I think the Teraflops are plummeting for a different reason. I'm getting "no work sent (but there was work for other platforms)" - all of my machines are getting this message. My average is dropping fast, but I didn't do anything and the machines only run ROSETTA. Is there an upgrade that I don't know about that has to be installed manually to run new units?

your computers are hidden, so don't know what you are running.
if the machines are below the memory requirements for the current tasks, that may be the cause of no work. there are other possibilities as well, maybe someone else can tell you some other reasons.

The other likely cause would be if they are running PPC-based Mac machines. The new minirosetta does not support them, so they will only be able to get work sporatically as work for the v5.98 version is available.

...the credit drop is apparently due to some halt in the assimilator. There are a pile of tasks with pending credit.

---

Rosetta Moderator: Mod.Sense

Users 241,757
46,305 active
Hosts 722,654
83,452 active


not a lot of active users and hosts but i guess it'll have to do.

i've only got 2 of my pcs running rosseta and my laptop is currently doing a cpdn wu... after that's done, i have that laptop on 3 projects.... Milkyway, Rosseta, and VTU@home.

not a lot of active users and hosts but i guess it'll have to do.

The problem is that BOINC has never managed to penetrate the actual target audience.

What we have, for the most part, are the techno geeks and a few passers-by. The intent was to actually get the average joe blow schmoe to run BOINC ...

If we could, the numbers would sky rocket even though the machines used by that population are, in general, slower than what we usually have.

When you think about it, 46,305 participants from the entire world is virtually no one ...

The reasons for this disconnect have been hashed and rehashed, but, until it changes you will not see those number change significantly.

I agree with what you have said above. If you take the top 10 countries in terms of the number of computers they have:

1 United States 164,100,000
2 Japan 49,900,000
3 Germany 30,600,000
4 United Kingdom 26,000,000
5 France 21,800,000
6 Italy 17,500,000
7 Canada 16,000,000
8 China 15,900,000
9 Australia 10,600,000
10 South Korea 10,600,000

Data taken from http://www.aneki.com/computers.html

And compare it to the 83,452 active hosts on Rosetta that is that is only around 0.022%.

Even compared with all of BOINC, with 514,912 active hosts, thats still only 0.14%.

Obviously, these are only very very rough estimates and the percentages would be significantly lower if you took all of the computers in the world but it does show to the extent that BOINC is still not mainstream. To be honest, I don't think it ever will be unless something major happens like Microsoft ship it with an os!

wow... is the TFlop counter wrong? Or did the TFlops drop by 80%?

---

got to be a bug in their system.

check out boinc stats for Rosetta and you will see the project has Average floating point operations per second 89,736.6 GigaFLOPS / 89.737 TeraFLOPS vs the home page which says TeraFLOPS estimate: 23.845.

Boinc stats data is from the 29th.

credit from the 29th is Total Credit 6,674,837,205 vs current credit as of this moment of Total credits: 6,677,937,660.

so credit is accurate just the TF estimator has a error in it.

As of now we have 102 tflops so we are again over 100!

i find the RAH terra flops estimate to be way off vs the boinc stats teraflops.
not sure if that is a estimate vs actual or what?
RAH says estimate is 102 whereas Boinc stats says Average fps of 91.638 teraflops.


so what is more accurate? the estimate or the average?

i find the RAH terra flops estimate to be way off vs the boinc stats teraflops.
not sure if that is a estimate vs actual or what?
RAH says estimate is 102 whereas Boinc stats says Average fps of 91.638 teraflops.


so what is more accurate? the estimate or the average?

I would like to know this as well.

It's 107TFlops... but that might be because of the outage we had two days back...

---

i find the RAH terra flops estimate to be way off vs the boinc stats teraflops.
not sure if that is a estimate vs actual or what?
RAH says estimate is 102 whereas Boinc stats says Average fps of 91.638 teraflops.


so what is more accurate? the estimate or the average?

I would guess BOTH are right, it is just in how you create the numbers that makes them different. Now your "more accurate" question is a bit harder, is it "more accurate" to say you make x amount of money per month or y amount of money per day? Are they the same at the end of EVERY month? NO THEY ARE NOT, why not you say...because some months have more work days than others. Which is "more accurate", per day salary or per month salary? So the answer to your original question is not an easy one at all, it comes down to how you use the numbers.

---

Keep in mind that the TFLOPs estimate is just the number of credits issued divided by 10,000. And we all know that credit is not the most precise measure.

From there, RAC is computed by a formula, and used to represent an average over time. But I believe the figure shown on Rosetta's homepage is strictly credits issued last 24hrs divided by 10,000. And I believe the figures shown on many stats sites are actually a RAC figure, based on each stats report's credit change.

So, the two figures are probably not quite as close as mikey suggests. But each is serving a different usage. The Rosetta figure being TFLOPs "right now" (which varies widely as things like the recent validator problems occur), and the stats sites figure being more of a sustained scale of computing over time.

---

Rosetta Moderator: Mod.Sense

IMHO the most meaningful chart about project growth rate is the following one:

http://boincstats.com/charts/chart_uk_rosetta_project_active_hosts.gif

It's a pity that it is only available for 60 days. Anyone knows a chart of active hosts for more than 60 days?

Now:

TeraFLOPS estimate: 95.676

crunch crunch crunch :)

---

WWW of Polish National Team - Join! Crunch! Win!

Teraflops is dropping like a rock. I hope it is a glitch and that the project didn't just lose a bunch of people.

---

strange numbers ......on 3-19-09


users 46132 hosts 84547 tflops 85.95 today 46089 users hosts 81974 tflops 93.62 i don't think more powerful computers were added in that short time something does not add up http://boincstats.com/stats/project_graph.php?pr=rosetta

How is the terraflops counted, is all power of the computers counted, or only half sicne same task needs to go to two different users to see if it is right?
In case that is not the case but all power is counted, how is it with super computers, do they have a need to send same task to two differen cores or?

What I am asking is is 100 TB equals to 100 TB of a super computer, or 50?

Well, some, like me, may have turned some attention to WCG to get their "gold" in the new sub-project ...

I got mine a little bit ago ... and am back to doing my normal project mix instead of almost 100% WCG ... still it took me almost a whole week to get gold ...

Maybe I need to upgrade again ...

{edit}

And I don't do more here because the pay stinks ...

Computational power is TeraFLOPS. Trillions of floating point operations per second. "TB" is TeraBytes, a measure of storage capacity.

Rosetta@home does not use redundancy. So the only time a task is sent to more then one host is when a deadline is missed or validation of the first result fails for some reason.

So, the estimate is comparable to a 100TF supercomputer. Not 50.

---

Rosetta Moderator: Mod.Sense