Dual-core processors. Overclocking a dual-core AMD processor with a faulty external cache
Extreme sports enthusiasts even go as far as liquid nitrogen, but that’s not our wayMany video game fans have probably tried to overclock their processor or video card. However, despite the fact that this procedure has long ceased to be overly complex and dangerous, it should be approached competently. Last time we told you aboutSafe video card overclocking , and now let's touch on the topic of processors.
Note: In this material, we only consider working with processors no older than five years. You can burn a modern processor only if you try to overclock it by more than 30%, raising the voltage by more than 25%, without having highly efficient cooling (enthusiasts even sometimes use a liquid nitrogen instead of coolers). If you act within reasonable limits, then in extreme cases, overclocking will simply automatically reset after a reboot.
Suitable processors and the feasibility of overclocking
Typically, processor overclocking is done for one of three good reasons:
1. The processor does not cope well with modern non-gaming tasks (video editing and rendering, modeling, transcoding, working with large amounts of data, etc.).
2. The processor performs poorly in processor-intensive games (Battlefield 1, Rise of the Tomb Raider, Company of Heroes 2, Dishonored 2, Mafia 3, Crysis 3, etc.).
3. The processor does not open the video card (read more about this case).
We are mainly interested in two latest reasons, since in both of them overclocking the processor will increase the number of FPS in games. And this is exactly what any gamer needs.
However, there are a couple of cases when there is no point in overclocking the “stone”:
1. If your processor is more than five years old.
2. If your processor has less than four threads (like dual-core Core i3) or four full cores (Core i5, i7, AMD FX-4300 or higher).
3. If your video card is one of the most budget models(GeForce GT 710, etc.) or generally represents a graphics core built into the processor.
It turns out that at the end of 2016, owners of processors no lower than AMD FX-4300 or Core i3 and sufficiently powerful video cards should engage in CPU overclocking. After all, only then will something worthwhile come out of this whole undertaking in the form of an additional dozen or two in your favorite “shooters” and strategies.
Stage one: preparing to overclock the processor
Now let's get started.
First you need to check the current processor frequencies and compare them with the factory ones:
1. download CPU-Z program
2. install and run,
3. Look at the Core Speed column.
It will be indicated there current frequency processor. Now open Google and enter the exact name of the model into the search bar (it is listed in the Name column). Find it in the characteristics clock frequency and compare with the one in the Core Speed column. If the frequency in CPU-Z is higher, then your processor is already overclocked (this happens if you buy a computer second-hand). In this case, you will need to make a reset (more on this below). If the processor is not overclocked, then the frequencies will either be the same, or the indicator in the program will be significantly lower ( economy mode, which is disabled during overclocking).
Now you need measure the number of FPS in one of the graphic benchmarks:
1. download and run the Heaven Benchmark program;
2. click the Run button in the window that appears;
3. After a beautiful video appears, press the F9 button to start the performance test;
4. After finishing the test, press Save button(“Save”) and write the results to any convenient place (for example, directly on your desktop) under the name “Before overclocking CPU.html”.
To be on the safe side, you need to run one of the CPU-intensive games: Rise of the Tomb Raider, Crysis 3, Dishonored 2, Company of Heroes 2 or Battlefield 1. Ideally, all of the above. To measure FPS in them, you can use the utility Riva Tuner Statistics Server or the corresponding function in the program Bandicam. Play each game for about 5 minutes (the main thing is not to be indoors, where the load on the system is always much lower) and record the average frame rates.
Now restart your computer and go to BIOS. In it you need to find a section with a name like Advanced Frequency Settings or CPU Performance (for different manufacturers motherboards names are different). This section should display all information about the current state of the processor: temperature, frequency, voltage, and so on. Write them all down on a piece of paper and move on to the second stage.
Stage two-A: Overclocking the processor by multiplier
The first type of acceleration. On modern processors it is not always available, because it requires an unlocked multiplier (hence the name). The latter is found only in some models of “stones” from AMD and in K-processors from Intel (Core i5-6600K, i7-6700K, etc.).
