Chia Coin (XCH) is one of the fastest growing countries in the cryptocurrency world. This is mainly because Chia is a storage-based cryptocurrency, unlike Ethereum, which relies on graphics cards. However, this is not the first time storage has been used for cryptocurrency. Filecoin (FIL) and Siacoin (SC) also rely on storage devices, but they don’t seem to get much attention. On the other hand, Chia has a good start.
Although Chia drawing (the process of creating profitable Chia files) places great emphasis on storage speed, other core components (such as the processor and memory) are equally important when creating drawings.Processors with more cores and faster SSDs can provide better results, as you have in our Best SSD Article, but most Chia farmers may not think twice about the impact on memory speed. However, there is ample evidence that memory frequency plays an important role in certain workloads, and we found that Chia drawing is no exception.
along with Recently implemented pooling, Chia farmers can now form teams to increase their chances of winning, which expands the number of users that can participate without having to invest heavily in new hardware, which is not a very profitable effort at this time. However, if you have your own existing system and some free storage space, you can easily create some graphs. In addition, the introduction of the MadMax plotter has changed the way the game is played, allowing farmers to reduce drawing time. Today we will use Chia’s official plotter and the new MadMax plotter to evaluate the impact of memory speed.
We make our testing method as simple as possible. Of course there is room for further optimization, but this is beyond the scope of this article.
For the Chia plotter, we measured the time required to create two k=32 plots in parallel with zero latency. We use the default settings for each graph with two processor threads, 3,390MB of memory, and 128 buckets.
For MadMax plotters, we used stotiks plotters to implement 0.1.5, which now includes a new plot manager GUI. We used the default value (256 buckets) and only increased the number of threads to the maximum of our processor to maximize performance.
|Intel System||AMD system|
|processor||Intel Core i9-10900K||AMD Ryzen 9 5900X|
|motherboard||ASUS ROG’s largest Apex 12||ASUS ROG Crosshair VIII Dark Hero|
|memory||G.Skill Ripjaws V DDR4-4000 C15 2x8GB||G.Skill Ripjaws V DDR4-4000 C15 2x8GB|
|Graphics card||MSI GeForce RTX 2080 Ti Gaming X Trio||MSI GeForce RTX 2080 Ti Gaming X Trio|
|Store||Toshiba PX04PMB080 800GB||Toshiba PX04PMB080 800GB|
|cool down||Corsair Hydro H115i Pro||Corsair Hydro H115i Pro|
|power supply||Seasonal X-1250 1,250W||Seasonal X-1250 1,250W|
We evaluated seven different memory frequencies from DDR4-2133 to DDR4-4000. We use the same memory timing for each frequency to ensure one-by-one comparison. We used the latest Windows 10 update in May 2021.
As far as our AMD test system is concerned, we always run Ryzen 9 5900X’s Infinity Fabric Clock (FCLK) at a ratio of 1:1:1, and use a unified memory controller clock (UCLK) and memory clock (MEMCLK).
MadMax plotter performance
The memory speed has a great influence on the drawing time using the MadMad plotter. Upgrading the memory speed from DDR4-2133 to DDR4-4000 on the Intel platform shortened 40 minutes, which is equivalent to a 43% reduction in the time required to complete the drawing.
On the contrary, the impact of memory speed on AMD platforms is not so obvious. There are still significant differences in graphics performance, but compared with Intel platforms, the gap is not big. The jump from DDR4-2133 to DDR4-4000 only shortened the drawing time by 6 minutes, or 15%.
Compared with DDR4-2133 with Intel system, even the drawing time of DDR4-3000 and DDR4-3200 are reduced by 29% and 32%, respectively. However, on the AMD platform, both memory speeds provide a similar 9% margin.
Chia plotter performance
We observed an 18% reduction in drawing time between DDR4-2133 and DDR4-4000 on Intel platforms. This number may seem small at first; however, if we convert the value to minutes, we will see a considerable margin of 53 minutes.
The performance gap on AMD platforms is roughly the same as Intel. Compared with DDR4-2133, the time required to complete the drawing process using DDR4-4000 is reduced by 15%. The difference is 43 minutes.
Compared with DDR4-2133, DDR4-3000 and DDR4-3000 perform the same on Intel platforms, because the completion time of both is shortened by 13%. For AMD systems, the drawing time of DDR4-3000 and DDR4-3200 were reduced by 11% and 12%, respectively.
One aspect to consider is that we use the same memory timing for comparison. Therefore, logically speaking, if you compare memory kits with different speeds and timings, the performance margin will be greater.