How much have your points increased?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
How much have your points increased?
How much do you figure your points have increased under Stanford's new points system put into place April 22? Pick the number that best typifies the change.
David
David
I really can't tell you -- I've been fluctuating so much lately due to problems unrelated to F@H.
Originally, I was thinking 30-40% when we first switched over. Lately, I'm thinking more in the 20-30% range. It all comes down to how many DGros they're passing out (since all but 2 of my boxes are SSE2).
Do you really want to go back to the EOC "change the color cutoff" flame war again? A few weeks back you upset a bunch of EOC'rs when you wanted to double the cutoff (although you did back off on that). Now you want to go back to that forum and ask to reduce the cutoff? You've got bigger ones than I my friend.
Dave
Originally, I was thinking 30-40% when we first switched over. Lately, I'm thinking more in the 20-30% range. It all comes down to how many DGros they're passing out (since all but 2 of my boxes are SSE2).
Do you really want to go back to the EOC "change the color cutoff" flame war again? A few weeks back you upset a bunch of EOC'rs when you wanted to double the cutoff (although you did back off on that). Now you want to go back to that forum and ask to reduce the cutoff? You've got bigger ones than I my friend.
Dave
Last edited by dasman on Thu May 06, 2004 9:19 am, edited 1 time in total.
-
- *Lifetime Patron*
- Posts: 1740
- Joined: Sun Dec 21, 2003 11:24 am
- Location: 'Sunny' Cornwall U.K.
Damn, got a mention in a thread and didn't get a chance to reply before it was locked, so I came here and have cast my vote.
I used the old technical method of throwing two darts at a dart board to get the result.
Anyone got anything likely to be more accurate?
Pete got the answer from my sig and not going to tell you it's ~36% Amer
I used the old technical method of throwing two darts at a dart board to get the result.
Anyone got anything likely to be more accurate?
Pete got the answer from my sig and not going to tell you it's ~36% Amer
My "double the points" propasal was intended only to "spark discussion" , it was never a serious propasal. However, it did look to me like the points had jumped around 80% immediately after the change. And ultimately it was Jason who chose the new cutoffs, not me. I did get flamed pretty badly, but there was no merit in the argument behind the flame. It was more like a special effects flame, where the flame is added in post-production.dasman wrote:Do you really want to go back to the EOC "change the color cutoff" flame war again? A few weeks back you upset a bunch of EOC'rs when you wanted to double the cutoff (although you did back off on that). Now you want to go back to that forum and ask to reduce the cutoff?
I do hope to stay away from the emotional response this time by presenting my evidence up front, and keeping it simple. Ultimately I am just looking for fair and reasonable numbers, not promoting any sort of personal agenda. [Except to make myself maroon again! ]
David
Last edited by haysdb on Thu May 06, 2004 11:21 am, edited 1 time in total.
My PPD is all over the place, so I went by PPW, which is a bit more stable for me.
The point change happened in the middle of an EOC stats week, so I ignored that week. The 5 weeks before that averaged about 2200 PPW. The one full week after the change is about 2700. This week seems to be on track for 2700 or a little more.
That's about a 22% increase for me. I went with the 25% vote.
Scott (apparently safe from layoffs at the moment and contemplating a small blade farm )
The point change happened in the middle of an EOC stats week, so I ignored that week. The 5 weeks before that averaged about 2200 PPW. The one full week after the change is about 2700. This week seems to be on track for 2700 or a little more.
That's about a 22% increase for me. I went with the 25% vote.
Scott (apparently safe from layoffs at the moment and contemplating a small blade farm )
Agreed, it was emotional and you're probably a marked man at this point. You will have one thing going for you tho, they all want to be RED so I'm sure they'll support any decrease in the cutoffs.haysdb wrote:I did get flamed pretty badly, but there was no merit in the argument behind the flame. It was more like a special effects flame, where the flame is added in post-production.
