Many mistakenly believe that the rapid memory is easier to install it, it is not necessary to set it up, but it makes no sense to force it. In fact, everything is much more complicated and now I will tell you in the form of questions and answers how to squeeze the maximum performance from RAM.
The editors thanks the company and, kindly granted memory sets and motherboards for testing.
Is it possible to combine memory of different models, brands and frequencies?
In the theory for PC, you can use several RAM modules not only from different manufacturers, but also with different frequency. In this case, all memory will work at the frequency of the slowest module. But in practice, incompatibility conflicts may arise: the PC may not be launched at all, or periodic OS failures may occur. Therefore, the RAM is better to immediately buy a set of two or four modules, especially if you plan to engage in overclocking. In modules from one set, chips from one batch, which have identical acceleration potential are used.
How good is the multichannel memory mode?
All modern Intel processor platforms and for desktop PCs support at least two-channel memory mode. In turn, processors Intel Core I7 Gulftown and Intel Xeon Nehalem and Westmere support three-channel mode, and AMD Opteron 6000 series, Intel Core I7 LGA 2011 and Xeon E5 and E7 - generally four-channel (eight memory slots).
The processor two-channel memory mode adds from 5 to 10 percent of performance, while the integrated graphical accelerator is up to 50 percent. That is why when assembling on aMD processor A8-7600 With built-in Radeon R7 graphics, we recommend using two memory modules.
If there are only two memory modules and the motherboard with four DIMM slots, it is important not to be mistaken with the order of installation. So as to use two-channel mode, the modules must be stopped into the connectors through one, that is, the first and third or the second and fourth. More universal is perhaps the second option, since the first slot can overlap with a large processor cooler, like that. However, for memory and with low-profile radiators it is not a problem.
Check if the memory really earned in two-channel mode, you can use the AIDA64 application (Menu item "Test Kesha and Memory"). The same program will help measure the speed of memory before and after overclocking.
How to configure frequency and memory timings?
Immediately after installation, the RAM often operates at its minimum frequency, or at a frequency that the processor officially supports. For example, 2400-MHz Hyperx Savage. on the intel processor Core i3-4130 by default earned at a frequency of only 1600 MHz. You can set the maximum memory frequency in the settings. Bios maternal Cards: either manually or using Intel XMP technology (even supported by AMD motherboards).
If you manually select 2400 MHz, then the memory will work with standard timings for this frequency (delays) 11-14-14-33. But in practice, Hyperx Savage can work stably at the same frequency at smaller timings. But it is precisely the ratio of high frequency and low timings guarantees high speed speed.
In order not to select the value of each timing manually, intel. Developed technology called Extreme Memory Profile. It allows you literally in two clicks choose the optimal memory profile, prepared by the manufacturer. So, our version of Hyperx Savage supports two XMP profiles: 2400 MHz 11-13-14-32 and 2133 MHz 11-13-13-30. The first is relevant, for example, for the motherboard with support for overclocking memory to 3300 MHz, and the second - for a motherboard, in which the frequency of the RAM is limited to 2133 MHz.
How to dispersed memory?
Acceleration of something (processor, video cards, memory) is always a lottery: one copy can accelerate well, the second is exactly the same - bad. Be afraid that the memory during overclocking fails. Do not stand: if you install too high frequency, it simply does not start.
If the motherboard does not have the automatic overclocking settings function after several unsuccessful PC start attempts, you can reset the settings manually using the Clear CMOS jumper (another JBAT name).
In cases of RAM to select the experimental method, not only the frequency and supply voltage, but also timings. Moreover, it is not a fact that it will be possible to choose the ratio better than what is provided for by the maximum XMP profile. In the case of Hyperx Savage, this is exactly what happened: dispersed the memory managed to the frequency of 2600 MHz, but the timings had to be raised to 12-14-15-33.
