Hey guys,

I'd like to know if there's a way to know exactly (or close) my total runtime spent on the project. I don't find such information in the BOINC manager or in my account on the site here. By total runtime, I mean in months/days/hours etc. I know there's that kind of feature on other projects lie World Community Grid, where we can precisely know the total runtime.

Thanks in advance for your answer.

Cheers.

Yes this is a great idea, it would also be good if there was away to tell how many decoys we have contributed to the project.

---

Have a crunching good day!!

Hey guys,

I'd like to know if there's a way to know exactly (or close) my total runtime spent on the project. I don't find such information in the BOINC manager or in my account on the site here. By total runtime, I mean in months/days/hours etc. I know there's that kind of feature on other projects lie World Community Grid, where we can precisely know the total runtime.

Thanks in advance for your answer.

Cheers.

Most of those kind of things, I do not know if anyone tracks that exactly, is done thru 3rd party stat sites, like BoincStats, etc. Go to your account and near the bottom of the page you will see links to a few. There are many, many others! Personally I like the Starfire site..

---

Yes this is a great idea, it would also be good if there was away to tell how many decoys we have contributed to the project.

Now your request seems more project oriented to me, meaning the project itself may have to keep those stats.

---

This has been suggested in the past and has been rejected as not needed.

One simple way to know, run your computers 24/7 and just count the days ... :)

This has been suggested in the past and has been rejected as not needed.

I was unaware that this idea had been rejected.

I was thinking if there was this sort of information on our account page it might encourage more people to help out. People would be able to say I have contributed x number of models/decoys to Rosetta.

Best way for me to keep track of how much I have contributed is to go by GigaFLOPS in Boinc Spy, unless someone can think of a better way? I have had a look on Boinc Stats I can't find a GigaFLOPS reading for my computers only.

---

Have a crunching good day!!

This has been suggested in the past and has been rejected as not needed.

I was unaware that this idea had been rejected.

I was thinking if there was this sort of information on our account page it might encourage more people to help out. People would be able to say I have contributed x number of models/decoys to Rosetta.

THe whole idea behind the design of the credit system was to allow this very thing. Sadly this has been obscured and corrupted because of, well, it is enough that it has ... the reasons matter not ...

We can tell how well your last point works by you asking yourself how inspired you are by my contributions ... my suspicion is that my work has had little to no influence on your efforts.

I believe that most people, even within the BOINC community, would not have any idea what a model is. And from one protein to the next there can be more then a 10x variation in how long it takes to run a single model. And so it is more meaningful to create an accurate credit system. Since, even within the same protein, there is variation on the runtime per model, but the models are where the science work really lays, it was decided to devise the current system where credit is issued on a per model basis, which is adjusted for each type of work being done.

So, looking at your credit is the best measure of your contribution. And this is something that people from other BOINC projects can grasp as well. Not that there is perfect credit parity between projects, but it is close enough that you get a sense of scale on your work.

FYI, I believe the original post was suggesting that the computing effort be stated in computational years. Since they still have zero total credit, I can safely convert that to zero computational years. Otherwise, you might take your credit as compared to TeraFLOPS, and convert back to seconds based on your machine's benchmarks. Divide credit by 100,000 to get TeraFLOPS.

---

Rosetta Moderator: Mod.Sense

FYI, I believe the original post was suggesting that the computing effort be stated in computational years. Since they still have zero total credit, I can safely convert that to zero computational years. Otherwise, you might take your credit as compared to TeraFLOPS, and convert back to seconds based on your machine's benchmarks. Divide credit by 100,000 to get TeraFLOPS.

Thank you for the information Mod Sense. Is their away to convert my total credit into TeraFLOPS & how do we convert back to seconds based on our machine's benchmarks? Thanks in advance for the answer

---

Have a crunching good day!!

Well... literally a FLOP is a single floating point operation. It would be too much overhead to actually track and measure exactly how many FLOPs a given task takes. But you can estimate based on the machine's benchmarks, and the number of seconds a task took you to complete.

Speedy, you presently have 60,000 credits (give or take). And (conveniantly), you only have one host, which gets 2618.56 million ops/sec.

Divide by 100,000 credits per TeraFLOP and you get .6 TFLOPS. The huge supercomputers generally are rated in TFLOPS per second. It took your machine more then a second to earn that credit, so be careful when comparing.

