We have fed GPUGRID with some new WU to keep you guys busy this christmas. This time we are simulating the binding of a small ligand to a protein called CXCL12, member of the family of chemokines. This protein appears to be altered in cancer and other inflammatory diseases. Happy Chrunchmas!

GPU queue 	Unsent		In Progress
Short		0		129
Long		508		3,065

...

GPU queue 	Unsent		In Progress
Short		497		264
Long		507		3,063

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Merry Christmas Gerard & skgiven ! :)
And Merry Christmas and Happy Crunchmas to everyone !
Keep crunching !

---

[CSF] Thomas H.V. Dupont
Founder of the team CRUNCHERS SANS FRONTIERES 2.0
www.crunchersansfrontieres

Merry Christmas peeps.

---

Team Belgium

Merry Christmas to everyone, everywhere.

Merry Christmas to you (and all crunchers) as well to the people who make GPUGRID a noble project.

Thanks Gerard, this certainly saved our cards from starving during the holiday :)

As a side note: since the molecules are rather small (about 30k atoms), the GPU load is less than for most other long runs. On my GTX970 this results in 7x % GPU load and I expect similar values for other high end cards.

To combat this I now run 2 WUs concurrently using this app_config.xml:

<app_config>
<app>
<name>acemdlong</name>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>
</app_config>

This yields a GPU usage of 91%, which is pretty good under Win 8.1. Note that cpu_usage is 1 since I have set SWAN_SYNC = 0, so each tasks really uses a full core. The result is very nice: previously I was doing one WU every 20.4 ks, i.e. 2 WUs in 40.8 ks. Now each WU takes about 35.5 ks. This includes some time where only 1 WU runs, because the finished one is uploaded and a new downloaded. So let's say I return 2 WUs every 36 ks, which means a free throughput increase of about 12.5%. The power draw is not increased by this, as my card is power-limited (i.e. it runs at slightly lower clock speeds when the utilization is higher). The only drawback is loosing one more CPU core.. but one doesn't have to use SWAN_SYNC = 0 (which drops my GPU utilization by ~2%).

MrS

---

Scanning for our furry friends since Jan 2002

Hi.

Mery Christmas!!

The file: tcl84.dll is detected by the avast antivirus as suspicious file... and don't allow to work with it.

So... no WU in progress...

Bye!!

Hi.

Mery Christmas!!

The file: tcl84.dll is detected by the avast antivirus as suspicious file... and don't allow to work with it.

So... no WU in progress...

Bye!!

tcl84.dll has been in use for many months without suspicion or complaint. I strongly believe this to be a false positive introduced by an Avast signature update late on 31/12 or early on 01/01 - see my thread in Number Crunching.

Avast provides tools for temporarily disabling the download shield (so you can download a fresh copy to replace the one you've probably had since last April), and for marking the file as trusted and safe to use.

Thanks Richard!

I tought that it was a problem with a new kind of WU, more than to a problem with the new avast version.

I hope they will fix it soon.

We have fed GPUGRID with some new WU to keep you guys busy this christmas. This time we are simulating the binding of a small ligand to a protein called CXCL12, member of the family of chemokines. This protein appears to be altered in cancer and other inflammatory diseases. Happy Chrunchmas!

What is the reason for the simulation of this CXCL12 is taking so long?

We are testing the binding of 700 in-silico compounds to prove whether molecular dynamics (MD) is a suitable method for drug design. Several rounds of iterations are necessary to optimize both compounds and method.

In general is an extensive and computationally expensive work that has never been done before. So far, I can only tell you that we may have found some interesting compounds, but we have to validate them experimentally to prove their affinity. If we are right, this could have an enormous impact in the scientific community and in terms of publications.

Music to my ears. That is what fundamental research is for. Sometimes it works, sometimes it doesn't. But if you don't try new things, you never know.

Gerard wrote:

We are testing the binding of 700 in-silico compounds to prove whether molecular dynamics (MD) is a suitable method for drug design. Several rounds of iterations are necessary to optimize both compounds and method.

In general is an extensive and computationally expensive work that has never been done before. So far, I can only tell you that we may have found some interesting compounds, but we have to validate them experimentally to prove their affinity. If we are right, this could have an enormous impact in the scientific community and in terms of publications.

I hoped that you would answer something like that.

Music to my ears. That is what fundamental research is for. Sometimes it works, sometimes it doesn't. But if you don't try new things, you never know.

+1