Aida64 Cache & Memory Benchmark
| 28479 | 24721 | -15 |
| 36960 | 32572 | -13 |
| 31109 | 27343 | -14 |
| 55 | 55 | 0 |
Measurement The speed speed of the above-mentioned AIDA64 Cache & Memory Benchmark program before and after overclocking showed a drop of speed by an average of 14 percent. So the overclocking of the memory of 200 MHz above the nominal turned out to be spectacular in the theory, but useless in practice. But this is in the case of the top 2400-MHz version of the Hyperx Savage, and the lower frequency version, for example, 1600-MHz, the potential for manual overclocking is much better.
conclusions
As you can, correctly install and set up the RAM is not as difficult, especially if it supports ready-made XMP profiles. If you buy a memory kit, then you can get a speed gain not only from a two-channel mode, but also from successful overclocking. And so that there is no incompatibility with large processor coolersIt is better to choose a low-profile RAM, especially if you plan to use the memory slot closest to the processor.
) to 9. The bandwidth of the "Processor" section of the processor is largely dependent and, as a result, the speed of the main components of the system.
An example from practice: a memory system at a frequency of 100 MHz with timings 2-2-2 has an approximately the same productivity as the same system at 112 MHz, but with delays 3-3-3. In other words, depending on delays, the difference in performance can reach 10%.
Mera timing - tact. Thus, each digit in formula 2-2-2 means a signal delay for processing, measured in tactics system Tire. If only one digit is indicated (for example, CL2), then only the first parameter is meant, that is, CAS Latency. The rest at the same time are not necessarily equal to him! Practice shows that usually other parameters are higher, which means that the memory is less productive (i.e. it is a marketing stroke, specify one timing that does not give views of memory delays when performing other operations).
Sometimes the timing formula for memory can consist of four digits, for example 2-2-2-6. The last parameter is called "DRAM CYCLE TIME TRAS / TRC" and characterizes the speed of the entire memory chip. It determines the range of interval during which the string is open to transferring data (TRAS - RAS # Active Time), by the period, during which the full opening and update cycle is completed (TRC - ROW CYCLE TIME), also called the Bank Cycle (Bank Cycle Time).
Manufacturers usually provide their chips based on the memory plan, information on the recommended timing values, for the most common frequencies of the system bus. You can view this information for example a CPU-Z program.
From the point of view of the user, the timing information allows you to roughly estimate the performance of RAM, before purchasing it. DDR generation memory timings attached great importance, since the processor cache was relatively small and the programs often referred to memory. DDR3 generation memory timings pays much less attention as modern processors (For example, Intel Core Duo and Intel i5, I7) have relatively large L2 keachy and are equipped (again relative) a huge L3 cache, which allows these processors much less often to turn to memory, and in some cases the program is fully placed in the processor cache.
| Name parameter | Designation | Definition |
|---|---|---|
| Cas-latency | Cl. | Delay between sending to the memory of the column address and the start of data transmission. The time required for reading the first bit from the memory when the desired line is already open. |
| Row Address to Column Address Delay | T RCD. | The number of clocks between the opening of the string and access to the columns in it. The time required for reading the first bit from memory without an active row - T RCD + CL. |
| ROW PRECHARGE TIME. | T RP. | The number of clocks between the team to the preliminary charge of the bank (closing line) and the discovery next line. The time required for reading the first bit from memory when another line is active - T Rp + T RCD + CL. |
| Row Active Time. | T Ras. | The number of clocks between the team to open the bank and the team for a preliminary charge. Time to update the string. Opened on T RCD. Typically approximately equal to the sum of the three previous numbers. |
Notes:
|
||
Wikimedia Foundation. 2010.
