System configuration: Windows 10, two (2) GTX 1080 FE and two (2) Xeon E5-2697 v2

I'm having some difficulty running both GPU's optimally. Both GPU's have tasks. Only one GPU is running at > 2000 MHz for the full task, the other GPU runs one task at < 139 MHz, practically idle but does appear to be processing the task.

<app_config>
<app>
<name>period_search</name>
<gpu_versions>
<gpu_usage>1.00</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

if I add "<max_concurrent>4</max_concurrent>"

<app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>1.00</gpu_usage>
<cpu_usage>1.00</cpu_usage>
</gpu_versions>
</app>
</app_config>

all CPU's flag "waiting to run", I've never experienced this with other projects.

Would appreciate any insight.

I had this happen on a pair of HD7950. Your situation may be different but what happened was occasionally there was an ATI kernel reset and one of the HD boards did not recover. It was running at minimum freq according to gpuz. I discovered that I could suspend the task and then immediately resume it and that would fix the board that was not running at full speed.

At one time I looked into doing this automatically but I came up with a better cooling arrangement and the kernel resets stopped.

System configuration: Windows 10, two (2) GTX 1080 FE and two (2) Xeon E5-2697 v2

I'm having some difficulty running both GPU's optimally. Both GPU's have tasks. Only one GPU is running at > 2000 MHz for the full task, the other GPU runs one task at < 139 MHz, practically idle but does appear to be processing the task.

<app_config>
<app>
<name>period_search</name>
<gpu_versions>
<gpu_usage>1.00</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

if I add "<max_concurrent>4</max_concurrent>"

<app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>1.00</gpu_usage>
<cpu_usage>1.00</cpu_usage>
</gpu_versions>
</app>
</app_config>

all CPU's flag "waiting to run", I've never experienced this with other projects.

Would appreciate any insight.

If you want to do it this way, then you need to do it like this:

<app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>0.25</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

BTW, has anybody been working with app_config setting with their GPUs?
I have several, so would like to know what is the setting for:
- GT 1030
- GT 730
- GTX 1050Ti
- GTX 1650OC

Thanks

---

non-profit org. Play4Life in Zagreb, Croatia, EU

System configuration: Windows 10, two (2) GTX 1080 FE and two (2) Xeon E5-2697 v2

I'm having some difficulty running both GPU's optimally. Both GPU's have tasks. Only one GPU is running at > 2000 MHz for the full task, the other GPU runs one task at < 139 MHz, practically idle but does appear to be processing the task.

<app_config>
<app>
<name>period_search</name>
<gpu_versions>
<gpu_usage>1.00</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

if I add "<max_concurrent>4</max_concurrent>"

<app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>1.00</gpu_usage>
<cpu_usage>1.00</cpu_usage>
</gpu_versions>
</app>
</app_config>

all CPU's flag "waiting to run", I've never experienced this with other projects.

Would appreciate any insight.

In the 2nd one you are specifying one cpu core for each gpu workunit you run, that's 4 cpu cores, of course your cpu workunits are 'waiting to run' as they are tied up running the gpu workunits.

BTW, has anybody been working with app_config setting with their GPUs?
I have several, so would like to know what is the setting for:
- GT 1030
- GT 730
- GTX 1050Ti
- GTX 1650OC

Thanks

The only way is to try it on your system and see what happens, ie try 2 wu's at once and it is slower or crashes then go back to one wu at a time.

The Forums are only visited by less than 10% of the total users so the chance of finding someone with your specs is small. Most people just sign up and crunch, if it doesn't work they try another Project, if that too doesn't work most give up and stop crunching.

BTW, has anybody been working with app_config setting with their GPUs?
I have several, so would like to know what is the setting for:
- GT 1030
- GT 730
- GTX 1050Ti
- GTX 1650OC

Thanks

The only way is to try it on your system and see what happens, ie try 2 wu's at once and it is slower or crashes then go back to one wu at a time.

The Forums are only visited by less than 10% of the total users so the chance of finding someone with your specs is small. Most people just sign up and crunch, if it doesn't work they try another Project, if that too doesn't work most give up and stop crunching.

Well, it worked for some time & then one app crashed & got stuck with several hours in perpetual motion. Maybe someone can fix the app, so we can run 2 at a time.

As this is what it means to run a single WU:


& when the 2 of WUs are run on same card, it's constantly 10-15% under stress. & that's only 1030 you see here.
So imagine how 1050Ti or 1650OC are "free" for work... ?

---

non-profit org. Play4Life in Zagreb, Croatia, EU

Actually, why can't we see pictures on the forum?

With only 1WU at a time, GPU works like 30% of the time from 0% to 15%.
With 2WUs at a time, GPU works constantly from 5-15% of the time.
(observation have been done on 1030GT card)
?

---

non-profit org. Play4Life in Zagreb, Croatia, EU

Actually, why can't we see pictures on the forum? ?

My guess would be because they are using an older version of the software and it doesn't support them. Or it could because you used a png extension and it doesn't support those, your signature is a gif file and it shows.

Try it this way instead:

https://i.imgur.com/4brOg6t.png

Currently, my RTX 2070 Super and 2080 Ti is running at full capacity, using GPU = 1 and CPU = 0.25.

The new WUs might have fixed the low GPU usage.

If a 2070 Super and 2080 Ti can be utilized 100%, a 1030 should as well.