Announcement: SMP2 Cores

The servers are getting slammed right now.
Thousands of users trying to download the new A3 cores.
Will get better over the next few hours.
 
I'm getting about 1360-1370PPD on my E2140 according to HFM.NET. I was hoping for a little better, since it does about 2000PPD on a2 workunits. I'll be switching my Q9550 over as soon as it finishes its current unit, so hopefully that will do a little better.
 
Ok, is that 128mb/thread(core) or 128 total? (memory usage)

And am I reading this right that on a dedicated folder I can disable HT and use smp4, get the same ppd with less draw? What if I've got a gpu on the rig? Leave HT on?
Posted via [H] Mobile Device
 
Ok, is that 128mb/thread(core) or 128 total? (memory usage)

And am I reading this right that on a dedicated folder I can disable HT and use smp4, get the same ppd with less draw? What if I've got a gpu on the rig? Leave HT on?
Posted via [H] Mobile Device

I would turn HT off if this is a dedicated folder.

Im doing some testing to see if I can A) see a PPD diff on my GPU clients B) affect my use on my daily driver
 
Let me know what you find. I'm on orders now, but next week when I get back I'll use this to double my ppd

Think A1 will be gone by then and A3 willbe easy to get?
Posted via [H] Mobile Device
 
If that's the case, how long will 2 be around? My biggest hope for A3 was a simple app, like the UNI client

Regardless, this should bring me near 85k, get my next i7 and top 100
Posted via [H] Mobile Device
 
what about running the SMP2 client along with the GPU2 client? I'll be getting my i7 920 up and running this week along with a 2Gb 4890. Will i suffer a performance hit from running both clients simultaneously?
 
O ya, would 2 gigs be enough for A3, max of 2 gpu's, and XP64? If so, I could build 2 more i7rigs with extra memory from my bigadv build (1 2g stickin each case vs 3 sticks per case)
Posted via [H] Mobile Device
 
If that's the case, how long will 2 be around? My biggest hope for A3 was a simple app, like the UNI client

Regardless, this should bring me near 85k, get my next i7 and top 100
Posted via [H] Mobile Device
Do you mean how long will A2 be around, or how long will SMP2 be around? If you mean the former, it seems that the A1 and A2 cores will be phased out completely in the near future.
what about running the SMP2 client along with the GPU2 client? I'll be getting my i7 920 up and running this week along with a 2Gb 4890. Will i suffer a performance hit from running both clients simultaneously?
You should be able to run both simultaneously with no problems.
 
my A3 is using around 450MB of ram.

These A3s are not Bigadvs, when those are released then the memory usage will go back up.
 
Last edited:
Heres what i got with my first A3

SM X8DAL-i
2x E5530's ES(no turbo) with HT on
12GB(6x2) Gskill 1066MHZ
2x EVGA GTX 275's
Win7 Ultimate x64 Fresh install

Running the two GPU clients with priority set to high.
WinSMP 6.29 Project 6013(R0, C10, G123) = 2:44 TPF 14638.3 PPD based on last three frames according to HFM.net 0.4.8
 
I'm seeing ~80MB of RAM usage with a P6012.

Same here, about 80,000 K. Firefox takes about as much. Just read this thread about a half hour ago so I reinstalled Windows SMP, shut off my VM and I got an A3 right away. It's running without problems so far.

HFM.net 0.4.8 is reporting about 8,000 PpD so far at 3.8 GHz on the system in my sig. I just got my passkey a few minutes ago though so I won't be getting completion bonus yet, I'm not sure if HFM knows that or not. My TBF on P6012 is approximately 5 minutes.
 
OvrrDrive: your problem should be fixed now, there was one server @ Stanford that was missing the core file.
 
It picked it up during the night and I'm up and running. ;)

If HFM is right it should be good for an extra 1000 PPD.
 
The first SMP2 projects are (as usual) reruns of old projects, designed to test the validity of the results produced by the new scientific code. These first projects are:

  • 6012: 77,445 atoms, preferred deadline: 3 days, final deadline: 6 days, 470 points with a bonus multiplier of 2, studying the human antibody AFP3 (HAFP3). It seems that it has not yet been studied by SMP1, but we have no more information about this antibody.
  • 6013: 146,808 atoms, preferred deadline: 3 days, final deadline: 3 days, 380 points with a bonus multiplier of 3.36, restudying the series "IBX in Water" that formed SMP projects 2662 to 2677.
  • 6014: 77,484 atoms, preferred deadline: 3 days, final deadline: 6 days, 484 points with a bonus multiplier of 2.1, restudying the series 'Protein in POPC' that formed SMP projects 2651 and 2653.
  • 6015: 77,457 atoms, preferred deadline: 3 days, final deadline: 6 days, 484 points with a bonus multiplier of 2.1, restudying the series 'Protein in POPC' that formed SMP projects 2651 and 2653.
All of these projects are distributed from a new server situated in Stockholm, Sweden, at the PDC Center of the Royal Institute of Technology. The IP of the server is 130.237.232.140. Peter Kasson is in charge of the server and these four projects.
 
Currently got my X2 SMP crunching a 6012 using the A3 core. PPD is about half so far of what I got with the Linux VM client :(
 
Currently got my X2 SMP crunching a 6012 using the A3 core. PPD is about half so far of what I got with the Linux VM client :(

What monitoring program are you using to calculate PPD?
 
Hmm, so 2 gigs is enough for a system... That wont affect OC ability will it, since its supposed to be tripple chanel and all that
 
try HFMnet, some seem to have better success, plus it serves as a way to double check your findings
 
Okay, I'll give HFMnet a whirl, then :)

Cool, HFM net appears to be more like FahMon, can even import its list of clients. Gives 525 PPD for the A3 SMP2.
 
Last edited:
Anyone seeing any stability issues with the new wu's?

I built this machine about a month ago and have been folding 24/7 for about 3 weeks with it and have been completely stable. I updated the client on the day 6.29 came out and got my first new a3 core unit today and just got home for lunch and the machine was locked up. I have been running it slightly oc'd (3.0ghz on an unlocked PII 720) and today is the warmest day of the year so far but the cpu is at the same temp it has been at for days. I'll try finishing a few units at stock clock to see if it is stable for me there before I try oc'ing it again.

I'll keep an eye on it...
 
I will give these a try on my two rigs currently running linux VMs. It'll be Friday as I am on the road till Thursday evening. I am hopeful that this works as I would love to move away from VMs.

 
The best part for me is that i am getting the same PPD i was when running a bigadv VM.
And now i dont have to run a VM.
 
Back
Top