Although I don't know precisely how it will be implemented, I tend to agree with this. Factoring a performance gradient to the points allocation scheme will only foster additional elitism.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
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.
6.22R2 SMP client released
- Thread starter Xilikon
- Start date
- Joined
- Oct 12, 2004
- Messages
- 15,022
Yes, that's the goal. Keep in mind that the PandeGroup doesn't support running dual SMP officially anyway but I can predict that as long as the core is not scaling efficiently, it will still be worth running 2 clients to maximize production (they might probably not use a sliding ruler, more like a set point like if it get below the 75% to deadline will cut the points 10-25% and below 50% will cut it in half).
Until it's implemented, it's just speculations pulled out of our asses
However, the basic idea of using the performance factor as a element will help a lot for sure at reducing the injustice some people is feeling faced to the system.
Hmm that means dual smp's will finally get the boot... faster turnins most likely will net more points then running 2x smp. That would be pretty cool, I just hope they explain the scale so we know how to setup things. To many times things are kept somewhat mystical and that make it hard to understand how to setup things and where to invest.
OK, but how will this help people running hardware a year or two older than the bleeding edge? How will this new scheme help people like FLECOM running a farm of P4s entirely with console clients? To me, it appears even more like Stanford would like us all to sell our current hardware for Tesla and Quadraplex setups. Moreover, monitoring programs will all be rendered instantly obsolete the day Stanford implements the reforms because projected PPD values will be in error due to the variable point allotments based on actual time of completion.
OK, but how will this help people running hardware a year or two older than the bleeding edge? How will this new scheme help people like FLECOM running a farm of P4s entirely with console clients? To me, it appears even more like Stanford would like us all to sell our current hardware for Tesla and Quadraplex setups. Moreover, monitoring programs will all be rendered instantly obsolete the day Stanford implements the reforms because projected PPD values will be in error due to the variable point allotments based on actual time of completion.
FLECOM is running single clients, but 24/7 and on fast chips (for single core) I bet he would come out ahead.
The second point, if they use a 75% and 50% rule, it is not hard to do an update to the software and use the projected time to finish to adjust the ppd measurement.
- Joined
- Oct 12, 2004
- Messages
- 15,022
OK, but how will this help people running hardware a year or two older than the bleeding edge? How will this new scheme help people like FLECOM running a farm of P4s entirely with console clients? To me, it appears even more like Stanford would like us all to sell our current hardware for Tesla and Quadraplex setups. Moreover, monitoring programs will all be rendered instantly obsolete the day Stanford implements the reforms because projected PPD values will be in error due to the variable point allotments based on actual time of completion.
I understand your point but I never said it's just the performance factor which is added. The points will surely be rebalanced (if we want to get the console points in line with SMP, it must be benchmarked to 440 points / day while it's just 110 points currently. A more possible value would be like 350-400 points / day). Just this measure will help FLECOM for sure even if the performance factor can bite on some of the numbers.
Good, I was hoping for a general readjustment of the WU values. At least a more reasonable starting point for some WU values would ameliorate potential negative effects of a variable point scheme. Case in point, the P2665.
OTOH, there will be many people who will probably hate it. I don't know about you, but I would be very upset if I had issues with a system or my entire farm and it caused or required a shutdown, but the consequent hours of downtime was sufficient to drop all my clients' work from the maximum point 'bracket' to a lower one due to the lost time. What about maintenance that can take several hours or longer? I have to conduct maintenance at least once a week on my farm necessitating several of my systems to be shut down for extended periods.
This is an example of things that would weigh on many folders' minds after this new scheme is implemented. There's nothing I or anyone can say to change Stanford's intentions, I know this, but in my opinion a simple reallocation of the WU values is all that is required. The variable allotments would cause a new set of headaches. I don't like it.
- Joined
- Oct 12, 2004
- Messages
- 15,022
Problem is no solution will make everyone happy.
Correct, it's impossible to have the perfect solution or they would have implemented it by now.
However, If I understood Vijay correcty, he might be doing a kind of discussion thread asking for the output of the folders to determine what each one would like then pick the solution which make a consensus. This is actually what make the system fair and the users feeling good because Stanford listen to them.
Stanford needs to be careful... listening to the community could result in a large push toward a single solution which most users drop the cpu clients all together... I hope they make sure whatever solution won't ruin the balance among clients.
Stanford needs to be careful... listening to the community could result in a large push toward a single solution which most users drop the cpu clients all together... I hope they make sure whatever solution won't ruin the balance among clients.
You ain't just a' jivein' brother Sunin. It seems at times the total world population is made up of 98% of the people tryin' to be independant (still going for their own self interests) and 2% that are actually independant (could give a shite less about self interests and want a benefit for everybody, kinda' like the folders on team 33)
Remember that old dude (I think he was like the 16th pres.of the US) he said something like you can't always please everybody (err.., fool everybody or something like that). So what it all boils down to IMO is Stanford should, and probably will, take a stab at the points allocation thing, but when they start talking about a concensus I just loose all hope. (who decides the majority opinion (whatever that is, 51%?) and who ensures that the majority opinion will be adheared to if they ever figure out what the majority is, "you've gots to be kiddin'
FOLD ON whether it be with today's newest IT or the technology of yesterday (like most of us), please just fold for the cause and don't forget to fold for the [H]orde team 33
Oh yeah, if this sounds like a trek into bitterness about life, it's not meant to be, its 'just how I see it sometimes. All I have to do is take a look around and see the many selfless people, with different backgrounds, with different beliefs, with different nationalities, with different just about everything, participating harmoniously in this wonderful F@H project. (we's gonna' getcha' cancer )
I didn't read it the same way. My take is that Vijay realizes that the
current point system is not working well, especially (?) for those of us who
run the cpu client on small or LARGE farms. I think the discussion, prompted
by your good and valid questions, will help sort this out. I'm an optimist...
While I got into folding for the science, the reality is that the point system
supports the science mission. Science is what is important, but points count.
Fold on, [H]orde!