Watch what is "timings" in other dictionaries:
Acceleration, overclocking (from English Overclocking) Increase the speed of computer components by operating them in forced (abnormal) operation modes. Contents 1 Criteria of the standard computer mode ... Wikipedia
This term has other values, see DDR. Types of DRAM memory FPM RAM EDO RAM BURST EDO RAM SDRAM DDR SDRAM DDR2 SDRAM DDR3 SDRAM DDR4 SDRAM RAMBUS RAM QDR SDRAM VRAM WRAM SGRAM GDDR2 ... Wikipedia
Types of DRAM memory FPM RAM EDO RAM BURST EDO RAM SDRAM DDR SDRAM DDR2 SDRAM DDR3 SDRAM DDR4 SDRAM RAMBUS RAM QDR SDRAM VRAM WRAM SGRAM GDDR2 GDDR3 GDDR4 GDDR5 ... Wikipedia
DRAM Types of the FPM RAM EDO RAM BURST EDO RAM DDR SDRAM DDR2 SDRAM DDR3 SDRAM QDR SDRAM WRAM SGRAM GDDR3 GDDR5 DRAM (Dynamic Random Access Memory) is one of the types of computer memory (RAM), the most widely used in ... ... Wikipedia
Screenshot from the test program demonstrating acceleration processor overclocking, overclocking (from English. Overclocking) Increasing the speed of computer components by operating them in forced (abnormal) operation modes. Contents 1 ... Wikipedia
Screenshot from the test program demonstrating acceleration processor overclocking, overclocking (from English. Overclocking) Increasing the speed of computer components by operating them in forced (abnormal) operation modes. Contents 1 ... Wikipedia
This term has other meanings, see latency. This article should be vico. Please refer it according to the rules of articles. Latency (eng. Wikipedia
This term has other values, see DNS (values). DNS Title: Domain Name System Level (OSI Model): Applied Family: TCP / IP Port / ID: 53 / TCP, 53 / UDP Protocol Assignment: Domain Naming Resolution ... Wikipedia
Harvard architecture AUM architecture, the distinguished signs of which are: 1. The repository of instructions and the data storage are different physical devices. 2. The instruction channel and data channel are also physically separated. ... ... Wikipedia
Hello, guests of my blog.
I decided to write an article about what kind of rapid memory timings, as I noticed, few people pay due attention to this parameter when the device is selected. Although it is precisely on it that the capacity of RAM is determined at the same clock frequency and other identical characteristics.
I have already written on this topic, but this time I want to stay in more detail on it so that even novice users know that they mean "incomprehensible" figures indicated on the RAM. Thus, I will start with the azov.
Principle of operation of the RAM
It will be easier for me to explain the appointment of timings if you first realize how it functions. It has a dynamic character, that is, needs a constant supply of electricity. Therefore, with each computer reboot, you lose what was in the cache.
The microcircuit includes cells in the form of capacitors. They receive a charge when recording a logical unit and discharge when the zero is applied. All cells are structured in the form of two-dimensional matrices, and access to each opens by specifying the address of a specific RAS string (ROW Access Strobe) and the CAS column (Acess Strobe).
Their choice is done with the help of a strobe pulse, that is, changes in the level of voltage from large to small. Synchronized with a tacting pulse The activation signal is started alternately: first on the string, and then on the column. If an entry is recorded, then another impulse is given to it - WE (WRITE ENABLE), working on the same principle.
The essence of timing
These parameters show how long the RAM is required to perform certain columns and strings to record information into the cell or read from it. Timings are measured in the system bus tacks. As you understand, the fewer these values, the better.
If you considered the bar, then the designations of the type DDR3 1600 MHz 9-9-9-24 were noted. The last figures indicated through the hyphen show the number of clock pulses for 4 timing. The most important one is, so it can only be registered on the label. However, we will analyze all that you understand what they affect.
Cas Latency.
The first 3 letters are decrypted as Column Address Strobe (strobe signal address speaker). This is the parameter that is indicated in the beginning. It shows the clock delay required by the module to select the desired column in the memory string to read certain information.
I will try to simplify your explanation: cl - this is the time between receiving the command for reading and its execution. Query by this operation Enters RAM from the processor, and to him, in turn, from you.
That is why this parameter is the most important - it shows the speed of operations.
Ras-Cas.
With these two abbreviations, you have already met when I describe the principle of the memory module. Abbreviated this timing is called TRCD. It shows the number of clocks from removing the RAS pulse (selecting the desired string) to the CAS signal (finding a column in the string). In other words, it is a length of time between receiving the "Active" command and performing the "read" command incoming after it (read).
Ras Precharge.
