Purchasing high-speed RAM is a mistake!


The more the merrier, that seems to be the motto of every user when deciding to buy RAM for their PC. But the truth is that this is not fulfilled in any case, since like everything there is a limit of both speed and capacity. The latter is easy to know, since each platform sets the maximum number or number of modules, but in terms of high-speed RAM and latency, be careful.

RAM memory has always been critical in every computer or server. As we have consistently stated, it is the most difficult bottleneck we confront within the great “BIG THREE.” The CPU significantly enhances overall performance, particularly in the aftermath of AMD’s Ryzen processor launch, when the scaling of cores, speeds, and IPC has been overpowering. Graphics cards have improved and scaled significantly, but RAM has remained static.

Why not invest in high-performance RAM for Intel or AMD?

Although it may appear counterintuitive, and as previously stated, more is not always better, especially if we lack the requisite expertise. Brands sell us fantastic high-speed RAM modules and kits with ever-shrinking latencies, they show us screenshots of Runmemtest Pro at 500 percent, and it all seems to come out of nowhere.

This means that it appears that we are going to buy them, and all we have to do is click on them and enjoy. This does not work at all, and it does not work precisely because RAM is the aspect that limits performance the most, inside everything that comprises and connects it. Although motherboards frequently specify a maximum speed with overclocking, it is unsurprising that the first-generation models recently debuted on the market with the platform, do not support the speeds now available in modules for sale.

That is, our board allows DDR4 modules running at a maximum of 4800 MHz and we can already discover modules running at 5000 MHz. Yes, this is an extreme example, but it occurs with many speed ranges in various models of boards, where the lower the speed range, the more overclocking support there is, where the top model will have the most supported range, and the lowest range may not even support it.

CPU BMI, the most limiting factor!

The motherboard can be a problem, but it is usually not, since the average user does not pay for the extreme modules and therefore “saves himself” from the problem. However, there is a second component that is important: BMI. Both Intel and AMD stipulate a DDR4 speed of 3200 MHz, thus the BMIs are currently close to the serial limit, but there is possibility for overclocking.

This is incredibly relative, and we will explain why. The IMC is a particularly delicate portion of the CPU that connects the cores, registers, and caches to the RAM memory, where access times must be kept to a minimum and transfer rates must be as high as the modules allow. As a result of increasing its synchronization using faster RAM, its voltage increases, jitter is added into the signal, and it may be difficult to maintain synchronization at that speed, depending on the MHz of the modules.

Certainly not with all of the world’s electricity on it or the VCCIO. The overclocked BMI limit (greater than 3200 MHz theoretically) is currently hanging around 3600 MHz and will support more or less depending on the quality that you get from it on your CPU. The issue is that sophisticated expertise is required to make it stable when it exceeds the aforementioned 3200 MHz, which the great majority of customers are unaware of or are unwilling to spend hundreds of hours to raise 200 MHz or 400 MHz to 5000 MB/s more and 3ms less.

As a result, if we don’t want to complicate things and want to play it safe, we should stick at 3200 MHz, and if we have at least advanced ideas, we should go with DDR4-3600 MHz with medium or relaxed latencies. And now for the final point: latency effects the stability of the BMI precisely because it requires it to update cell information in less time. It is not the same C18 to C16 in DDR4-3200 or DDR4-3600 MHz; the stress sustained by the BMI increases with the increase in speed and decrease in latency, therefore a balance must be found.

BMI wear, what no one tells you!

The IMC and the cache are currently the two elements of the CPU that wear down at breakneck speeds. Adjusting the voltages and understanding where the sweet spot of each of them is critical if we do not want to see blue screens on our PC after 6 months, 1 year, or at most 2 years and have to re-adjust voltages and even change modules memory at lower speed or higher latency.

Applying more voltage or reducing it to the very minimum is not an option; the IMC voltage must be correct, the appropriate one, because only in this way, the least jitter in the signals is assured, and so it suffers less. Above 3200 MHz, there is currently little or no discussion of accelerated wear and tear. Additionally, any PC expert is aware that the IMC will eventually fail at certain speeds after typical PC use, and much faster if we are regularly passing stress benchmarks.

So be careful not to choose the wrong RAM modules, their speed and latency, not everything works and far from it is to click, enable XMP or DOCP, and enjoy (if we go over 3200 MHz).

Previous Post

Nintendo Switch vs OLED: First Images

Next Post

Nintendo 64 controller is coming for Nintendo Switch

Related Posts