Guide To Overclock
Overclock your CPU/mobo/memory
1) A mainboard (mobo) that allows you to manually change at least the front side bus (FSB), or multiplier.
2) Adequate cooling for the CPU. What is adequate depends on how far you want to push your system, what vcore (CPU voltage) you’re willing to use, and your specific chip model. You can overclock w/ a stock AMD/Intel heat sink (HS), but better cooling = higher speeds (generally). Use a program like MotherBoard Monitor to check your temps. If they're above say 55C don’t even bother overclocking your CPU until you have upgraded your cooling. Though you can overclock your FSB and lower your multiplier in order to keep your CPU at stock speed.
3) Common sense, general computer knowledge, luck, and good memory helps.
Terms/Definitions
FSB - is the speed at which your CPU, memory, and motherboard transfer data. A higher FSB means higher bandwidth, and performance. Your motherboard and memory will limit your max FSB when overclocking. Generally, your CPU will not, though there have been exceptions.
PCI Bus - is the frequency at which your hard disk drive (HDD), PCI slots, USB ports, etc run at. The stock speed is 33.3mhz. Increasing the PCI bus usually doesn’t increase performance noticeably. Running your PCI bus too high can cause PCI cards not to work, USB to stop functioning (rarely), or HDD corruptions (most common) where you could potentially lose all your files and have to reinstall everything. If your using a VIA chipset or any other motherboard where your FSB is multiplied by 1/4, 1/5, or 1/6 to obtain your PCI bus, make sure it never goes over 37mhz-39mhz (37mhz being safe, 39mhz being slightly risky) unless you don’t care if you lose everything. [PCI Bus] = [FSB] X [the divider]
AGP/PCI Express Bus - is the bus speed your accelerated graphics port and graphics card will run at. On motherboards w/ dividers, AGP/PCI Express bus = the PCI bus x 2. Higher bus speed mean more bandwidth. However, many graphics cards do not need anymore bandwidth than they already have at 66.6mhz(AGP)/100Mhz(PCI Express). Though you might see a very small increase in 3D benchmarking scores by it being higher.
Multiplier - is what your FSB will be multiplied by to get the end CPU clock speed. Most mobos allow you to adjust this by .5 increments.
CPU Cache - is the memory that a CPU has within its self (onboard). There is the level-1 (L1) and level-2 (L2) cache and level-3(L3). L1 is faster. Generally, the L2 cache is bigger (because it’s cheaper to manufacture). The size of the caches depend on the processor’s core.
Voltages:
Vcore (vcc) - is the CPU’s core voltage. The higher the voltage, the faster the clock speed the CPU will be able to run at, and the hotter the CPU will get. And yes, there is a point at which more voltage will hurt your overclock.
Memory voltage (vdimm) - The more voltage, the higher the FSB your memory will be able to hit; and/or the better the memory timings your memory will be able to run at. Some memory modules like higher voltages more than others.
Vdd voltage - is the voltage supplied to your chipset. Generally, the higher the voltage the higher the FSB that your motherboard can obtain.
AGP/PCI Express voltage - is the voltage supplied to your AGP/PCI Express slot/card. Raising the voltage to anything above stock (1.5v) is not recommended. This voltage will not help you overclock your GPU much further. It will only enable you to increase the AGP/PCI Express bus, which in most cases, doesn’t increase performance by more than a very small fraction. Increasing this voltage for an extended period of time can cause damage to your video card.
CPU/RAM Ratio, FSB/RAM Divider, etc - This allows you to run your mobo and CPU at a different clock speed than your memory. However, generally w/ AMD boards, a ratio of 1 (1:1, 3/3, 4/4, 5/5, etc) will yield the best performance. Running your mobo and CPU at 200mhz FSB w/ a memory ratio of 6:4 (memory running at 133mhz) will not be very advantageous.
Memory Timings/Delays - are how many cycles your memory is delayed between certain operations. Different memory have different stock timings. Lower timings are better, but the lower the timings, the lower the max FSB that can be obtained by the memory. Example of very good memory timings:
CAS Latency: 2
RAS/Row Precharge (tRP): 2
RAS-to-CAS Delay (tRCD): 2
Row-active-delay (tRAS): 6
CAS - CAS latency is the number of clock cycles between the memory receiving a "read" command and actually starting to read.
RAS/Row Precharge (tRP) - This Precharge to Active timing controls the length of the delay between the precharge and activation commands.
RAS-to-CAS Delay (tRCD) - This timing controls the length of the delay between when a memory bank is activated to when a read/write command is sent to that bank.
Row-active-delay (tRAS) - The Active to Precharge timing controls how soon after activation the access cycle will be started again.
You can think of memory timings and FSB like this: You're picking up boxes from different locations, organized in rows and columns, and moving them to another given location. The timings would be how long you pause between getting the info for your which box you need, finding the box's column, then the box's row, picking up the box, and leaving that area. Frequency would be how fast you walk/run from the time you leave the pick-up area to the time you return for another box (in other words, it has to do with how many times you could get from the pick-up area, to the drop-off area, and back in one minute if you didn't have to worry about finding/getting the box). The faster you run with the box you're carrying, the slower you are able to find your new box and everything with out freaking out. The more you hussle to find/get your new box, the slower you can hussle to drop off the box and get back. You need to find the best balance. With dual channel there are two of you.
