I have an I5-11400 with 32GB mem 4TB NVME drive(4.0) with win10pro and it is taking 13-14 hours to do 5 plots. What gives? This should only take 8ish hours.
Any help or suggestions?
Right now I have a 6 plot test going with a 60 minute delay and I am only getting 10% an hour.
What brand of NVMW do you have? Important.
I had a few Barracuda Q5 en they are not for Chia … To fast cache bottleneck.
Mem is enough, 6 cores 12 threads so you can use 12+6/2= 9 plots at a time at max.
# Sabrent 4TB Rocket 4 Plus NVMe 4.0 Gen4 PCIe M.2 Internal SSD Extreme Performance Solid State Drive R/W 7100/6600MB/s (SB-RKT4P-4TB) and the mobo is the Asrock z590MPhontom gaming 4.
I also have 2 10TB destination drives. Would it be worth doing 4 plots with a 60 minute delay to one drive then 30 minutes later do anther 4 plots with a 30 minute delay to the other drivee and keep that rotation up?
Drives looks goood… How many memory and what is your plotting config…threads etc
# G.SKILL Ripjaws V Series 32GB (2 x 16GB) 288-Pin DDR4 SDRAM DDR4 3200 (PC4 25600) Desktop Memory Model F4-3200C16D-32GVK
Right now I have 6 k-32 in parallel with a 60 minute delay. Default settings for the rest. 3390 mem, 2 threads, 128 buckets and bitfield on. The worst part is the approx. 2 hour hang at 31%. I started this at 9:30 this morning (about 6 hours ago) and the first plot is at 32%.
Everything looks ok.
I asume you use windows?
Check through taskmanager the real read and write speeds from the sabrent.
I thinks the write (boost) cache is used through the big files.
Please post the read and writing speeds
SSDs have ratings on the box that are all lightning fast, because they all promote the speed of the SSD’s cache.
The cache is very fast, and is usually between 25GB to 100GB (or more), depending on the SSD.
Chia will quickly use all of the cache, and then your plotting will hit the SSDs native speed, which could be slower than a mechanical drive.
SSDs are fastest to slowest, as follows:
SLC, MLC, TLC, QLC.
TLC and QLC SSDs are bad for plotting. They are low-end consumer level SSDs, designed for ordinary use, where the customer will never dump enough continuous writes to exceed the cache. So the drive will always perform for the customer at very fast speeds.
The cache is made from SLC or MLC NAND cells. That is what makes the cache fast. That is what the boxes’ ratings are based on.
There are high-end consumer level SSDs, and data center SSDs, and enterprise SSDs, that are made of 100% SLC or MLC. Such SSDs will not slow down, no matter how hard Chia pounds on them. Those SSDs are also more expensive.
It is hard to find out which NAND fabric is used. But if your 2TB SSD costs half of some other 2TB SSD, then you probably have a TLC or QLC SSD, and it will be slow for Chia.
Correct! for that reason i bought Samsung 970 pro with MLC … the 980 Pro is cheaper than the 970 pro because it uses TLC …
Sometimes brand sell it as 3 bit MLC but that is BS. MLC is 2 bit
Are you referring to Samsung?
I am not familiar with the Evo 970 Pro (or any Evo Pro SSDs from Samsung).
Samsung has, for example, a 970 Pro and a 980 Pro, but I never heard of an Evo 970 Pro or Evo 980 Pro.
Please provide a link to a sale or an article.
I mean the 970 pro … This oone is with MLC . This is the one you need.
First also try with a 90 minutes delay and see how everything is doing