If this is your case, then:
1. go to BIOS;
2. find the section with processor frequencies and a parameter with a name like CPU Multiplier or CPU Clock Ratio (this is the same multiplier; if it is blocked, then go to the chapter “Overclocking the processor via the bus”);
3. write down the current multiplier value on a piece of paper;
4. add 25-30 percent to it (NOT units);
5. save the changes and restart the computer (Apply changes and exit in the BIOS main menu);
6. if problems arise after rebooting, then go into the BIOS again and increase the processor voltage (CPU Voltage or CPU VCore) by 0.100-0.175 (for example, from 1.100 to 1.200-1.275);
7. if step 6 did not help, then go to the BIOS and reduce the percent multiplier by 5;
8. repeat step 7 until the problems stop;
9. if operating system loads and freezes (or blue screen) after starting any demanding game does not happen, then go into the BIOS again and reduce the processor voltage by 0.025 (to reduce its power consumption);
10. Repeat step 9 until problems begin, and then return to the previous voltage value;
11. Go to the “CPU overclocking test” stage.
Stage two-B: Overclocking the processor on the bus
Bus overclocking is available for any processor. This should only be done if the multiplier of your “stone” is blocked, since this method is considered more dangerous than the previous one. However, in reality you just need to take care of the frequency in advance random access memory, which will increase in parallel.
The algorithm of actions here is as follows:
1. go to BIOS;
2. Find the section with the parameters bus frequency (BCLK Frequency, Host Clock Value, etc.) and RAM frequency (Memory Frequency, DRAM Frequency, etc.);
3. write down the current frequency value on a piece of paper;
4. Reduce the RAM frequency by 25-30 percent;
5. add 25-30 megahertz to the bus frequency;
6. save the changes and restart the computer (Apply changes and exit in the BIOS main menu);
7. if problems arise after rebooting, then go into the BIOS again and increase the processor voltage (CPU Voltage or CPU VCore) by 0.100-0.175 (for example, from 1.100 to 1.200-1.275);
8. if step 7 did not help, then go to the BIOS and reduce the megahertz bus frequency by 5;
9. repeat step 8 until the problems stop;
10. if the operating system boots and a freeze (or blue screen) does not occur after launching any demanding game, then go into the BIOS again and reduce the processor voltage by 0.025 (to reduce its power consumption);
11. Repeat step 10 until problems begin, and then return to the previous voltage value;
12. Go to the “CPU overclocking test” stage.
Stage three: CPU overclocking testing
All that remains is to test the benefits of overclocking. Take the Heaven Benchmark test and play the same games as in the first stage. Compare FPS indicators- if they have increased by at least 10 points, then you can consider overclocking successful.
* * *
Overclocking a processor is a very useful activity for any gamer. And the computer will become faster, and the user will become more experienced. However, we do everything wisely. There is no point in overclocking very outdated hardware - new games won’t run well on it anyway (and if the old ones don’t work, then why are you still using it?). Trying to increase productivity by more than 20-30 percent without expensive and highly efficient cooling is naive.
You can (and should!) share your impressive overclocking results in the comments below.
Come play cool blockbusters like Dishonored 2, The Witcher 3, GTA 5 or Total War: Warhammer to our main site http://playkey.net. Games run without overclocking on any PC!
Hello dear blog readers. Nowadays, probably every second person has quad-core processors. Of course, if earlier two cores were good, today 4 cores in the system are generally excellent.
But the more cores in the processor, the worse its work with the tasks that we perform on the computer is organized, which means the optimization of the processor leaves much to be desired.
This is mainly due to the fact that not all programs are well designed for multi-core processors, that is, in some applications, programs and games, the main productive power of your processor may simply not be used and is in idle mode.
I think few people are happy with this state of affairs, especially when demanding game or the program slows down on a powerful quad-core processor.
Today we will talk about effective processor optimization using a simple but useful programCPU Control .
To optimize the processor viaCPU Control , we don’t have to overclock it as we did in the article -. By the way, I recommend reading it.
Download the programCPU Control ( ) and launch it. The program is very simple, free and in Russian.
After installation, you will see a shortcut, launch the program and see this window.
Let's start optimizing the processor withCPU Control . By default, the program is turned off. First, go to settings and select Russian language.
Next, select the second processor optimization itemCPU Control — manual .
To select a processor core for a specific task, click right click mouse over the process and select one of the processor cores.
Also, you can select several processes for one core or one process for several cores.
We configure the second (third and fourth) kernel for all other processes.
If a game or some application is slow or performance is clearly not enough.
Try to independently assign all processes except the one that is slowing down to the fourth or second core for processing. And let all the other processor cores take care of one of your tasks.
If you don’t want to dig around and understand the settings, you can simply select the modeauto and see productivity gains.
