How to solve the problem of SSD heat generation?

首页    News    How to solve the problem of SSD heat generation?

Most of the new M.2 SSDs do not have heat sinks. How to solve the problem of heat generation?

From mechanical hard drives to SATA SSDs to M.2 SSDs, we have witnessed the step-by-step improvement of storage hardware performance. Now the latest PCIe 4.0 M.2 SSDs have a sequential read speed of 7000MB/s and a sequential write speed of 5000MB /s or above, the improvement in file transfer speed and program loading time is quite obvious.



While everyone enjoys the high-speed transmission brought by M.2 SSD, the problem of high heat generation is also very worrying. On the one hand, the M.2 SSD has strong performance and high heat generation; on the other hand, the M.2 standard was originally created for lightweight and portable ultrabooks. The small size is the advantage of M.2 SSD, but this also leads to heat accumulation. If the heat on the SSD cannot be dissipated in time during use, the M.2 SSD will slow down very quickly, or even cannot be used normally, which will seriously affect the performance.



In order to enhance the heat dissipation of M.2 SSDs, many products in the past few years will have their own heat sinks, or users will also buy a heat sink after buying back the product. But I don’t know if you have noticed that the M.2 SSDs newly launched this year, whether they are PCIe 3.0 products or PCIe 4.0 products, are rarely equipped with heavy heat sinks except for individual models. Instead, graphene heat dissipation stickers, plating Nickel coating, nano-copper foil composite material heat dissipation stickers... I can't help but wonder: Is it because the heat problem has been solved on these new M.2 SSDs?

The thickness of the SAMSUNG 980 PRO SSD with Heatsink is about 8.5mm, and most laptops and SSD enclosures cannot be installed without space.



The comparative study found that for some products, a certain degree of heat reduction has indeed been achieved, which is most obvious on the PCIe 4.0 M.2 SSD. The main reason is that the main control process has been significantly improved. The first batch of PCIe 4.0 M.2 SSDs to be launched in 2019 all use the Phison PCIe 4.0 M.2 SSD solution. In addition to the inability to fully utilize the PCIe 4.0×4 bandwidth, the 28nm process also results in relatively high power consumption and heat generation high.

Today, the PCIe 4.0 M.2 SSD solution has entered the second generation. For example, the Seagate FireCuda 530 adopts the new Phison second-generation master PS5018-E18, which uses an advanced 12nm process to reduce power consumption and achieve the highest continuous read and write speed, They are 7300MB/s and 6900MB/s respectively.

ADATA XPG GAMMIX S70 adopts SMI's SM2264, which also adopts advanced 12nm process to reduce power consumption. The continuous read speed can reach 7400MB/s, the continuous write speed can reach 6800MB/s, and the random read and write speed can reach 1,000,000 IOPS.

WD Back SN850 adopts the main control of its own Sandisk, the model is SanDisk 20-82-10034, adopts 16nm process, provides sequential read and write performance of up to 7000MB/s and 5300MB/s, and random read performance of 1,000,000 IOPS. This is also a high-performance Gen 4 SSD controller.

And the Elpis master control like Samsung 980PRO even uses 8nm process, which is better than TSMC's 12mm process, better performance, and naturally greatly reduced heat generation.

The following is a comparison table of several major PCIe 4.0 SSD Controllers in the current market, which can help you have a deeper understanding of the performance of the current PCIe 4.0 SSDs and make better judgments when choosing to buy SSDs.



Why is the temperature of NVMe SSD so high?
There are many reasons why SSD temperature rises: reading and writing data for a long time is the biggest dominant factor.
1. The temperature of SATA SSD is much lower than that of NVMe M.2 SSD, so don't care too much about the temperature of SATA SSD.
2. For an NVMe M.2 SSD, 60 degrees is a comfortable working temperature for it, and the overheating threshold is around 80 degrees. If the temperature exceeds 80°C for a long time, it is quite dangerous. Once the heat is accumulated due to poor heat dissipation, the performance of the SSD will be degraded at light level, and the data transmission will be lost or even the SSD will be damaged in severe cases.

Even though the current SSD Controller using advanced technology has greatly reduced the heat generation, the heat generation of the M.2 SSD during continuous reading and writing is still considerable. Some users have tested that the temperature of the main control chip and flash memory particles on the Samsung 980 PRO is at least 80°C after half an hour of uninterrupted writing. But if you add a heat sink to it, the temperature will drop a lot immediately.

How to solve the problem of NVMe SSD overheating?
In fact, many brands of SSDs are not equipped with original heatsinks. At the same time, there is no space for installing heatsinks on hard disk boxes or laptops that use SSDs. Therefore, in the face of overheating, the solutions given by SSDs are actually the same as those of CPUs and laptops. The graphics card is the same, that is, the overheating protection mechanism. There are temperature sensors in the main control, flash memory, and DRAM cache on the SSD. Some of the sensors may be built into the PCB. These temperature data will be transmitted to the main control, and some data can be fed back to the system through SMART. As for whether the temperature is too high? The judgment is made by the SSD master controller. The SSD manufacturer can set the critical temperature in the firmware. If it exceeds this line, the frequency of the master controller and the flash memory will be reduced, so as to reduce its own heat generation, thereby reducing the temperature.

Of course, the critical temperature of overheating protection set by different brands of SSDs is also different, some are 75 degrees, then the SSD will have a speed drop problem when it exceeds 75 degrees. Some settings have a critical temperature of 80 degrees, and after 80 degrees, the temperature threshold is triggered to reduce performance in exchange for low temperature. This is also the reason why the speed drop setting caused by overheating stops working after the SSD has been working continuously for a long time.

