Dear crunchers!

Today, on August 11, we proudly celebrate the first birthday of RakeSearch project.

It is a joy for us to have managed such project for already a year. This is 1316 years of CPU time - which means that on average, 1316 cores perform computations.

By now, we have processed 27.5% of all possible workunits. This resulted in 90 found graphs of ODLSs.

On the peak load the project reached 60.6 Teraflops.

Many thanks for the participation and good wishes to all of you!

With best wishes, RakeSearch team:
hoarfrost and Natalia.

https://www.youtube.com/watch?v=inS9gAgSENE :-)

Glad to help, onward to the next birthday.

Conan

---

Awesome! :D

On the peak load the project reached 60.6 Teraflops.

On which day did this happen?

---

Hello Tom!

Awesome! :D

On the peak load the project reached 60.6 Teraflops.

On which day did this happen?

Value of 60.6 TFLOPS was registered on July 31th - due to large interval of averaging. If we calculate performance during only sprint days (1 day before, sprint, 1 day after) - we get a more realistic value ~205 TFLOPS.
But the metric from Server status page is very useful for tracking very large changes of project performance.

So am I right in saying that in less than four years we will have completed the current search? If I am right it's about rank 9 latin squares?

I wonder if it's possible to have the current code to run on GPUs, from what I know it should be doable: the memory requirements of the project are very low and the search is about comparing latin squares, nothing too complex.

So am I right in saying that in less than four years we will have completed the current search?

Yes. Time of the end of research - another interesting thing.

If I am right it's about rank 9 latin squares?

Yes.

I wonder if it's possible to have the current code to run on GPUs, from what I know it should be doable: the memory requirements of the project are very low and the search is about comparing latin squares, nothing too complex.

Perhaps, but there is no news on this subject yet.

Thank you for participation!

I wonder if it's possible to have the current code to run on GPUs, from what I know it should be doable: the memory requirements of the project are very low and the search is about comparing latin squares, nothing too complex.

Perhaps, but there is no news on this subject yet.

Thank you for participation!

Square checking can be moved to GPU quite easily, and it is blazing fast there. Unfortunately square generator running on CPU is too slow to feed GPU and keep it loaded - I tested my code on 1080 and its load was only about 3%. So generator also must be moved to GPU, what is more complicated, as you have to run multiple generators in parallel. Here things gets complicated - search space has to be divided somehow in smaller pieces, so few hundreds of them could run in parallel, Also keep in mind that every generator instance needs some memory (about 370 bytes), plus some memory for output buffer. And this everything must somehow fit into local memory on GPU, which is very limited (only 32KB per compute unit is guaranteed).

---

By now, we have processed 27.5% of all possible workunits. This resulted in 90 found graphs of ODLSs.

Is it possible to implement a progress bar into the server status page?

Felix, hello!

Is it possible to implement a progress bar into the server status page?

We added a "Research progress" section in "Computing status" column on Server status page.
This value shows a fraction (in %) of completed workunits (with canonical result!) relative to total number of workunits that can be generated in this research.

Happy crunching!

P.S. Found a small mistake and correct it. :)

great work, big thanks, i will switch soon to your project

Thank you, Felix! Good luck!

Oh thank you, a simple and yet effective idea! And I see it's constantly updated!

Do you have plans after finishing those possible workunits hoarfrost?
i don't know that much about latin squares, but is it possible to search another batch/ another rank?

Felix,

Do you have plans after finishing those possible workunits hoarfrost?

Future is unknown. If processing rate is stay current, research will be completed in ~ 550 days. This is a large amount of work by results processing and project maintenance. But very interesting work! Now we focused on it.

i don't know that much about latin squares, but is it possible to search another batch/ another rank?

Technically - yes. But research space for case of Rank 10 (for example) in many times larger and search can be only as partial. But now we fascinated by results of current research and focused on them.

Thank you for attention and participation!

1st September 2018 project reached a 30% milestone!