Do we need them?

Comparison of HDD and SSD drives in terms of ease of use. Part 1

Introduction

Let's move away from the reviews of the laptops themselves for a while and turn to their components, namely ?? storage devices. Until the last moment, drives on hard magnetic disks, aka "Winchesters", reigned supreme here. However, relatively recently, they have a strong competitor ?? flash drives, SSD (Solid State Drive).

SSD is a fundamentally different type of drive, it is built on the same technologies that are used in flash memory, and is similar to flash drives in organization and cells, and the drive as a whole.

Detailed information about the speed and functional characteristics, as well as the results of testing modern drives can be found in the following materials website:

• Overview of one of the first SSD drives,.
•   in which participates SSD Corsair. This SSD-drive participated in our tests.
• The latest at the time of testing fast SSD drives and the study of the effect of SSD capacity on performance.
• Other materials on the performance of SSD and hard drives can be read on the site.

At the same time, most testing is designed for savvy readers and is a comparison of the performance characteristics of selected drives. And although they contain a lot of interesting information about specific products, a large number of drive properties (especially those that are difficult to measure unambiguously) remain behind the scenes. Therefore, the potential buyer can not always determine whether he needs a particular device.

In this series of materials we will try to move away from the traditional method of testing drives (you can see its description on our website) and focus on subjective impressions of use. First of all, this study should answer the question: what does a regular user get from switching to SSD, what are the advantages of a new type of drives in everyday work, is it worth switching to them, or is it better to stay with traditional hard drives? And in what cases these or other drives are more profitable.

Basic storage requirements

Any user has two basic requirements for a storage device: reliability  (so that you can not be afraid for the safety of their data) and speed. Of course, there are other requirements, but they play a minor role and are unlikely to be taken into account if reliability or speed are unsatisfactory.

Reliability ?? key requirement, the importance of which can not be exaggerated. Losing a laptop is not so scary: in the store you can buy the same. But if you lost your main laptop with the entire personal archive or the hard disk failed, then everything is much sadder: you lose unique information, which is often impossible to recover. It is obvious (and has long been emphasized in all presentations) that information in a corporate laptop can cost several times more than the whole notebook with giblets. However, the preservation of information is not only important when it comes to business secrets: there is still the concept of subjective value. It is difficult to evaluate your photos or documents in money, but they mean a lot to the author. Of course, there is backup, online storage, etc., but not always using them is possible and convenient.

With this reliability of storage systems for laptops ?? very difficult and painful question. Due to the design features of hard drives are afraid of vibration and shock. When working head soars very close to the surface of the magnetic disk. A shock or shaking can cause it to touch the surface and either damage itself or scratch the surface. data in this location will be lost.

And with laptops this happens all the time. Got hooked on the wire ?? and he flew from the table or sofa, worked on his knees and dropped, even a simple shake-up could damage the device. Very often, careless or unqualified users themselves shorten the life of their disks. To take at least a typical example, when a user, holding a laptop on his lap, presses the hibernation button, the screen goes out (for some reason, this happens in new Windows systems, although XP showed on the screen that the hibernation process is still underway) and the user is completely sure that the system has shut down, throws the laptop on the sofa ?? and at this time the system intensively writes the state of the operating system to disk.

Most manufacturers in corporate models (where information integrity is the most important factor) began to introduce active protection of the hard disk, which should park the heads (take them away from the surface) if the laptop jerked or struck. Manufacturers in the development of new models of mobile hard drives are trying to make them more resistant to external influences. However, this stock is not always enough.

The second most important requirement ?? speed  drive operation. And here it should be noted that modern hard drives (especially mobile) are already close to the ceiling of their capabilities. A radical increase in the speed of work is not to be expected, one can only hope for some evolutionary growth, and even then ... Besides, due to the design features, a hard disk can not always work at maximum speed. First, the speed of reading and writing data strongly depends on whether it’s the beginning or the end of the disk, and secondly, although with linear reading or writing (when a large amount of information is read and written in a row), the disk can provide good speed, but when working out of tune "speed drops to indecently small values, 1-2 MB / sec. And most often the main hard drive of the laptop works in this mode. Therefore, for example, laptops are loaded for a long time: you need to read many small operating system files from different places.