This timing tells about the number of cycles between the signal to pre-charge one line of the data and receiving access to the next one. Simply put, it shows how much time passes between the completion of the processing of one row and the transition to another (from the "Precharge" command to "Active").
Row Active.
You can also meet such a name as TRAS (Time of Active to Precharge Delay). This parameter determines the delay during which one line is active.
Learning the number of clocks of your module
I wonder how many timings has a RAM installed in your computer? You do not have to disassemble it, since the label on the bar is not the only option where you can see these values.
I already wrote about it in, but I repeat for those who missed her. You need to download utility CPU-Z.. When you run it, go to the MEMORY tab and see all 4 timing and even more. Generally, that's useful program, so do not be lazy to install it.
Change timing
If you want to dispersed the RAM, probably wondering if you can change the values \u200b\u200bof temporary delay? Can.
Usually they are installed automatically when you connect the module to motherboard. But going to the BIOS, it is possible to configure them manually. To do this, you need the "Advanced Chipset" section and the "DRAM TIMINGS" option. However, if you are not sure about your actions, it is better to leave an AUTO option, for unpleasant consequences may occur if the settings may occur.
Perhaps someday I will write about this topic in more detail.
Subscribe to updating my blog so as not to miss the most interesting.
Hi, GT! We all love new iron - it's nice to work for fast computerAnd not to look at all sorts of progressbars and other sandy hours. If everything is more or less understandable with processors and video cards: here's a new generation, get your 10-20-30-50% of performance, then everything is not so simple with the RAM.
Where progress in memory modules, why the price of gigabytes almost does not fall and how to please your computer - in our iron liberation.
DDR4.
The DDR4 memory standard has a number of advantages over DDR3: large maximum frequencies (that is, bandwidth), less voltage (and heat generation), and, by itself, twice the tank per module.Committee of Engineering Standardization of Semiconductor Products with Electronic Industries Alliance (more known as JEDEC) works above that your rAM Kingston approached the maternal asus board Or Gigabyte, and on these rules they play everything. According to the parts of electricians, physics and connectors, everything is rigidly (it is clear, it is necessary to provide physical compatibility), but in relation to working frequencies, the volume of modules and delays in the work of the rules allow some volatility: you want to do better - do, most importantly, in order for standard settings Users had no problems.
This is how the DDR3 modules with a frequency are higher than 1600 MHz, and DDR4 with frequencies above 3200 MHz: they exceed the basic specifications, and can work both on "standard" parameters that are compatible with all motherboards and extreme profiles (XMP), tested at the factory and stamped in the BIOS memory.
Progress
The main improvements in this area are carried out immediately in several directions. First, manufacturers directly with memory chips (Hynix, Samsung, Micron and Toshiba) are constantly improving the internal architecture of chips within one process. From revision to revision, the internal topology is brought to perfection, providing uniformity of heating and reliability of work.Secondly, the memory is slowly moving to a new technical process. Unfortunately, there should be no improvements here quickly, as they do (made the last 10 years) manufacturers of video cards or central processors: Rough decrease in the size of the working parts, that is, transistors will require an appropriate reduction in operating stresses that are limited to the JEDEC standard and built into CPU memory controllers.
Therefore, the only thing that remains is not only to "undermine" production norms, but also in parallel to increase the speed of operation of each chip, which will require an appropriate increase in voltage. In the end, both frequencies grow, and the volumes of one module.
There are many examples of such development. In 2009-2010, Normal was the choice between 2/4 gigabytes DDR3 1066 MHz and DDR3 1333 MHz per module (both were performed according to the 90-nm technical process). Today, the dying standard is ready to offer you 1600, 1866, 2000 and even 2133 MHz operating frequencies on modules in 4, 8 and 16 GB, the truth is already 32, 30 and even 28 nm.
Unfortunately, a similar upgrade costs a lot of money (first of all for research, the purchase of equipment and debugging of the production process), so wait for a radical decrease in the price of 1 GB of RAM before the release of DDR5 does not have to: well, and there we are waiting for the next doubling of useful characteristics with the same Production price.
