Overclocking
Overclocking
We have come to expect Pentium E5200 and E5300 processors to top out at a much lower FSB speed than other Wolfdale-based chips, such as the Core 2 Duo E7300 or E8400. Luckily, an 800 MHz FSB equates to a high available multiplier and the ability to exceed 4.0GHz core speed at FSB-1333. An apparent wall was reached at 340 MHz (FSB-1360) on this system, which then eliminated the need to later raise voltage settings normally associated with a high FSB overclock.
As we have often stated, overclocking is luck of the draw and results vary from chip to chip. The much higher voltage ID (VID) of this E5300 was an early indication that the processor would not scale as well as the 1.125V E5200 we received for the $600 gaming PC build back in May.
Jumping straight to FSB-1333 and a multiplier of nine, we began stability testing at each multiplier increment, finding that 3.33 GHz was fully stable at stock voltage, but a jump to 3.5 GHz failed to load Windows. Finding stability required a 50mV bump, and an additional 50mV bump at 3.66GHz meant voltage scaling for this Pentium’s core speed was running about 200 MHz behind the gem from back in May.
We pressed on by increasing the multiplier in half increments and bumping up the CPU voltage until finding stability. When all was said and done, this E5300 successfully completed a 20 minute stress test at 4.0GHz, but required a whopping 1.488V idle (1.456V load) to do so. This was more voltage than any of the E5200s I have tested required at this core speed, and above what we would utilize for long-term use, so no attempts were made to completely stability test the chip at this frequency.
After numerous presumably stable overclocks failed to pass a lengthy array of stress testing, tweaking time was running out and we settled on using 3.83 GHz at 1.424V idle (1.392V load). Further efforts after all performance testing had been completed found 3.9 GHz stability at 1.456V idle (1.424V load).
Overclocking with the DFI BI P45-T2RS was an interesting experience. On a positive note, the board rarely failed to post, and successfully recovered each time it did fail without the need to clear the CMOS jumper. The trouble we encountered was in the amount of time it took to dial in a fully stable overclock. Voltage scaling with this E5300 was unpredictable, as putting an hour into overclocked settings that eventually were found to be unstable chewed up a lot of time. Unfortunately, little time was left for maximizing the speed and timings of the system RAM, and no such luck was found when even booting at DDR2-1066, never mind passing Memtest 86+ stability testing.
The full-featured Genie BIOS is packed with settings and voltages, and pressing F9 reveals additional adjustments normally reserved for enthusiast-class motherboards. I have little doubt that enough tweaking time would have brought out more potential from this system. The overclock we settled for, while maybe not ideal, did provide a fully-stable test environment.
And for those hesitant to dabble too much in the BIOS on their own, DFI even includes preconfigured profiles and an auto-boost system (ABS) for saving, loading, and sharing BIOS settings. The profile for FSB-800 chips was easily able to run our E5300 at 3.46 GHz (266 x 13), although it applied more CPU voltage than was necessary at such a speed.
Overclocking the two XFX Radeon HD 4870s was a painless and rewarding experience. Incremental increases to the GPU core and then the GDDR5 memory revealed that this pair of cards was capable of running at the maximum 790 MHz/1,100 MHz settings available in Catalyst Control Center. The graphics card fans were manually set to 100% from the start, but then lowered to 85% for final stability and performance testing.