How to overclock Ryzen processors (part two – the end)


How to overclock Ryzen processors (part two – the end)

How to overclock Ryzen processors (part two – the end)

 

  • How to overclock Ryzen processors (part one)

In the first part about BIOS settings, overclocking software in the environment We talked about Windows and BCLK frequency overclocking, in the second and final part we will discuss other issues of Ryzen processors overclocking.

Ryzen 1700, 1700X and 1800X overclocking

Overclocking using equal voltages

Knowing that all these three processors are the same and produced by the same production line, is there a reason for the 1800X processor to boast of its working frequency and power? Do other processors not have the ability to reach these values? For this purpose, we have compared the base frequency and the maximum frequency that can be achieved in overclocking mode and under heavy load of all eight cores. Memory is set to 3200 MT/s and other options are at default. The following table shows the basic voltage and frequency of Ryzen 7 processors.

Model Frequency (MHz) Voltage (V) Temperature (°C)
Ryzen 7 1700 3200 1.07 35
Ryzen 7 1700X 3500 1.16 52 / 32
Ryzen 7 1800X 3700 1.23 58 / 38

The Ryzen 1700 seems to have the edge with a core voltage of 1.07V (compared to 1.16 for the 1700X and 1.23 for the 1800X). Of course, CPU temperature is affected by core voltage; But according to the offset provided by EMD (-20°C), the obtained temperatures are ambiguous.

We have used this offset to correct the temperature; But we remind again that this offset is not exact. The difference between reported and actual values ​​is not constant. This difference should vary as a function of several parameters. Applying this offset brings us closer to the appropriate temperature; But there is no guarantee that it will be accurate. The following table shows the frequency and voltage of Ryzen 7 processors in overclock mode.

Model Frequency (MHz) Voltage (V) Temperature (°C)
Ryzen 7 1700 OC 3975 1.35
Ryzen 7 1700X OC 3950 1.35 63 / 43
Ryzen 7 1800X OC 4050 1.35 66 / 46

1800X in base mode has clear advantages; But this difference has decreased after overclocking. If you are not afraid of adjusting and changing the parameters of your processor, we recommend you to buy the 1700 processor due to the lower price. The graph below shows the scores obtained by the Cinebench benchmark software in basic and overclocked mode.

Considering the 1700 processor, this processor in basic mode and working frequency of 3.2 GHz , has achieved a score of 1438 and is far away from the 1800X processor with a score of 1640. Considering the 1800X’s 3.5 GHz frequency, it is not difficult to explain the reason for this difference.

After overclocking the processors, this difference disappears. The 1700 processor even outperforms the 1700X. The reason for this can be attributed to the difference in the quality of the processors at the time of manufacture, and it can also be seen in the Kaby Lake Intel processors. The choice is to perform more tests.

In this section, we will examine the behavior of the processor using different core voltages. This case starts with a voltage lower than the base voltage and ends with an increase in voltage and overclocking the processor. Due to lack of time, we used the relatively short test of Cinebench R15; Therefore, the reported value is probably not the ceiling of their stability. However, under heavier loads, the observed improvement should be similar.

Frequency%

Voltage (V) Frequency (MHz) Point (pt) Temperature (°C) Score%
1.0 3450 1540 46 0.0 0.0
1.1 3700 1642 52 7.2 6.6
1.2 3850 1710 56 11.6 11.0
1.3 4000 1770 63 15.9 14.9
1.4 4100 69 18.8 18.3
1.4 SMT Off 4175 1318 62 21.0 -14.4
  • 1.0V core voltage: we had to reduce the frequency to 3450 MHz. This point is the starting point.
  • Voltage 1.1 V: The frequency reached 3700 MHz with a jump of 250 MHz. A 7.2% increase in frequency resulted in a 6.6% increase in score.
  • Voltage 1.2V: This voltage is slightly lower than the default core voltage. The frequency can be increased up to 3850 MHz. The amount of overclocking, considering the frequency of XFR, is not high in 1800X.
  • Voltage 1.3 V: with this voltage we reached the frequency of 4 GHz.
  • The last move to reach the frequency of 4100 MHz It is at 1.4V.
  • Since there are applications that do not use SMT, we tried to increase the frequency by disabling this option. Without SMT, we reached 4175 MHz. The temperature also decreased by 7 degrees Celsius.

At the beginning of work, with a slight increase in voltage, we experienced a high frequency jump; But this jump becomes less in the next steps (this rule is valid in most processors). An increase in voltage from 1 to 1.1 V resulted in an increase of 250 MHz in frequency; While this increase is 100 MHz by changing the frequency from 1.3 to 1.4 V.

According to the conditions, the suitable voltage for daily work is 1.3 to 1.35 V. For the benchmark, the voltage can be increased to 1.4 volts.

Cinebench performance is highly dependent on clock rate, so naturally there is a strong relationship between frequency and visual score. will be But the processing of crisps is also involved in this score. Disabling SMT helps us increase the frequency; but it affects the results.