CPU optimization withCPU Control This is a vital manipulation for all multi-core computers, especially quad-core ones. Of course, after all, the performance increase due to processor optimization withCPU Control can reach one and a half times. On dual-core computers the increase will also be noticeable, but perhaps less than with 4 cores.
In the case of single-core processorsCPU Control will not be able to do anything, since the program is designed to optimize at least two processor cores.
Does your computer have a single-core processor? Then I recommend that you read this powerful article -. After completing 7 simple steps you will be able toCPU Control will make working on your computer more convenient and faster :)
Now you know how to optimize the processor withCPU Control and get a productivity boost for free, in no more than 5 minutes. By the way, to speed up your multi-core computer I recommend this one. This way you will make your computer work even faster. Also don't forget to subscribe to. This is the only way you can be the first to know about new articles on the blog. That's all for me. I wish you to smile more often and look at the world more positively 😉
Synchronicity, apparently, did not bring any outstanding results, which, in general, is a familiar picture for the Athlon64.
Returning to 3.3V on memory partially solved the issue proper operation in tests. True, we also had to install a weak 80 mm fan to blow the memory, otherwise errors would start somewhere in the 3rd minute of operation. SnM: running, temp. in the processor test - up to 67 degrees, 3DMark06: graphics do not start (after loading the first sample). Increasing the effect to 3.4 V with 3DMark did not change the effect (switching to 2-3-2-5 did the same, lowering the FSB frequency to 236 did the same; more precisely, the system began to hang at this point; at FSB = 232 M the picture did not change, but SuperPi grew up to 1m14.7s, which is close to 1m15.0s from the setting with the latest running 3DMark), which is a pity - good speed memory performance greatly affects computing power Now. Perhaps adjusting minor timings would help with the work, but there is little point in doing this - the system’s capabilities are assessed, not configured for operation. (It was later discovered that increasing one Tcl to 2.5 solved the issue with overclocking to 240 without errors.)
3DMark not launching is an indication that the system is defective, but even if we got it to work, it also does not check all errors. If it were possible to run several games and compile a complex project, the sum of such tests would almost guarantee stability, but for now this is not required, the system is on an open bench. We will assume that the system is stable at the previous full measurement:
| 240M, 11x, 1.275V | 2-2-2-5 (7-12-3-3-2-3), 2T, Dual, 180 MHz, 3.2V |
| 2640M, 220M RAM(1/12) | SPi1M =1m15.0s, SnM : working; 3DMark06 : 4632(1681,2158,1547) |
This data suggests that in typical computing (SuperPi) the system lags behind overclocked processors with cache approximately doubled (75/35). In graphics-related calculations, where the impact of the cache is apparently less due to large quantity data - lags behind by 30% (1-1547/2000). (The average runtime slowdown will likely be even less, which can be verified later by running PCMark.)
The same or slightly better readings are recorded in the screenshot.
The memory worked in dual mode. The resulting string in the accepted abbreviations will look like this:
SuperPi calculation speed drop: 10%
(1-82.5/75).
Reducing the total 3DMark06 score: 3.6%
(1-1492/1547).
Reducing CPU Score in 3DMark06: 3.6%
(1-1492/1547).
Finally, let's give the mark's readings in non-overclocked mode.
| 200M, 11x, 1.275V | 2-2-2-5 (7-12-3-3-2-3), 2T, Dual, 200 MHz, 3.0V |
| 2200M, 200M RAM(1/11) | SPi1M =1m25.3s, SnM : working; 3DMark06 : 4319(1615,2040,1319) |
It turned out to be very interesting too. Speed depends more on memory timings than on processor overclocking. Works faster than single ordinary memory at overclocking 2661 MHz. And the same according to 3DMark06 tests, as with regular memory overclocked at 2530 M (235 M FSB).
Addition dated June 25, 2008. With cooler Zalman CNPS9700 Cu and 2 fans on it (for the sake of not increasing noise), this is what we managed to get:
| 253M, 11x, 1.275V | _2.5_ -2-2-5 (7-12-3-3-2-3), 2T, Dual (180M, 3.0V RAM - bios) |
| 2861M, 232M RAM(1/12) | SPi1M =1m12.3s; SnM: working; 3DM06 : 4684(1688,2171,1609) |
With overheating errors in the FPU test, it operates up to 260 MHz (2861M) at 1.30V, that is, for water cooling it still has a very large overclocking margin. The processor was good. (I couldn’t find a way to change the cache size using jumpers on the Internet. If it exists, maybe someone can tell me?) For clarity of results Let's summarize the obtained benchmark data into a table. (Readers want spectacle, we need to come up with something.) We have 4 approximate speed values different processors in both benchmarks and 4 measurements of the processor under study in different modes. Arranging them in a row of the histogram fully shows what has been achieved, and we can move on to conclusions. With one caveat: in order to clearly show the performance when calculating SuperPi, a scale was taken according to the formula 100000/SuperPi1M.
