Mention low-end solid state What will everyone think of? Is it a garbage black chip particle lottery? Or once you slow down, you will become a large USB flash drive? To tell the truth, everyone knows that low-end solid state is not good But what's wrong with it In addition to the vague concepts of speed and safety, It seems that I can't say any specific problems A hard disk that is very poor in running points How much slower will it be than high-end solid state in actual use Can you make do with it? In order to thoroughly understand the entry and high-end SSD, Difference under real use conditions This time, I bought three hard drives with different positioning at my own expense Test their actual performance in so many scenarios I hope everyone can give me a free compliment We'll start right away First of all, introduce the players and the test environment In terms of capacity, I chose 512G (500G) As for why not use the 1TB with higher cost performance Mainly because Disks with weak performance are more likely to show differences under small capacity And 512G is also the mainstream configuration in previous years There are a large number of stock users Model aspect I chose SOLIDIGM's P44 Pro for high-end solid state It was after Hynix acquired Intel's solid-state business Established independent brand The nominal speed of this hard disk can run to PCIE 4.0 × 4 There is also a stand-alone cache It can be regarded as the "top solid state of the first-line big factory" in the eyes of the public Getting started solid here I was going to choose some familiar products But I feel that it is not enough to "get started" So I finally bought a Quintech Fast Tiger TP3000 The price of this plate is less than half that of P44Pro The nominal speed is only about PCIe 3.0 × 2 No external cache Using an unknown "high-speed 3D particle" The elements of the low-end solid state are all gathered by it Should reflect entry-level products Especially the general level of Taobao complete machine Finally, considering that the 2.5-inch SATA disk still has a certain audience We also brought a piece of Yingruida MX500 It should be regarded as the above-medium level in SATA plate By comparing its achievements with those of the other two solid states, You can infer that when upgrading from SATA to NVME, How much has the actual experience improved Test platform aspect To reduce bottlenecks caused by other hardware This time, a relatively high-equipped desktop platform was selected And both M2 hard disks are equipped with heat sinks Two sets of memory are selected according to different test items The reasons will be explained in detail later Except in the empty disk state We also tested the results in the near full set Closer to the actual use Okay, before the actual application test, Let's look at the theoretical running points first The first is Crystaldisk mark 1GB small file test The MX500 is limited by the rate of the SATA bus Sequential reading and writing can only run to more than 500 MB s Random reading and writing are also the lowest Read less than 40MB s for 4KQ1T1 Then there is the Kingtec TP3000 PCIe 3.0 Hard Disk as Entry Level It is indeed much faster than the MX500 However, this writing speed seems to be a little far from the nominal value And it drops a little bit in mixed reading and writing This may indicate that the performance of this disk under complex working conditions is average Finally, it is the P44Pro as the flagship disk The results are far ahead without accident In fact, the 512G is not a full blood version But it should be enough for testing Then we will increase the test file to 64G Under the condition of increasing pressure, The speed of all three disks decreased slightly Among them, the random reading and writing speed of Jintek TP3000 There are more declines On the whole, it is still in line with expectations Then I filled them all to 85% capacity Simulate the situation near the full plate The performance of MX500 is very stable and has little change TP3000 has a large decline in 4K reading P44Pro has a significant decline in mixed reading and writing Finally, the full disk + 64G high voltage test All three discs have a certain drop speed The MX500 is fairly robust TP3000 is directly returned to its original shape Read less than 400 There are only two digits written It's totally different from when it's empty This result is not as good as a better USB flash drive As for P44Pro The main reason is that the writing speed has a relatively large decline Other projects are relatively controllable After reading the results of CDM Run another HD Tune Look at the cache 100G file test The first 32G of MX500 can be maintained at nearly 500MB s After slowing down, it dropped to 400 Then enter the state of writing while recycling The writing speed is about 320MB s In the full plate state Cache capacity and speed change little TP3000 in empty disk state The simulated SLC cache is very large We had to run the 150G test It maintained a write speed of 1300MB s in the first 110G And then struggled for a short time It directly falls below 100MB s This … It's no surprise, is it It is worth mentioning that After such violent writing, Its reading speed will also be affected It was only 300MB s at the beginning And the results close to the full set are a bit ugly The SLC cache is reduced to only 25G After writing, it also plummeted below 100MB s Then the reading speed is not very good The situation of P44Pro is a bit special Its speed is very fast SLC Cache under empty disk is nearly 100G There is also a writing speed of about 700 But the results don't seem to be very stable There are many serrations In the full disk state, the cache will be reduced to about 13 G There is no change in speed On the whole, The theoretical performance of these three hard disks is really quite different P44Pro is a high-performance solid state in the first echelon MX500 Speed Stable But limited by SATA bandwidth The upper limit is very low TP3000 can also have the level of entry-level PCIe solid state when it is empty But when the pressure is great, it is a bit terrible So the three hard disks with such a wide gap in running points How different is the experience in actual use? Let's go to the application test When we switched from the mechanical disk to the solid state, The somatosensory gap is the most obvious It is a startup speed several times different So let's do a boot