Memory efficiency

Overclocking the processor increases system efficiency; But if you really want to increase the performance of the system, you can’t ignore the effect of memory. . On the Crosshair VI Hero motherboard, the base transfer rate for memory is 2400 MT/s (may be different on other motherboards).

Cinebench R15

Configuration Score
2400 1639
3200 1663

Geekbench 4

This test affects the processor and memory system. By increasing memory frequency, single-core and multi-core efficiency increases by 5 and 6%, respectively. The results after 3 consecutive executions are shown in the table below.

ConfigurationSingle-CoreMulti-Core >

Mem. CopyMem. Latency

Mem. Bandwidth

2400 4417 20786 6229 4697 5568
3200 4635 22150 7546 5651 7094

Time Spy

Graphics are not affected by increasing memory bandwidth. The overall score shows a slight increase. While the processor-based benchmark shows a difference of 343 points.

ConfigurationGraphic card

Processor

 

 

Score

2400 7204 8010 7314
3200 7217 8353 7367

BCLK frequency

using JSkill Flare X memory kit , and the Crosshair VI Hero motherboard, we had five different options in the D.O.C.P (short for Direct Over Clock Profile) menu. This menu is actually the AMD/Asus version of Intel XMP.

The maximum data rate of each profile is 3200 and the maximum delay is 14. In the following, we have provided explanations about each mode:

  • D.O.C.P. 1: This option is most likely the optimal system mode that provides the best compatibility along with performance.
  • D.O.C.P. 2: A frequency ratio of 2133 for DRAM and 150 for BCLK is used to achieve a data rate of 3200 MT/s.
  • D.O.C.P. 3: 2400 frequency ratio for DRAM and 133 frequency for BCLK.
  • D.O.C.P. 4: 2666 frequency ratio for DRAM and 120 frequency for BCLK.
  • D.O.C.P. 5: Frequency ratio of 2933 for DRAM and 109 for BCLK.

AMD has locked the RAM latency numbers in the BIOS; Therefore, these numbers cannot be accessed and changed by the user. Of course, it is better to say that these numbers cannot be changed directly by the user.

In the above picture you can see different timings using different rates, which It is marked by yellow color. The main delay times are marked in blue. This change works as a function of memory rate, but can be applied manually in the BIOS.

 

    Introduction to AMD Ryzen Pro processors

  • AMD denied the news of price reduction of Ryzen 7 processors
  • Guide to buy a personal computer with a Ryzen processor
  • Can AMD deflate Intel with Ryzen?
  • AMD Ryzen processors price revealed

Other timings are placed in the green box and these items are not accessible by BIOS. If you look closely, you can see that a higher transfer rate increases the values. Higher delays decrease performance; Therefore, we try to keep these values ​​as low as possible.

To make sure you understand the above, we have given two examples of cases where the RAM transfer rate is equal to 3200 MT/s:

  • 2133 rate and BCLK frequency is equal to 150 MHz frequency -> RAM frequency is equal to 150 times 21.33 = 3200 MT/s. In this case, tRC is 51, tRTP is 8 and tFAW is 23.
  • Rate 3200 and BCLK frequency is 100 MHz Frequency -> RAM frequency is 100 times 3200 = 3200 MT/s. So, tRC is 75, tRTP is 12, and tFAW is 34.

No doubt you would say the first setting should be faster. Now, the reference frequency has changed, so we manually adjust the processor’s multiplier so that its final frequency is as close to 4050 MHz as possible. We had no issues with system stability during these changes.

Cinebench R15

We start again with Cinebench R15. However, as we mentioned, this software is not sensitive to memory frequency. Of course, a difference of 10 points can be very important for an overclocker, but it is not an impressive figure for a normal user. By manually changing the timing settings to 12-12-12-12 instead of 14-14-14-14, we gained five more points.

Configuration
3200 D.O.C.P.5 – BCLK Freq. 109 1803
3200 D.O.C.P.3 – BCLK Freq. 133 1810
3200 D.O.C.P.2 – BCLK Freq. 150 1812
3200 C12 – BCLK Freq. 150 1817

Geekbench 4

Usually less sensitive single-core results have RAM frequency; Therefore, from 2 to 5 D.O.C.P., we see a 2% increase in efficiency. Reducing latency increases performance by 1%.

We had a 4% improvement in the multi-core test. Assuming that the score increase is completely linear, this 4% increase is equivalent to a 200 MHz increase in clock rate.

3200 D.O.C.P.3 – BCLK Freq. 1334726243978845589974563200 D.O.C.P.2 – BCLK Freq. 1504741246038505592674703200 c12 – BCLK Freq. 150478624634870661067456

Configuration Single-Core Multi-Core Mem. Copy Mem. Latency Mem. Bandwidth
3200 D.O.C.P.5 – BCLK Freq. 109 4649 23640 7730 5677 7062

If by increasing the RAM transfer rate from 2400 MT/s to 3200 MT/s, we have a 6% increase in performance, the total performance increase is approximately 10%, which is equivalent to an increase of 500 MHz is in frequency.