The price of improvements, overclocking and searching for balance
The growing volume and the speed of work directly affects another parameter of RAM - delays (they are timing). The operation of the microcircuits at high frequencies still does not want to violate the laws of physics, and on various operations (search for information on the chip, read, write, cell update) requires certain time intervals. The decrease in the process gives its fruits, and the timings grow slower than working frequencies, but here it is necessary to follow the balance between linear reading speed and response speed.
For example, memory can work on profiles 2133 MHz and 2400 MHz with the same set of timings (15-15-15-29) - in this case, the acceleration is justified: with a greater frequency of delay in several clocks only decrease, and you will not only get the increase in linear speed. Reading, but also response speeds. But if the next threshold (2666 MHz) requires an increase in delays by 1-2, and then 3 units, it is worth thinking. Cut out simple calculations.
We divide the operating frequency for the first timing (CAS). The higher the ratio - the better:
2133 / 15 = 142,2
2400 / 15 = 160
2666 / 16 = 166,625
2666 / 17 = 156,823
The value obtained is the denominator in the fraction of 1 second / x * 1 000 000. That is the higher the higher the number, the lower the delay between receiving information from the memory controller and sending data back.
As can be seen from the calculations, the greatest increase - upgrade from 2133 to 2400 MHz with the same timings. The increase in the delay of 1 tact required for stable operation at a frequency of 2666 MHz still gives advantages (but not so serious), and if your memory works at an increased frequency only with an increase in the timing by 2 units - performance even slightly decreases relative to 2400 MHz.
It is true and the opposite: if the modules do not want to increase frequencies (that is, you spoiled the limit for the specifically of your memory kit) - you can try to play a little "free" productivity, reducing the delay.
In fact, the factors are somewhat larger, but even these simple calculations will help not express with the acceleration of the memory: it makes no sense to squeeze the maximum speed of modules if the results become worse than on averages.
Practical application of memory acceleration
In terms of software from such manipulations, first of all won the tasks that constantly operating memory are not in streaming mode, but twinkling random data. That is, games, photoshop and all kinds of programmer tasks.
The hardware system with a built-in graphic processor (and devoid of own video memory) is obtained by a significant performance gains both when delays decreased and with increasing operating frequencies: a simple controller and low bandwidth is very often becoming a bottle of integrated GPU. So if your favorite "tanks" barely crawl on the built-in schedule of an old computer - you know that you can try to make to improve the situation.
Mainstream
As not strange, the average users benefit from such improvements. No, unconditionally, overclockers, professionals and players with a full wallet get their 0.5% of performance, using extreme modules with exhaust frequencies, but their share on the market is small.What under the hood?
White aluminum radiators Remove quite simple. Pitch zero: We are grounding about the battery or some more metal contact with the earth and give static static - we do not want to give a ridiculous accident to kill the memory module?Step One: Warm the memory module with a hairdryer or active loads on reading-record (in the second case, you need to quickly turn off the PC, de-energize it and remove the RAM while it is still hot).
Step 2: Find the side without stickers and carefully draw a radiator in the center and at the edges. Use pCB As the base for the lever can be, but with caution. Carefully choose the plot point, try to avoid pressure on the fragile elements. It is better to act on the principle "Slow, but right."
Step Three: Open the radiator and disconnect the locks. Here they are, precious chips. Screw on one side. Manufacturer - Micron, 6xA77 D9SRJ chip model.
8 pieces of 1 GB each, the factory profile - 2400 MHz @ CL16.
True, houses are not worth removing the heat distributors - the fillings will be tightened and your lifelong 1 warranty cried. Yes, and native radiators are perfectly coping with functions entrusted to them.
Let us try to measure the effect of overclocking the RAM using the example of the Hyperx Fury HX426C16FW2K4 / 32 set. Decoding the name gives us the following information: HX4 - DDR4, 26 - factory frequency 2666 MHz, C16 - CL16 delays. Next, it goes the color of the radiators (in our case - white), and the description of the K4 / 32 kit is a set of 4 modules with a total volume of 32 GB. That is, it is already clear that the RAM is slightly dispersed in production: instead of regular 2400, the profile of 2666 MHz is stitched with the same timings.
