So has anyone else seen this before where Boinc sees both cards and identifies them as d0 and d1, yet both WU's were running on d0, more than a little odd.

So now I have a GTX 1060 3Gb that I can't use. My only spare machines are 10 year old Dell's that cost too much to run all the time.

I have had that problem on two different Windows machines with both AMD and Nvidia video cards.

For me, the fix was to back grade to a significantly older driver using a clean install. I too had 2 tasks running on one gpu.

Tom

---

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

Well I had an "interesting" afternoon yesterday.

The new motherboard arrived and after the machine had finished its tasks I started the rebuild. All went very well until the front panel header cable. Now I have fitted this same board in a 11 year old HP case and all cables and plugs matched 100%. Not so with a 7 year old ASUS case, the plug was twice as long but only had 4 wires, in totally different places than the new Gigabyte MB or the 11 year old HP MB. So little "surgery" was required, but I had the machine ready to test within an hour.

It booted perfectly even to still having the correct drivers for the 970!! Of course Windows 10 detected different hardware and Windows wasn't activated. I had prepared for this by checking before hand how to get round this and the answer was to register the machine on a Microsoft account. This I did and it appeared on my MS account "digitally" registered.

So now all I had to do was go to the activation screen and click on the " I changed hardware on this device recently," After a while it came up with the message that it was unable to connect to the activation server, please try again in a few minutes. So I tried a few more times, same message, finally looking up online, to find what this message actually means is you had a "digitally updated" copy of Windows, updated from Win 7 or 8, and this cannot be re-activated after a hardware change, contact support. So I went to the support page, that suggested if activation failed, click on the Microsoft App store and purchase a new licence. Not an option I wanted but in the end decided to take, however the App store only offered the option to "upgrade to Pro for £219, no thanks. I was not happy, So I looked up the MS phone number and tried that. Well my first few tries just ended up with an automated system asking for the re-activation numbers on the screen, took around 6 tries to actually get through to a human who, whilst obviously being in a "non UK" call center was actually extremely helpful, she listened as I explained what I needed and asked what I had upgraded to Win 10 from, when I said Win 7, she asked if I had any record of the Win 7 key, well as it happened I did, so she just "re-registered" the old Win 7 key as the new Win 10 key, and it activated. best part is she said that key can be used to re-activate Win 10 in the future.

So I now have a nice new Ryzen 3 2200G with the GTX 970, odd thing is the 970 is now taking around a minute longer per WU, the positive side is the old AMD CPU was taking around 3 hours, whereas the Ryzen does a WU in around 60 minutes. And I noticed my power usage has dropped as well.

Thanks for the summary. That jibes with some information I had retrieved about the Win10 registration process using the old Win 7 key. A ex-coworker I built a computer for just asked about how to upgrade to Win10 now that he is getting notices from M$ about Win 7's impending cutoff.

---

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

A proud member of the OFA (Old Farts Association)

You told him to move to Linux right?? hahahaha...

Well I had an "interesting" afternoon yesterday.
---edit---
So I now have a nice new Ryzen 3 2200G with the GTX 970, odd thing is the 970 is now taking around a minute longer per WU, the positive side is the old AMD CPU was taking around 3 hours, whereas the Ryzen does a WU in around 60 minutes. And I noticed my power usage has dropped as well.

I don't have a good idea about why the gtx 970 would be slower except you may need to set the cpu% in the local Boinc Manager to 75% and then use 1 core dedicated to drive the 970.
This should maintain or speed up the cpu processing while speeding up the 970. But you will only have 2 cpu's running cpu tasks.

HTH,
Tom
--edit-- I notice your period iterations could be a lot lower, especially if you use the builtin APU to drive your monitor. period_iterations could go down to 1. I would also experiment with -tt 1500.
-edit--

---

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

I don't have a good idea about why the gtx 970 would be slower except you may need to set the cpu% in the local Boinc Manager to 75% and then use 1 core dedicated to drive the 970.

