LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited November 2006
Yes, out of eight cores in four computers (see signature) I've got four (4) 600-pointers running right now. That's the most I've seen at one time in probably three months.
Flags: -verbosity 9 -forceasm
-advmethods is turned on in the client settings, but not in client properties
Yes, out of eight cores in four computers (see signature) I've got four (4) 600-pointers running right now. That's the most I've seen at one time in probably three months.
Flags: -verbosity 9 -forceasm
-advmethods is turned on in the client settings, but not in client properties
Cool thanks ...so basically you're pulling them from standard servers and not the semi-betas from -advmethods. I'll give those servers a shot!
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited November 2006
I would imagine they are all dowloading because of the -advmethods flags. BTW, most of the servers can serve up a variety of work units. But the servers will not distribute the "semi-beta" work units if a Folding client is not marked for advmethods. But you probably already knew that.
Every instance picks up the 600 pointers, just now and again they get sent to a different Server.
Datsun 1600
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited November 2006
Advmethods does not direct your machine to specific servers, it only directs whatever server your computer connects with to serve semi-beta work units, if they are available at the time.
I found with any instance that had any reference to the -advmethods flag, it would pick up anything but the 600 pointers, so I reconfigured every instance just to check that the -advmethods flag was not turned on in the 'Change advanced options'.
How Stanford work out which Boxen gets which Protein, only they know, all I know is I have done 268 x 600 pointers.
On my machines I had to remove the advmethods flags to recieve the 600 pointers, though this week I seem to have been swamped with the 5-way "patty melts". (lg wu's= yes)
On my machines I had to remove the advmethods flags to recieve the 600 pointers, though this week I seem to have been swamped with the 5-way "patty melts". (lg wu's= yes)
And aren't those 5way_melt Proteins slooooooooooow, but there is a better one, the p1815_collagen takes half as long again as the p1809_Collagen for the same points. Way to go Stanford.
I've also had better success getting the 600 pointers without using -advmethods than with it. But you can still get the occasional 600 pointer with the -advmethods flag. I really think all those patty melts suck big time, myself.
And it seems to be impossible to get anything from server 128 any more. The cheaters who have been filtering for certain servers like 128 and 162 have caused Stanford to lower the priority of server 128 to where it almost never gets assigned any more it seems.
I tell you what, with these server filtering cheaters now, I'm ready for Stanford to just plain drop all bonus points now, so there is no advantage to filtering for particular servers. And if you know my position on the bonus points from the past, you would know that this is a 100% turnaround in attitude from me.
I have been getting a steady flow of Ribo Proteins from Servers 142 and 162.
The bonus system, well Stanford thinks it is a reward for Bandwidth.
Stanford's point system needs overhauling, as it sucks big time.
If you think it's bad now, wait for Version 6 of the client, when you can pick which server you connect to, I hope Stanford increase the number of Servers for the good stuff, as they are going to be hammered.
-advmethods is turned on in the client settings, but not in client properties
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited November 2006
The advanced settings in the configuration setup turn on advmethods by default. It is redundant to enter "advmethods" in the properties window for the Folding@Home start file.
The advanced settings in the configuration setup turn on advmethods by default. It is redundant to enter "advmethods" in the properties window for the Folding@Home start file.
Ah ...that's what I thought you said. I didn't think that setting it in the configuration even worked. I've never seen the evidence of it initialized once the client is started so I assumed it had no effect.
Also ...even if I have it set in the config file I was still getting 10 wu's back when they were being issued. But if I had it set in the startup I would only get one.
Still picking up ribo's here, but i am starting to get a pattern on some pc's. It will get a ribo then it will get a pmd cored unit then it will get one of those 99-130 point Dgromacs. "Not to mention a few instances of my oh so favourite 5 way melts" NOT!
Except for putting in my machine ID for my dual core, I don't set any other advanced options. The only others I set are to launch FAH as a service and allow receipt of work assignments and return of work results greater than 5Mb in size. I've had almost nothing but 600 pointers on both cores since the day I built it and I've got 600 pointers on both cores right now.
My machine processes these pretty quick, but they are pretty memory expensive (my current ones are about 70MB each). Hmmm- need enough memory?
I just set the mem limit to whatever fah detects (full boat) ie. fah sees 1024, that's what both cores are set too. All my machines have at least 1 gig of ram.
