It only has one DVI-D connector as there barely was any need for monitors. Also fans on these might be different from the regular models... and other minor changes in specs.
Yes, this was exactly what I looked at on Newegg. The $100 ones are out of stock now.
I changed the AMD Radeon Driver workload settings for my RX570 from "graphics" to "compute" and it seems that I am getting less errors than before. I've been running 1 task at a time, usually around 600 seconds per task.
Since I've made the switch, I have 6 "error" tasks, 6 "valid" tasks, and 25 "validation pending" tasks which should become valid, since the "error in computation" happens before the task is completed.
To give you an idea of my task totals, I now have 195 valid, 2 validation error, 53 error in computation, and 67 pending tasks (and 164 aborted tasks that I'm ignoring for these calculations). Assuming that all of the pending tasks will be valid, I will have (53 computation error + 2 validation error)/(195+2+53+67) total tasks = 55/317= 17.3% of tasks ending in errors.
I used to run this project with a RX560 a few months back with almost 0 errors. I'd like to help out this project, so I'm planning to continue running Einstein tasks. There will probably be around the same error rate in the future, but at least I'll also have many valid tasks too.
I've been continuing to get computation errors. They seem to come in groups. For example, 6 out of the 8 recent tasks were errors. Then it will go a while without errors. Most errors end the task early, before the 10 minutes that they usually take. Sometimes a task will get "stuck" at a given percent, and the elapsed time will keep running. I suspend these tasks when I catch them.
That may resemble to what you get with too much undervolting. You mentioned you have at least 2 tools in use for your GPU settings. Could those be conflicting? Or, in Windows, some improper setting left over if that's a former mining machine?
Richie wrote:I think they are
)
Yes, this was exactly what I looked at on Newegg. The $100 ones are out of stock now.
mikey wrote:DanNeely
)
I did. It also wasn't just me, the issue with 4xx/5xx series cards was widely discussed here at the time.
I changed the AMD Radeon
)
I changed the AMD Radeon Driver workload settings for my RX570 from "graphics" to "compute" and it seems that I am getting less errors than before. I've been running 1 task at a time, usually around 600 seconds per task.
Since I've made the switch, I have 6 "error" tasks, 6 "valid" tasks, and 25 "validation pending" tasks which should become valid, since the "error in computation" happens before the task is completed.
To give you an idea of my task totals, I now have 195 valid, 2 validation error, 53 error in computation, and 67 pending tasks (and 164 aborted tasks that I'm ignoring for these calculations). Assuming that all of the pending tasks will be valid, I will have (53 computation error + 2 validation error)/(195+2+53+67) total tasks = 55/317= 17.3% of tasks ending in errors.
I used to run this project with a RX560 a few months back with almost 0 errors. I'd like to help out this project, so I'm planning to continue running Einstein tasks. There will probably be around the same error rate in the future, but at least I'll also have many valid tasks too.
Thanks everyone for your help!
I've been continuing to get
)
I've been continuing to get computation errors. They seem to come in groups. For example, 6 out of the 8 recent tasks were errors. Then it will go a while without errors. Most errors end the task early, before the 10 minutes that they usually take. Sometimes a task will get "stuck" at a given percent, and the elapsed time will keep running. I suspend these tasks when I catch them.
That may resemble to what you
)
That may resemble to what you get with too much undervolting. You mentioned you have at least 2 tools in use for your GPU settings. Could those be conflicting? Or, in Windows, some improper setting left over if that's a former mining machine?