I do hope to stay away from the emotional response this time by presenting my evidence up front, and keeping it simple.
Dave
I took the 5 oldest weeks of PPW data on the EOC stats and averaged it for an "old" PPW number, and I used the 4/25/04 week as a "new" PPW number. Here's what I found for SPCR overall, the people who have posted heren and a few other "folders of interest." (Sorry shens/isp, not enough data to figure out where you used to be. Way to start up strong, though!)
So why does my increase suck so bad?
Scott
Code: Select all
Folder Old PPW New PPW %change
===========================================
SPCR 171336 294578 +72%
mas92264 7355 14397 +96%
dasman 4603 7571 +65%
wgragg 10568 17381 +64%
Pete_Amer 547 860 +57%
David_Hays 7021 10714 +53%
TRC-13 13122 19218 +47%
captain_android 1862 2300 +24%
sbabb 2222 2717 +22%
Scott
Some of my increase is due to additional machines (I added two OC'd Barton 2500+'s in that time frame as well as upgraded a box). Also, anyone with P4's would have seen a pretty big bump that first week or two of the new scores due to all of the Dgros floating around (I think I had 6 of them at one point). Now that the Dgros and their 1300+ppw scores are scarce, scores are going to go down.
The couple of weeks prior to the change I think I was consistently a little below 700ppd. Since the disappearance of the Dgros, I'm probably going to be around 900ppd.
Dave
The couple of weeks prior to the change I think I was consistently a little below 700ppd. Since the disappearance of the Dgros, I'm probably going to be around 900ppd.
Dave
-
- Friend of SPCR
- Posts: 96
- Joined: Fri Dec 05, 2003 11:03 am
- Location: Seattle, Washington, United States
I keep pretty detailed records on my 16 folders (yes I know I'm a geek), charting out a PPW average every month. Comparing my historical data to records since the change, my production is up 23.7% on average. I tossed out the high and low machines (low is doing much worse than before, high is going almost 70% better so clearly there are some odd stats things going on there).
Tracking my sig line PPW number shows almost exactly the same increase.
Tracking my sig line PPW number shows almost exactly the same increase.
One premise of my argument that the cutoffs were raised too much is that the first week following the points change was artifically high, and should therefore not be including in computing the "new" PPW number.sbabb wrote:I took the 5 oldest weeks of PPW data on the EOC stats and averaged it for an "old" PPW number, and I used the 4/25/04 week as a "new" PPW number.
I am not presenting this as a fact, since it's possible that this last week has been the exception, but I do believe three factors worked together to inflate the numbers for that first week:
- Queued Tinkers caused a spike in production
- An unusual number of DGromacs, which are worth 3x other work units on P4's, and 2x on some other processors
- A spate of "high value" Tinkers
So longer as server .108 stays up.bkh wrote:My points rate more than doubled, from about 650-700 to about 1500. It's because most of the machines are "tinkers only", so they are getting much more credit now.
Server .108 is the only server issuing the Timeless Tinkers assigned to machines with a client preference of "GAH". It was down for a couple of days last weekend, leaving a lot of machines without work.
David
Tip: Set it to download work in batches, it will keep the queue filled with WU's. Mines are still folding despite 108 being down (again).haysdb wrote:So longer as server .108 stays up.bkh wrote:My points rate more than doubled, from about 650-700 to about 1500. It's because most of the machines are "tinkers only", so they are getting much more credit now.
Server .108 is the only server issuing the Timeless Tinkers assigned to machines with a client preference of "GAH". It was down for a couple of days last weekend, leaving a lot of machines without work.
