Differences

This shows you the differences between two versions of the page.

--- guides:plotting:virtualplots [2024/02/07 15:11] – created page on Virtual Plots vleesbanaan
+++ guides:plotting:virtualplots [2024/02/09 06:55] (current) – changed some wording - was underselling it a little :P vleesbanaan
@@ Line 1: / Line 1: @@
 ===== Virtual Plots =====
-Although k-32 is by far the [[https://dashboard.chia.net/d/CL1X4UWnk/blockchain-state?orgId=1&viewPanel=41|most used k-size]], chia allows us to plot big-k plots. As the number of proofs is directly proportional to the plot size in bytes, there is limited incentives on the revenue side to using larger k-plots. On the cost side however, using big k-plots offers a small advantage in terms of HDD power-saving. A k-32 has double the disk reads per byte and is accessed by the plot-filter twice more than a k-33, and four times for than a k-34, and 16 times more than a k-36, etc. One problem in leveraging these benefits has been that in order to fill up a drive to make best use of the capacity, a mixed bag of multi-k plots is required (see image below), resulting in a cost-saving that is insignificant enough to warrant a replot.
+Although k-32 is by far the [[https://dashboard.chia.net/d/CL1X4UWnk/blockchain-state?orgId=1&viewPanel=41|most used k-size]], chia allows us to plot big k-plots. As the number of proofs is directly proportional to the plot size in bytes, there is limited incentives on the revenue side to using larger k-plots. On the cost side however, using big k-plots offers the possibility of an advantage in terms of HDD power-saving. A k-32 has double the disk reads per byte and is accessed by the plot-filter twice more than a k-33, and four times for than a k-34, and sixteen times more than a k-36, etc. One problem in leveraging these benefits has been that in order to fill up a drive to make best use of the capacity, a mixed bag of multi-k plots is required (see image below), resulting in still too much disk io to use certain power saving modes.
-|  Drive Size  |  Plotmix  |  Available GB  |  Chiapos Space  |  GB Free  |
+|  Drive Capacity  |  Plotmix  |  Available GB  |  Total ChiaPos GB  |  GB Free  |
 |  5TB  |  35,37  |  4655.82  |  4615.43  |  40.39  |
 |  8TB  |  32,33,34,35,36,37  |  7449.96  |  7182.09  |  267.86  |
@@ Line 11: / Line 11: @@
 The Virtual Plot method tries to enhance these cost savings to the fullest, by leveraging the fact that both the plotID seed and the k-size are an input in the deterministic proof-construction of the plot format. This way, a single drive can host a variety of multiple (strictly) different k-sizes that share a single plotID, which is consulted by the plot-filter as if it were a single (hence, “virtual”)  plot.
-This means that any HDD using this matter is accessed only ~18 (9216/512) times per day with the current plot filter. This allows us to put HDDs into standby_z (spin-down) or idle_c (reduce rpm), saving 80% and 50% respectively compared to drives filled with only C0 k-32s.
+This means that any HDD using this method is accessed only ~18 (9216/512) times per day with the current plot filter. This allows us to put HDDs into standby_z (spin-down) or idle_c (reduce rpm), saving 80% and 50% respectively compared to drives filled with only C0 k-32s.
-Whenever a signage point arrives, you need to check the plot filter for plots passing. This plot filter check is done purely against the plotID, and is done in memory/cached, so you don't need any disk io for it. When a plot passes the plot filter, you'll need to look up qualities from the plot, that's when the disk io starts. With the current plot filter, a single plot is expected to pass only once every 512 signature points, or once every 80 minutes.
+Whenever a signage point arrives, the harvester process checks the plot filter for plots passing. This plot filter check is done purely against the plotID, and is done in memory/cached, so no disk io is needed for it. When a plot passes the plot filter, the harvester will look up qualities from the plot, and that's when the disk io starts. With the current plot filter, a single plot is expected to pass only once every 512 signature points, or once every 80 minutes.
-This means that if you have an 18TB drive filled with 165 k-32s, you have 165 different plotIDs per disk and you'll have a plot passing every 30 seconds which will require disk io. The disk will not be able to sleep. If, however, you have a single plotID per disk, your disk will only have to wake once every 80 minutes. It can sleep the rest of the time, which is much more power efficient.
+This means that if you have an 18TB drive filled with 165 k-32s, you have 165 different plotIDs per disk and you'll have a plot passing every 30 seconds or so requiring disk io. The disk will not be able to sleep. If, however, you have a single plotID per disk, your disk will only have to wake once every 80 minutes. It can sleep the rest of the time, which is much more power efficient.
 <WRAP center round important 90% em>
 <color #001e37>
-Note that standby_z leads to increased wear on spin-up. Most HDDs are rated at a limit of 50k spin-ups, so with 18 start/stop cycles a day (plot-filter 512) that would constitute ~7,6 years and with 36 start/stop cycles (plot-filter 256) you would hit the limit in ~3,8 years.
+Note that standby_z leads to increased wear on spin-up. Most HDDs are rated at a limit of 50k spin-ups, so with 18 start/stop cycles a day (plot-filter 512) that shoud last ~7,6 years and with 36 start/stop cycles (plot-filter 256) one would hit the limit in ~3,8 years.
 </color>
 </WRAP>
@@ Line 28: / Line 28: @@
 The virtual plot method may still be considered by a diverse set of farmers.
+It offers an alternative to the direction of compressed plots by using less, rather than more electricity.
 A good fit would be with farmers on small drives, with high electricity prices.
 But it also allows for more complex methods, such as using idle_c during winter, while switching to standby_z in summer to reduce electricity used for AC.
 Admittedly, the plotting requirements make this method prohibitive to many.
 It is in recognition of the large variety of farmers that we try to bring attention to this method.