HWBOT Rev.4 - The Plan.

[|ron]

Hi rbuass, as Pieter was saying in another topic, Richbastard had to re-write some parts of Rev4 code, so it's under development

[Christian Ney]

Hi rbuass, as Pieter was saying in another topic, Richbastard had to re-write massman's parts of Rev4 code, so it's under development

 

Corrected

[|ron]

Rofl, I see Pieter is paying hard the April Fool's joke he did :D :asd:

[[GF]Duke]

I havn't seen anything on the Rev. 4 lately. How is everything going?

[Massman]

 

All Rev 3.9.X are small steps that lead up to Rev4. The final R3 version you currently use is 3.8.0. None of the 3.9.X will be public.

[[GF]Duke]

Thanks for the update.

[Chiller]

i hate the ,update, 2years ago it was good, much better

[Massman]

Why?

[Chiller]

with a Server 4x 7500 xeon super wprime time gets less points then a

 

single core doing Wprime,

 

i find that not realistic.

[Massman]

Having expensive server should yield more points than pushing cheap single core Sempron?

[Chiller]

no, the fastest score ,no matter witch hardware is used should get better points.

[Massman]

So, like I said, having the expensive hardware is better than pushing cheap stuff? It's better because it gets more points.

[Chiller]

it's more normal that a faster score gets better points, no matter what hardware is used.

[knopflerbruce]

Then only 4-way GPU and giga-servers get boints. How interesting is that? Not so much.

 

The ranking isn't perfect (most of the gold cups with 10+ more submissions that get 2 boints pr run are alot more difficult to beat than the 50th place 20 boint scores in the popular categories is another example), but we can only try to improve it.

 

Perhaps if 300 people benched those servers it would be different... Then you'd get alot of boints for those results.

[Massman]

it's more normal that a faster score gets better points, no matter what hardware is used.

 

Why is that 'more' normal?

 

If I beat 1 million other Sempron 145 overclockers, is that worth less than someone beating 1 other 980X overclocker?

[TaPaKaH]

Chiller, current score algorythm awards points mostly based on "how much people you beat" rather than "how good your score is" hence multi-CPU systems score low since not much people have them.

[Massman]

Sam, please help to define the characteristics of "good score"

[TaPaKaH]

Sam, please help to define the characteristics of "good score"

I think we've already discussed that it's a difficult thing to measure hence we use the current scoring system, no ?

[Warrior_oF_Byte]

I think that a "good score" (by Chiller) is when you are more close to a TEWROAMTOXB... (The Extremest World Record Of All Time Of X Benchmark), so for that you need the fastest and expensive hardware never seen.

 

But, I love Rev 3, It was a good step from Rev 2.

 

So have we will Rev 4 in one week?? :D

[K404]

Holy grail would be an algorithm that could roughly calculate score efficiency Rivatuner graphs could show the GPU MHz, so downclocking for the screenshot wouldn't work

[voidn]

 

All Rev 3.9.X are small steps that lead up to Rev4. The final R3 version you currently use is 3.8.0. None of the 3.9.X will be public.

 

So is this still a go? Rev 4 in two days?

[TaPaKaH]

so when the rev4 is actually coming, Pieter ?

[Massman]

When delays stop being unforeseen

[[GF]Duke]

So i'm guessing its not happening today then.

[OptyTrooper]

When delays stop being unforeseen

You've watched Star wars one too many times Massman. That answer was very Yoda.