In addition to the aesthetic pleasure from the contemplation of four "Snow Wholes" in the case of your PC, this set is ready to offer weighty 32 gig memory and aimed at users of ordinary processors, not particularly indulging in the CPU acceleration. Modern Intels without the letter K at the end finally lost all possible methods obtaining free performance, and there are practically no bonuses from memory with a frequency above 2400 MHz.
As test stands, we took two computers. ONE BY intel database Core i7-6800K and the Motherboard ASUS X99 (it represents a platform for enthusiasts with a four-channel memory controller), the second with Core i5-7600 inside (this will be fade by the mainstream iron with integrated graphics and missing). On the first check acid potential Memory, and on the second we will measure real performance in games and work software.
Acid potential
With standard JEDEC profiles and factory X.M.P. Memory has the following modes of operation:DDR4-2666 CL15-17-17 @ 1.2V
DDR4-2400 CL14-16-16 @ 1.2V
DDR4-2133 CL12-14-14 @ 1.2V
It is easy to note that timing settings for 2400 MHz make memory not as responsive as profiles 2133 and 2666 MHz.
2133 / 12 = 177.75
2400 / 14 = 171.428
2666 / 15 = 177.7(3)
Attempts to start memory at a frequency of 2900 MHz with an increase in delays to 16-17-18, 17-18-18, 17-19-19 and even with a voltage rise to 1.3 Volta did not give anything. Without serious loads, the computer works, but Photoshop, archiver or benchmark spit errors or dump the system in BSOD. It seems that the frequency potential of the modules is chosen to the end, and the only thing that remains to be - reduce delays.
The best result that managed to achieve with a test set of 4 modules is 2666 MHz during Timing CL13-14-13. This will significantly increase the speed of access to random data (2666/13 \u003d 205.07) and should show a good improvement in the results in the game benchmark. In two-channel mode, the memory accelerates better: specialists from OCLAB managed to bring a set of two 16 GB of modules to a frequency of 3000 MHz @ CL14-15-15-28 with a voltage rise to 1.4 volts - an excellent result.
Satural trials
For our i5 with embedded graphics as a benchmark, we chose GTA V. The game is not young, uses the DirectX 11 API, which has long been known and is excellent in Intel drivers, likes to consume RAM and loads the system at once all the fronts: GPU, CPU, RAM reading from disk. Classic. At the same time, GTA V uses the so-called. "Deferred rendering", thanks to which the time of calculating the frame is less dependent on the complexity of the scene, that is, the test technique will be cleaner, and the results are visually.For the average FPS, take the values \u200b\u200bstacked in the normal course of the game: the aircraft will fly, riding in the city, the destruction of the sacrament has a uniform load profile. According to such scenes (dropping 1% of the best and worst results from the data array) and get medium-gaming FPS.
Drawers Determine on scenes with explosions and complex effects (waterfall under the bridge, sunset landscapes) in the same way.
Passing and unpleasant friezes with a sharp change of the environment (switching from one test case to another) even on the Montstrual GTX 1080Ti, we will try to mark them, but we will not take them to the results: it does not occur in the game, and it is rather a cant of Benchmarck himself.
Demo Stand Configuration
CPU: Intel Core i5-7500 (4C4T @ 3.8 GHz)
GPU: Intel HD530.
RAM: 32 GB Hyperx Fury White (2133 MHz CL12, 2666 MHz CL15 and 2666 MHz CL13)
MB: ASUS B250M.
SSD: Kingston A400 240 GB
To begin with exposed standard frequencies X.m.p.-profile: 2666 MHz with timings 15-17-17. The built-in GTA V benchmark produces identical FPS and the same drawdowns on the minimum and medium settings in the 720p resolution: in most scenes, the meter ranges around 30-32, and in heavy scenes and when changing one location to another FPS sends.