time test first Adopt Win11 pure version system Then install common software such as WeChat, QQ, WPS and steam And set the boot boot Record the average time of multiple boots Note that this is only for testing It is definitely not recommended that you set so many startup items As a result... The P44Pro is undoubtedly the fastest The TP3000 is 10 seconds slower than the P44Pro This gap is bigger than I expected It is estimated that too many startup items have opened the gap MX500 is at the bottom as scheduled However, compared with TP3000, it is not much worse It took 54 seconds Compared with the mechanical disk, it takes more than 3 minutes The promotion is still obvious But during the test, I found a problem That is, the startup time of MX500 fluctuates greatly The other two discs There is a difference of two seconds at most And the fastest and slowest boot of MX500 There is a difference of 14 seconds This also led us to have to repeat the test 10 times To get a relatively representative average In the case of near full plate, The boot time of the three hard disks is basically unchanged TP3000 is even 1 second faster It may be caused by random fluctuations In short, the remaining capacity of the hard disk should have no effect on the boot speed In addition to the boot time, Usually, it is most likely to encounter hard disk bottlenecks It is the scene of file copying That is, copy and paste This is also the favorite test used by merchants when promoting First, look at the copying of a single large file Write using a 48G compressed packet In the empty disk state All three players are running in the cache As a result, the MX500 takes two minutes TP3000 took 37 seconds P44Pro only takes 12 seconds Almost instantaneously completed The results of the three are similar to the running speed of the previous theory But by the time it was nearly full, The situation becomes very different The first is this TP3000 It took almost 7 minutes to finish copying The average speed is not even one tenth of the empty disk You can see it in the actual test This plate dropped directly to the level of 60MB s after slowing down This speed is not as fast as a mechanical disk Even QLC is just like this The speed of P44Pro is also a lot slower after slowing down But there are still about 700MB s The actual time is extended to 50 seconds MX500 is still as stable as an old dog Although it is not fast But the results have not changed much compared with the empty set In addition to a single large file such as a compressed package, Copying multiple small files It is also one of the scenes that are often encountered at ordinary times Here we test it with an unzipped Stable diffusion integration package It contains a large number of fragmented files with only a few KB The speed of copying this time is obviously much slower than that in front Even the fastest P44 Pro often drops to single digits In this scene, The gap between the three hard drives has narrowed a lot The MX500 is less than 20 seconds short of the fastest P44Pro Considering that P44Pro is written sequentially or randomly, They are several times that of MX500 I guess the bottleneck here It should be mainly on the reading speed of the copying disk 980Pro The performance of TP3000 is quite outrageous It's actually a little slower than the MX500 You know, this is an empty disk TP3000 won't slow down at all Moreover, all the running points are obviously rolling MX500 The situation is similar near the full plate All three have a small slowdown The ranking has not changed In general If you often copy large files, The high-end solid state It can really bring immediate results But if it's a small file, The gap is less obvious There is also an "interface bottleneck" that needs to be considered here In actual use It is less to copy directly between the hard disks in the computer More often, it is an external mobile hard disk Or copy files between NAS servers Then here comes the problem The most common USB 3.2 5G rate interface at present The actual speed is more than 400 M 900M for 10G The 2.5 G network port is 300M Unless it is a hard disk such as TP3000 that slowly drops into slag Otherwise, most of the time Bottlenecks are all on external interfaces Of course, if you are the boss of a full set of lightning equipment, Then forget I said it Finish copying the file Let's have some more relaxing projects Measure the loading time of the game Here I am divided into two parts One is the time to enter the main interface of the game The other is to load the archive Time for reading pictures The first is Assassin's Creed: Illusion Start by starting the game Crazy press space to skip animation As a result, the speed of the three hard disks actually appeared upside down The TP3000 became the fastest The MX500 came second P44Pro is at the bottom instead But the gap between the three is really very, very small No impact In the case of near full plate, The MX500 will have a slightly noticeable slide TP3000 and P44Pro are almost identical There is no difference at all In terms of archive loading, Under the empty disk is P44Pro > TP3000 > MX500 But it was only 1 second away at most The gap opened to 2 seconds under the full set It can only be barely perceived In addition, we all tested the time when the archive was loaded for the first time after entering the game If it is loaded multiple times There will also be caching to play a role The gap between the three disks will be smaller Let's change the game Horizon 5 In terms of game startup, It may be because there are too many animations that can't be skipped There is almost no difference in startup time between the three hard disks Only when it is close to the full plate The MX500 will be a little slower Archive loading opens a little gap TP3000 is still the fastest under empty disk P44Pro is next There is a difference of 2 seconds between the fastest and the slowest Under the full plate, MX500 and TP3000 both lost speed P44Pro became the fastest loading But no matter how you measure it, These three hard disks are in terms of game loading Can't open the quality gap I know that many viewers hoard hard drives in order to install more big games This result is good news to some extent If only for the purpose of using the game library That's enough to buy an entry-level or mainstream price hard disk It is well known Game loading is almost a pure read scenario Very little writing So