Alright, I have an evga 680i motherboard paired with some Corsair Dominator PC800 and an e4300. Some would say the e4300 is an odd choice but its basically the same chip is an e6300. Both are 1.8 stock but the FSBs are 800 and 1066. The idea here is that I can push a e4300 further without stressing the motherboard as much. At the stock FSB of the 6300, the 4300 is at 2.4ghz. Plus its a very cheap chip. Really, the main benefit is the 9x multiplier. Currently I'm sitting at: CPU 2592Mhz FSB 1152 Memory 806 The goal is to get the CPU a bit higher (around 2.8 would be nice) However, I'm running into a wall and the limits of my knowledge. The math is simple enough CPU Multiplier = 9x Stock FSB 800 Quad Pumped, 200 actual 200 x 9 = 1800 1152 / 4 = 288 x 9 = 2592 So if my goal is 2.8 (2808) then 2808 / 9 = 312 x 4 = 1248 1248 is not a stretch considering how far many people are pushing this motherboard. And seeing as nvidia claim this board will do 1333, I should have no problems. I'm willing to accept that I may be hitting the limit of my CPU but I want to try everything first. I had kick more voltage to the CPU in order to get to where I am now. I increased it a bit more but I was unable to get more out of the CPU. It does go a few Mhz higher (refering to FSB here) but doesn't survive stability tests. One thing that I'm not 100% on is the SPP and MCP clockspeeds. At stock, they are both at 2500Mhz. Overclocked, I get 3125 on the SPP and the stock 2500 on the MCP. I'm not really sure about where these speeds are derived from. I know its the HT speed. I didn't really notice it until a few days ago when I was checking my GPU temp with ntune. Since the SPP is pretty much the northbridge, I guess it would make sense that as the FSB clock speed increased, so would the HT speed. The FSB and SPP HT speed are both running roughly 20% faster. Is it possible that the increased HT speed of the SPP is my wall? Another question is that in the BIOS, there are adjustable HT multipliers for the MCP -> SPP and SPP -> MCP. Both are stock at 5x. So, given that, if the SPP HT speed goes up, should the MCP as well? I feel like the answers are going to be something embarrassingly and painfully obvious that I've overlooked. I'm keeping a good eye on all my temps so I doubt overheating.
well, so as not to misinform the noob's that might read this...the differences are a little more than that...1.86 ghz for the e6300, plus a few optimizations the e4300 doesnt have, but nothing worth the extra cost imo...now, ur post is a little confusing...when you say 1248, i assume you are talking about ur memory, and yes, even with dominator, you are going to have significant trouble getting there...you need to use the divider, and lower your mem speed...that will probably allow you to get past ur wall...
I have to run my CPU vCore at 1.4375v to attain 3.2Ghz (1424FSB) on my E6600. You'll probably need to set your vCore to around 1.4v aswell when your FSB is around 1333.
1248 refers to the FSB speed after the "quad pump." The true FSB speed will be 312 Mhz. 312 x 9 = 2808. 9 is the CPU multiplier. The memory is separated by a divider. I plan on attacking the memory down the line but for now I'm making sure it isn't a bottleneck. But yeah, I was generalizing about the CPUs. The voltage of the core is currently a 1.4. So is the FSB. I'm fairly certain voltage isn't holding me back. When I did the initial overclock, I set up the system in a bench testing situation. I had good control over the temperatures and pushed the voltage to some risky levels. I've talked to a friend over at intel and his position is that well, basically he's not happy I went with in nvidia based board instead of a BX2 . Not much help there. He's not down with overclocking. Maybe it would be clearer if I gave a summary of my method to get to my current speed. When I received the parts, I set the motherboard up on a motherboard tray I had. I lowered the room temperature with my air conditioner to about 75F. Initially, I used a stock intel 775 heatsink with some artic silver 5. Added the memory and my 8800. The power supply is a 750w server PSU. One the first boot, I went into the BIOS and made sure everything was well and good. After that, I used a spare SATA drive and installed winXP, some benchmarks, ntune, cpuz, and a few stability tests. Now I ready to start the OC. Around the test area, there was a number of fans to keep the air circulating. Everything was done in the BIOS. I dropped the divider on the memory and disabled the the unnecessary stuff. I began to increment the FSB, 3 mHz at a time. On each successful boot, I check CPUz to confirm the settings, and then ran prime95 to quickly check stability. When I first ran into stability issues, I increased the voltage. I continued on. After every extra 100 mHz on the CPU, I ran 3dmark06 to stress the system as a whole. Stability is important as this is my main system. The math, as I understand it, for overclocking non EE C2D's is pretty simple. Since you can't increase the multiplier on non-Extreme Edition parts, the FSB must change. This should help anyone not on the same page as well. The CPU derives its clock speed form the true FSB speed. If it were a forumla where 'a' is the FSB, 'b' is the CPU multiplier, and 'c' was the CPU clock speed, you would get 'ab=c' Or if you prefer to draw it out a bit 'a * b = c' This is easily verified by looking at clock speeds of C2Ds: E4300- FSB= 800 Multiplier= 9x CPU clock speed= 1800 E6300- FSB= 1066 Multiplier= 7x CPU clock speed= 1862 E6400- FSB= 1066 Multiplier= 8x CPU clock speed= 2128 E6600- FSB= 1066 Multiplier= 9x CPU clock speed= 2394 E6700- FSB= 1066 Multiplier= 10x CPU clock speed= 2660 800 and 1066 are the effective FSB speeds. Intel achieves this by "quad pumping" the bus. So the actual mHz speed are a quarter of the effective speed. So 800 mHz effective is 200Mhz actual and 1066 effective is the 266 actual. The actual speed is what the multiplier is used with to determine the CPU clock speed. What I'm trying to figure out is how the FSB and HT speed relate and if the high HT speed of the SPP could be my wall. If it is, I can either increase the HT voltage or try and reduce the clock. Or if perhaps there is something else I ahve not considered that is holding my OC back.
Quite a few people seem to be hitting a wall around that sort of area on with the 4300's. It _looks_ to be an FSB bug, similar to that seen on the 7x5 Dothan's and the Merom/Yonah CPU's. Good way to test this is drop the multiplier and see if you can push things any further. Try dropping the HT multi to 4x and see if that helps, also give the NB and SB a small voltage bump and see if that helps you at all.