SSD is a fundamentally different type of device, so most of the shortcomings of the HDD are unusual for it. Briefly recall the main consumer benefits SSD:

• High read and write speed, the same anywhere in the drive.
• At times lower delays when working with data compared to hard drives.
• Lack of moving parts: SSD is not afraid of shaking, vibration and shock, i.e. less likely to lose data.
• SSD is not heated, does not make noise, does not vibrate itself.
• Lower power consumption.
• Large operating temperature range.
• The best weight and size indicators compared with the hard disk (the drive can be made smaller and lighter).

The main disadvantages of SSD:

• Very high price.
• Limited capacity.
• Dependence of price on storage capacity, high cost of additional capacity.
• Perhaps the limited life of the memory cells.

Let's try to assess how much these pros and cons of the SSD are weighty in and of themselves and in comparison with modern hard drives during their regular work.

Testing sections

The main objective of our testing ?? Understand the difference in operation between an SSD and a regular hard drive. First of all, it concerns the speed characteristics: it is interesting to see how much the difference in speed between the hard disk and the SSD drive is noticeable in the usual work of the laptop user. However, our testing is not limited to this.

All testing is divided into four large parts. In the first part, we talk about testing participants, methods, etc.

In the second part ?? We will look at the performance of the test participants in synthetic applications, and we will also estimate by the example of one of the participants how much the workload of the operating system is affected by data and third-party programs.

In the third part, we will compare the performance of test participants in real work. These are the basic operations associated with the operation of the operating system (boot, shutdown, entry and exit from hibernation), as well as the speed of copying files. And on a clean system, and on a system with installed applications. In addition, we will look at such an important parameter as the speed of copying files.

Finally, in the fourth part, we summarize the subjective sensations of using SSD and HDD during normal work on a laptop. Plus, we compare parameters such as heat and noise, as well as battery life.

However, even this will not end our testing. For I have both drives, an operating system with a set of applications (this is my working system, so it is constantly in work and is gradually degrading), as well as software for cloning. So it is possible to return to the tests at any time and at the same time to see if the system performance will deteriorate after a long time (there are persistent rumors about this). Therefore, we invite readers to actively participate in the discussion, ask questions, offer their own tests and point out moments where a particular type of drive is different for the better or, on the contrary, the worst.

Test participants and methods

It should be noted that fate has made some adjustments to the testing program. Initially, we planned to compare six drives: four hard drives and two SSD drives. However, in the middle of testing, our test bench broke, so only three drives, but the most interesting, took part in the testing core. If our readers have a great interest, you can try to test other drives using a similar method.

So, in testing participate:

Seagate Momentus 5400.6 with a capacity of 500 GB;
   Seagate Momentus 7200.2 with a capacity of 160 GB;
   SSD CORSAIR CMFSSD-128GBG2D with a capacity of 128 GB.

Let's look at the characteristics of the test participants in more detail.

We’ll dwell on this now: we have a capacious disk, there is a fast disk and there is an SSD drive with good performance: not top, but close to it.

Testing method

All tests were conducted on a laptop ASUS K52Jr. The laptop is relatively modern and fast, built on the new Intel HM55 Express chipset.

For testing, we took the standard configured Windows 7 system that comes with the ASUS K52Jr with installed drivers. Only programs (such as Norton, etc.) have been removed from the system. I want to draw attention to this. Theoretically, in the updated versions of the drivers, the performance of the SSD could have improved. However, we decided to take the drivers from the package (especially since the model is relatively fresh). So you can "fix" the state of the system so that all drives work in more or less the same conditions. In addition, not all users (especially those who work on a laptop and not experiment) love games with drivers.

