How accurate is the PPD generator in EMIII?

[nahmus]

Hi all,
Since I'm about to get dropped from the top 10 any day now (nice folding MLMIB) I've been reviewing my home rigs to see if there are any optimizations I can make till I get some more borgs under my belt.

having all of my home rigs under EMIII the PPD generator is holding steady at around 1000 ppd. I'd love to be dumping 7k a week from just me home systems but thats not happeing.

Any ideas?

[TollhouseFrank]

what are your borgs at home? CPU/RAM? what flags ya usin?

[ChasR]

QMDs are about the only WUs that give the ppd calculator problems but we're only talking about ~20ppd or so.

You could gain some ppd by sneakernetting QMDs to your AMD rigs. On an A64 you could go from 60ppd per GHz on a Gromac to 130ppd/GHz on a QMD.

[nahmus]

this is what I have running currently at the house.

1) dual xeonHT 2.0 with 1.5 gig of memory. running 2 qmd 2 non QMD clients -advmethods bigpackets=yes
2) dual xeon(nonHT) 2.0 with 512rdram running running 1qmd 1nonqmd -advmethods bigpackets=yes
1) P4 2.8 1gigddr QMD -advmethods bigpackets=yes
1) XP2200 1gig -advmethods bickpackets=yes
1) xp2200 1.5gig -advmethods bigpackets=yes

these are just my home rigs. I'm startting to thing I should be posting more that 3.5k per week.

[ChasR]

I think you'd make more ppd on the first rig running one QMD and two non QMD instances. My Sig rig # 2 makes about 590 ppd on two QMDs but has made as high as 770ppd and averages 680 ppd on one QMD, one or two big WUs and one tinker.

I think you ought to be making at about 1500ppd off of those rigs.

[TollhouseFrank]

do what Chasr said for rig #1. For both Athlon rigs, take off -advmethods. Right now, the WU mix for AMD's when getting wu's that use the -advmethods flag is very poor. You 'generally' get higher point WU's without -advmethods on AMD machines (this is just testing i've done here at home the past month and a half with my A64 and my wife's sempron). if ya get 600 pointers on those like my wife's sempron, then you'll be makin' very good ppd.

[nahmus]

changes made. lets see what happens. thanks all

[LandShark]

however, I've been having problem getting anymore 600 pointer the past few days on the Athlon and PentiumM..... make me wondering if the 600 pointer is drying out..... :shrug:

[ChasR]

I did a little checking and found the EMIII total ppd calculator is highly inaccurate on startup. If running continuously it pretty close, but it doesn't get close until every WU has finished one complete frame after EMIII startup.

I've gotten a few 600 pointers this week including at least one this morning so they're not gone just yet.