Overclock your CPU/mobo/memory
1) A mainboard (mobo) that allows you to manually change at least the front side bus (FSB), or multiplier.
2) Adequate cooling for the CPU. What is adequate depends on how far you want to push your system, what vcore (CPU voltage) you’re willing to use, and your specific chip model. You can overclock w/ a stock AMD/Intel heat sink (HS), but better cooling = higher speeds (generally). Use a program like MotherBoard Monitor to check your temps. If they're above say 55C don’t even bother overclocking your CPU until you have upgraded your cooling. Though you can overclock your FSB and lower your multiplier in order to keep your CPU at stock speed.
3) Common sense, general computer knowledge, luck, and good memory helps.
Terms/Definitions
FSB - is the speed at which your CPU, memory, and motherboard transfer data. A higher FSB means higher bandwidth, and performance. Your motherboard and memory will limit your max FSB when overclocking. Generally, your CPU will not, though there have been exceptions.
PCI Bus - is the frequency at which your hard disk drive (HDD), PCI slots, USB ports, etc run at. The stock speed is 33.3mhz. Increasing the PCI bus usually doesn’t increase performance noticeably. Running your PCI bus too high can cause PCI cards not to work, USB to stop functioning (rarely), or HDD corruptions (most common) where you could potentially lose all your files and have to reinstall everything. If your using a VIA chipset or any other motherboard where your FSB is multiplied by 1/4, 1/5, or 1/6 to obtain your PCI bus, make sure it never goes over 37mhz-39mhz (37mhz being safe, 39mhz being slightly risky) unless you don’t care if you lose everything. [PCI Bus] = [FSB] X [the divider]
AGP/PCI Express Bus - is the bus speed your accelerated graphics port and graphics card will run at. On motherboards w/ dividers, AGP/PCI Express bus = the PCI bus x 2. Higher bus speed mean more bandwidth. However, many graphics cards do not need anymore bandwidth than they already have at 66.6mhz(AGP)/100Mhz(PCI Express). Though you might see a very small increase in 3D benchmarking scores by it being higher.
Multiplier - is what your FSB will be multiplied by to get the end CPU clock speed. Most mobos allow you to adjust this by .5 increments.
CPU Cache - is the memory that a CPU has within its self (onboard). There is the level-1 (L1) and level-2 (L2) cache and level-3(L3). L1 is faster. Generally, the L2 cache is bigger (because it’s cheaper to manufacture). The size of the caches depend on the processor’s core.
Voltages:
Vcore (vcc) - is the CPU’s core voltage. The higher the voltage, the faster the clock speed the CPU will be able to run at, and the hotter the CPU will get. And yes, there is a point at which more voltage will hurt your overclock.
Memory voltage (vdimm) - The more voltage, the higher the FSB your memory will be able to hit; and/or the better the memory timings your memory will be able to run at. Some memory modules like higher voltages more than others.
Vdd voltage - is the voltage supplied to your chipset. Generally, the higher the voltage the higher the FSB that your motherboard can obtain.
AGP/PCI Express voltage - is the voltage supplied to your AGP/PCI Express slot/card. Raising the voltage to anything above stock (1.5v) is not recommended. This voltage will not help you overclock your GPU much further. It will only enable you to increase the AGP/PCI Express bus, which in most cases, doesn’t increase performance by more than a very small fraction. Increasing this voltage for an extended period of time can cause damage to your video card.
CPU/RAM Ratio, FSB/RAM Divider, etc - This allows you to run your mobo and CPU at a different clock speed than your memory. However, generally w/ AMD boards, a ratio of 1 (1:1, 3/3, 4/4, 5/5, etc) will yield the best performance. Running your mobo and CPU at 200mhz FSB w/ a memory ratio of 6:4 (memory running at 133mhz) will not be very advantageous.
Memory Timings/Delays - are how many cycles your memory is delayed between certain operations. Different memory have different stock timings. Lower timings are better, but the lower the timings, the lower the max FSB that can be obtained by the memory. Example of very good memory timings:
CAS Latency: 2
RAS/Row Precharge (tRP): 2
RAS-to-CAS Delay (tRCD): 2
Row-active-delay (tRAS): 6
CAS - CAS latency is the number of clock cycles between the memory receiving a "read" command and actually starting to read.
RAS/Row Precharge (tRP) - This Precharge to Active timing controls the length of the delay between the precharge and activation commands.
RAS-to-CAS Delay (tRCD) - This timing controls the length of the delay between when a memory bank is activated to when a read/write command is sent to that bank.
Row-active-delay (tRAS) - The Active to Precharge timing controls how soon after activation the access cycle will be started again.
You can think of memory timings and FSB like this: You're picking up boxes from different locations, organized in rows and columns, and moving them to another given location. The timings would be how long you pause between getting the info for your which box you need, finding the box's column, then the box's row, picking up the box, and leaving that area. Frequency would be how fast you walk/run from the time you leave the pick-up area to the time you return for another box (in other words, it has to do with how many times you could get from the pick-up area, to the drop-off area, and back in one minute if you didn't have to worry about finding/getting the box). The faster you run with the box you're carrying, the slower you are able to find your new box and everything with out freaking out. The more you hussle to find/get your new box, the slower you can hussle to drop off the box and get back. You need to find the best balance. With dual channel there are two of you.
Last edited:
)
