When we reviewed AMD’s FX-8350 we weren’t able to find the time to really explore it’s overclocking capabilities. But now that things have slowed down a bit we can finally remedy that oversight by clocking our Vishera sample to the moon.
We’ve decided to use two cooling solutions for our overclocking attempts today. The first of which is a stock heatsink from retail boxed version of AMD’s old Phenom II X6 1055T, and the second is Corsair’s H80 closed loop water cooling solution. By using two different cooling solutions we’re trying to show what kind of benefit you can expect to see if you chose to go with a mid-range aftermarket cooler over AMD’s stock solution.
After disabling all of AMD’s power saving features, and any other BIOS settings that could impede our overclocking attempt, we began our testing with AMD’s stock cooler. Using AMD’s Overdrive app we clocked our FX-8350 up by 100Mhz increments. Things went fairly smoothly as we tried to find the limit of how much voltage our chip could stand, and how high we could clock the chip before it quit booting into the Windows desktop. Increasing the voltage of the CPU past 1.6 Volts resulted in a system crash during the Windows boot cycle, and clocking our chip above 5.4 Ghz resulted in an immediate system lock up.
Now that we’d found the hard maximums of our testing platform, it was time find the maximum stable clock speed that our stock cooler could support. Unfortunately, this turned out to be an exercise in begrudgingly walking clock-speeds and voltages back down. After bringing clocks down to 4.5Ghz and 1.45 Volts, we were finally able to get our chip stable in an hour long Prime95 stress-test.
So we switched over to the H80 to try and reach higher clock speeds.
Our overclocking efforts topped out at cool 5Ghz using 1.6 Volts. 5.1 would boot to Windows but would crash after about ten minutes of Prime95′s “blend” stress-test. So we turned to tweaking the integrated north-bridge frequency up from 2.2Ghz to a modest 2.6Ghz using 1.3125 Volts.
The biggest limiting factor to our overclocking exploits in this article was cooling. With a higher performance cooling solution the FX-8350 probably would have clocked into the low 5Ghz range without any problems. Alas, the H80 is all we have. On that note, I think it’s worth mentioning that the FX-8350 was running pretty hot, particularly when we were using the stock AMD cooling solution. I doubt that I would’ve have been able to fry an egg on the thing, but it was none the less a very effective foot warmer.
To quantify the benefits of our overclocking efforts we quickly ran Cinebench and wPrime.
With a 500Mhz overclock we can see that gained about ten percent more performance, and by doubling our overclock to a Ghz we saw linear scaling up to a 20 percent gain over stock performance.
Moving to the only gaming benchmark that we ran, we’ve include a stock Core i7-3770K for comparison. This is admittedly a rather cluttered comparison, but the take way here is that at 5Ghz the FX-8350 is neck and neck with the i7-3770k in-terms of gaming performance.
It’s clear that there is quite a bit of extra performance up for grabs if you’re comfortable with turning up the voltage dial and willing to splurge on an aftermarket cooling solution. Granted the same is pretty much universally true of any modern microprocessor. But in light of how aggressively AMD clocked its FX chips, it’s nice to see that there’s still a Ghz of head room available for price conscious buyers to utilize.S|A
Latest posts by Thomas Ryan (see all)
Follow these categories: Microprocessors