Conclusions.
The conclusions turned out to be rich and multifaceted. The processor was looked at from a different, unconventional angle. Several interesting facts immediately emerged.
1) A processor with a disabled external cache can be fully used _ to temporarily replace another processor _ in case of its absence and even work fully on the computer, creating only a slight slowdown general work(in the worst case - up to 40% of the rated speed of a basic working processor).
2) Disabling the cache has a beneficial effect in _ lower core voltage _ (you need to check the output - maybe it's because of the kernel Toledo, but logically, consumption at least decreases). Consequently, overclocking to higher frequencies is possible than with cache. The consequence is a possible “not entirely fair” achievement of frequency overclocking records. But the increase in frequencies and decrease in heat dissipation also slightly compensates for the loss of performance when disabling the cache, if compared equal frequencies, and the limits for acceleration (in air, where heat dissipation is critical) in both cases.
3) Apparently, a processor with cache disabled allows _ slightly more warm up _. In any case, for this instance there was operating temperature(more precisely, its indicator in the test) was 67 degrees, while 3200+ processors on the same board usually froze at more than 60 degrees. An experiment with a working processor with a deliberately disabled cache can prove this. If this is the case, then disabling the cache is another reason to get extreme overclocking with limited cooling.
4) Certainly, _ choice fast memory _ and its overclocking greatly affects the computing speed of such a processor. If we consider 50 seconds in SuperPi 1M to be the “dream” of a cacheless processor in achieving the minimum usefulness of an unoverclocked A3000+, then in experiments with DDR440 memory, 2-2-2-5, it lags behind it by only 30% (1-50/72). Quantitatively, the memory improvement gave an increase: 95(s)/75 - by 26%, which is significantly more than the usual results with cache (of course, in in this case I came across another successful test result at a memory frequency of 220 MHz, and the first number was obtained in the mode Single Channel). If such memory is significantly more expensive (for example, the cost of a working processor), then there is no point in chasing it. It will be useful if it is “at hand”.
5) Usage dual channel memory will be more efficient than the Athlon64 processor with cache (which has an average gain of 6% in SuperPi). Winning - 10%. This is not so noticeable in all calculations. For example, according to the Cpu Score test in 3DMark06, the acceleration is 3.7% (1547/1492-1), as well as an increase in the overall test score (3.6%).
6) It is possible to overclock a partially working processor; the stability of the tests does not suffer from this.
The results show that even in the cache-intensive SuperPi tests, the processor was only slightly behind the standard 3000+ without overclocking. This means that even with the most difficult calculations (due to the absence of a cache), it will not lose much to normal Atlons (of course, due to the resulting overclocking and fast memory), and in dual-core work it will surpass them. General conclusion: if you come across such a processor as an alternative to a single-core one, then most likely it will satisfy the user (at least as a temporary solution). If there is a single-core processor without a cache, the lag in calculations will be 25-50%, which is tolerable, but the lag behind an overclocked single-core processor with SuperPi of about 35c will be 2-2.5 times, which can only suit temporarily. Because there will be “strange” freezing of calculations on familiar simple things: archiving, decoding, rendering, game calculations, and the system will behave a little differently, although the average slowdown is unlikely to be more than 50%.
Discuss the article or make comments, additions, links, please click here:
It's probably no secret that the speed of a computer can be increased not only by replacing a part with a more efficient one, but also by overclocking the old one. If it’s still a secret, then I’ll explain.🙂
Overclocking, overclocking- this is an increase in the performance of PC components (processor, , and ), due to an increase in their standard characteristics. If we are talking about a processor, then this means increasing the frequency, multiplier coefficient and voltage thereof.
2 Frequency increase
One of the main characteristics of a processor is its frequency. .
Any processor also has such a parameter as a multiplier (number), which if multiplied by the FSB bus frequency, you can get the real processor frequency.
Therefore, the simplest and safe way overclocking the processor via bios means increasing the frequency system bus FSB, due to which the processor frequency is increased.
