Welcome MagicPhysicitst!

Welcome!
 
http://stats.free-dc.org/stats.php?page=teambycpid&team=Crunching%40EVGA  You'll find a list of BOINC projects we have a team for.  If there are any you're interested in and we don't have a Crunching@EVGA team for, make a post and one of us will get it created for you.
 
http://www.dc-vault.com/showteam.php?team=539  Most of our efforts are focused on some of the projects here, not all are BOINC....

Nevermind, content deleted.

So I've been running The Dnet client for you guys for a month or two now.
 
Seems I'm the only one lately?
 
Anyway.. ideas on gnodes per day or blocks per day on CPU/GPUs. 
 
Incase anyone was thinking to run it again.  Some benchmarks on what can be run:
 
Running OGR  (all 64bit client):
-i7 3930k @ 3.8---   ~60,000 units/day  aka ~655,000,000 nodes/sec
-i7 920 @ 2.8---  ~25,000 units/day aka ~265,000,000 nodes/sec
-Phenom II X4 920 @ 2.8 ~15,000 units/day aka 185,000,000 nodes/sec
 
AMDs are the Video card to use for RC5... but nVidias will work.
 
Require CUDA client for 3.0:
-GT440 stock  52,000,000 keys/sec (unsure on daily)
-GTX285 stock 285,000,000 keys/sec    ~7,000 units a day (think I even had it running with 197 drivers and the 3.0 client with no issues)
 
Require CUDA 3.0 client or OpenCL beta client, but give about the same output with both client types:
-GTX580 stock ~800,000,000 keys/sec ~ 17,000 units a day
-GTX 680 stock ~580,000,000 keys/sec ~11,000 units a day
 
Require the Stream client and 11.11 driver or older (being 4800s) otherwise they do 2/3 this output given with any of the 12.x drivers.   10.2 did the most output, but made the cards real hot compared to 11.11, and they use less CPU each with 11.11:
-HD4850 stock ~485,000,000 keys/sec or ~10,000 daily
-HD4850 in crossfire stock ~ 965,000,000 keys/sec ~20,000 daily
 
Only works with Opencl beta client (as of this posting, it may go release soon):
-HD7870 @1000  1,900,000,000 keys/sec ~40,000+ (I run mine at 800 getting 1,500,000,000 / 30,000 per day due to blowing up the first HD 7870 in a week).

All are personal benches!
 
Both CPU and GPU clients can run together.   I just set the affinity in the task manager  for the CPU client not to use 1 core, or 2 cores if running Crossfire.
 
  I ran both projects for a long time now.   RC5 I started with Pentium 233 MMX's.. which did maybe 250.000 keys/sec.. lol   I took a break from RC5 when OGR-26 came out, and ran with OGR-27 to this day.   RC5, I just got back into it when someone gave me a HD4850, and it couldn't run Folding so I tried and got back addicted to running RC5.  
 
My GTX 680 is my dedicated folder, and on the first few days of the month, I try to burn through the "bucks" points with the 3930k running with the 680.   The last TZC I ran .. all I could, even the 7870 with its 7300 pdd, until EST finish lined anyway.    
 
 
 
 

Welcome, carldxpc!  You have been tearing it up and gaining the team quite a few places.    There's not a ton of dnet specific talk in the forum, although from time to time it will pop up in the "Plundering the DC Vault" thread.
 
That's some very useful info you posted.  I'll have to look at getting the OpenCL prerelease client, as I did try to benchmark a 7770 and it immediately threw an error and reset the card.
 
I'm like you, I cut my teeth on RC5 back in 2000 and then have participated in OGR-24 and on up.  I like the definite endpoint of OGR, plus I think the math discoveries are useful - even in the biomedical field.  I love the client, it's like Boinc before Boinc.  And there are still things which I think are easier to do with the dnet client.   As usual for me, there are too many projects and not enough hardware.  (But what is enough hardware?  )

Welcome carldxpc!  Nice to have new blood aboard.  Hope you find your stay here fruitful and fun.

Hi all!  I've been folding now for about 3 years, the last year and a half with the EVGA team.  I'm looking to branch out into Crunching and was surprised/delighted to learn that EVGA also has a Crunching team.  I have not really selected a project of choice yet, although I did try for a while.  Are there certain projects where I can best help the team out rather than just doing my own thing?

Howdy

Welcome aboard!
If you have any questions, just holler out!

Hello everyone New to this. How do I get started?

Hey Punchy good to see you again.
 
Thanks, will go throug the guild to get started.

boballee

Well, decided to toss my hat in to the Crunching realm (just started up the other day).  Currently working on Einstine and LHC.  Nice to support something that is more of a passion than helping to find cures.  I'm using my "weaker" rig (an i3-2120 and 580 SC).  I do have to say, I like that I can set it to Crunch only during certain times.  Comes in handy since it's the 'family' computer. 

Looking forward to learning more about this side of DC.

 
Welcome and thanks for joining. For running your GTX 580 under Einstein, it helps performance to have one physical CPU core available for the GPU application as the application is a hybrid type that does some processing on the CPU and GPU. You can also run Process Lasso and give the GPU application higher priority. Also, having the 580 installed in a x16 2.0 slot will give the best performance and can be up to 30% faster than when installed in an x8 2.0 slot.

Thanks, linux. Made a few tweaks and we'll see how it goes. The card is in a x16 slot. Question though. I'll soon be switching this rig over to a GT 640. I assume I'll still be ok to Crunch on this, albeit with a decrease in PPD?

Joing Crunching@EVGA-
 
User: Sha Dizzle  - add me!
Docking@home  - 430864
Rosetta@home  - 305855
GPUGRID - 20532100
 
 
How do I get one of those stat counters at the bottom of my postings?