• 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.

F@H MySQL System/PPD database.

Patriot said:
great for those that cant do under 10.... shitty to make the incentive less for more powerful machine owners...

No dont make it hard on yourself.. linear for all....
Click to expand...

Linear across the board would need some serious revamping of deadlines, otherwise there is no real incentive for shaving 30 seconds off of a wu - it might be worth 0.05% extra points. The exponential then linear attempts to motivate people to the high end - go from a 1p to 2p you might get 40% more per WU,go from 2p to 4p again a 40% increase, for from 4p to 4p + 20% you might get 10% more, but you still get progressively more - on this WU anyway. There has to be a throttling mechanism in place, otherwise Moores law and the exponential curve will just run away from sanity. Go from 4p to 4p x 2 frequency, we'll give you 10x as many points. Go to 4x freq we'll give you 40x the points - that is the nature of the exponential bonus. Throttle it somewhere, or it runs away from you and sanity.

H.
 
um thats the point.... the whole system needs revamping.... more base pts less bonus...

base should represent computation required... bonus should be a max of 10-20% of the total wu base
 
But the bonus is for the quick return. A WU returned in one day enables so many more WU's to be made available sooner than one that is bogged down for 5-6 days. More/quick "science" on the WU completed, even more opportunities for more WU's be processed the quicker it is returned. Its not just the science in that one WU, its the subsequent work that it enables

H.
 
Haitch said:
But the bonus is for the quick return. A WU returned in one day enables so many more WU's to be made available sooner than one that is bogged down for 5-6 days. More/quick "science" on the WU completed, even more opportunities for more WU's be processed the quicker it is returned. Its not just the science in that one WU, its the subsequent work that it enables

H.
Click to expand...

so cut the deadlines shorter
make it more cores

if your wanting to rewrite a broken system scrap it... target current ppd roughly as "worth" 2p stock system...

and fix the system... enough with the freakin bandaids...they aren't helping
 
"Fixing" the points system requires one thing - Stanford to tell us how much more valuable to them the various relevant units - GPU, SMP, BIGADV, BigAdv-12 are over the historical ones. Uniprocesssor and PS/2 units have been around for almost a decade - they aren't worth more now, and they shouldn't be worth more now, than they were back then - the question is "Assign a numerical value to how much more relevant these units are" and "How much does it impact the science if a WU is returned quicker".

Without knowing the scientific value in those answers, any attempt to propose a solution is an exercise in futility. Points are supposed to reflect scientific value, but if the values are unknown, a solution is impossible.

H.
 
Got your e-mail. Super swamped. I have a few good pivots setup at home that I can send you.
 
It would be helpful to see WU cound on the "average min/max for each WU type" query

It would also be helpful to separate multi socket boxen from single socket boxen. It's clear that we have a pretty high end selection of rigs running these, so the "average" is skewed really high.

It appears something isn't updating properly. My quad 6172 rig is here: http://www.fahbench.net78.net/AllDataWU.php?wu=6904

But not here: http://www.fahbench.net78.net/AllDataCPU.php?cpu=0

Other than that, I think these are really the only queries we will actually need, but it would be useful to see a summary of the data along with the raw results. (Average TPF, PPD, etc).
 
jebo_4jc said:
It would be helpful to see WU cound on the "average min/max for each WU type" query

It would also be helpful to separate multi socket boxen from single socket boxen. It's clear that we have a pretty high end selection of rigs running these, so the "average" is skewed really high.
Click to expand...

Can do on the first, and on the second I'll looking at splitting then out by CPU count and by core count.

It appears something isn't updating properly. My quad 6172 rig is here: http://www.fahbench.net78.net/AllDataWU.php?wu=6904

But not here: http://www.fahbench.net78.net/AllDataCPU.php?cpu=0
Click to expand...

It's an issue with the query - where CPU="" need to become where CPU is null. Will get it fixed.

Other than that, I think these are really the only queries we will actually need, but it would be useful to see a summary of the data along with the raw results. (Average TPF, PPD, etc).
Click to expand...

Thanks - will play with them next week, out of town at the moment - might get a chance to work on them.

H.
 
jebo_4jc said:
It would be helpful to see WU cound on the "average min/max for each WU type" query

It would also be helpful to separate multi socket boxen from single socket boxen. It's clear that we have a pretty high end selection of rigs running these, so the "average" is skewed really high.
Click to expand...

Something like this - incorporates by core & the raw data:

http://www.fahbench.net78.net/AllDataWUbyCore.php?wu=6900

Also by OS Platform:

http://www.fahbench.net78.net/AllDataWUbyCoreByOS.php?wu=6901

It appears something isn't updating properly. My quad 6172 rig is here: http://www.fahbench.net78.net/AllDataWU.php?wu=6904

But not here: http://www.fahbench.net78.net/AllDataCPU.php?cpu=0
Click to expand...

Fixed:

http://www.fahbench.net78.net/AllDataCPU.php?cpu=

H.
 
Last edited:
As a thought, darker header row with white text. It is hard to read blue/ black on some displays.
 
You think I know how they got theat color ? ;-) Will see what I can do.

H.
 
I think I have the Linux speed/cores issue resolved, but I need someone with an overclocked dual Intel processor box running Linux to test a couple of test apps to verify. The apps just read some system info and spit out text - no harm will come.

any volunteers, please email me at hamish@haitch.net

Thanks.

H.
 
You must log in or register to reply here.
Back
Top