Synthetic tests

HDD Tune 4.01 and Ashampoo HDD Control were installed on a clean system, with which we evaluated the performance of the drives in synthetic tests. It should be noted that the tests were done in order to understand what approximately can be expected from the drive. They were not decisive.

For HDD Tune, we did a basic test of reading from the surface of the disk. I want to draw the attention of readers that these results can not be completely objective, because The test was performed on the system disk on which the operating system is installed. The system and applications may well work with the disk directly during the test, with the result that the results may differ on different runs and will not be completely objective and repeatable. For the same reason, there is no test record. Full-scale testing can be found in other materials on our website, where they are carried out in accordance with an objective and comprehensive methodology.

In addition, we decided to remove additional tests in HD Tune: additional tests for searching and reading, tests for random reading in the file system. Finally, we measured how long a program takes to scan a disk and display its structure (the number of folders and their size).

HDD Control was used more as a program to check the results.

We also paid attention to the temperature data of the drives obtained during the test. Except, of course, SSD, which is not heated.

The real tests of the operating system

We decided to see how big the difference is in the use of different types of drives during normal work with a laptop. For this, several test groups were conducted.

Operating system

At first, we measured the speed of the operating system for performing basic actions: boot, hibernate, exit hibernate and shutdown(in that order).

Judging by the sensations, Windows 7 is optimized much better than the previous Vista. In particular, in everyday work, you constantly feel that it is much less “twisting the disc” during work and in its absence. What does Vista do with it ?? a separate question, because a laptop can, standing in absolute idle time, thresh with a disk for half an hour, which has a great effect on performance, and on the response speed of applications, and on battery life. Plus, as it seemed to me, the new system is less “slowed down” with active work with the hard disk, i.e. With the active work of the system with the hard disk, you can continue to work, the open application responds. Although sometimes there are situations that cause some irritation.

We tried to measure the time when the system was completely “loaded”, i.e. stopped loading data from hard disk. When you start Windows 7, it completely turns on the desktop and shows with its whole appearance that it is ready to perform tasks, but the hard drive continues to work, adding something. When measuring, we tried to take into account this time. Although at the start the situation is more or less decent, after the system is completely “ready” (the hourglass disappears next to the cursor), the disk spins for about half a minute, but when it leaves the hibernate this process continues for two or three minutes. You can work (I specifically checked), but, for example, new applications run with difficulty.

Speaking about the start of the system, it is worth noting one important point. When booting the hard drive is constantly actively working, and it seems that the download speed depends only on him. However, in a system with SSD from time to time the disk access indicator goes out, i.e. data loading is not 100% of the operating system load time.

Data copy speed

Copy and transfer data ?? important characteristic, and in many ways (if not all) it depends on the drive.

For this part of the test, we prepared four groups of files.

First, a 700 MB movie (folder size is 734,486,528 bytes). Secondly, a set of zip files ?? a set of drivers for ASUS laptops. Folder size is 811 742 316 bytes, inside 53 files, from 2 MB to 102 MB. Thirdly, and this is the most interesting ?? set of documents. Folder size is 943,813,860 bytes. This includes MS Word documents and saved web pages (each of them has a folder with graphic files used on the page). All files are very small, from 2 KB to 40 KB. As a rule, such files are the most difficult to copy, because they are located on the disc "discordant" and copying takes a long time.

Fourth set ?? folder with a capacity of 4,532,507 KB. Inside ?? 24 rar file. We took a 4.5 GB archive and archived it into a multivolume archive with a 200 MB volume. In addition to the tests for copying, we used it in the test for unarchiving.

For tests, the disk was divided into two sections, approximately equal in capacity. Next, we copied the files from drive C to drive D and back. Those. Files were copied within the same drive, and it worked simultaneously for both reading and writing.

We also measured the time during which Total Commander erases files (a set of documents) from drive C and D. It should be noted that large files are erased very quickly (which is understandable) and there is nothing to measure, only erasing the document folder is of interest. I also want to note that the standard Conductorwhich erases only the information about the location of the file, almost instantly reports about the removal of anything.

