OGR-26 questions

[vss1980]

Does anybody have any idea how much longer OGR-26 WU's are going to take to complete compared to OGR-25 WU's?

I remember when D.net first started sending out OGR-25 blocks, people were quite stunned at how much longer they took.

 

[ViRGE]

I don't know exactly, but I expect it to grow exponentally unless Dnet hands out much bigger partial rulers.

 

[imported_Thunder]

vss1980,

This is just my speculation, based on a few things that some of the D.Net folks said in .plans, posts here, conversatons on IRC, etc. I think they realized that increasing the final ruler size without increasing the "pre" done part of the stub ruler, was not practical. Even considering the growth in speed of processors, that was still much too much work for a single WU. The 5 value OGR-25 stubs that they tried initially were waaaay too large and I don't think they'll make that mistake again. Even when they did increase the stub size to 6 values for OGR-25 stubs, it still took longer to complete (on average) than for a 5 value OGR-24 stub.

I'll see if anyone at D.Net can address this, but my best guess is that we'll see 7 value OGR-26 stubs and even (moderately) longer times for WU completion.

-Brian

 

[JHutch]

Personally, I'd like to see 8-value stubs for OGR-26. That would keep the times somewhere between the speeds for OGR24 and OGR 25.

JHutch

 

[vss1980]

OK, next question, node count.

At the moment, the 6 value OGR 25 stubs are giving average node counts in the range of 150-200 Gnodes. What will the newer average be?