That's an idea too. I saw too much system lag when I attempted -hp in the past. I have now reduced all projects to a single gpu task per card for the experiment. My observation is that the tasks are taking about half the time to finish which is what I expected or maybe 5% faster?? Of course the fact that I am crunching through tasks faster means I will run out even earlier come the weekly outage. I guess that is normal for people that are running Petri's app too.

. . Oh yeah! My Core2 Duo with a single GTX1050ti has a 100 Wu cache and empties it in anything from 4 to 7 hours (depending on workload) but so far I have not had a 100% guppi cache. Today it had a 90% halfling cache and that was the 4 hours ... (the dozen guppis were the only thing that slowed it down)

Stephen

:(

For seti 2 is the best for the 1070, at einstein it is 3

Zalster, is that exclusively running only Einstein tasks on the card. I find the systems sluggish when there is a SETI and Einstein both running on the same card.

Yes those were exclusively with each project respectively.

Also Grant make a good point. I'm still running r3500 on some machines as it was the fastest app of the current ones out. You have to remember that it was tested toward my 980Tis. The later revisions were changing the app to work with older GPUs. So as Grant pointed out, newer app didn't seem to work as well with higher end GPUs in processing more than 1 at a time.

---

Aha! Makes sense.I can't remember that far back. Did r3500 have all the tuning parameters we enjoy with r3584?

---

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

A proud member of the OFA (Old Farts Association)

My experiment isn't going so well on the FX systems. I have dropped 2K in RAC since changing to single task running. My Ryzen system is still climbing with single task configuration. I'm guessing the high CPU count is helping.

---

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

A proud member of the OFA (Old Farts Association)

Aha! Makes sense.I can't remember that far back. Did r3500 have all the tuning parameters we enjoy with r3584?

I'm pretty sure those tuning parameters will still work with that version. Thou to tell the truth, I run a lean commandline

-sbs 1024 -hp -period_iterations_num 1 -tt 1500 -high_perf -high_prec_timer

You can try it if you like. If you do run v3500, see what you think when doing 2 then compare it to 3. I currently have 3 with v3500. With newer versions I only run 2.

---

Aha! Makes sense.I can't remember that far back. Did r3500 have all the tuning parameters we enjoy with r3584?

I'm pretty sure those tuning parameters will still work with that version. Thou to tell the truth, I run a lean commandline

-sbs 1024 -hp -period_iterations_num 1 -tt 1500 -high_perf -high_prec_timer

You can try it if you like. If you do run v3500, see what you think when doing 2 then compare it to 3. I currently have 3 with v3500. With newer versions I only run 2.

Looks like all I can do is go back to r3525 after looking in the oldBackup directory the installer moves the old apps to. Was there a significant change in performance from r3500 to r3525? I looked in Crunchers Anonymous and Mike's World but all they have are the current. I haven't tried to dive into the SETI download directory. I tried that a while ago and found a lot of apps corrupted.

---

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

A proud member of the OFA (Old Farts Association)

I'd say it's a 2 minute difference when running 3 at a time. Doesn't sound like a like but that 8 minutes in each hour. So works out to be an extra 36 BLC a day.

I still have it in a download file on my computers. If I could figure out how to send it to you I would.

---

Hi Keith,

Here is the download direct from Raistmer's e-mail page. SoG-r3500.

Good Luck

---

Bruce

Hi Keith,

Here is the download direct from Raistmer's e-mail page. SoG-r3500.

Good Luck

Hey thanks Bruce. Didn't think of Raistmer's Cloud

---

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

A proud member of the OFA (Old Farts Association)

Well I just looked over my completed tasks over at Einstein and MilkyWay. The single task is definitely not the way. My two task aggregate time is only 25% longer than just the single task.

The same seems to be applied here at SETI. There is more variation because of the mix of Arecibo and BLC along with VLAR or non-VLAR, but the principle is basically the same. The aggregate time for two tasks concurrent is not just twice the single task time. It is significantly less.

I've gone back to two up per GPU on all projects. I shall see if I can recover my RAC level before the experiment.

---

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

A proud member of the OFA (Old Farts Association)

A few days ago swapped from ssse3 to sse4.1 I started to get random errors, for example this task https://setiathome.berkeley.edu/result.php?resultid=5688193418
There was no noticeable performance improvement with sse4.1, but I didn't like the look of this group of lines in the stderr file:

Pulse: peak=7.804996, time=45.86, period=21.53, d_freq=2638122186.4, score=1.001, chirp=16.964, fft_len=1024 
SIGILL: illegal instruction
Stack trace (4 frames):
[0x6e3682]
[0x776ba0]
[0x40e9c0]
[0x160af10]

Exiting...

I'll be watching see4.2 for similar effects.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Are you going to report that back to Urs over at Lunatics?

---

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

A proud member of the OFA (Old Farts Association)

Thanks for reminding me.

Edit
Just tried - I can't log in to send him a PM - do you mind doing it for me please?

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Sorry Bob, neither can I. Never was invited to join when registration was open. I would suggest contacting Raistmer here at Seti via PM so he can relay the info or at least point to the thread and error task. I do believe it is Urs who is responsible for the Linux ports.

---

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

A proud member of the OFA (Old Farts Association)

That makes two of us!
Let's try the Raistmer route.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

I can send him a note if you want me to.

---

With each crime and every kindness we birth our future.

Thanks Mike - cut and paste my post if you could please.
CPU is a Ryzen 1700, 14 cores crunching two set aside for the GPUs.
No problems I can remember on SSE2, SSE3, SSSE3 (I think I've got hose right...)

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Thanks Mike - cut and paste my post if you could please.
CPU is a Ryzen 1700, 14 cores crunching two set aside for the GPUs.
No problems I can remember on SSE2, SSE3, SSSE3 (I think I've got hose right...)

Will do.

Edit:

Done.

---

With each crime and every kindness we birth our future.

Thanks Mike - cut and paste my post if you could please.
CPU is a Ryzen 1700, 14 cores crunching two set aside for the GPUs.
No problems I can remember on SSE2, SSE3, SSSE3 (I think I've got hose right...)

. . Hi Rob,

. . Care to enlighten us with the results so far? I am presuming that SSE3 has given better results than SSE2, but is SSSE3 better still? I know I was getting better results with SSSE3 in Linux than with SSE3 in Windows (Core2 Duo, 3 GHz E7600).

Stephen

??

I was going to ask the very same question. What are the results so far in trying out all the Linux flavors that are available. Is there a noticeable difference in tasks done with the FX or in Windows on the Ryzen with Linux?

---

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

A proud member of the OFA (Old Farts Association)