A Tera prefix means a thousand billion. So a Tera is a thousand giga. A Giga prefix means billions. So, you have run roughly (.6) * (1000) * (1,000,000,000), or 600,000,000,000 floating point operations. Divide by your 2618.56 million ops per second, and you get about... 229 seconds?? I see the host was just created 4 days ago... but still, I'm sure it's crunched must more then 4 minutes in that time. Also, that host didn't produce all of the credit for your user account either.

I know credit is not directly a match to TFLOPS, but I would have expected my estimate to come much closer. Perhaps others can shed light on my error here.

---

Rosetta Moderator: Mod.Sense

A Tera prefix means a thousand billion. So a Tera is a thousand giga. A Giga prefix means billions. So, you have run roughly (.6) * (1000) * (1,000,000,000), or 600,000,000,000 floating point operations. Divide by your 2618.56 million ops per second, and you get about... 229 seconds?? I see the host was just created 4 days ago... but still, I'm sure it's crunched must more then 4 minutes in that time. Also, that host didn't produce all of the credit for your user account either.

I know credit is not directly a match to TFLOPS, but I would have expected my estimate to come much closer. Perhaps others can shed light on my error here.

Hi Mod Sense Thanks for you explanation
I no longer use This host As it's not as stable as My quad
As for the 229 seconds issue you described above is odd because my run time preference is set for 6 hours. My machine is generally running for more than 12 hours a day.

---

Have a crunching good day!!

Hi Mod Sense Thanks for you explanation
I no longer use This host As it's not as stable as My quad
As for the 229 seconds issue you described above is odd because my run time preference is set for 6 hours. My machine is generally running for more than 12 hours a day.

Hi Mod Sense acording to Boinc Spy PC1 has contributed 3.26 gigaFLOP & 38.12% of my total credit. It's just become clear as to where you got the 229 seconds from, I had to reinstall Boinc as I was having a issue with another project witch has been sorted out. I hope that explains why you got such a low GIGAFLOP reading.

---

Have a crunching good day!!

OK, looking into it a bit and using the Q6700 as the example.


1 credt = 860,000,000,000 flops (figured from base scoring for certificates, though few projects actually use this model for scoring so it's very inaccurate)

flops per credit * credits = total flops
860,000,000,000 * 24,134 = 20,755,240,000,000,000

total flops / flops per second (host reports 2593.36 million ops/sec)
20,755,240,000,000,000 / 25,933,600,000= 800,322.36172378690193417034272141
(seconds run time)

seconds run time / 60
800,322.36172378690193417034272141 / 60 = 13338.706028729781698902839045357
(minutes run time)

minutes run time / 60
133387226953.95 / 60 = 222.31176714549636164838065075595
(hours run time)

222 hours run time is what I get :P

I probably screwed up somewhere though.

And I think my calculator has fantasies of doing horrible things to me while I sleep.

Still seems low, but not sure how much that machine has been used... just took a stab at it.

---

"Every passing hour brings the Solar System forty-three thousand miles closer to Globular Cluster M13 in Hercules -- and still there are some misfits who insist that there is no such thing as progress." - Kurt Vonnegut

Thank you Steven For your workings below. Does anyone know how our RAC is worked out?

---

Have a crunching good day!!

In a nutshell, RAC is just the average of the past X days of granted credit. If you stop crunching today, there is a "decay" function to get your RAC to eventually drop to zero. I believe "X" is 14 days. So, RAC is like the speed your car is going, and total credit is like the total miles driven. Since a machine may not report in to the project for several days, they average it over a longer period of time to provide a rather stable number rather then one that bounces up each time you complete work, and then drops to zero for 3 days while you work on completing more. Many machines only have an internet connection on the weekend or etc. This system still affords a pretty comparable RAC figure between that machine and one with a full-time network connection.

For more details, see the description in the unofficial BOINC wiki.

---

Rosetta Moderator: Mod.Sense

If my computer has two cores each running at 2500 Mflops or 2.5 Gflops.

2.5 x 2 = 5.

5 x 60 x 60 x 24 = 432000Gflops. Or 432 Tflops.

And i Have 26000 credits, and the computer earns on average 900 credits in a whole day.

Then 26000/900 = 28.8888 days.

432 x 28.8888 = 12479.9616 Tflops in total :)

EDIT - please note the the "s" on the end of flops is to exemplify the plural, not to mean seconds, which i assume is obvious to most people :)

---