In addition, we created a 1 GB virtual disk in RAM and tried to copy files to and from it. In this test, the drive works only for reading or only for writing, so, theoretically, it should show the best results.

Finally, we tried to see how the unzipping process depends on the hard disk. To do this, we have unzipped a large file from a multivolume archive on drive C.

Tests on the working system

After completing the tests on a clean system, we installed a large amount of software commonly used during operation on the disk. This includes antivirus, office applications, applications for working with a mobile phone, organizer and much more. The applications are fairly standard, plus I tried to pick up several applications that have an “agent” that starts with the system and works with it all the time.

After installing the applications, we once again measured the time required for the basic actions of the operating system. We also measured the time of copying again.

After that, we tried to install two games on a disk (Crysis Warhead and H.A.W.K.). We also indexed the music collection using Helium Music Management and opened the photo collection using XnView.

Finally, we measured the launch time of some applications, for example MS Word.

And in the final, we measured the time of the “parallel start”. To do this, we turned on antivirus checking, started the process of copying files from D to C, and started MS Word to see how difficult it would be to start in this mode.

I note that the OS tests (start, shutdown, etc.) were measured by a stopwatch, so that a small error is possible there. Time has been measured since the start of the laptop, i. including the time spent checking and starting the BIOS. Here I want to note that the presence of a disk in an optical drive greatly increased the BIOS loading time (20 seconds instead of the usual 4), tests were performed without an optical disk in the drive.

Testing in the OS was carried out through the recording of actions from the screen, then we watched how long the action took on the timeline of the recording program and rounded up to whole seconds. Anticipating the objection that this program could also work with the disk, I will answer: yes, it could. Just like any other resident program. Since this is a working system, not synthetic testing, the additional effects, which are more or less stable, help illustrate the work in real conditions more clearly.

Well, let's move on to the tests themselves!

Synthetic performance

Synthetic tests were needed first of all in order to place the applicants, to see what to expect from them in greenhouse conditions. In addition, if there was something wrong with one of the drives, it would be revealed already at this stage.

First, let's look at the easiest test ?? reading from the surface. First of all, we decided to quickly see how things are going with repetition, i.e. whether the test will give out similar numbers at several starts. Between restarts, the system was overloaded, but the article did not present all the diagrams obtained. So...

And now we will try to remove the data after some time, when the disk has thoroughly worked under load.

As you can see, the two tests give very similar results. However (just an example, why not test on the system disk) ...

With some launches such failures appeared. Either these are accesses to the system disk, or disk problems due to overheating (look at the temperature, it is very high). Finally, we shot the results the next day (but also at the end of the test):

Thus, the results are fairly stable (only the reading indicator from the buffer jumps strongly). Although in general, this disc did not show good speed even for notebook drives. Also worth noting is the very high temperature, which can even lead to bad consequences for the disk.

Let's look at the graphs of the second program:

As you can see, the second “run” is worse, the search time was particularly affected. Because of the heat? Let's see what the next day:

The graph is relatively even, there was a failure, apparently, the system at the time of the test turned to the disk.

Thus, in general, the repeatability of HDTune 4.01 is good, while HDD Control is worse. Later we drove the test three times and chose a picture without failures. Except for 5400.6, where the second run failed.

What do we see? 5400.6 gives a much higher linear performance figure compared to 7200.2. The access is worse (on average, in tests it was 18.5 ms), which is generally understandable. Hence the conclusion: in linear operations, 5400.6 is noticeably faster, in random read and write operations (just small files) it may be slower ... Or it may not be. We'll see below how this is done in real life. For now we will recheck results in other program.

"Parallel counting" confirms the data. Those. we assume that the linear speed of 5400.6 is faster. Moreover, it is also much more capacious, i.e. the probability that the entire system partition will be in the fastest part of the disk is higher. Although it should be noted that the variation in random reading at 7200 is smaller.

And finally, let's look at the star of today's testing: SSD Corsair.

