Memory overclocking instructions

Memory overclocking instructions For DIY players, the highest realm of overclocking is to put everything to the end! Therefore, in addition to overclocking the processor and the graphics card, the memory inserted on the motherboard is still superb. However, the success of overclocking, how effective, but also depends on the method and method of overclocking. The author observes that many game players have many misunderstandings in the process of overclocking the memory. The consequences of these incorrect practices are minor impact on the machine performance, and serious will endanger the life of the hardware, so we must not take it lightly. Below for everyone to analyze the memory of overclocking the most common six errors, and correct it!

Misconception 1: Different brands of the same chip do not mean the same overclocking capability. Sometimes the two chips (different brands) use the same chip, but the overclocking performance shows a significant difference. This may have a certain relationship with the PCB design of the product and the quality of the selected particles. As many known memory brands use the particles provided by South Korea's Samsung Electronics, but these particles (the same number) are also graded, Samsung will put it The best quality is left to your original memory, or to the best memory manufacturers. For the rest of the manufacturers, or for smaller manufacturers, the purchased particles may be of inferior quality. This may also happen on other products, so the chip is important, but it may only determine the lower limit of overclocking, and the upper limit will vary from manufacturer to manufacturer.

Samsung Gold Bar is the original original memory of South Korea’s Samsung Electronics.

Misunderstanding 2: CSP packaged chip does not mean good overclocking Although CSP packaging is more advanced in technology, it has little to do with the final memory performance. Memory using CSP packaged chips is not necessarily better than conventional TSOP-II chipsets in terms of overclocking and memory optimization. We should not overly obscure the CSP package. Although it can achieve higher frequencies than TSOP-II itself, The requirements for peripheral circuits and changes in the electrical environment are also more demanding. With TSOP-packaged memory, it is more common for overclocking capabilities to be stronger than CSP packages.

Myth #3: Products with heat sinks are not necessarily overclocked. In fact, I have always suspected the necessity of installing heat sinks on memory. For now, heat sinks play a more important role in beautifying product images. Samsung's low-profile products even more than a lot of high-profile products with heat sinks can explain the problem. Therefore, when purchasing, there is no heat sink is not a factor (even secondary factors can not be ranked), and sometimes the heat sink may also play the role of the manufacturer's "Fashion cloth", people can not see The chip used must know that the memory chip is the basis for determining the lower limit of overclocking.

Like Samsung's gold bars and other brand memory, each chip is an excellent model that has been carefully selected, has a small amount of heat, and leaves room for frequency. Even if it is properly overclocked, it will not cause any problems, and there is no need for heat dissipation. worry. Even if you do not use any means of heat dissipation, Samsung Gold Bar memory can also have a good performance.

Misunderstanding 4: It is not difficult to understand that single-sided memory has more inherent advantages than double-sided memory overclocking. This phenomenon mainly depends on the drive load capacity of the chipset. The single-sided memory is mostly 8 chips, such as the current mainstream 512MB products, double-sided 16 chips, such as the current mainstream 1GB products. This means that the latter requires more drive capability (although a P-bank operation will be selected by a chip select signal at work, the inactive P-Bank will still increase the address and control signal lines). Obviously, 8 chips are always easier to drive than 16 chips. In addition, the increase in the number of chips will also double the number of signal lines on the PCB. If the design is poor, the probability of mutual interference will increase significantly. This may also affect the ability to overclock, but this is relatively minor. The reason for the overclocking ability of the module itself depends more on the chip.

Myth #5: Memory with high specifications is not necessarily better than memory with less memory than theoretical memory. Theoretically, DDR-500 is faster than DDR-466, but I tell you that it is not absolute. Because there are a lot of high-specification memory is all at the expense of timing optimization capabilities in exchange for. Although the memory bandwidth of DDR-500 is higher than the nominal frequency, it does not mean that it has better optimization capability under DDR-400. The reason is simple. The chips used by DDR-500 may be the same as those of ordinary DDR-400 chips, and they are not dominant in the timing optimization ability. The attainment of DDR-500 is more achieved by relaxing the timing requirements. of. This approach has little significance for users, but manufacturers are very fond of it, and use the same product to sell a higher price, why not do it, huh, huh! On the other hand, although some memory can work with the fastest timing under DDR-400, this does not mean that it can also be overclocked. Conversely, a particular memory can be overclocked, but it does not mean that it can achieve the fastest timing in DDR-400. There is no direct proportional relationship between the two. Therefore, we must pay attention to this point, which also reflects the two different purchasing tendencies, is to get the best DDR-400 performance or the highest operating frequency, only one of two options.

The Samsung Gold Bar DDR2-533 using Samsung ZCD5 particles can operate in DDR2-667 mode with 4-4-4-X timing without voltage overclocking, and it also has the challenge of DDR2-900.

Most of the common Samsung GCCC and GCD5 particles have overclocked to DDR2-800 levels under the 5-5-5-15 parameters.

Myth #6: Some product parameter settings are not necessarily the best. Overclocking uses DDR400 memory as an example. When 3-4% is not used, it should be replaced with By SPD. Some default CL = 2.5 product, By SPD but will be better than the 3-4-4 over, but some are not, the default CL = 3 products, By SPD and 3-4-4 are similar, there may be 1,2MHz The difference. The author estimates that this should be the difference that the chip shows. CL is the only parameter that is written to the on-chip mode register when the memory is initialized (tRCD, tRP only need to control the north bridge), so its value and the inherent adjustable range of the chip need to match, if the chip itself is limited Larger, then increase the CL value will cause instability in the memory subsystem and can not boot. So, although 3-3-4 is a commonly used setting for overclocking, if you encounter a point that does not time out, it is recommended that you use By SPD to try it out. There may be surprises. In addition, many PC3200 memories set to 3-3-3 at 200MHz are automatically reduced to 3-4.4 when the By SPD is overclocked, which is estimated to be the automatic setting of the P4C800 motherboard.