Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 40
Posts: 40   Pages: 4   [ Previous Page | 1 2 3 4 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 218514 times and has 39 replies Next Thread
robertmiles
Senior Cruncher
US
Joined: Apr 16, 2008
Post Count: 444
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

That's what I had in mind. On some other BOINC projects, I've seen significant speed differences reported for 32-bit vs 64-bit, Windows vs Linux, and even Intel vs AMD CPUs.
[Dec 21, 2009 1:17:01 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Sekerob
Ace Cruncher
Joined: Jul 24, 2005
Post Count: 20043
Status: Offline
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

~14% I get from Vista 32 bit to W7, 64 bit on the same hardware with the mixed set of WCG projects. Still to install in dual boot the 32 bit W7... found the manual to do that, but wont be doing that until next year as it requires the repartitioning of the drive. The supposed easy W7 repartition allows me presently to reduce the C drive by 7Mb, when having 150Gb free on a fully defragged, drive and contiguous free space... whatever happened to partion magic?
----------------------------------------
WCG Global & Research > Make Proposal Help: Start Here!
Please help to make the Forums an enjoyable experience for All!
[Dec 21, 2009 1:30:27 PM]   Link   Report threatening or abusive post: please login first  Go to top 
robertmiles
Senior Cruncher
US
Joined: Apr 16, 2008
Post Count: 444
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

I wouldn't be surprized if W7 has something I've read about for Vista - the swap space file is kept near the high end of the C: disk space even when everything else is contiguous and at the low end. In that case, it MIGHT help to reduce the size of the C: partition as much as you can, reboot, and try this again.

Also, it's likely to help to create another bootable Windows partition, boot that one, then find what's normally your C: partition and try to reduce its size.
----------------------------------------
[Edit 1 times, last edit by robertmiles at Dec 22, 2009 3:32:19 PM]
[Dec 22, 2009 3:29:05 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

Thanks a lot. Nvidia and ATI will soon be leaders in supercomputers ... or I hope so.


Judging by what I see, I'd say that Nvidia will soon be a leader in supercomputers. ATI might eventually be a leader also, if they ever catch up in providing compilers for computer languages such as C++ and Fortran.


You should check out ATI Stream Technology
[Dec 28, 2009 2:46:22 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

Possibly somebody would like to make it reality and graduate!

Graduate Fellowship Program
NVIDIA´s Graduate Fellowship Program provides funding to PhD students researching topics that will lead to major advances in the graphics and high–performance computing industries by leveraging the power of the GPU. Applications due by 2/3/10.

Learn more: http://is.gd/5WqTZ
[Jan 18, 2010 3:57:27 PM]   Link   Report threatening or abusive post: please login first  Go to top 
robertmiles
Senior Cruncher
US
Joined: Apr 16, 2008
Post Count: 444
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

Thanks a lot. Nvidia and ATI will soon be leaders in supercomputers ... or I hope so.


Judging by what I see, I'd say that Nvidia will soon be a leader in supercomputers. ATI might eventually be a leader also, if they ever catch up in providing compilers for computer languages such as C++ and Fortran.


You should check out ATI Stream Technology


I did. Looks like they recently started offering an OpenCL compiler, but with no sign that they have one for either Fortran or C++. Still looks like they have a lot of catching up to do in compilers.
[Jan 18, 2010 7:16:12 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

Accelerating the Pace of Drug Discovery using GPUs

http://blogs.nvidia.com/ntersect/2010/01/acce...discovery-using-gpus.html

Nice talk, but no real details
[Jan 31, 2010 11:06:14 PM]   Link   Report threatening or abusive post: please login first  Go to top 
fablefox
Senior Cruncher
Joined: May 31, 2010
Post Count: 168
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

I think they replied to (probably) your question. They give this link

http://www.nvidia.com/object/tesla_bio_workbench.html
----------------------------------------
[Jun 2, 2010 3:45:09 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

I will also become a full time WCG cruncher when it gets a GPU Client or use the GPU capability of the BOINC client (v. 6.10.58), meanwhile, If you want to do some protein research in the same style as WCG then you can join F@H GPU Cleint

F@H Support Forum


to configure the GPU ATI Client (I have 2x Radeon 4670HD):
1) create the following environment variables:

- BROOK_YIELD = 2
- CAL_NO_FLUSH = 1
- CAL_PRE_FLUSH = 1
- FLUSH_INTERVAL= 128 to 256
on ATI Radeon 4000 series, similar or higher values for 5000 series.

