Good to know! i guess i'll have to be patient =)

I don't know if i can break into the 100k with both my host, we'll see later on..
6700K@4.5 + 1x1080
R7 1700@3.8 + 1x1060(6g)

On any gtx 1080(ti?) system that is not running the CUDA90/91 (Linux only) you can run upto 3 tasks in parallel. This increases your production.

FYI, incase you haven't caught it yet. The following files live in the Windows Hidden directory: \ProgramData\BOINC\projects\Setiathome

You use a "app_config.xml" file for this. In Windows, notepad has a tendency to save it as "app_config.xml.txt" which doesn't work.

Try this:

<app_config>
<app>
<name>setiathome_v8</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>1.0</cpu_usage>
</gpu_versions>
</app>
<app>
<name>astropulse_v7</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>1.0</cpu_usage>
</gpu_versions>
</app>
</app_config>

Also assuming you are using the stock Seti apps with your Neon Linux rather than the CUDA90/91 version, I can recommend a parameter list for your gtx 1060.

It goes in a text file that starts "MB*SOG.txt"

 -sbs 1024 -period_iterations_num 10 -spike_fft_thresh 4096 -tune 1 64 1 4 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 64 -oclfft_tune_cw 64

Under Linux, you can put these kind of files in the /Setiathome folder. While the "app_config.xml" works under Linux, under CUDA90/91 [speedup] you might only want to use it to reserve a cpu core for each gpu card. The CUDA90/91 heavily loads the card speeding up the calculations to the point where it is often doing them in less than a minute on a 1080 and a 2 1/2 minutes or so on a 1060.


HTH and doesn't confuse you.

Tom

---

A proud member of the OFA (Old Farts Association).

I am using the special cuda91 on the 1060 and cuda92 on the 1080, only thing i changed is added the -nobs parameter, im only using windows app rarely in between some gaming sessions..

Running 6cpu + 1gpu and 1 free thread on the i7 and 14cpu 1gpu 1free on the ryzen7, does that sound good?

---

I am using the special cuda91 on the 1060 and cuda92 on the 1080, only thing i changed is added the -nobs parameter, im only using windows app rarely in between some gaming sessions..

Running 6cpu + 1gpu and 1 free thread on the i7 and 14cpu 1gpu 1free on the ryzen7, does that sound good?

Yes, it sounds good. How are you getting the free core? I used the 2 cpu for 1 gpu trick in the app_config.xml file but I think you could also use the local configuration settings to free up a core.

Might use the 3 task trick on the Windows box. You can also setup locally for your Windows machine, that it stops processing when you start running any particular application. You should also make sure that the website says "don't leave anything in memory" when the seti app pauses. It might let it "sneek" in a little more processing :)

Tom

---

A proud member of the OFA (Old Farts Association).

Im only limiting cpu usage in the gui until it uses the number of desired thread, it does not free a core per se, but it seems to help stabilise the time variance on the gpu task.. so far my 1080 runs 1min12 for the most common work unit, and the 1060 6g is running 1min58

When the room is heat soaked the time goes up by about 5sec on each, i guess the card boost to a lower speed :)

Im only limiting cpu usage in the gui until it uses the number of desired thread, it does not free a core per se, but it seems to help stabilise the time variance on the gpu task.. so far my 1080 runs 1min12 for the most common work unit, and the 1060 6g is running 1min58

When the room is heat soaked the time goes up by about 5sec on each, i guess the card boost to a lower speed :)

You would be better off using a <project_max_concurrent>#</project_max_concurrent> in your app_config.xml just before the last </app_config>

# being the max number of work units total being both cpu and gpu

Using the "use at most" doesn't do what most people think it does. I've had this argument many a times over what it actually does and what people says it does.

For all my computers, I have use 100% of all cpu and 100% all times for dedicated crunchers. Then limit how many cores are used by the <project_max_concurrent> ie..1 core for each work unit be it CPU or GPU.

my 2 cents.....

Z

---

When you say 1core for each WU do you mean physical core or counting hyperthreading too? So for a i7 would you use 4 or 8?

Sorry, 1 thread for each work unit.

Edit..

Use 8. Some people will tell you not to use HT, but that is usually for lower end CPU Chips.

---

i had no app_config.xml so i created one... using only this for my ryzen system:

<app_config>
<project_max_concurrent>16</project_max_concurrent>
</app_config>

I don't need anything else in there right? or would there be something else useful to add?

我想插入一个图片，但是不知道怎么插入图片，谁能帮我一下呢，谢谢[/img][/quote]

我想插入一个图片，但是不知道怎么插入图片，谁能帮我一下呢，谢谢[/img]

[/quote]


Sorry, didn't quite get that, could you try something like english?

---

A proud member of the OFA (Old Farts Association).

i had no app_config.xml so i created one... using only this for my ryzen system:

<app_config>
<project_max_concurrent>16</project_max_concurrent>
</app_config>

I don't need anything else in there right? or would there be something else useful to add?

That will work in any project folder.

If the goal is to keep 1 thread free, I think you would want the # to be 15. I think your Ryzen is an 8c/16t machine?

Tom

---

A proud member of the OFA (Old Farts Association).

Google translate gave this...

I want to insert a picture, but I don't know how to insert the picture. Who can help me? Thank you.

I want to insert a picture, but I don't know how to insert the picture. Who can help me? Thank you.

English should mean this.

If the goal is to keep 1 thread free, I think you would want the # to be 15. I think your Ryzen is an 8c/16t machine?

Tom

The difference seems to be pretty minimal between both, ill leave it a 15 for now

---

If the goal is to keep 1 thread free, I think you would want the # to be 15. I think your Ryzen is an 8c/16t machine?

Tom

The difference seems to be pretty minimal between both, ill leave it a 15 for now

Actualy no, you need to test on your particular host what is the best number.
In a 6850 like the one i use the sweet number is 4 GPU + 6 CPU (total of 10 tasks) even if the cpu could run 12 tasks.
That's because the way the CPU works, each CPU has is best number of active WU at the same time.
Above this number you just waste heat & electric power you run more tasks but each task takes more time to crunch so the total output is less. With a smaller number your host crunches faster each WU but produces less WU/hr.
So as allways test is the best way to be sure.

---

Juan!! Woo hoo

---

Zalster! Woo hoo. 500K RAC!

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

The sky is the limit! Z Congrats for the 500K RAC. And not sure if you see the latests Free-DC today stats.

1 8559302 Zalster 81,906
2 7475713 petri33 76,837
3 8480062 Keith Myers 72,660
4 8396902 juan BFP 71,748	

I know that will not hold until the end of the day, but for few hours you has the top Seti cruncher host!!!

LOL

---

I want to insert a picture, but I don't know how to insert the picture. Who can help me? Thank you.

Ah, sorry, SLAPS forehead (again). I seem to be doing that a lot lately.

I am told to post a picture you need to "host" it someplace that has a publicly accessible URL, like Google drive or Dropbox.

Then you use the [img]http://put-the-url-here[/img]

And apparently it will display here. Or simply post the publicly accessible http://URLhere and we can click on it.

Hope this was right :)

Tom

---

A proud member of the OFA (Old Farts Association).