Advancing and expanding Help Fight Childhood Cancer research

[PoppaGeek]

@WCG

By: Dr. Akira Nakagawara, MD, PhD
President Emeritus, Chiba Cancer Center
2 Jul 2014

Summary
Dr. Nakagawara, the lead researcher for Help Fight Childhood Cancer, provides an update on his continued efforts to advance his neuroblastoma drug candidate discovery, his transition to a new role and his hopes for expanding his research on World Community Grid to other pediatric cancers.


I recently accepted a position as CEO of the Saga Medical Center KOSEIKAN, located at Saga City on the island of Kyushu, Japan. The Saga prefecture is my home region, and Saga's governor asked me to come back to manage the region's biggest general hospital. A new research institute is being set up within the hospital, which will be affiliated with Kyushu and Saga universities in the near future.

My association with the Chiba Cancer Center will continue - I am now President Emeritus and will keep my research position at Chiba's research institute as an official councilor. The collaboration between the Chiba Cancer Center Research Institute and Chiba University will also continue. In this capacity, I aim to build on the success of the first phase of Help Fight Childhood Cancer (the discovery of several drug candidates which in lab testing killed cancer cells, with no apparent side effects). While we will be closing out this phase of the project on World Community Grid, we will continue our efforts to seek pharmaceutical collaborations to advance these promising drug candidates through the drug development process.

I also plan to expand our research. We have started planning for a second phase to the project, possibly in collaboration with a newly formed pan-Asian oncology group, of which we are a founding member. While Phase I of Help Fight Childhood Cancer focused on neuroblastoma, we hope to expand the scope of Phase II to cover multiple pediatric cancers. We are in early discussions to define the scope of Phase II and look forward to working with our oncology collaborators and with the World Community Grid team to bring this project to World Community Grid volunteers.

We are so grateful of the support of World Community Grid volunteers and the role you've played in powering our Phase I research. We hope we can count on your continued support and participation as we prepare for Phase II.

[Otis11]

Awesome news!!! Not only more crunching, but progress! (And potentially more resources from the scientists side!)

[PoppaGeek]

Thanks Otis, nice to see someone has something to say around here.

[Otis11]

Haha, yeah, it has been fairly quiet... where'd everybody go? Sound off!

[Evantaur]

hopefully it will run on GPUs

[Otis11]

hopefully it will run on GPUs

That would be awesome! I'm hoping the get something figured out soon in that area as I'm looking to completely revamp my farm in the next 2 years... would be nice to have some stability in GPUs to show it's going to continue before I jump in and invest significantly...

Plus, harnessing the extra work potential is just good for the project in general! (in the mean time - Nvidia go to GPUGrid, AMD I'm not quite sure where's best?)

[PoppaGeek]

IF they do release a GPU project it will be CUDA only. Why? Because I have 4 AMDs and 1 CUDA GPU.

Seriously, a GPU project would put this team back toward the top again.

[Otis11]

IF they do release a GPU project it will be CUDA only. Why? Because I have 4 AMDs and 1 CUDA GPU.

Seriously, a GPU project would put this team back toward the top again.

It actually is very likely to be CUDA - why? Because these are actually complex problems that are difficult to vectorize... CUDA has lots of support and great development features, OpenCL, well, doesn't. From a development and support perspective, they're much better off with CUDA, even if it sacrifices half their user base, they sadly have more power than they can use since they're likely to be the only GPU project on WCG for a while. (Likely to run out of work units to keep the GPUs all busy consistently anyway - unless they're seriously prepared for it)

[PoppaGeek]

No they will pick CUDA because I have AMD. Science and math have nothing to do with it. They hates us precious.

A WCG GPU project would be enough to get me to get that new GTX 970 I have been drooling over

[Otis11]

No they will pick CUDA because I have AMD. Science and math have nothing to do with it. They hates us precious.

A WCG GPU project would be enough to get me to get 10 GTX 980s I have been drooling over

I think you had a few typos...

[yojimbo197]

I think you had a few typos...

This from Mr. "Role Call"?

*cough cough*

[Mumak]

I think in news some time ago there was something about porting MCM1 to GPUs...
The new UGM1 (Uncovering Genome Mysteries) will be CPU only (at least for the start).

[Otis11]

This from Mr. "Role Call"?

*cough cough*

Huh... I had to google that. What'd'ya know...

(And I wasn't picking on him for grammar - I changed it to be 10 higher end GPUs instead of 1 high end.)

I think in news some time ago there was something about porting MCM1 to GPUs...
The new UGM1 (Uncovering Genome Mysteries) will be CPU only (at least for the start).

Aww... sad day...

Any idea how the first GPU based project went? Did they get good results and it all go smoothly? Or is there a reason they're not having another one in the mean time (Other than the scientists simply don't want to code it)?

[Mumak]

I haven't seen any further info about GPU..
There's about 1 year left for MCM1, if they would launch a GPU version we'd complete it in a month or so...
So maybe it's also a bit of politics - if all projects on WCG would be GPU, they all would be finished already and WCG would be empty

[Otis11]

I haven't seen any further info about GPU..
There's about 1 year left for MCM1, if they would launch a GPU version we'd complete it in a month or so...
So maybe it's also a bit of politics - if all projects on WCG would be GPU, they all would be finished already and WCG would be empty

But I feel like they would just accept more projects... I don't think there's a shortage of projects interested. They just limit the new projects in efforts to maintain enough resources for the current scientists. (at least to my understanding - they're concerned with getting a bunch of new projects and then old projects starting a phase 2 or 3 at the same time...)

[Sparky]

My 7950 stays idle far too much...

[stoneageman]

My 7950 stays idle far too much...

Bring it over to Einstein@home. We're doing quite well there. Currently #1 team by RAC

[jeanguy2]

Bring it over to Einstein@home. We're doing quite well there. Currently #1 team by RAC

Sam, What is the Team's name? Do you have link to its home page? Can you crunch WCG CPU WUs with a rig that has its GPU(s) committed to Einstein?

What are the steps to follow to join you guys and set it up?

[Mumak]

Team is XtremeSystems.
Sure, you can crunch WCG CPU WUs along with E@H GPU WUs. For AMD GPUs make sure to leave one core spare for each GPU/WU (which is not default). They have CPU WUs too, so if you want to crunch on GPU only, make sure to select BRP4G or BRP5 tasks.
Some more info here: http://www.xtremesystems.org/forums/...=1#post5238591

Also, their GPU tasks utilize the GPU only up-to 75%, which is a safety feature. So if you want a better RAC make sure to run multiple WUs (2-3 / GPU). You don't need an app_config for this, you can define this via their Preferences ("GPU utilization factor of BRP apps").

BTW, if your machine finds a pulsar (BRP4 or BRP5), you'll get a certificate from them

[PoppaGeek]

Einstein@home

[jeanguy2]

Mumak and PG,

Thank you very much for quickly answering my post. This is exactly what I was looking for.