@ Oriane, there should have been an option in the config to "change advanced options". (I think ?)
Oriane, there should have been an option in the config to "change advanced options". (I think ?)
Yes, I know- you have to in order to set up the second core. My point is- that's all I change in the Advanced options.
I suppose my answer as to why is mostly philosophical. I don't care about the points and I let the Project decide what I'm going to fold as much as possible.
Thinking out loud - I question the value of messing with advanced options in this fashion. If I read it right, the Project can override some preferences to fold what it needs and I think sometimes they can cause trouble (remember deadlineless WUs?- I lost a bit of folding time on that one). It seems I'm having more success with 600 pointers leaving the advanced options alone. This might be worth the debate.
At any rate, I think these 600 pointers could be sensitive to memory and deadlines and might not get assigned to less capable machines. I also leave mine on as much as possible, so I do one normally in about a day and a half. As for memory usage- see the attached.
I've been getting mainly 600 pointers on my 1.6 P-M Lappy since the beginning of Oct with no -advmethods. Using 130mb of mem now of 1.25gb total.
I took -advmethods out of the client.cfg last night to see what I get next. I haven't had a 600p in about a month or so and I've been completely stable (mostly). I have no flags set in the registry either. Let's see what happens this time.
csimon
Acadiana Icrontian
Comments
Flags: -verbosity 9 -forceasm
-advmethods is turned on in the client settings, but not in client properties
Cool thanks ...so basically you're pulling them from standard servers and not the semi-betas from -advmethods. I'll give those servers a shot!
Datsun 1600
Datsun 1600
How Stanford work out which Boxen gets which Protein, only they know, all I know is I have done 268 x 600 pointers.
Datsun 1600
Datsun 1600
And it seems to be impossible to get anything from server 128 any more. The cheaters who have been filtering for certain servers like 128 and 162 have caused Stanford to lower the priority of server 128 to where it almost never gets assigned any more it seems.
I tell you what, with these server filtering cheaters now, I'm ready for Stanford to just plain drop all bonus points now, so there is no advantage to filtering for particular servers. And if you know my position on the bonus points from the past, you would know that this is a 100% turnaround in attitude from me.
The bonus system, well Stanford thinks it is a reward for Bandwidth.
Stanford's point system needs overhauling, as it sucks big time.
If you think it's bad now, wait for Version 6 of the client, when you can pick which server you connect to, I hope Stanford increase the number of Servers for the good stuff, as they are going to be hammered.
Datsun 1600
Also ...even if I have it set in the config file I was still getting 10 wu's back when they were being issued. But if I had it set in the startup I would only get one.
You sure it works when set in the client.cfg?
No -Advmethods flag here either.
Later
I jinxed myself.
Later
Except for putting in my machine ID for my dual core, I don't set any other advanced options. The only others I set are to launch FAH as a service and allow receipt of work assignments and return of work results greater than 5Mb in size. I've had almost nothing but 600 pointers on both cores since the day I built it and I've got 600 pointers on both cores right now.
My machine processes these pretty quick, but they are pretty memory expensive (my current ones are about 70MB each). Hmmm- need enough memory?
@ Oriane, there should have been an option in the config to "change advanced options". (I think ?)
Yes, I know- you have to in order to set up the second core. My point is- that's all I change in the Advanced options.
I suppose my answer as to why is mostly philosophical. I don't care about the points and I let the Project decide what I'm going to fold as much as possible.
Thinking out loud - I question the value of messing with advanced options in this fashion. If I read it right, the Project can override some preferences to fold what it needs and I think sometimes they can cause trouble (remember deadlineless WUs?- I lost a bit of folding time on that one). It seems I'm having more success with 600 pointers leaving the advanced options alone. This might be worth the debate.
At any rate, I think these 600 pointers could be sensitive to memory and deadlines and might not get assigned to less capable machines. I also leave mine on as much as possible, so I do one normally in about a day and a half. As for memory usage- see the attached.
I think it checks out your available memory, and give you the big WU's only if you have the memory for it?
...or will it just go ahead and max out your mem start using the swap file?
seems like the first one makes more sense...
[settings]
username=csimon
team=93
asknet=no
bigpackets=yes
machineid=1
local=118
[http]
active=no
host=localhost
port=8080
usereg=no
[core]
checkpoint=30
ignoredeadlines=yes
[clienttype]
memory=1024
type=2
nonet=yes