David
[10:42:53] Printing Queue Information
00 *ACTIVE "Folding@Home" (65) 171.64.122.108:8080 April 28 00:00:26
01 READY "Folding@Home" (65) 171.64.122.108:8080 April 28 00:00:29
02 READY "Folding@Home" (65) 171.64.122.108:8080 April 28 00:00:32
03 READY "Folding@Home" (65) 171.64.122.108:8080 April 28 00:00:35
04 READY "Folding@Home" (65) 171.64.122.108:8080 April 29 16:08:42
05 READY "Folding@Home" (65) 171.64.122.108:8080 May 4 09:57:52
06 READY "Folding@Home" (65) 171.64.122.108:8080 May 4 09:57:54
07 FETCHING
08 FETCHING
09 READY "Folding@Home" (65) 171.64.122.108:8080 May 7 10:43:15
(yes, it is a slow folder)
edit: 108 worked again when I posted this, hence #9 being filled again.
Ain't no doubt about it.Queued Tinkers caused a spike in production
An unusual number of DGromacs, which are worth 3x other work units on P4's, and 2x on some other processors
A spate of "high value" Tinkers
There were a bunch of the 249 point Tinkers going around then; pretty rare now - these seem to be worth 2x ppw over the dreaded p638.
M
P.S. Home cable modem or router on the fritz yesterday and several folders ran dry for a few hours, resulting in ppw = 0.
David,
I'm not broadband at home so I've been searching for a way to do this - turned up UD Monitora couple of days ago but haven't tried it yet -- maybe this weekend...
Dave
EDIT Imagine it would play he!! with your perf fraction though...
I'm not broadband at home so I've been searching for a way to do this - turned up UD Monitora couple of days ago but haven't tried it yet -- maybe this weekend...
Dave
EDIT Imagine it would play he!! with your perf fraction though...
Weeee, that even you didn't know that.haysdb wrote:How is that done?Wrah wrote:Tip: Set it to download work in batches, it will keep the queue filled with WU's.
It's the next question it asks right after the project type. You have to switch to gah for it to appear, unfortunately it's not there with fah. In the winclient it's under the advanced tab in the config window. There appears a checkbox right below the project when you choose gah.
I think this is usefull to add to your guides. It can be of *great* use for modemmers. A WU is allways finished when you're not around for half a day to connect, plus you have to connect at least once-twice a day to keep it folding. But with this setting all you have to do is connect every few days when it suits you, stop and restart the client twice, and you got a enough work again for days. You won't have to wait till something is finished anymore. And why twice? Well, the first time it starts uploading finished WU's, but at the same time it will download new WU's, concurrently. Loading usually goes faster then sending, and obviously it can't load WU's into slots in the queue that are still filled with finished WU's it hasn't finished sending yet. In my experience you have to start it twice sometimes to get a really full queue.
I use it at work since a few weeks because the proxy shuts the internet connection down at nights and in the weekends for some reason.
Perhaps it's also usefull for offline folders, cause you'll get a *guaranteed* large amount of work instead of some puny gromacs that it finishes in 5 hours after which the whole hassle starts again of moving the WU's back and forth to the upload pc.
David,
Regarding getting the cutoff values changed again -- while I think the teal color cutoff is a little stiff, you may want to hold off visiting the EOC forum again and let the points stabilize for a couple more weeks. I started picking up Dgros again late yesterday and my points are waaaaay up today (I'm gonna be around 1400 after 1200 yesterday). That's a good 400-500 ppd increase just due to Dgros.
I think the first change in the cutoff may be been a little premature (after about a week) and it could be premature to change again after only one down week.
Dave
EDIT -- Geez, two posts in a row!!! Isn't that what the edit button is for???
Regarding getting the cutoff values changed again -- while I think the teal color cutoff is a little stiff, you may want to hold off visiting the EOC forum again and let the points stabilize for a couple more weeks. I started picking up Dgros again late yesterday and my points are waaaaay up today (I'm gonna be around 1400 after 1200 yesterday). That's a good 400-500 ppd increase just due to Dgros.
I think the first change in the cutoff may be been a little premature (after about a week) and it could be premature to change again after only one down week.
Dave
EDIT -- Geez, two posts in a row!!! Isn't that what the edit button is for???