Memory copy speed results are a bit difficult to interpret. The workflow is clear, but hundreds of points make the difference between the best and worst execution. So even if the results were calculated from 3 consecutive runs, they still cannot be trusted. The best performance increase in this part is equal to 12% and other increases are equal to 7 and 5%. . But does reducing the timing affect the results? According to the results of the first column in the table below, as the memory becomes faster, the scores decrease!

Of course, the half percent difference is not noticeable. For example, if the best result is 30 fps, then the worst result is 0.2 less. These results are very interesting; Because in contrast to the reduction of graphic results, processor results increase!

ConfigurationGraphic card

Processor

 

 

Score

3200 D.O.C.P.5 – BCLK Freq. 109 7198 8893 7410
3200 D.O.C.P.3 – BCLK Freq. 133 7180 9291 7433
3200 D.O.C.P.2 – BCLK Freq. 150 7177 9336 7434
3200 c12 – BCLK Freq. 150 7161 9485 7427

Game Ashes of the Singularity

We made the last comparison using the Ashes of Singularity benchmark and selected the “CPU-oriented” benchmark. This test was done before the game optimization patch for Ryzen processors; But in relative terms, its tools are similar.

In comparison to the data rate of 2133 MT/s, increasing the memory speed to 2400 MT/s results in a 4.1% increase in Points are awarded. Higher frequency, but slower timing; Therefore, the two available parameters neutralize each other’s effectiveness.

Changing to 3200 MT/s increases the scores by 13%. Therefore, the results show the sensitivity of this test to memory bandwidth; Our guess was that since large amounts of data are transferred from one CCX to another CCX, given the interface between CCXs that is affected by RAM frequency, it is natural that the impact of this bottleneck will be reduced as memory performance increases.

Using the mentioned methods, optimizing the memory delay by increasing the transfer speed (by increasing the BCLK frequency), the points obtained were at least 31% higher.

Ryzen 7 1800X Configuration Within-Core Latency Range Within-CCX Core -to-Core Latency Range Cross-CCX Core-to-Core Latency Range Cross-CCX Average Latency Bandwidth Std. Deviation
1333 MT/s 14.6 – 14.8ns 39.5 – 41.7ns 230 – 243.4ns 237.65ns 43.74 GB/s 2.84ns
2666 MT/s 14.6 – 14.8ns 39.6 – 42.3ns 117.8 – 124.6ns 120.4ns 50.16 GB/s 1.86ns
3200 D.O.C.P. Standard – BCLK Freq. 100 14.6 – 14.8ns 40.1 – 42.1ns 108 – 114.6ns 114.66ns 52.02 GB/s 1.69ns
3200 D.O.C.P.3 – BCLK Freq. 133 14.6 – 14.9ns 39.4 – 42.0ns 108.4 – 112.4ns 111.51ns 55.24 GB/s 0.90ns

Maximum memory frequency

News indicating the low ability of Ryzen processors to are memory overclocks. We set out to find this out ourselves. However, reaching the rate of 3200 MT/s on CAS12 was not that difficult. You change the settings in BIOS and reboot the system. At least the Crosshair VI Hero motherboard had such conditions. It may be more difficult to reach this rate on other motherboards; But fortunately, the bugs in this new platform are gradually being fixed.

After reaching the rate of 3400 MT/s, it becomes more complicated to continue overclocking. The result seems to vary from processor to processor and is influenced by the motherboard. Between 3400 and 3650 MT/s, we saw successive crashes and restarts. It was not possible to boot the system. However, at the rate of 3800 MT/s we saw a “miracle zone”! At this rate, the system was usable. The timing was automatically set to 14, but manually setting the timing to CAS12 did not cause any problems.

We tried to run our tests under these conditions, but the system was completely unstable. After 3 hours of unsuccessfully trying to score from Cinebench, we gave up.

It seems to arrive in the future with numerous updates that will be provided for this platform. It is not difficult and inaccessible to users at higher rates.

Conclusion

Many hours spent testing and reviewing the new AMD platform, we found that even if the Crosshair VI motherboard Use Asus Hero, It’s always difficult to work with a new platform. Due to the rush to provide this platform, AMD has not helped in this regard and has not provided much information to users.

This platform is improving. Regular BIOS updates are provided to improve performance and fix early bugs. We did not think that the basic problems would be solved so soon; But it happened, and Thoms Hardour of France published this article.

However, at a certain point, with all the data and tutorials available, overclocking Ryzen in the future will be like child’s play.

Previous Intel offers specially shaped SSDs for servers
Next How to overclock Ryzen processors (part one)

No Comment

Leave a reply

Your email address will not be published.