| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Hardware Chat Room
Thread: Do WUs keep getting larger for same Project over time? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 18
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Moore's Law? That's so 1960s... Two years ago it was posted:
Next month, the worldwide semiconducter industry will formally acknowledge what has become increasingly obvious to everyone involved: Moore's law, the principal that has powered the information technology revolution since the 1960s, is nearing it's end. The Semiconductor Industry Association (SIA) in Washington DcC, which represents all major US firms, has already said that it will cease its participation in the road-mapping effort once the report out... |
||
|
|
KLiK
Master Cruncher Croatia Joined: Nov 13, 2006 Post Count: 3103 Status: Offline Project Badges: 
|
If you had one device crunching one project only, would the amount of work accomplished tend to level out over time or be highly variable? It would be highly variable, as the size of molecules tested against target are different in size, number of atoms, etc.  But also, after every test, like BETA for FAH2 is doing...you can see that the BETA FAH2 is twice longer to compute on same machine than normal FAH2.  |
||
|
|
KLiK
Master Cruncher Croatia Joined: Nov 13, 2006 Post Count: 3103 Status: Offline Project Badges:
|
Moore's Law? That's so 1960s... Two years ago it was posted: Next month, the worldwide semiconducter industry will formally acknowledge what has become increasingly obvious to everyone involved: Moore's law, the principal that has powered the information technology revolution since the 1960s, is nearing it's end. The Semiconductor Industry Association (SIA) in Washington DcC, which represents all major US firms, has already said that it will cease its participation in the road-mapping effort once the report out... Well, it was also 2015 when NYT posted this: https://www.nytimes.com/2015/09/27/technology...p;pgtype=article&_r=0 Article is expecting that the Moore law will end up in 2025...it's still going strong & speed of the chips is going not through the speed of clock rate, but through the vastness of size of the michips & number of cores! So yes, it's still speeding up...even though you can't see it in getting faster crunching time, but all the cores in total get more work done - Moore law in practice! But some of the people here think that "Moore law, in which experts say it's still on work till 2025, is not present here because WCG grows only 10-20%." Sorry, but followers of greek god Mom are not comparing new systems to the old ones, they are talking about the used computers all of us use in giving power to some grid called WCG. Of course it's not correlated, any Vulcan child could tell you that. What they don't do is ask themselves: When do we buy new computers? Every 5 years? So check the users that are here more than 10 years, to see their growth in those 10 years (2 circles of computer purchases). I bet is in correlation with Moore law. ---------------------------------------- [Edit 1 times, last edit by KLiK at Nov 4, 2017 4:33:21 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If you had one device crunching one project only, would the amount of work accomplished tend to level out over time or be highly variable? It depends what you mean by "work"! That's part of the problem with volunteer crunching -- what should be measured and rewarded? That is why WCG has three measures: run time, points, and results. It should be that, for the case you describe, these should all level out to a fairly constant rate if you average over a week or a month, but they might vary by quite a few percent over the short term. Run time will vary because the time is not credited until the work unit (WU) is validated, so there will be some waiting for wingmen. Points will vary as the rate of granted credit goes up and down due to errors in estimating the processing power needed to process a WU, and as your scores are adjusted to average with those of your wingmen. And results will vary due to the same processing power requirement estimates varying for each WU so, for the projects where WU length can be varied, these will go up and down as well. Don't get hung up on this. Pick a sub-project (or projects) that you want to volunteer your computer time for and let it run. Any and all work that you do is benefiting science. |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7231 Status: Offline Project Badges: 
|
Just a couple of comments here.
----------------------------------------1. The Semiconductor Industry Association is ceasing giving its roadmaps mainly because there are only 4 main foundries for the production of computer chips - Intel, Global Foundries, Samsung and Taiwan Semiconductor(TSMC). These four are in cutthroat competition with each other and no longer want to participate in communicating with equipment suppliers as a group. See this article . 2. Many of us use machines which have been repurposed from their original purpose which means they are a few generations behind what is the leading edge. 3. As far as Moores Law is concerned, it is only concerned with how densely transistors can be packed. There are several ways the engineers have at their disposal to continue this for quite some time.  4. What I can say, speaking only for myself, is the machines I am using now are processing much more work, more quickly than the machines I started with - PIII Pentiums. 5. As far as I know the most stable run times per work unit are with MCM1. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
KLiK
Master Cruncher Croatia Joined: Nov 13, 2006 Post Count: 3103 Status: Offline Project Badges:
|
+1
----------------------------------------from 2nd to 4th point from Sgt.Joe Though, as an engineer we have to see what are the challenges of cooling for 3D chips (1st point). ---------------------------------------- [Edit 1 times, last edit by KLiK at Nov 5, 2017 12:17:13 PM] |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Quoting the 'engineer' words
----------------------------------------"One more time, on every estimate no Moore law is given into account...so even though the processor calc strength is going up / doubling (or it used to, until later processors), WCG has seen an increase of 5-10% growth in calculation speed per year...so let's say it's pessimistic 20% growth, so any project with later date can be calculated down by those 20% speed up...if you want, those can be implemented in next reports also!" LMHO [Edit 1 times, last edit by SekeRob* at Nov 5, 2017 1:05:08 PM] |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1664 Status: Offline Project Badges:
|
Even if everyone says "Moore's Law" it is not a "law" but a "conjecture".
----------------------------------------It was based on observations, extrapolations, and assumptions, without any formal physical/mathematical/natural rational behind it. It would be fine if the people would use the terms a little bit more rigorously. Enjoy, Yves ---------------------------------------- [Edit 1 times, last edit by KerSamson at Nov 5, 2017 8:29:48 PM] |
||
|
|
|