Hello!

Yes of course, it´s sad that this was happened...

"We must do Simon in a room and we must lock the door!"

I´m very happy that we have Simon and the others to get optimized apps!

I use "optimized" for the science not for the Credits!

Greetings!


EDIT:
...after a look to my signature... I leave show only my Credits (in the boincstats- picture), that the others can calculate how many WUs my PCs had crunched...

---

Now for the 5.3.12tx36, it is outdated and holds no advantage with the release of Seti Enhanced.

I continue to run tx36 on one hyperthreaded machine (Gallatin). I value it for helping me run one thread Einstein, one thread SETI, which I don't know how to come very close to doing consistently otherwise.

I leave calibration turned off, and am not aware of active harm to me or others in this choice.

I confess that my measurements show a smaller benefit for the strict one thread SETI/one thread Einstein situation than was true many months ago, but still show some.

To give a common basis, I'll express this in credits.

SETI+SETI 18.9 cobblestone/hour
Einstein+Einstein 18.8
SETI+Einstein 22.5

This was run a few weeks ago, using Simon's SETI ap, and the then stock Einstein ap. I've not checked to see how Bernd's 4.24 revision at Einstein may have altered this, but since the mixed application HT situation has been moderately to strongly favorable for over a year and several ap changes, I think it likely still true.

Off the current topic, but on the ever-popular "does HT do any good?" question, the same runs gave non-HT output thus:

SETI 11.8
Einstein 15.7

So a very strong yes for HT benefit on SETI (as of Simon's enhanced V5.15) and a still appreciable yes for Einstein.

Obviously these results are machine dependent. My Gallatin has a 2 Mbyte LIII cached perched atop the Northwood LI and LII arrangement. My specific host has a slower RAM than I suspect most purpose-built Gallatin boxes have. (mine is 4x133 MHz, while I suspect most are 4x200)

---

Evening Everyone

On further investigation, Simon has not released 5.17 in Beta. I would caution everyone that what happens in Seti Beta should stay there... When it is ready to be realeased here, Simon will let everyone know.

Misunderstandings and hate wars caused the lose of Crunch3r and what help he could have provided... That is now water under the bridge and never to be seen again.

"WE" are here to Find ET... Some go an extra step!

Pappa, Tips his Hat to Simon!

As archae86 mentions he is running a Trux BOINC Core Client.. It works for him... I have a machine that is still running the same version for Statistics... When I look at what it turns out, it causes problems with numbers that are being sent to Eric for Cross Project Calibration... Einstien and Rosetta are both trying to do something about presenting Fair Credit for the work that users that produce work... So now "we" have to figure out what Trux did to take care of the HT issue and get it incorported into the BOINC Core Client...

Thank You For Your patience

Keep Crunching!

Pappa

---

Please consider a Donation to the Seti Project.

One way to achieve cpu affinity, the main reason I would see for running the Trux client, is to set a very low 'connect to network' setting.
On my Dual P3, not running at mo due to fan gone noisy and no time, is set to 0.025 days. It has been at this setting for about 15 months since I realised it actually did more work this way, about 15% faster on Seti. My sons and his friends P4 HT machines are also set this way.

Andy

One way to achieve cpu affinity, the main reason I would see for running the Trux client, is to set a very low 'connect to network' setting.
On my Dual P3, not running at mo due to fan gone noisy and no time, is set to 0.025 days. It has been at this setting for about 15 months since I realised it actually did more work this way, about 15% faster on Seti. My sons and his friends P4 HT machines are also set this way.

Andy

Hello!

I think that you get only short AR- WUs...
...because that cannot accelerate your app...

But what will you do when a long server outage come?
...and for example every Tuesday...


Greetings!

---

As archae86 mentions he is running a Trux BOINC Core Client.. It works for him... I have a machine that is still running the same version for Statistics... When I look at what it turns out, it causes problems with numbers that are being sent to Eric for Cross Project Calibration.

Pappa

Could you be more specific regarding "problems with numbers"?

My tx36 host claims the same credit as others in its quorum. As I have calibration turned off it does not revise the CPU time. I don't know what other numbers you might have in mind, nor the reporting channel, so don't see how to check unless you'll be more specific.

---

I have noticed that the stock 5.17 app is faster than Chicken's 5.15 app here.

I crunched two WU's, one from each project, with almost similar AR, totally average, about 0.45, and the Beta 5.17 finished with a time about 20 minutes less than the Chicken 5.15 here.

---

"I'm trying to maintain a shred of dignity in this world." - Me

I have noticed that the stock 5.17 app is faster than Chicken's 5.15 app here.

I crunched two WU's, one from each project, with almost similar AR, totally average, about 0.45, and the Beta 5.17 finished with a time about 20 minutes less than the Chicken 5.15 here.

I would be a bit sceptical unless you can test both apps with the exact same test wu to verify what you think you are seeing. Sometimes 2 'very similar' appearing wu's don't crunch the same due to some content in the wu that you cannot see on the surface.

---

"Time is simply the mechanism that keeps everything from happening all at once."

Ok, downloaded the dev version, 5.5.13, ran my SETI wu's dry, changed the app info back to where it would reflect 5.15 on the WU's, and its crunching happily. So they fixed this issue with the 2 projects.

Jeremy

---

As archae86 mentions he is running a Trux BOINC Core Client.. It works for him... I have a machine that is still running the same version for Statistics... When I look at what it turns out, it causes problems with numbers that are being sent to Eric for Cross Project Calibration.

Pappa

Could you be more specific regarding "problems with numbers"?

My tx36 host claims the same credit as others in its quorum. As I have calibration turned off it does not revise the CPU time. I don't know what other numbers you might have in mind, nor the reporting channel, so don't see how to check unless you'll be more specific.

When I was running tx36 to compensate for Akos's optimized Einstein S4 app, I noticed that even with calibration active only for Einstein, it bumped up my integer ops benchmark enough to overclaim credit by about 60% on projects using time*benchmark scoring. If you are running only Seti Enhanced and Einstien S5, though, I don't think the tweaked benchmarks matter much, if at all.

[edit] The only exception I can think of is if benchmark scores are being taken into account when determining the baseline credit/hour for a given computer type.

Regards,

-- Tony