קרא Read original article on ASCII.jp
AMD's mid-range GPU "Polaris 20", "Radeon RX 580" (RX 580) was announced in April 2017, and one and a half years passed in October 2018. Also more than a year has passed since the release of "Radeon RX Vega 56" 56) which adopts the "Vega 10 XT" core in August 2017.
Meanwhile, graphics managers are updating their versions, and at this moment "Radeon Software Adrenalin Edition 18.9.3" is the latest version. However, after making a PC once, driver updates tend to be forgotten unexpectedly.
In general, it is said that graphics drivers are optimized for each upgrade version and improved performance. This example should also apply to the RX Vega 56 and the RX 580, but what is the performance improvement it fills?
So I really want to install old and new drivers with these GPUs and see if the changes can be seen in benchmark scores and frame rate games.
Performance comparison between 18.9.3 and 17.8.1
The impact of the PUBG driver update is amazing
Let's quickly explain the testing environment. For this time, we used ASUS ROG – STRIX – RX 580 – O 8 G – GAMING & # 39; With the Vega RX 56 reference card and RX 580 as a graphics card. For the graphics driver, we used 18.9.3, which is the latest version at the current time, and 17.8.1, which is the GPU – compatible version released after the Vega RX 56 was announced.
In other words, we compare the performance of the two versions of the driver in early August 2017 to the end of September 2018. For the processor, "Core i7-8700K" is used to not become as bottleneck as possible, and the other test environments are as shown in the table.
Let's look at the result of "3DMark" (version 2.5.5029) which becomes a standard performance test. First of all, from the general score of "Fire Strike", the Vega RX 56 has a difference of 1 to 4% between 18.9.3 and 17.8.1, and the RX 580 has an opening of about 6% in both cases.
Although 3DMark has implemented great optimization to the driver at the time of GPU announcement in many cases, it is a good impression that the score is still growing steadily. Also, for the Vega RX 56 and RX 580, the last point is that I want to remember that the growth of the score is great.
Let's proceed to the result of the overall score of "Spy Time" which is a DirectX 12 test.
In Spy Time, the score of the Vega 56 RX is only about 2% at most from 17.8.1 to 18.9.3, and the score of the RX 580 has not reached 1%. In other words, as of August 2017, it seems that the Spy's optimization time of the driver was almost finished.
The following is the result of the classic benchmark "Final Fantasy XIV: Crimson's Liberator Benchmark". In this case, benchmarking is performed with each of 3840 × 2160 points, 2560 × 1440 points, 1920 × 1080 points. For setting the option, Highest Quality is selected from preset settings.
The climax of the Crimson Benchmark is a test that the Radeon series is not good, but if it does, the Vega RX 56 can only extend the overall score by 1% at 18.9.3.
As for the RX 580, the overall score due to differences in drivers did not reach 1% and there is no significant improvement in performance. The average frame rate follows the overall score trend, and Final Fantasy XIV: There seems to be little optimization in Crimson's liberating test.
So, how does the game's performance vary with the driver? First of all, I want to test this "Far Cry 5". In Far Cry 5, the quality is set to "highest" quality, and the benchmark mode is ready for the game being executed. Three resolutions of 3840 × 2160 points, 2560 × 1440 points, 1920 × 1080 points were selected.
First, although it is the average frame rate, the Vega RX 56 increased by 3 to 5% by a change from 17.8.1 to 18.9.3. The RX 580 also improved by 18% from 17.8.1 on 18.9.3, and increased performance due to driver update can be confirmed. On the other hand, at the minimum frame rate, the Vega RX 56 of 1920 × 1080 points are up by 9%, significantly improving performance.
Including the RX 580, the minimum frame rate increases by 18.9.3 and even other resolutions, and there are certainly benefits of updating the driver.
By the way, when 17.8.1 was used with RX 580, there was a problem that the screen was interrupted at 3840 × 2160 points. This problem may be due to individual cards, but since Peugeot 5 has been released in March 2018, it can not be helped to cause problems on 17.8.1 more than six months ago. Where are you? Because the problem does not occur on 18.9.3, bug fixes are applied well.
Then in "Fortnite", performance has changed with clear differences between drivers. Even Fortnite actually played the game with three resolutions of 3840 × 2160 points, 2560 × 1440 points, 1920 × 1080 points. The average frame rate and minimum frame rate per minute were purchased by Fraps. In addition, the quality is defined as "epic".
The average frame rate of the Vega RX 56 is increased by 5 to 9% by a change from 17.8.1 to 18.9.3. As the resolution decreases, the tendency to increase its elongation can be verified.
On the other hand, this tendency is the same at RX 580, and the opening of 18.9.3 and 17.8.1 is 2 to 7%, which is the largest at 1920 × 1080 points. For the minimum frame rate, the RX Vega 56 increased by 6% to 8% on 18.9.3, and the RX 580 improved by 4% at most.
The surprising result was "PAYERUNNNNN & BATTLEGROUNDS" (hereinafter PUBG). In PUBG, set all the "Ultra" quality and actually play the game for 1 minute. The average frame rate and the minimum frame rate during that period were purchased with "Fraps" (version 3.5.99). As with Far Cry 5, the resolution is three, 3840 × 2160 points, 2560 × 1440 points, and 1920 × 1080 points.
First of all, although it is the average frame rate, the Vega RX 56 is 18.9.3, which is 25-27% more than 17.8.1, which is quite significant. Although the RX 580 is not like the Vega RX 56, the difference between 18.9.3 and 17.8.1 is 19-21%, the opening performance by the driver is quite considerable.
This tendency is also the same at the minimum frame rate, especially at 2560 × 1440 points of the Vega RX 56 and 1920 × 1080 points of the RX 580, both dividing 60 fps at 17.8.1, while at 18.9.3 and the benefits of driver updates are great. PUBG seems to be an advanced driver optimization significantly only the title which is popular all over the world, AMD too.
In "Shadow Tomb Raider", set "Best" preset and make the benchmark mode ready for the game. The average frame rate and the minimum frame rate shown in the result are summarized in the graph. As in previous tests, the resolution is 3 templates from 3840 × 2160 points to 1920 × 1080 points.
In the shadow of the Ryder grave, the Vega RX 56 was unable to start the game because an error occurred at 17.8.1. For this reason, we can not compare the frame rates between 18.9.3 and 17.8.1, but since there was no problem with 18.9.3, problems were made in solving problems firmly.
Looking at the RX 580, note the result of 3840 × 2160 points here. The average frame rate of the RX 580 is 32% at 18.9.3 and 29% at the minimum frame rate, respectively, and you can confirm a significant improvement in performance. It is reasonable to think that the tuning focused on the 4K chart, considering that only about 3 to 4% was changed in other decisions.
By updating the driver
Game performance will definitely improve
As can be seen from the results of the above test, performance improvement was confirmed by updating the driver in most scenes. Especially the PUBG performance improvement is quite strange and really surprising if it will only change by optimization.
Even with Far Cry 5 and Fortnite, the frame rate was increased by 18.9.3, and AMD knows that optimization is being applied steadily. In addition, since the attached software also includes functions that have been added or can be improved, if you use the Radeon series and you are neglected to update and you are using an old administrator, you can download the latest version for whatever you do not try it.