Assuming that it lasts that long, fine.

Personally, I would not go any higher than the CPU can handle with it's stock voltage and with decent temps in the range of 60-70°C (air of course).


Anyway, my overclocking times are long ago, don't have any more PC's in the house besides my server and I definitely won't overclock that thing.

 

[nanofrog]

Assuming that it lasts that long, fine.

Actually, I'm not worried it won't. Keep in mind, the processing has swapped out aluminum for copper due to EM, and there's a greater safety margin now, so it's not as critical as just a couple of years ago. EDA tools have improved for EM analysis as well (I use National Semi's suite).

Personally, I would not go any higher than the CPU can handle with it's stock voltage and with decent temps in the range of 60-70°C (air of course).

I prefer to stay in that range, but I'm still comfortable at 10C over stock temps (67C is listed for the LGA1366 parts per Intel).

Anyway, my overclocking times are long ago, don't have any more PC's in the house besides my server and I definitely won't overclock that thing.

OCing a server does feel risky, but that may change out of necessity. The current parts are actually capable of it. It's the rackmount cases and no HVAC system found in data centers that would make me nervous. But for pedestal systems, limited budgets may end up dictating it. To me, this is going to be an interest of SOHO and SMB, not large scale enterprise environments.

Can't get around the economics...