are there any things that need to be read What about the scenes that need to be written in large quantities? In fact, this is the case with file decompression We may not pay much attention to the decompression time at ordinary times But in this scene, Hard disk bottlenecks are often very obvious We will test it with the 48G compressed package copied earlier In the empty disk state The SATA-rate MX500 is significantly slower than two PCIe hard drives It took almost 6 minutes to decompress Although P44Pro is still faster than TP3000, But the lead is much smaller than when copying large files before Estimated at PCIe 4.0 speed CPU and memory become more major bottlenecks And when it comes to the full plate, The situation has reversed again TP3000 became the bottom And it takes nearly twice as long as the MX500 The main reason is that the write is too low after bursting the cache Writing from the third minute is only fifty or sixty TP3000 also takes up more than the other two hard disks when decompressing Almost always 100% I feel overwhelmed The other two hard drives The results of MX500 are basically unchanged Although it is a little slow, it is stable P44Pro has increased the duration by 50% It's acceptable In short, if you often encounter file decompression in your work, That high-speed, stable hard disk It can still improve a lot of efficiency (compression is mainly the CPU bottleneck) Now that the work purpose is mentioned, Then we have to test the content of the retouching and cutting video The first is PS We used the 8.8 billion pixel galaxy map released by NASA for opening test There are 24 G's in this project file alone The result is really strongly related to the performance of the hard disk The MX500 takes a minute to unfold TP3000 takes 21 seconds The P44Pro takes only 14 seconds In the full plate state MX500 is basically unchanged TP3000 extended by 13.4 seconds P44Pro performance is still stable It was only extended by 1.4 seconds Next is the video test Here we use a large project file from a previous period to open it Because of the randomness of the results, Therefore, this item also adopts the method of taking the average value five times As a result, although the three hard disks are different, But the gap is not big The fastest P44Pro compared to the slowest MX500 It's only less than 7 seconds faster The situation at the full plate is similar The difference between the three disks is only three or four seconds And compared with the empty disk, the change is not big During the test, we also found that When dragging the timeline, The hard disk will also have obvious actions Theoretically, there is a relationship between hard disk speed and fluency But it is difficult to measure accurately in this scene In our actual experience, There is no obvious difference between the three disks In fact, if you really want to do more serious content creation, That compares to a hard disk Other hardware is more prone to bottlenecks In the previous test, We all use 32G of memory But when it comes to PS and PR, 32G memory will be directly full I had to change to 64G to run out of the hard disk gap In fact, everyone can understand it this way When we use it, we often read the data from the hard disk into the memory Therefore, scenarios with high hard disk demand generally have high memory requirements When matching, you must balance it Otherwise, the barrel effect will be obvious All right, after measuring so much data, Finally, we can give a relatively accurate answer to the first question In actual measurement The items with the biggest performance gap among the three hard disks are large file copies Compression package decompression and PS large-scale engineering opening If you have a need for this continuous large-capacity reading and writing situation, It's really worth spending more money to upgrade the hard disk While small file writing, boot speed, and video project opening belong to "a little improvement" But not much " If the budget is sufficient You can really buy a higher-end model Don't try to be brave when the budget is short But cutting video may be a special case After all, there is a need to copy materials Still need to keep a little bottom line As for game loading... It can only be said that the gap between hard disks is too small Unless your other configurations have been pulled to the top Otherwise, there is no need to spend more money on the hard disk for game purposes In fact, in addition to the data shown above, We also tested some other projects For example, software installation time Large EXCEL table opens and so on But these scenarios are either too random Or there is no difference at all After changing from the mechanical disk to the solid state, The improvement of hard disk in actual experience has encountered a bottleneck In most cases Top-level and entry-level solid state There is really no obvious gap So I suggest For the average consumer Buying some mainstream speed hard drives is enough At the time of purchase Mainly pay attention to avoid two pits One is the hard disk with various unknown particles and unstable firmware This kind of disk is mainly a data security hazard Among my colleagues alone, there have been several cases of losing orders The second is a disk with too weak external performance like TP3000 in this issue In some projects, it can't even beat the SATA rate MX500 According to my experience in testing computers for so many years, Choose PCIe3.0 model of first-line big factory, mainstream level and time-tested You can basically avoid the pit above More than 90% of the demand has been met The money saved can upgrade other more important configurations Change the peripheral that takes advantage of it Or go out for a good meal with your friends The utility obtained is much higher than that of upgrading the flagship disk Ok, that's all the content of this video This issue really cost us a lot of time Light is repeated test and frame-by-frame recording time It took nearly two weeks So if it helps you I hope I can have a long press and praise One key and three connections support us This is really important to us If you want to interact with me every day You can pay attention to the official WeChat official account of the pen bar evaluation room Exchange the experience of buying this book to help computer Xiaobai This is the pen bar evaluation room I am the King of Pigs See you in the next video
