• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

A3 core/ SMP2 PPD & config thread

nitrobass24 said:
Anyone else noticed significant slowdowns on 6015's?
Click to expand...

6015 is a weird WU. It does have a lower ppd.

This will improve as the core revisions go on.
There is much work to be done on the way the threads interact with each other (Windows & Linux)
 
Last edited:
Elledan said:
Yeah, same experience here. Running an X2 4450e I'm getting ~550 PPD compared to the ~950 PPD previously with the Linux SMP VM client.
Click to expand...

I believe they are using an Intel Compiler for windows based systems - Same as with the protomol core (Core B3)
That would explain the lower ppd on AMD systems in a windows system.
(Widely documented that the current Intel Compiler does not work well with AMD processors)
If available, you can try the Linux A3 and report an difference in A3 performance.
Would be good information for the team to have.

I think Intel is working on a new complier per the Intel vs AMD settlement of last month.

The bonus does come into play as well.

SMP2 is a complete rewite of the rules of Folding@Home Points Per Day.
Going to take a little while to adjust.
 
These A1 core WU are horrible.

I hope they don't have too many of these stacked up.
 
EvilAlchemist said:
I believe they are using an Intel Compiler for windows based systems - Same as with the protomol core (Core B3)
That would explain the lower ppd on AMD systems in a windows system.
(Widely documented that the current Intel Compiler does not work well with AMD processors)
If available, you can try the Linux A3 and report an difference in A3 performance.
Would be good information for the team to have.

I think Intel is working on a new complier per the Intel vs AMD settlement of last month.

The bonus does come into play as well.

SMP2 is a complete rewite of the rules of Folding@Home Points Per Day.
Going to take a little while to adjust.
Click to expand...

Is Notfred already updated to the new Linux SMP2 client? Else I'd need instructions on setting up the Linux client from scratch :D
 
Elledan said:
I'd need instructions on setting up the Linux client from scratch :D
Click to expand...

Setting up a VM from scratch is fairly easy.
When you start a new VM point it at your Linux64.iso
When it installs you can strip out as many modules as you like, but make sure the Windoze share/samba is installed.
Set the samba share up <- I found that was the hardest part first time around .
Once your version of Linux is up and running, download Finstall and use that to install the SMP client.
Set the client up as a service and mod the default flags as needed.
Now your ready to fold ......... :p

Luck ................ :D
 
Finstall?

I'll try to find some time tomorrow to try it :) Just got done fixing an EUE issue with a 9800GX2 core (I hope). One issue at a time :D
 
Elledan said:
Finstall?
Click to expand...

Look in the first post of the bigadv sticky for more info on how to use Finstall to install the smp client.
It makes installing the client as a service sooooooo easy.

Luck ............ :D
 
Elledan said:
Yeah, same experience here. Running an X2 4450e I'm getting ~550 PPD compared to the ~950 PPD previously with the Linux SMP VM client.
Click to expand...

I'm seeing the same thing also. Running a X2 3800+ I'm getting ~525 PPD compared to ~1200 PPD previously with Notfred. I'm going to stick with Notfred until the A1 and A2 units run out.

SMP2 obviously favors Intel ... looks like the party's over for AMD :(
 
Last edited:
Mr. Pedantic said:
I think I'll stay with my Notfred VM until the A1s have been completely phased out.
Click to expand...

Plan to wait until Q3 2010 if you are going that route.
It will take months to get user over to SMP2 and start winding down the older projects.
 
Last edited:
We still need to finish all the A1 projects. They are important like the uniprocessor WUs. I've folded my share of the A1 units and will continue to fold them. The only concern is receiving them on 8-core (or higher) systems. Since the A1 WUs are limited to 4 core processing, it would be great if the servers can distinguish which clients should be sent these WUs to prevent idle cores. That has happened to me on many occasions forcing me to start secondary SMP clients since the A1 units take so long to complete, and don't want to have idle cores.
 
APOLLO said:
...it would be great if the servers can distinguish which clients should be sent these WUs to prevent idle cores. .
Click to expand...

The AS servers were upgraded a while back to do things simular to this.
The client on the other hard does not have those "reporting" features built in yet.
 
EvilAlchemist said:
The AS servers were upgraded a while back to do things simular to this.
The client on the other hard does not have those "reporting" features built in yet.
Click to expand...

It will also be nice when they package the SMP client with a handy front end like the single core and GPU client.
 
AgrFan said:
I'm seeing the same thing also. Running a X2 3800+ I'm getting ~525 PPD compared to ~1200 PPD previously with Notfred. I'm going to stick with Notfred until the A1 and A2 units run out.

SMP2 obviously favors Intel ... looks like the party's over for AMD :(
Click to expand...

I found these numbers on techpowerup:

Phenom II x4 FX-5000 @ 3.2Ghz - 5900 ppd
Phenom II x4 FX-5000 @ 3.0Ghz(feeding 2 GPU's) - 4800 ppd
Phenom x4 9850 @ 3.0ghz(feeding 4 GPU's) - 4100 ppd

AMD does well with A3. It's just because AMD is usually slower than Intel. With the bonus that makes the lower CPUs look crappy.
I'll try A3 on my Core2Duo e1200, I'm sure the score will be pretty low too.
 
Okay, seeing ~6220 PPD on my Q6600 @ 3.1 GHz. Definite improvement over the Linux VM by ~1k or so :)

<<-- Happy kitten :p
 
Just switched my q9450 over.
Of course the VM I was running that I tried to "gently" :D quit to add the -oneunit flag just ate itself and shat all over the WU it was running that had less than an hour to go (@ 89%). [sigh]

Oh well....running an A3 right now. w00t! :D
 
zero2dash said:
Just switched my q9450 over.
Of course the VM I was running that I tried to "gently" :D quit to add the -oneunit flag just ate itself and shat all over the WU it was running that had less than an hour to go (@ 89%). [sigh]

Oh well....running an A3 right now. w00t! :D
Click to expand...