Or just use app_config.xml (in your Seti project folder) and make full use of the CPU & allow it to crunch on all cores of the GPU runs out of work, and support the GPU when it gets work again.

<app_config>
 <app>
  <name>setiathome_v8</name>
  <gpu_versions>
  <gpu_usage>1.0</gpu_usage>
  <cpu_usage>1.0</cpu_usage>
  </gpu_versions>
 </app>
 <app>
  <name>astropulse_v7</name>
  <gpu_versions>
  <gpu_usage>0.5</gpu_usage>
  <cpu_usage>1.0</cpu_usage>
  </gpu_versions>
 </app>
</app_config>

Each to their own.


Edit- since you are running Anonymous Platform i'd suggest you re-run the installer, and this time select AVX for the CPU (make sure to re-select SoG for the GPU) and you should see a significant reduction in it's CPU WU processing times.

Edit- with a reserved CPU for the GPU it should improve GPU crunching times, reduce the difference in CPU time & Run time (which is bugger all anyway), and since you have a more powerful CPU you might try lowering the period_iterations_num=25 to 15 or (even lower if that doesn't have any effect on system responsiveness. Generally lower it till it becomes noticeable, then bump it up by 2 or so).

---

Grant
Darwin NT

OK so I changed period iterations to 2 as currently I am using the monitor on the 970 and occasionally need to use the machine "in gaming mode" so need the 970. (I stop crunching then)

I only ever crunch at 50% CPU on all machines so that should not be a problem. The machine does little else except occasional web surfing when I am gaming.

I also did as suggested and reinstalled lunatics and selected AVX

Completed this at 10.30 pm(UTC) last night.

Looking just now at the overnight results, nothing seem to have changed GPU around 12 mins and CPU around 60 mins.

Will see how results go during the day.

Here is a screen shot of task manager, so as you can see it is not exactly being stressed ;-)

Looking just now at the overnight results, nothing seem to have changed GPU around 12 mins and CPU around 60 mins.

Disappointing- it looks like AMDs AVX implementation still lags Intel's by a fair margin, but the processing times when using the SSE3 application show that their non-AVX performance has improved a huge amount compared to Bulldozer.

The GPU is using very little CPU resources even with the aggressive settings- Giving the app_config.xml settings a go may improve processing times by giving it a dedicated core all the time. Or it may not.

If you do feel the urge, i'd be interested in hearing the results.
As it is, it's humming along very nicely.

---

Grant
Darwin NT

So I added an app.config.xml with those lines in.

First thing that happened is it stopped crunching the second CPU WU, I assumed that it was due to my having set it to 50% CPU, so I upped that to 75% and the second WU restarted.

Also changed the period iterations to 1 earlier.

Will see if all that has any affect

Will see if all that has any affect

Doesn't appear to be, CPU time is still remaining low.
It looks like Raistmer's efforts at tuning the application for better performance on lower end hardware, while reducing system impact, have been very successful.

About the only difference between your configuration & mine is I have set the TT value to 1500 (the maximum) in the MB config file- if you look at my GPU results you'll see each GPU result having a very similar CPU time. It does improve output, however the side effect is increased impact on system responsiveness- hence the need to increase the period_iterations_num
With my RTX 2060 it's no longer noticeable, with my GTX 1070 it was quite noticeable (but tolerable for me). With the GTX 750Tis, the system would be completely unusable if one of them were driving the display, I don't use the iGPU for crunching & it just drives the display allowing the GTX 750Tis to run with the highest possible values.

While it hasn't made things worse, it hasn't made things any better either so i'd suggest just removing the app_config.xml, exit BOINC & restart & re-set you CPU usage back to 50% from 75% (even at 75% you probably wouldn't notice any effect on your system's performance due to the low need for CPU support of the GPU).
Thanks for your efforts.

---

Grant
Darwin NT