The aftermath of a failed upgrade
When, several months ago, I decided to upgrade my laptop with a more performing processor, I never thought that some time later I would have been forced to go back on my steps: the system was going well, everything worked as it should with no problems of sorts. And yet the idyl with the X9100 CPU has lasted just three months, whereupon the PC has started to misbehave forcing me to put hands on several internal components, to waste time and precious money in useless purchases before I was persuaded that in the end the upgrade had been a failure.
At first, I was saying, the CPU behaved remarkably well but soon I’ve experienced what is the most clear sign of the fact that something is wrong with the hardware: the PC has started to shut down unexpectedly during particularly intensive computing sessions – with video games and 3D emulators, above all. And if the PC didn’t shut down, the performance of the aforementioned video games and emulators was substantially reduced. At this point I’ve attepted to rationally evaluate the situation, surveying any possible culprit and going on by exclusion.
The first suspect of my survey was the integrated cooling system for CPU, GPU and motherboard chipset, in particular the fan devoted to eject the heat generated by the aforementioned components. Before the upgrade the fan worked in a different manner, I thought, and maybe the new processor forced it to an overwork that in the end damaged it in some way. So I’ve purchased an original replacement from my usual digital stuff dealer (eBay), I’ve pulled apart the cooling system, I’ve reassembled everything applying a new layer of thermal paste and I’ve crossed my fingers.
Results: zero, the PC was still shutting down discouraging me and preventing the most intensive tasks of my computer tool – video games, but the “trivial” transcoding of a DivX video file in DVD format too. Ruled out the cooling system, my suspects focused on the memory chips: after all the unusually high temperatures (100° and even beyond!) recorded by the sensors were related to the motherboard, and maybe the two DDR3 memory banks with 2 GB each didn’t keep up with the higher frequencies the X9100 CPU forced on the front-side bus (FSB) for the communication with the Northbridge (1066 MHz against the former 800 MHz).
I have used Memtest86+ to confirm that something was really wrong with memory: before the forced shutdown the tool generated many errors between the first 5 and 10 minutes of the test. So I went on eBay again, I spent other money to purchase two DDR3 RAM banks by CORSAIR – better to be on the safe side, I thought, and pick a brand that is always a guarantee – I replaced the old banks and ran the Memtest86+ test again. The errors were gone but the PC was still shutting down within the first 10 minutes of benchmarking.
At this point I started to turn my attention to the CPU, even if indirectly: thinking that the issue was the bigger amount of heat generated by the system on its whole, I’ve purchased a cooling mat equipped with fan made by Targus. The on-line reviews weren’t very encouraging – a few more dissipated degrees than a barebone laptop – and my gullible hope to solve the problem soon showed to be misplaced because the PC continued to shut down even though it was placed on the cooling mat.
The mat has been later adopted for good because of the comfort to be able to use an inclined keyboard even on the laptop, while the next step in my troubleshooting efforts has been replacing the X9100 “Extreme” CPU with a new Core 2 Duo – a T9900 processor, with the same 3.06 GHz frequency but with a lower maximum TDP. It was my last, useless purchase (255 €): the performance was horrible and the PC shut down as usual, so I couldn’t help but surrender and put the laptop’s original CPU back: a Core 2 Duo T6400 at 2.00 GHz.
The nefarious effects of the automatic CPU thermal throttling after playing Mirror’s Edge for some minutes
Luckily for my flimsy finances and my heart, the downgrade operation has solved the overheating issue and the PC has returned to be functional: 1 GHz less powerful but better than nothing. As you can see in the lower right corner of the two following screenshots, full loading the CPU hasn’t the same devastating effect on the temperatures recorded by the several PC sensors – HDD, GPU, CPU core 1 and core 2, overall CPU, motherboard – anymore.
I still have to understand the reason why the system held the X9100 CPU just for a few months before developing this bizarre hardware “idiosyncrasy”, and at this point I don’t even care anymore: the lesson of this painful story is that I have lost any faith in the opportunity to make a substantial hardware upgrade to a laptop, I will advise anyone against the practice and I will patiently wait to be able to afford a kick-ass desktop to tinker as I wish.
P.S.Yes, Sir Arthur is back blogging. Let’s hope it will last 😛