As the saying goes, "started." Let me remind you that in the characteristics of this utility it was that the active mode of UDMA-5, the results of which we observe. From this, at least, we can conclude that it is not always enough to buy an SSD and insert it in place of the hard disk.

Check the second program:

As you can see, here the speed is shown much higher ?? and closer to the stated figures.

It remains only to repeat the conclusions of almost any article. In benchmarks, showing maximum performance, in greenhouse conditions SSD goes very far forward in performance from traditional hard drives. At the same time, again, unlike hard disk drives, it can maintain the same high speed anywhere on the disk without “dropping” by the end. The third conclusion: the SSD has a very short access time, i.e. The drive responds instantly to commands from the operating system. This option is almost an order of magnitude better than hard drives.

However, these findings have long been known. Let's see how things are with more detailed tests.

Extended Random and Linear Tests

Let's start with 5400.6.

We have already seen the second four indicators: linear reading at the beginning, middle and end of the disk, plus reading from the buffer. It is much more interesting to look at the beginning of the table. The first two numbers characterize a random search, and the picture is bleak. It is in this mode, when the disk constantly has to move the head and look for a small piece of information, and then move the head again, performance suffers the most. This is evident from a huge drop: the read speed is less than a megabyte / sec. And the next two numbers show the difference between working with small and large blocks.

Let's compare the results with 7200.2.

As you can see, when searching for a positioning, heads are slightly faster, which gives a tiny advantage. As soon as a large block appeared in the work (i.e. it took at least a little to count in a row), the 5400 instantly took the lead, and his advantage was quite significant. With consistent reading, everything is clear and true.

Now compare hard drives with SSD.

In absolute numbers, SSD is far ahead. Access times are always the same (except in one case, but this test was also not easy for hard drives). With random reading, the speed also fails very much, although it remains very high compared to hard drives. We have already discussed linear reading, there are no surprises here and in general cannot be.

As you can see, SSD is much ahead in speed. However, there is a drop in performance when working with small blocks and volumes of information, and it is also very large. The SSD remains in top positions, but the absolute numbers are a bit disappointing. By the way, note that with linear reading, the results of access time are not so different. There is nothing surprising, but still pay attention to it.

Random access operations

Since we started about the access time, let's take a closer look at the corresponding test. We begin, as usual, with 5400.6.

The test was removed when the hard disk was already warm enough (37 degrees Celsius, i.e. + 5 ° relative to the idle temperature). I want to note that in this mode the number of sectors is growing, access to which takes a long time, and this happens with both disks.

The graph looks different, because there is a different price of division, and on both scales. If you look at the numbers, then with small blocks of 7200.2 it is faster (just 14.5 milliseconds versus 18.5), but with a block size of 1 MB it is already losing. And what about SSD?

With such operations, hard drives up to SSD as before the moon. What is the speed of reading, what is the access time. The difference is just at times. Since the SSD reads information in blocks, the larger the block, the larger the gap grows. In numbers, this drive is far ahead, but it doesn’t like small blocks just like ordinary hard drives. Those. it will work faster with respect to the disk, but it’s an equally sad picture with respect to its maximum results.

Work with file system

Let's look at the work in the file system, i.e. benchmark, a little closer to real life.

For comparison, let's see what the result is 7200.2

It can be seen that 5400.6 is everywhere faster. Let's see how things are in this test for the SSD.

In SSD, the graph is smoother and the characteristic features become apparent: dislike for small blocks and a stable lag in the write speed from the read speed. The overall speed is very high, much higher than the disks. I also have to say that the test results of the SSD are smoother, there is no such spread from launch to launch, like hard drives.

Preliminary findings

The obvious conclusion: the SSD speed is much higher, and the access time is much better. However, these are not new conclusions at all; they have been repeated in one form or another for quite some time. Of course, the SSD has its own characteristics, which, by the way, can be read in objective testing, the link to which we cited.

