Each of the eleven tasks currently showed in the error list for that machine include this key string:
CL_MEM_OBJECT_ALLOCATION_FAILURE
That is what people have been consistently seeing when a card tries to run a GW task that needs more VRAM than it has.
You can compute what we have been calling here the DF or Difference Frequency by subtracting the first (base) frequency field in the task name from the second. For example, the task with the name:
has a DF of .75, which is high. At a quick glance, tasks for which your machine returns a successful result appear to have generally lower DF values than do your failing ones. I think it is still true that lower DF tasks tend to have lower VRAM memory needs.
If you wish to avoid this problem, you may de-select Gravity Wave GPU tasks in your preferences. Otherwise you can just take your chances on what fraction of the tasks sent to you will actually be within your capacity to run.
treninoburu wrote: Hi, I
)
Your pc's are hidden so it's hard to know for sure but my guess is you are running the GW tasks on a gpu with less than 4gb of onboard memory.
2021-02-22 23:50:11.1774
)
Each of the eleven tasks
)
Each of the eleven tasks currently showed in the error list for that machine include this key string:
CL_MEM_OBJECT_ALLOCATION_FAILURE
That is what people have been consistently seeing when a card tries to run a GW task that needs more VRAM than it has.
You can compute what we have been calling here the DF or Difference Frequency by subtracting the first (base) frequency field in the task name from the second. For example, the task with the name:
h1_0611.05_O2C02Cl4In0__O2MDFS3a_Spotlight_611.80Hz_738_0
has a DF of .75, which is high. At a quick glance, tasks for which your machine returns a successful result appear to have generally lower DF values than do your failing ones. I think it is still true that lower DF tasks tend to have lower VRAM memory needs.
If you wish to avoid this problem, you may de-select Gravity Wave GPU tasks in your preferences. Otherwise you can just take your chances on what fraction of the tasks sent to you will actually be within your capacity to run.