Having said that, I think the power consumption per cuda core on the 980s is pretty well in line with the newer stuff.

Yep.

GTX 980         165W   2048 CUDA Cores
RTX 2060 Super  175W   2176 CUDA Cores

However the work done by a given CUDA core on the current series is way, way more than on the older ones (almost, but not quite, double)

---

Grant
Darwin NT

Interesting I think I have to agree with you in regards to the power consumption of the 980 which surprised me

Considering I've paid ~$100 each for the most recent of these 980s used on eBay, it takes it into the realm of the doable while letting me get close to competing with the "big boys".

All good. I was just surprised that the power consumption was around the same that was all

---

Interesting I think I have to agree with you in regards to the power consumption of the 980 which surprised me

Considering I've paid ~$100 each for the most recent of these 980s used on eBay, it takes it into the realm of the doable while letting me get close to competing with the "big boys".

All good. I was just surprised that the power consumption was around the same that was all

There is also the price difference between the 980 and a 2060.

I did find one of Vypers machines running 2 x 2060 and 1 x 1660ti running at 285K Rac, so roughly over 90k average for each card.

I just averaged it, but more likely the 2060s are pulling 100k each and the 1660ti is around the 85k.

Just some food for thought for those who want to dream as well.

Cheers and Happy Crunching.
Mark

I thought I heard months ago that Samsung B-Die was no longer in production. That is apparently not true.: https://youtu.be/7QvMzuFfkgk. The first few minutes sums it up, the rest is just wandering on PC Part Picker.

---

Seti@home classic: 1,456 results, 1.613 years CPU time

Down in the Dumps over Seti@Home?








Try a helium balloon (Helium is in short supply too).



Tom

---

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

I thought I heard months ago that Samsung B-Die was no longer in production. That is apparently not true.: https://youtu.be/7QvMzuFfkgk. The first few minutes sums it up, the rest is just wandering on PC Part Picker.

Thanks for the link. All the tech magazines picked up that story and it spread. Probably also the rumour was helped by the dearth of any CL14 memory which was the exclusive domain of B-die. Also when memory stick timings are dissimilar, that was also a giveaway that the memory was not B-die. Seems none of that is true. Just that the memory vendors have changed all their XMP timings to be much higher.

---

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

A proud member of the OFA (Old Farts Association)

Temporary Alternate gpu projects advice?
1) I am looking for a "well-behaved" gpu project.
2) One with less than 4 hour gpu tasks?
3) Last time I tried Einstein@Home it tried to takeover my Boinc environment.
I had to perform NNT to throttle it into a semblance of order.

Any suggestions?

---

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

Temporary Alternate gpu projects advice?
1) I am looking for a "well-behaved" gpu project.
2) One with less than 4 hour gpu tasks?
3) Last time I tried Einstein@Home it tried to takeover my Boinc environment.
I had to perform NNT to throttle it into a semblance of order.

Any suggestions?

If you go to E@H you need to adjust your number of days cache to a small number. Something like 1/24 day so you will DL work for only 1 hour. Then when SETI returns you adjust again to you normal SETI settings.

<edit>You could set the resource share at 0 but in this case your WU cache will be limited to the actual crunching WU. So you will be forced to wait about 1 min when you end the crunching to ask for a new one. Not sure is that is good with your lots of GPU's hosts.

---

I use Milky Way as my backup project. Their GPU work units are short, about 4 minutes on my GTX 1060. Their CPU tasks can be ornery and many hours at times so I stay away from them. Set Resource Share to 0 for MilkyWay so it only asks for tasks when SETI is completely out of GPU work. Works for me, YMMV.

Temporary Alternate gpu projects advice?
1) I am looking for a "well-behaved" gpu project.
2) One with less than 4 hour gpu tasks?
3) Last time I tried Einstein@Home it tried to takeover my Boinc environment.
I had to perform NNT to throttle it into a semblance of order.

Any suggestions?