In all variants the processor frequency will be 2 GHz
— bus 166 and frequency multiplication factor 12;
— bus 200 and frequency multiplication factor 10;
- bus 333 and frequency multiplication factor 6.
The simplicity lies in the fact that the FSB frequency can be changed directly in the BIOS or programmatically in 1 MHz steps.
If earlier, this method could easily end sadly for the processor (burning out). Then for today kill multi-core processor, simply increasing the frequency is very problematic.
As soon as a novice overclocker goes too far with the processor frequency, the system will immediately reset its settings to default and after a reboot everything will be fine.
To change the bus frequency you need to go to BIOS and find there CPU value Clock as shown in the picture.
Press Enter on this value and enter the bus frequency. next to it you can see the processor multiplier and the effective processor frequency of 2.8 GHz.
Please note that the processor multiplier in the example is quite high - 14x at FSB 200MHz; in this case, I would recommend increasing the FSB in steps of no more than 5-10MHz (that is, the frequency will increase by 70-140MHZ).
In the case of other multiplier and frequency values, increase the bus frequency in increments of no more than 10%. There is no need to rush when overclocking, and with this step it will be easier for us to calculate the most optimal frequency for your CPU in tests.
If you want to achieve tangible results when overclocking. Then you cannot do without a good cooler, pay attention to the Zalman cooler.
We carry out tests with temperature measurements and maximum load per processor. This can be done with programs such as Everest, 3D Mark.
If the temperature at maximum load is above 65-70 C, then it is necessary to either increase the cooler speed to the maximum or reduce the FSB frequency.
3
The processor multiplier can also be changed. This will affect the increase in CPU frequency. For example, at frequency:
- bus 133 and frequency multiplication factor 10; (1.33 GHz)
you can change the coefficient to 15 and as a result you will get 2.0 Ghz instead of 1.33 Ghz. Not a bad increase, right?
There's just one thing, your processor must be unlockedmultiplier, such processors are usually labeled as Extreme if Intel processor and Black Edition AMD processor.
But even if you don’t have the extreme version, you shouldn’t be upset. After all, when the right approach The first option can achieve excellent results. Although, most likely, you cannot do without...
4 Voltage increases
The principle is simple. If you apply more voltage to a light bulb than it needs to glow, it will burn brighter. A processor is a more complex thing than a light bulb, but the meaning is approximately the same.
Increasing the voltage allows you to more seriously overclock the processor. To achieve stable processor operation, with more high frequencies, it is necessary to increase the voltage on it. There are several points to consider here:
- be sure to install a good cooler.
- do not increase the voltage by more than 0.3 V.
To do this go toBIOS (Del key when starting the PC), after that go to the Power item Bios Setup=> Vcore Voltegeand increase the value by 0.1 V. Next, set your cooler to maximum and set the FSB frequency higher.
We test, if everything is fine and the performance suits you, then you can stop there.
When you reach a critical level of processor performance (that is, if you increase the frequency by 3-5%, a reboot will occur), I advise you to reduce the frequency by 5%, this way you will secure your overclocking stable work for a long time.
As you understand from the title, we will talk about how speed up the processor , or more precisely (just some inexperienced users called system unit so ;)) computer and Windows.
That is, today I decided to update and supplement one of the old and incomplete articles on optimization, and not just update it, but by inserting part of the material from one of the lessons into it ;)
It is important to enable core load monitoring in it, which is done by clicking the “ View” - “Kernel time output" And "" - " According to schedule for each CPU”.
Ideally, do all this not in idle mode immediately after starting the computer, but in the loaded system mode, that is, with resource-intensive applications running, like games, photoshop, or whatever else is consuming your resources.
By the way, here on 4 -x cores even in automatic mode The increase in productivity is noticeable, but for some reason it’s not always possible to achieve power with two, even manually.
In general, there is a huge field for experimentation here, which I hope you will explore, because everyone has their own system, set of programs and so on. By the way, the field is a minefield, because you can easily achieve the opposite effect instead of optimization, but it’s interesting to dig deeper :).
Try using different profiles in auto mode or manual setting in order to achieve maximum performance on your computer.
Afterword
In a nutshell, something like this. Someone, of course, will say that you shouldn’t bother with all this for the sake of some percentage increase and smoothness of work - your right.
I love tinkering with the system and bringing its speed, convenience and all that stuff to new heights, even if not always great And x, but then b O more, that is, speed up the processor at least somehow.
Let's see what happens for you. In general, I recommend at least trying it.
