There is lots of chatter regarding the limitations associated with increasing the fsb using a kentsfiled cpu. As I understand the issue the limitation is the GTL voltage is too low on all current 975 and 680i chipset boards to sustain a fsb much above 300. Because quad core cpus use the bus to communicate between the 2 dual cores instabilities due to this inadequete GTL voltage are more prone to cause an error when trying to overclock, even modestly, a quad core cpu. I realize that intel does not "support" overclocking, however since EVGA is working on a fix for their 680i chipset board, and DFI's RD600 and soon to be released 680i board address this issue, I am curious what Intel is going to do with the bx2?
Any thoughts out there?
975bx2 support for higher fsb using a kentsfield cpu
I saw an article in the inquirer using an Asus Striker 680i board to overclock a 3000 series xeon. Yes it did reach a good fsb, a similar OC was achieved using the new DFI board with the rd600 chipset, which addresses the GTL issue in the bios. An article at the techrepository shows how the bx2 has to be modded to alter the GTL voltage. The issue here is that the Qx6700 which is in the top seller list at many places will need a solution, and I do not think people who have purchased this cpu should have to look to spend another $1000 on a 3000 series xeon. Why is it an issue, because at a fsb of 266x4 you achieve a bottleneck pretty quickly with a cpu that can turn as many cycles as the kentsfield, opening up the bus allows the cores to move things in and out of the 2 cores at a rate that is noticeable in applications.
Intel has elected to have 1300 as their new target fsb for the next generation of cpus, so I suspect there is less motivation to develop a solution for the QX6700 and bx2 combination. Time will replace both at less cost. I do not own and EVGA MB, but to be honest I am looking hard at their solution when the post it as well as waiting for the DFI 680i board. My bx2 was my first intel MB. I guess my response to intel is that if I bought their board and cpu as a tenable solution to do a modest fsb overclock which is implicit in their touting of both products, then it should be important to intel to make sure this will happen. Intel does not support overclocking, is a cop out. The question is would intel gain more from making products that work in this manner than by making stock speed only products? And in that respect why name any product an "Extreme" product if you can only use it at stock speeds? Why have an unlocked multiplier? Yes you can crank up the MHZ using the multiplier and realize only marginal peformance gains in most applications due to the fsb bottleneck. I have tested this. There are very few applications that Windows users have on their machines that take advantage of 4 cores. Encoding apps are probably the most common. As OS's move towards supporting more multicore applications (slow to come), having a quad core cpu and a higher FSB would be a terrific combination. Something many of us who purchased a QX6700 and BX2 combination thought about.
Intel has elected to have 1300 as their new target fsb for the next generation of cpus, so I suspect there is less motivation to develop a solution for the QX6700 and bx2 combination. Time will replace both at less cost. I do not own and EVGA MB, but to be honest I am looking hard at their solution when the post it as well as waiting for the DFI 680i board. My bx2 was my first intel MB. I guess my response to intel is that if I bought their board and cpu as a tenable solution to do a modest fsb overclock which is implicit in their touting of both products, then it should be important to intel to make sure this will happen. Intel does not support overclocking, is a cop out. The question is would intel gain more from making products that work in this manner than by making stock speed only products? And in that respect why name any product an "Extreme" product if you can only use it at stock speeds? Why have an unlocked multiplier? Yes you can crank up the MHZ using the multiplier and realize only marginal peformance gains in most applications due to the fsb bottleneck. I have tested this. There are very few applications that Windows users have on their machines that take advantage of 4 cores. Encoding apps are probably the most common. As OS's move towards supporting more multicore applications (slow to come), having a quad core cpu and a higher FSB would be a terrific combination. Something many of us who purchased a QX6700 and BX2 combination thought about.