The mutual results of the hard drives surprised me: I did not expect such a lag 7200.2. Although it is clear that the model is older and the recording density is lower (this has a bad effect on the speed of work with the disk), nevertheless it seemed to me that it should overtake 5400.6. In practice, 7200.2 is almost everywhere inferior in speed in absolute numbers. In addition, this is a very important operational fact: it heats up much more strongly, i.e. use it is not so nice. 5400.6 retains moderate heat. SSD, by the way, is not heated at all, but since there are no temperature sensors (and they don’t need them), then this is subjectivism and we will leave it for the part with subjective measurements. Thus, when choosing a drive for a laptop, you need to be guided by the freshness of the model, but is the rotational speed ?? not a very important indicator.

On this we will stop. And in the next part we are waiting for tests from real life ?? time to start and shut down the system, copy files, launch applications and much more.

Hi admin! I decided to buy a solid-state drive the other day! I came to a computer store and said to the seller:

Sell ​​me the fastest SSD!

and they answer me:

Here you are, Kingston HyperX 3K (120 GB, SATA-III) speed 555 MB / s, excellent SSD, does not happen faster.

Prove it!

Apparently they wanted so much to sell me this SSD, that they installed it on a computer and ran a test in the CrystalDiskMark program, then showed the result of the test, here is the screenshot:

The speed of sequential reading is 541 MB / s and writing 493 MB / s, I even took it to the phone.

In short, I bought this SSD, came home connected to my computer, then downloaded and launched the program "CrystalDiskMark" and performed the same test, but the result was worse!

489 MB / s sequential read and 127 MB / s write speeds. Why?

In the store, the test was performed on a computer with an Intel® Core ™ i5 processor and 4GB of memory, my computer is more powerful and is based on the Intel® Core ™ i7 processor and has 8GB of memory.

Explain the admin, what's the catch, but I won’t sleep, because this SSD costs 3 and a half rubles.

Hello to all! Yes, it can be friends, you just need to be able to use the program CrystalDiskMark. Now I will show you everything.

• Note: You may be interested in our other articles on SSD solid-state drives.

SSD test will be carried out in the program CrystalDiskMark 3 0 3

The program can be downloaded on the official website. http://crystalmark.info/download/index-e.html

CrystalDiskMark tests our solid state drive in this way:

All: All 4 tests are carried out (Seq, 512K, 4K, 4K QD32);

Seq: Sequential write / read test (block size = 1024Kb);

512K: Random write / read test (block size = 512Kb);

4K: Random write / read test (block size = 4Kb);

4K QD32: Random write / read test (block size = 4Kb, queue depth = 32) for NCQ and AHCI;

The final result.

First, test your SSD or any other hard drive correctly! The fastest SSD will read and write information to a section filled with zeros. To do this, in CrystalDiskMark, select the file from the menu File-\u003e Test Data-\u003e All 0x0000 (Fill).

I also have this SSD Kingston HyperX 3K SSD (120 GB, SATA-III) and now I will do a simple test.

In the operating system, the SSD drive under the letter D :, then in the program settings, select the letter D: and click

Begins the test of our solid-state drive for sequential read and write speed!

In a minute we get the result. Sequential read and write speed 543 MB / s (read), 507 MB / s (write)

Now we test differently. File-\u003e Test Data-\u003e Default (Random)

After a minute, we get the result is not at all the same as when testing the All 0x0000 variant (Fill). Sequential read and write speed 499 MB / s (read), 149 MB / s (write)

It is also important for a good SSD to properly connect it to your motherboard. All solid-state drives have a high-speed SATA 3.0 interface (6 Gb / s), and your motherboard probably has such connectors. For example, my ASUS P8Z77-V PRO motherboard has four SATA 6 Gb / s ports and they are appropriately labeled SATA 6G, which means we connect the SSD according to the marking.

To connect the SATA 6 Gb / s SSD interface, use the native SATA 6 Gb / s data cable!

And the last. For normal operation of the SSD, the disk controller must be enabled in BIOS AHCI.

Useful article on this topic: