[nextpage title=”Introduction”]
Recently, we published an article, “Does dual-channel memory make difference in gaming performance?”, where we concluded that, in a gaming computer with an independent video card, using dual-channel memory mode has little or no influence in gaming performance. But, what about CPU integrated video performance? Will the conclusion be the same? That’s what we will check today.
To understand the theory under the dual-channel architecture and how to enable it, please read our tutorial “Everything you need to know about the Dual-, Triple-, and Quad-Channel Memory Architectures”.
Buy what is the impact of the dual-channel technology in a system that uses the integrated GPU? When you use an independent video card, the video memory, where the textures and other data related to the 3D video are stored, is located at the video card and is independent from the main RAM. The speed of this memory depends only on the video card characteristics. As the performance of a game usually depends mostly on the video card performance, in this case using a faster main RAM makes little difference, since it is not used to store video data.
However, when you use the integrated video, the GPU (“Graphics Processing Unit”) is located on the processor and uses the main computer RAM as video memory. So, it gets benefit from the higher bandwidth (due to the dual-channel technology or to higher memory clock). Besides that, we must remember that this bandwidth is shared between the GPU and the CPU, which, in theory, raises the need for a high memory bandwidth.
We can easily check how using dual-channel memory affects the GPU bandwidth using GPU-Z on an A8-7670K (which we tested recently) with two DDR3-2133 memory modules, first installed in single-channel, and then in dual-channel. Check Figures 1 and 2. In single-channel, the integrated video uses a 64-bit bus, with a bandwidth of 17.1 GiB/s, while with dual-channel, the bus is 128-bit wide and the bandwidth doubles.
Figure 1: video memory bandwidth in single-channel
Figure 2: video memory bandwidth in dual-channel
In order to measure the impact of installing the memory in dual-channel over the gaming framerate, we used two different hardware configurations. In the next page, you will see the configurations we used in this test.
[nextpage title=”How We Tested”]
In order to achieve a generic conclusion about the impact of the dual-channel technology over the integrated video performance, we decided to use two different systems, one with an AMD processor and one with an Intel CPU. Besides this, those two systems use different amounts of RAM and operating systems.
So, you must keep in mind that those tests do not compare the performance of the two CPUs (even because they have very different price points), but the impact of the dual-channel in each configuration.
During our benchmarking sessions, we used the configurations listed below.
Hardware configuration (AMD)
- Processor: A8-7670K
- Video: Radeon R7 (integrated to the processor)
- Motherboard: ASRock FM2A88X Extreme6+
- CPU Cooler: AMD stock
- Memory: 8 GiB DDR3-2133, two G.Skill Ripjaws F3-17000CL9Q-16GBZH 4 GiB memory modules configured at 2,133 MHz
- Boot drive: Kingston HyperX Savage 480 GB
- Video Monitor: Philips 236VL
- Power Supply: Corsair CX500M
Hardware configuration (Intel)
- Processor: Core i7-5775C
- Video: Iris Pro 6200 (integrated to the processor)
- Motherboard: ASRock Z97 Extreme 4
- CPU Cooler: Intel stock
- Memory: 16 GiB DDR3, two G.Skill Sniper F3-1866C10D-16GSR 8 GiB modules configured at 1,600 MHz
- Boot drive: Kingston HyperX Savage 480 GB
- Video Monitor: Philips 236VL
- Power Supply: Corsair CX500M
Operating System Configuration
- Windows 7 Home Premium 64-bit (AMD)
- Windows 10 Home (Intel)
- NTFS
- Desktop resolution: 1920 x 1080
Driver Versions
- AMD driver version: 15.7
- Intel driver version: 10.0
Software Used
Error Margin
We adopted a 4% error margin. Thus, differences below 4% cannot be considered relevant. In other words, products with a performance difference below 4% should be considered as having similar performance.
[nextpage title=”3DMark”]
3DMark is a program with a set of several 3D benchmarks. Fire Strike benchmark measures DirectX 11 performance and is targeted to high-end gaming computers, while Sky Diver also measures DirectX 11 performance, but is aimed on average computers. The Cloud Gate benchmark measures DirectX 10 performance, and the Ice Storm Extreme measures DirectX 9 performance and is targeted to entry-level computers.
On the Fire Strike benchmark, the memory performance with dual-channel improved by 61% on the A8-7670K and 38% on the Core i7-5775C.
On Sky Diver benchmark, the performance with memory in dual-channel was 42% higher than with single-channel on the A8-7670K. On the Core i7-5775C, the performance gain was 25%.
On the Cloud Gate benchmark, the performance gain with the use of dual-channel was 70% on the AMD processor and 19% on Intel’s.
On the Ice Storm Extreme, the performance with dual-channel was 62% higher on the A8-7670K and 23% higher on the Core i7-5775C, compared to the single-channel performance.
[nextpage title=”Gaming Performance (part 1)”]
Battlefield 4
Battlefield 4 is one of the most popular games of the Battlefield franchise, being released in 2013. It is based on the Frostbite 3 engine, which is DirectX 11. In order to measure performance using this game, we walked our way through the first mission, measuring the number of frames per second three times using FRAPS. We ran this game at 1920 x 1080, setting overall image quality at “medium.”
The results below are expressed in frames per second (fps) and they are the mean between the three collected results.
In this game, the performance gain by using dual-channel memory, compared to the single-channel configuration, was of 87% on the A8-7670K and of 22% on the Core i7-5775C.
Dirt Rally
Dirt Rally is an off-road racing game released in April 2015, using Ego engine. To measure performance using this game, we ran the performance test included in the game, in 1920 x 1080 (Full HD) resolution and image quality configured as “low” and MSAA off.
The results below are expressed in frames per second (fps).
In this game, the performance with dual-channel on the AMD processor was 74% higher than with single-channel. ON the Intel CPU, the gain was of 17%.
Dragon Age: Inquisition
Dragon Age: Inquisition is the most recent game from the popular action RPG franchise Dragon Age. It was launched in November 2014 and uses the Frostbite 3 engine with SpeedTree.
We ran the game at 1920 x 1080 (Full HD), with all quality options at “low”, measuring three times the framerate with FRAPS.
The results below are expressed in frames per second and represent the arithmetical mean of the three collected results.
On Dragon Age: Inquisition, the performance gain by using dual-channel memory, against single-channel, was of 90% on the A8-7670K and of 29% on the Core i7-5775C.
[nextpage title=”Gaming Performance (part 2)”]
Dying Light
Dying Light is an open-world horror game launched in January 2015, using the Chrome Engine 6. We tested the performance at this game with all quality options at minimum and 1280 x 720 resolution, measuring three times the framerate using FRAPS.
The results below are expressed in frames per second (fps) and they are the mean between the three collected results.
In this game, the performance with dual-channel on the AMD processor was 89% higher than with single-channel. ON the Intel CPU, the gain was of 34%.
Grand Theft Auto V
Grand Theft Auto V, or simply GTA V, is a open-world action game released for PCs in April of 2015, using the RAGE engine. In order to measure the performance on this game, we ran the performance test of the game (the plane part), measuring the framerate with FRAPS. We ran GTA V at 1280 x 720, with image quality set to the minimum.
The results below are expressed in frames per second.
On GTA V, the performance with dual-channel was, on the A8-7670K, 93% higher than with single-channel. On the Core i7-5775C, the gain was only 7%.
Life is Strange
Life is Strange is a choice-based drama adventure game released in 2015, using the Unreal Engine 3. We tested the performance in this game playing the first scene with the quality options set as “low” and 1920 x 1080 resolution, measuring the number of frames per second three times using FRAPS.
The results below are expressed in frames per second (fps) and they are the mean between the three collected results.
In this game, the performance with dual-channel on the AMD processor was 111% higher than with single-channel. On the Intel CPU, the gain was of 205%.
[nextpage title=”Conclusions”]
In this test, we had a very different result from when we tested the performance gain using dual-channel memory with an independent video card: using dual-channel memory on an integrated GPU provides a great performance improve in games.
The use of dual-channel memory made a big difference in those tests because the main RAM is used as video memory and, differently from the “real” video cards, the video memory is not exclusive. Using dual-channel memory configuration, and not single-channel, is like exchanging a video card with 64-bit memory bus for another one, with the same GPU but 128-bit bus. It is the same performance impact.
The difference is even bigger in games that make intensive use of both video card (by the GPU) and of the RAM (by the processor), since in this case both units must share the same bandwidth.
Another result we found is that the performance gain by using dual-channel memory was, in most cases, bigger on the A8-7670K than on the Core i7-5775C. A possible reason is the fact the GPU present on the Core i7-5775C has a 128 MiB embedded memory, which works as a memory cache for the video engine, thus reducing the impact of the smaller bandwidth of the single-channel configuration.
It is also necessary to keep in mind that our tests don’t make a direct comparison between the Radeon R7 GPU present on the A8-7670K and the Iris Pro 6200 present on the Core i7-5775C, since, beside the results are being influenced by the CPU power in each chip, other configurations (amount of memory and operating system) were not the same, since our objective was to measure the performance gain in two different situations. Besides that, the CPUs used in the tests have very different price tags, so it makes no sense on comparing both results.