Re: iNav causing hard reset - SOLVED -- PCTECH

Buck wrote:

Bad is a funny word. No, one is not likely at all to damage the chip by bumping the frequency, but internal timing (cache that doesn't reliably answer with a shortened cycle time, etc.) can cause the software to not run reliably. It's the whole system that we're talking about, really.

In that case you're talking about not upgrading any hardware then, and sticking with OEM machines. And that's a fine option if you have the money to do that. There are lots of companies who buy their Dells or whatever and never upgrade the machine.

I was talking about the companies where it's acceptable to put a faster disk drive in a machine, or add memory. If your IT department is capable of handling those types of changes, they're probably capable of handling overclocking responsibly.

The other issue is the user, of course. Most 5250/browser/Office users don't need a lot of horsepower, so they really don't need overclocking. Overclocking makes the most sense for power users - people who use graphics programs or RDi-SOA and the WebSphere Test Environment. Most of those folks would probably be willing to try out a new configuration to see if it was stable in order to gain a 20, 30, even 50 percent throughput gain.

There aren't many of us left who write x86 assembly code anymore, and a darned sight fewer who own an oscilloscope that runs at a high enough speed to watch the signals on the pins. My Tek 475 is marginal, at best. :-) The point is that while the CPU might overclock in isolation happily enough, the rest of the system as a whole might not. Most PC users are interested in running some application, and if they can't until they dial the frequency back down then we can reasonably say that the system has failed at the overclocked speed.

Not damaged, but not operational, either.

Personally, I've never seen situations where timing-related issues didn't come up immediately, or at least as soon as a specific piece of hardware or software was used. CPUs don't "wear out" or "degrade"; they either run at a given speed or don't. If errors occur, even if they're literally one in a million, they will show up quickly. Since modern CPUs run billions of CPU cycles a second, "random" timing issues manifest themselves immediately, typically at boot. A good motherboard will detect a CPU timing issue and reboot to a safe speed.

I don't know, maybe you've run into flakiness with overclocking - some sort of error that only occurs once in a trillion times, and causes a catastrophic failure in an application. I'd be interested in the specifics of the CPU, motherboard, RAM and bus frequencies you were running. But if your position is that overclocking is "bad" because it might cause intermittent timing issues, then we have to agree to disagree and move on.

Joe