The NVMe M.2 SSD has two temperatures, one is the main control temperature and the other is the memory temperature. The temperature displayed by the temperature monitored by different brands of SSDs is also different. Some SSDs just detect and show relatively low memory temperatures, while higher master temperatures are hidden. For example, Samsung can know that there are two temperatures, while Intel SSD does not have a main control temperature Sensor. Even if the main control temperature is very high, the speed-down mechanism will only be executed according to the temperature of the acquired memory.



Temperature conditions of Samsung 980 PRO 1TB in standby



This is the temperature of the Samsung 980 PRO 1TB after half an hour of uninterrupted writing. The temperature of the main control and flash memory are very high. at this time
If the Samsung M.2 SSD does not add a heat sink, the temperature will indeed be very high, and the temperature performance will be much better after adding a heat sink.
After using the heat sink, the maximum temperature of the Samsung 980 PRO 1TB main controller dropped from 100°C to 68°C, and the flash memory temperature dropped from 82°C to 61°C. The heat sink is still very useful.

The solution to the high temperature of the solid state drive
Faced with the problem of overheating, the solution provided by SSD is actually the same as that of CPU and graphics card, that is, the overheating protection mechanism. There are temperature sensors in the main control, flash memory, and DRAM cache on the SSD, and some may make the sensor into the PCB. Inside, these temperature data will be transmitted to the main controller, and some data can be fed back to the system through SMART. As for whether the temperature is too high, it is determined by the SSD main controller. The SSD manufacturer can set the critical temperature in the firmware. If the line is connected, the frequency of the main control and the flash memory will be reduced, so as to reduce its own heat generation and thus reduce the temperature.



The SSD speed reduction mentioned above is the actual performance after the overheating protection mechanism takes effect. As for how to drop and how much it is reduced, it depends on how the firmware is written. We tested the Western Digital Black Disk SN750 M.2 SSD to see how WD's overheating protection strategy is. The factory has no heat sink and has a capacity of 1TB. The test method is to use the TxBENCH software to perform a full-disk write test, and use HWInfo software to record the temperature of the SSD at the same time, so you can see under what conditions the SSD will trigger overheat protection.



As can be seen from the above picture, the overheating protection mechanism of the Western Digital SN750 is relatively aggressive, and the performance fluctuation is also the most severe. The node that activates the overheating protection may be at 84°C or 85°C, but the speed will drop quite sharply after the protection is activated, the write speed will drop from 1500~1800MB/s to less than 100MB/s at once, and it will quickly recover to the original speed after cooling down, so you can see the picture that the write speed range of SN750 after overheating protection is very large, and at the end It will even drop directly to 0, and its overheating protection mechanism is the most radical.

Let's take a look at the dynamic thermal protection DTG (Dynamic Thermal Guard) promoted by Samsung SSD

While Samsung's NVMe SSDs offer impressive performance, they also generate a good amount of heat. To keep them from overheating, Samsung has implemented what they call Dynamic Thermal Guard (DTG). This technology monitors the temperature of a drive and will reduce its performance once it reaches a certain point.

The 980 PRO is also equipped with Dynamic Thermal Guard. However, Samsung has tried to delay the inevitable drop in performance by dissipating heat more efficiently. The 980 PRO has a copper heat spreader built into the label on the back of the drive as well as a nickel coating on the Elpis controller that helps dissipate heat faster. 


Seeing the above test results, do you think that the SSD will be easily triggered by overheating? In fact, this is not the case, Such usage scenarios rarely occur in daily use. Generally, the read operation of SSD will account for more than 2/3, Due to the working principle of flash memory, the read operation power of SSD is lower than that of write operation resulting in heat generation. Naturally, it is also lower, and the power of the more common mixed reading and writing is actually lower, so there are not many cases where overheating protection will be encountered in daily use. At the same time heat sinks can be added to the SSD to help the SSD reduce heat.
However, for users who use notebook computers or SSD hard disk boxes, there is generally no space for heat sinks inside. At this time, thermal pastes are needed to directly transfer the heat of the SSD to the product shell, even if it is a plastic shell. The laptop is also better than simply letting the SSD run naked, which can reduce the temperature of the SSD. If it is an aluminum alloy shell product, the heat dissipation effect of the product is better.



For example, the Ankmax USB4 NVMe Enclosure UC4M2 with thermal paste installed, the temperature of the Samsung 980 Pro SSD does not exceed 70℃, which basically does not trigger the temperature protection. Even if it works continuously for a long time, it will not be slowed down due to high temperature, and the reading and writing speed will be much more stable.
In fact, the purpose of SSD's overheating protection mechanism is to protect the safety of SSD and user data. These things are far more important than speed, and to be honest, most of the SSDs used in daily life are in light load or standby state, and the temperature is very low. In fact, it is difficult to trigger the temperature protection for the load of read or mixed read and write operations. It is only triggered when large data is written, and it is necessary to continuously write tens of GB of data before the slowdown protection. Moreover, the power consumption of the SSD itself is actually not high. It can be said that it is lower than the PCH on the motherboard, and the load power is very different from the standby power. Therefore, the temperature of the SSD will drop soon after the writing operation is stopped, and the normal high-speed reading will resume. write performance.

NVMe PCIe Gen 4 SSD Controller Performance Comparison


2022年5月11日 12:01