Below is a summary of questions pertaining to the Texas 4 Client Q&A call from Tuesday, April 25.
Why did units hash fine, curtail and then fail?
Many units failed after curtailment because certain components of units were needing servicing already. Once offline, the unit has failed to locate chips on the units and has therefore failed to come back online.
(Please note failure does not mean the machine is broken, but that the unit fails to hash unless serviced).
Why are other companies not reporting XP issues?
Other companies we’ve spoken with are reporting issues with S19 XPs as well, typically around 10% failure rates. Some companies we’ve spoken with are even higher, around Compass’ failure rate at TX4.
What does the repair procedure look like for these units, if required?
The first thing to note is that the repair process will take some time. First, all miners are diagnosed remotely to confirm it’s not a simple reset or firmware issue and identify the faulty component. The XP models have proven to be less robust so we are testing them across multiple sites to see where they run best.
If the unit has failed or is still underhashing, the unit will go to Denver Server Center or to a Bitmain Warranty repair partner where a Bitmain certified technician will perform advanced diagnostic and perform the needed repairs. These technicians can normally 10 miners in one day per person, depending on the severity of the repair. The XPs are proving more difficult and four units per repair tech per day is more realistic.
If a unit is fixed, it’ll be redeployed. If the unit fails to be repaired, we will work with another repair company to try another repair technique which we’ve had great success with, but is more expensive and not covered by warranty.
What is the timeline for redeployments and new deployments?
Functioning XPs are already palletized and moving to Ohio 2 for redeployment by the facilities online date of May 15th. Other units will go in waves, as repairs are processed. We expect units to take 6-8 weeks to be repaired and redeployed, after arriving at the repair facility.
Why were hashing XPs and S19as removed?
We’ve removed units that were not performing well, beginning with the XPs and some S19as. Other S19as are performing well, but we may remove them depending on the performance.
Why were non-hashing units not removed more quickly?
Compass’ operations team performed a standard two-step approach to tackle the increase in offline units, beginning by bringing in a Bitmain certified team to tackle the large volume of units offline at Texas 4 in February. As the repair push went on, and through conversations with the repair partner and our own technicians, it became apparent that the issue was more on the miner level than facility level.
The combination of a less robust hardware architecture with a unique climate has caused the S19 XP units to not be successful at this location while other S19 variants hash without issue.
Why did units that were repaired not stay online?
There are two reasons. Some units do not take repairs well, even after testing (standard is twice).
The other reason is that these units are more sensitive to weather conditions than anticipated, so placing a fixed unit back into those conditions led to a relapse in mining experience.
How does the Bitmain warranty process work in general? What about with XPs?
Bitmain gives warranty on units for the first 12 months of a miner, from the date Bitmain releases the unit from its warehouse. Bitmain can either take the machine to a certified RMA provider in the US or China, or send a team on-site to perform repairs.
For XPs, our official Bitmain partner stopped repairing XP units with over 10 failed bad chips, awaiting further instructions from Bitmain.
Is Compass Mining facing issues at other sites with XPs?
Compass Mining has seen elevated offline units at our Indiana 2 location for XPs, likely relating to the warmer temperatures coming into the springtime. The facility is working on moving units to a different location within the complex in order to keep miners cool and hashing.
Is the problem with the facility, or with the units, or both?
Certain adjustments were made to the Texas 4 site and continue to be made, based on the warming temperatures in the area.
However, most units at Texas 4 continue to hash fine. We’ve noticed that only XPs and S19as–both with similar board level designs–have had higher failure rates at Texas 4. While weather-related, other units handle the conditions well and hash optimally.