P8104 is coming down!

-alias-

Limp Gawd
Joined
Jul 15, 2012
Messages
374
Then I got my first P8104 on the 4P 3.0GHz 6276 rig, TPF is 6:53 with a PPD of 635K which almost the same as the 8103 on my 6276 with TPF on 9:05 give, so this looks good so far.
 
Last edited:
Beta or general?

I need to know if I have to shut off the slower machines.
 
10.5 hour! Prefered deadline is 1.8 days, final is 3.

Edit: Sorry, 11.5 hour is correct!
 
Tonight I'll turn on the beta flag on the two 16-core machines: 2.9 and 3.0ghz

When I pick one up, I'll see if the minimum equip req has went up.
 
I'm folding one on a 32-core 4P and another on a 48-core machine. The PPD for both machines are within 3% of 8103 performance on the same computers. Looks like benchmarking was spot on, at least from the two units I've seen so far.
 
Last edited:
Then I got my first P8104 on the 4P 3.0GHz 6276 rig, TPF is 6:53 with a PPD of 635K which almost the same as the 8103 on my 6276 with TPF on 9:05 give, so this looks good so far.


Got one on my [email protected] (with no kraken :D) ... tpf is 8:09 --> PPD is 493k.

With kraken turned on, TPF is now 7:12 --> PPD 577k
 
Last edited:
on 2P 6274@3GHz (with no kraken) -- TPF is 14:30 > PPD is 200K
 
Last edited:
Then I got my first P8104 on the 4P 3.0GHz 6276 rig, TPF is 6:53 with a PPD of 635K which almost the same as the 8103 on my 6276 with TPF on 9:05 give, so this looks good so far.

00:06:53 vs 00:09:05

1.8 vs 2.4

Looks to be a wash. Interesting.
 
Picked up 3 of these:
Dual E5-2650 (2.0 Ghz Stock Clock): 12:32 TPS 258kppd
Dual E5-2650 (2.0 Ghz Stock Clock): 13:18 TPF 236kppd
Dual E5-2420 (1.9 Ghz Stock Clock): 18:33 TPF 143kppd
 
Last edited:
Picked up 3 of these:
Dual E5-2650 (2.0 Ghz Stock Clock): 12:34 & 13:18 TPF
Dual E5-2420 (1.9 Ghz Stock Clock): 18:35 TPF

HFM isn't giving me any credit or ppd calculations for these WU's yet.

Change your psummary to psummaryC and then have hfm download projects.
 
Minimum LGA1366 requirements are dual hex's at 2.26. That should give a deadline of 1.67 days. 2.13 is doable but you need very fast upload time.

If you really must run quads you need a clockspeed of 3.2 but again very fast internet, a more comfortable margin would be 3.4
 
Edit>Preferences>Web Settings>Project Download URL (middle of window)
 
Haven't seen one yet, but those with marginal P8101 TPF's need not worry. If you were 33:00 TPF on P8101 (34:30 limit), you should be 22:00 TPF on P8104 (25:55 limit) based on reported times.
 
Here's the final stats from my first 8104 that just finished this evening:

Rig = 4P, 6180, 2.5GHz, dedicated Folding, only F@H running.
OS = HardOCP 12.04 LTS with script.

Final TPF across all 100 frames = 6:59
Final Points = 300,864
PPD = 619,806

Same Rig with an 8103 WU:
Final TPF across all 100 frames = 9:16
Final Points = 395,306
PPD = 613,406

With these two WUs, the 8104 wins the PPD race by a margin of 6,400 per day.
 
Well after getting 4 8103's in a row I got a 8104, with a bit of downtime in the day I still managed to get 210K points from it, got another 8104 to chew on for now.

I get 14:10 TPF on the 8104's with 2P 6276 @ 3Ghz (no kraken)
 
My SR-2 with 2x5645 at 3.6GHz got down a P8104. TPF = 13:38 with a PPD at 228K. Compared with P8103 that has a TPF = 18:50 given a PPD of 212K, this is very good. It looks to me that Intel CPUs maybe is folding the P8104 better then AMD CPUs, but it is possible to early to conclude.
 
8104 is horrible on my 2x6274 ES.

8101 = 17:10 TPF = 285k PPD
8103 = 13:28 TPF = 325k PPD
8104 = 10:15 TPF = 246k PPD

This is with only 1xUnit (Run 0, Clone 70, Gen 0), but I'm not liking the look of it so far.
 
8104 is horrible on my 2x6274 ES.

8101 = 17:10 TPF = 285k PPD
8103 = 13:28 TPF = 325k PPD
8104 = 10:15 TPF = 246k PPD

This is with only 1xUnit (Run 0, Clone 70, Gen 0), but I'm not liking the look of it so far.
8104 @ 10:15 = 349.8k ppd
246k is your points for less than a days work.

All your figures are wrong for PPD, those are points earned for the work unit, not PPD.
Use this calculator for PPD: http://www.linuxforge.net/bonuscalc2.php
 
Last edited:
P8104 will easily make it with the slowest P8101 machine, with 20% to spare. Look at Tyan:
bqRSxTp.jpg
 
Even my i7 980X can run the P8102, P8103 and P8104 easily with a little help of Langouste!


The problem is that 980X can not manage to fold the P8101 within the deadline, so then the bigadv is no option for this CPU. It is only about 2 minutes over the deadline for 8101, so if I can manage to OC it until 4.6GHz it will do the 8101 too.

Meanwhile the is the 980X folding on standard smp`s.


Uploaded with ImageShack.us
 
The only thing I was concerned about was if the P8104 changed the minimum equip req since the deadline was moved up.

The quad 8389 machines with the default 7.3.6 install, and the default Ubuntu 10.04 or 12.04 install (EXT4) will barely make the P8101 deadline, but they do make it. So a rough guideline for today is 16 cores x 2.9ghz to play it safe. The P8104 changed nothing.

This is only important to the 8300 guys (sub $400 machines).
 
Finally the P8104 came down on my Xeon E5-4650 and the rig runs it with a TPF = 5:06 which almost gives a million (996,722) PPD. A TPF = 5:05 will give a PPD = 1001628,3

Uploaded with ImageShack.us

I love this toy.:D

Update:
And there was over the one million achieved, yess:D,
TPF = 5:04 PPD = 1.006.574,6
 
Last edited:
Just thought I'd update, My SR-2 is running 16:09 on them

8101s ran 26:30 to 27:20
8102s ran 21:20
 
Back
Top