Hardbowl scoring

[marty9876]

It's been awhile since I really got into the whole points/WU mess. My thoughts below, let me know yours!

Premise: Intel vs. AMD. Take a given Intel 3.2 running QMD's vs. and AMD (3700+ ?) running big packets.

Assume Intel will loose on non QMD front- normal vs. normal packets or big vs. big packets.

Assume Intel will do 1 QMD/day and on one day do 2 QMD's per a given 7 day period.


10 pts. Highest points for the week
-- Intel +10

14 pts. Highest number of WU's for the week
-- AMD +14

5 pts. Points per day average (game week only)
-- Intel +5

5 pts. WU per day average (game week only)
-- AMD +5

------------------------------------------------------

At this point AMD +4 points over Intel


6 pts. Highest points in one day
-- Intel +6

10 pts. Highest WU's in one day
-- AMD +10

-------------------------------------------------------

At this point AMD +8 points

These two sets get messy-

3 pts. Points per WU (game week only)

-- AMD will out score based on doing more WU's in a given week

1 pt per. Daily highest WU count and Point count (14 pts. Total available.)

-- Wash, Intel will take 7 for highest point count, AMD will take 7 for highest WU count



AMD wins.

 

[Kvothe]

3 pts. Points per WU (game week only)

-- AMD will out score based on doing more WU's in a given week

Actually, if its a QMD box, P4 will win that battle every time as the points per work unit would be 450 per as opposed to the 80 to 100 depending on what the AMD gets, so swing those 3 back to P4.

 

[AtomicMoose]

Marty, all those numbers look pretty but is there a point here?

 

[marty9876]

Actually, if its a QMD box, P4 will win that battle every time as the points per work unit would be 450 per as opposed to the 80 to 100 depending on what the AMD gets, so swing those 3 back to P4.

Ohh, I read that as 3 point for a given WU being completed.

1 WU= 3 points
2 WU= 6 points

and so on. Regardless of points assigned by Stanford for a given WU.

So you saying it's 3 points for the highest points average for total number of WU's completed in a given week?

 

[Kvothe]

Ohh, I read that as 3 point for a given WU being completed.

1 WU= 3 points
2 WU= 6 points

and so on. Regardless of points assigned by Stanford for a given WU.

So you saying it's 3 points for the highest points average for total number of WU's completed in a given week?

Exactly, just takes the the points per work unit average for the span of the game week.

 

[marty9876]

Marty, all those numbers look pretty but is there a point here?

This could be said for the whole DC forum could it not?

Na, I'm just trying to understand the scoring system and help others understand it also (ok, maybe me only dumb one).

I think it's a good scoring system, just looking for flaws.....gotta cheat somehow ya know.

You should see the mess I assembled to win this thing, and cooked....

 

[AtomicMoose]

This could be said for the whole DC forum could it not?

Na, I'm just trying to understand the scoring system and help others understand it also (ok, maybe me only dumb one).

I think it's a good scoring system, just looking for flaws.....gotta cheat somehow ya know.

You should see the mess I assembled to win this thing, and cooked....

Bored today, aren't you?

 

[marty9876]

Bored today, aren't you?

On vacation.... Time to go fix the tile in the shower I guess.

Well anyways, per p[H] correction AMD still at +5.

See, this just all proves my point of the great AMD conspiracy known as [H].

 

[Kvothe]

Actually the conspiracy theory isn't AMD-centric, its work unit biased. With the way people love the QMD, thats all alot of people run (not calling anyone out as I run QMD's too). But we wanted to change the slant to work completed and reward that sufficiently.

If you look at your break down, its all assuming that the AMD will turn in more finished results and thats where the lead is pulled from. Actually there were 2 or 3 games this week where the QMD's got beat by the normal WU's because more work got done.

 

[AtomicMoose]

Actually the conspiracy theory isn't AMD-centric, its work unit biased. With the way people love the QMD, thats all alot of people run (not calling anyone out as I run QMD's too). But we wanted to change the slant to work completed and reward that sufficiently.

This was all my idea....blame me.

 

[joeb0b]

I've been trying to work this out as well. I think I have so I'm not telling anyone.

-joeb0b

 

[Astral Abyss]

The only thing about running QMD is that's all you'll get, which I guess is good (lots of points) and bad (not many WUs). Out of 4 weeks running QMDs on my Intel box, I only received 1 non-QMD unit, a 133 pointer that still took a long time to process.

I think marty is just mad us rebels running Intel are kicking so much ass (and didn't have to overclock to do it). Blame Stanford, they like Intel. I'm not running the OCd box in my sig for the HardBowl, but instead using my generic Dell at work, because my A64 box isn't as competitive.

 

[Mohonri]

Blame Stanford, they like Intel.

It's not a matter of Stanford liking Intel, it's a matter of Intel refusing to allow Stanford to optimize a library for AMD processors. This issue has been brought up before. Stanford puts a large bonus on QMDs also because they require much more memory and SSE2.

 

[Astral Abyss]

It's not a matter of Stanford liking Intel, it's a matter of Intel refusing to allow Stanford to optimize a library for AMD processors. This issue has been brought up before. Stanford puts a large bonus on QMDs also because they require much more memory and SSE2.

Ok, their folding program likes Intel, same difference.