Yeah, I had one VM do that to a WU at ~73% as well :(
 
I feel much better about the Windows SMP client. It feels a lot more stable than the Windows SMP client that made me switch to Linux (though to be honest, a doubling in ppd probably helped that decision), the PC has crashed twice and it hasn't lost a work unit yet, I've had no problems with having to run install.bat for minor problems, I don't need to work out how to run two clients in parallel for my computer to fully load, I could go on and on...
 
I'm not sure if something got changed or not but I've had about 3 days perfectly stable running a3's finally. I'm getting just over 5000ppd on my unlocked PII 720BE. Between that and the PS3 I'm finally making pretty good time in the ranks. ;)

I was just about to give up...
 
OvrrDrive said:
I'm not sure if something got changed or not but I've had about 3 days perfectly stable running a3's finally. I'm getting just over 5000ppd on my unlocked PII 720BE. Between that and the PS3 I'm finally making pretty good time in the ranks. ;)

I was just about to give up...
Click to expand...

Never give up. For all the talk in here about big ppd numbers, we know that the lower ppd folder makes up the backbone of out team. They are the reason we are the #1 team in the world.
Fold on!
 
True that, never give up, there is always someone here who can help if you need it
Posted via [H] Mobile Device
 
Never gonna give you up,
Never gonna let you down,
Never gonna run around and desert you.
Never gonna make you cry,
Never gonna say goodbye,
Never gonna tell a lie and hurt you.
 
Ok.

I've got installed on a C2D E8400 and it is producing about 3K PpD (accoding to FHM)
I've got installed on a C2Q QX9300 and it is producing 825 PpD (according to FHM).

The C2Q is a mobile processor and is at 2.53 (4 core) vs 3.0 (2 core), but I would not have expected the huge difference in PpD. When I was running Linux VMs, I definitely got better PpD than that. I am running WinAFC to ensure that priority is normal when I don't neeed anything and below normal when I do. This WU just started and I am getting TPFs of 20 mins (Project 2021).

Am I missing something?
 
My Q9550 (desktop model) is only averaging 1194 PPD on this new client, Tobit told me it goes by a bonus system. I dunno.

I am in the same boat as capreppy I think.
 
Tobit said:
Never gonna give you up,
Never gonna let you down,
Never gonna run around and desert you.
Never gonna make you cry,
Never gonna say goodbye,
Never gonna tell a lie and hurt you.
Click to expand...

AAAACK! We've been Rick Rolled!
 
I think that you have to be running some of the last of the A1 cores.
 
I would have thought that, but it's currently running an A3 core :(

edit: The other laptop (C2D P8600 @ 2.4) is getting better PpD than this laptop!!!!
 
Last edited:
My stock Q6600 is averaging 9min per frame on A3 so something is definitely wrong with your QX9300.

Is the laptop throttling itself?
 
My Q9550 is averaging 15-17 minutes each, sheesh, something defenitly must be wrong if your averaging 9min per frame!
 
My Q9550 averages between 5 and 6 minutes per frame depending on whether I'm using my PC or not.
 
FrEaKy said:
My Q9550 is averaging 15-17 minutes each, sheesh, something defenitly must be wrong if your averaging 9min per frame!
Click to expand...
Hmm... dunno what I'm doing right then. I thought 9min was pretty slow but normal for my situation.

Q6600 @ 2.4GHz (Dell T3400 so no O/C), Windows 7 Enterprise x64, Centos 5.4 VM in VMware WS 6.5 w/ 4 vCPU.

Snippet from beginning of current WU:
[01:15:31] Folding@Home Gromacs SMP Core
[01:15:31] Version 2.13 (Tue Aug 18 18:28:33 CEST 2009)
[01:15:31]
[01:15:31] Preparing to commence simulation
[01:15:31] - Assembly optimizations manually forced on.
[01:15:31] - Not checking prior termination.
[01:15:31] - Expanded 1769888 -> 1971489 (decompressed 111.3 percent)
[01:15:31] Called DecompressByteArray: compressed_data_size=1769888 data_size=1971489, decompressed_da
ta_size=1971489 diff=0
[01:15:31] - Digital signature verified
[01:15:31]
[01:15:31] Project: 6020 (Run 0, Clone 10, Gen 15)
[01:15:31]
[01:15:31] Assembly optimizations on if available.
[01:15:31] Entering M.D.
[01:15:38] Completed 0 out of 500000 steps (0%)
[01:24:25] Completed 5000 out of 500000 steps (1%)
[01:33:07] Completed 10000 out of 500000 steps (2%)
[01:41:54] Completed 15000 out of 500000 steps (3%)
Click to expand...
 
NO NO, Im saying something is wrong with ME

If I am averaging 17 min with a faster CPU, and your averaging 9 minutes with a Q6600, than something is wrong on my end, I should be running better than 9 minutes.

uhoh.jpg
 
FrEaKy said:
NO NO, Im saying something is wrong with ME

If I am averaging 17 min with a faster CPU, and your averaging 9 minutes with a Q6600, than something is wrong on my end, I should be running better than 9 minutes.
Click to expand...
That's what I thought. I just posted my log just to confirm I wasn't crazy if it was an absurd number for a Q6600. ;)
 
I don't know how you guys get more than 5 minutes per frame for a quad-core. My i7 gets around 3 minutes per frame.
 
You must log in or register to reply here.
Back
Top