1.1) To configure, Right-click on the GPU client and click on 'Configure...'
- If you want to join ATI Team join team #51394.
- On 'Connection' tab, Check the allowance of 10MB WUs.
- On 'Advanced' tab, check 'do NOT lock cores to specific CPU' for the GPU client to use any core.

With these values only 5% max of CPU usage is taken (on a 4 core CPU), allowing your GPU to work in parallel with your CPU.

2) If you have more than one video card in a non-CrossFire/SLI mode, if you have have only 1 Video card skip this step:
2.1) For the first GPU (Video Card), modify the 'direct access' with something like this:
"C:\Program Files (x86)\Folding@home\Folding@home-gpu\Folding@home.exe" -gpu 0 -advmethods -forcegpu ati_r700

+ use '-forcegpu ati_r700' to force the ATI GPU Client to recognize/use a Radeon 4000HD series GPU, this will prevent some VPUs (GPU hangs).
+ '-advanced methods' will help more to the project, solving more complex proteins with advanced algorithms.
+'-gpu 0' is the default setting, specified here for clarity purposes.

2.2) In the 'direct access' 'Start in' text box you'll have something like this:
d:\ProgramData\Folding@home-gpu


2.3) so for the second GPU you need to copy-paste the direct access and modify it like this (mods are in bold ):
"C:\Program Files (x86)\Folding@home\Folding@home-gpu\Folding@home.exe" -gpu 1 -advmethods -forcegpu ati_r700

in the 'direct access' 'Start in' text box you'll need to modify it to have something like this:
d:\ProgramData\Folding@home-gpu1

Now you need to go 'd:\ProgramData\' and create the actual data folder/directory for the second GPU: 'Folding@home-gpu1' and copy-paste the content of the first/original GPU Client data folder/directory. Yeah, both will crunch the same first work unit, but they'll get different new ones once they've finish with it.

In Windows 7, You'll need to have a dummy monitor plug or an actual monitor connected to the second Video card in order to have that Video card ready for use. So you only need 3 resistors connected like this:


What it does is to get each color (red, green, blue) to their respective grounds.

The latest GPU Client for nVidia is 'GPU3' (for ATI is still GPU2) and it looks like it doesn't need a dummy connector.

2.4) Once the second GPU Client is running, right-click configure like you did with the first GPU Client.

For nVidia & ATI installation guide

Related: standalone memory tester for ATI+NV GPUs

I hope this post is not outside the Terms of Service.

Cheers.

----------------------------------------
[Edit 6 times, last edit by Former Member at Nov 5, 2010 12:00:14 PM]
[Nov 5, 2010 11:27:06 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Sekerob
Ace Cruncher
Joined: Jul 24, 2005
Post Count: 20043
Status: Offline
Reply to this Post  Reply with Quote 
Re: Autodock & CUDA

WCG for 99.999999999999999999999999999999% sure will only run research apps within the BOINC framework. Personally, I think the client is still not ready for the not tech savvy public. Continue to see alpha mail list reports of scheduler issues... idling GPU cards... idling CPU cores, overscheduling and whatnot.

And when I did it briefly to test, my case and CPU fans really started running in hoover howling max mode. Not for whimps and Saul's post shows so... not out of the box.
----------------------------------------
WCG Global & Research > Make Proposal Help: Start Here!
Please help to make the Forums an enjoyable experience for All!
[Nov 5, 2010 11:47:33 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 40   Pages: 4   [ Previous Page | 1 2 3 4 | Next Page ]
[ Jump to Last Post ]
Post new Thread