I do Einstein@home, though I do use NNT to manage it a bit.
Worth noting something about it.
While many folks used to set up Seti for 10 days tasks with 10 days extra, it didn't matter as SETI enforced max tasks/cpu and /gpu.
Einstein would just chuck out max tasks. Once SETI relaxed the max tasks per device to a more realistic value, I needed to set the max tasks to a more realistic value. In my case, 1/3 day with up to an additional 1/3 day. Einstein seems to respect this, as it should, and after the initial burst of work this time I'm only getting another 8-9 tasks per update request. Still more than I need, so I get a few hundred in queue and do NNT. A good app_config.xml in the project directory is a must as well, using the max_concurrent parameters to manage stuff.

---

Here at SETI I use the Home and Work venues for a 3 day/0.1 cache, but I just use the Primary (default) preferences for all of my backup projects which uses a 0.5/0.1 cache setting and their resources are set to 0 (or 0.01 if the project won't accept 0).

Cheers.

Temporary Alternate gpu projects advice?
1) I am looking for a "well-behaved" gpu project.
2) One with less than 4 hour gpu tasks?
3) Last time I tried Einstein@Home it tried to takeover my Boinc environment.
I had to perform NNT to throttle it into a semblance of order.

Any suggestions?

If you go to E@H you need to adjust your number of days cache to a small number. Something like 1/24 day so you will DL work for only 1 hour. Then when SETI returns you adjust again to you normal SETI settings.

I got to looking and besides setting things "way down low" I noticed that I could also ask for "gpu-only" which would help keep E@H from sinking much into my cpu cache.

TY everyone for helping me "tame" E@H.

Tom

---

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

I got to looking and besides setting things "way down low" I noticed that I could also ask for "gpu-only" which would help keep E@H from sinking much into my cpu cache.

Task types also can help there. The Gamma-ray pulsar binary search (#1 for NVIDIA, #5 for CPU SSE) is pretty well behaved and with OK run times. The Gravitational Wave search O2, on the other hand, has hugely long CPU run-times (~16 hr here) which I try to avoid.
[edit] Let me know if you'd like a copy of my app_config.xml for there via PM.[/edit]

---

The Gravity Wave O2MDF gpu application is good also. The Gamma-Ray gpu app is solid also and the oldest venerable gpu app. Those are the only ones I do on the gpu. The Gravity Wave cpu applications are ball-busters and run for 10-12 hours.

---

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

A proud member of the OFA (Old Farts Association)

Thank you. I have toggled off the ones (E@H) mentioned that will be "hogs" of resources.

Tom

---

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

Since I didn't change over to Linux until after the special sauce, did the old SoG run on Linux? I would think it would....asking for a friend.

---

Since I didn't change over to Linux until after the special sauce, did the old SoG run on Linux? I would think it would....asking for a friend.

Yes, it should. I don't know what the download directory fanout URL for the SoG app is but an older version of the app is at Lunatics.
http://lunatics.kwsn.info/index.php?action=downloads;sa=view;down=503

---

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

A proud member of the OFA (Old Farts Association)

Hopefully, I shrunk this down enough to not be annoying.
Note that this is a basement with an open stairwell.
Enough heat rises that I only used 21 therms of gas heating the place; the 980s did the rest :)

@Jim Good for the gas bill not so good for the gas company :) How much do you think cooling would cost you in the summer?

---

Hopefully, I shrunk this down enough to not be annoying.
Note that this is a basement with an open stairwell.
Enough heat rises that I only used 21 therms of gas heating the place; the 980s did the rest :)

@Jim Good for the gas bill not so good for the gas company :) How much do you think cooling would cost you in the summer?

I don't have a/c here, but since the power co throws a summer 100% surcharge on all electric usage over 500 kwh/month, I usually ramp down a bit from Jun1 thru Sep 30.
I'll have to play it by ear this next summer, as I've added a ton of capacity here since last Oct 1. I might relocate the servers near an open window and away from the stairs for the summer.

---

Is it normal for tasks to be in the pending state for a very long time?

I have only 1 in this state. Maybe it somehow got missed?

8035028691 3646008451 8754065 9 Sep 2019, 19:35:06 UTC 11 Sep 2019, 17:36:25 UTC Completed, waiting for validation 16,764.61 9,371.88 pending SETI@home v8 v8.08 (alt) windows_x86_64