The reason is obvious - the GPU capacities are enough, but the rasterization blocks simply do not have time to collect and draw a greater number of frames per second. At the "high" graphics settings, the results are rapidly deteriorating: the game begins to rest directly into the modest computing capabilities of integrated graphics.
2133 MHz CL12.
There is no own memory from GPU, and he is forced to constantly twink system. Bandwidth DDR4 In two-channel mode at a frequency of 2133 MHz, it will be 64 bits (8 bytes) × 2 133 000 000 MHz × 2 channel - about 34 GB / s, with small (up to 10%) overhead losses.For comparison, the bandwidth of the memory subsystem at the most modest discrete card NVIDIA GTX 1030 - 48 GB / s, and GTX 1050 Ti (which easily produces to GTA V 60 FPS on maximum settings In FullHD) - already 112 GB / s.
In the background, the same waterfall under the bridge leaning FPS in the in-game benchmark is visible.
The benchmark results were asked to 28 FPS on average, and lags when changing locations and explosions of their unworthy drawings turned into unpleasant microphrises.
2666 MHz CL13
The decrease in timings significantly reduced time waiting time from memory, and we already have standard results with this frequency: you can compare three benchmark and get a visual picture. The capacity for 2666 MHz is already 21.3 GB / s × 2 channel ~ 40 GB / s, comparable to the youngest NVIDIA.
The maximum FPS almost did not grow (0.1 not an indicator and is on the verge of measurement error) - here we still rest in the modest possibilities of Rops, and all the drawdors have become less noticeable. In scenes with a waterfall due to a high computing load, the result has not changed, in all the others - that is, on loading, explosions and other joys, slowing down the work of the video card rose by an average of 10-15%. Instead of 25-27 frames in episodes loaded with events - confident 28-29. In general, the game began to be felt much more comfortable.
TL; DR and results
It is impossible to evaluate the speed of operational memory on the frequency alone. DDR4 has sufficiently large clock delays, and other things being equal is worth choosing memory not only satisfying the needs of your iron at work frequency and volume, but also pay attention to this parameter.
Tests have shown that computers based on Intel Core I-series with embedded graphics receive a noticeable performance gain when using high-speed low-delays memory. Video catalog does not have its own resources for storing and processing data and uses systemic perfectly responds (to a certain limit) on the frequency growth and reduction of timings, since the recruitment of the frame with a multitude of objects directly depends on the speed of access to memory.
The most important! FRUY line is produced in several colors: White, red and black - you can find not only fast memory, but also suitable for style to the other components, how experts do from
How to change the timing of memory?
Master's answer:
If you want to increase the performance of RAM without establishing new RAM strips, then you should simply reduce the timings of existing ones. This operation should be done very carefully, because you can damage the computer devices.
Initially, you need to check the installed memory schedules. Windows Seven has a built-in program for the implementation of such a process. So you need to open the control panel and select the "System and Security" item there. There, select the "Administration" tab, then the item "Check windows's memory" There you need to select the "Run a Reboot and Memory Check" parameter.
Now restart the computer and click Delete to open menu Bios.. Press the Ctrl and F1 combination in order to open the menu additional parameters PC works. There, select the Advanced tab. Now look at the data that are located under the MEMORY FREQUENCY string. There you will see Cas Latency, Ras Pracharge Delay, Ras To Sas Delay and Active Prachrage Delay.
Here you need to reduce timings. It is necessary to do it very carefully, constantly change the parameter only to a minimum "unit". Start from the first CAS Latency item. There it needs to be reduced by 0.5. Return to the BIOS menu. There, select Save & Exit and press ENTER. After the computer restarts, then log in again in the RAM testing menu.
If the program has shown that the indicators have improved, then continue to further reduce timings by changing the value of the next item - Ras Pracharge Delay. In order not to constantly restart your computer when checking the memory, you can use special programs.
You can install the Riva Tuner utility or Memtest. With their help, you can check the stability and performance of your RAM. Riva Tuner also has such a function as a decrease in timings. Please note that this process is recommended to perform exactly through the BIOS, because if some failure occurs, you can quickly restore the factory parameters.
