[Poll] 4 cores / 8 threads significance in -bigadv folding

[mdk777]

My question in the OP was intended for officials.

And I agreed with you.
It would be interesting to know.

My point was that it was presumptuous to base your 80% estimate on the number of people who posted their stats in a thread, When the entire project was only opened to donors as a second thought.

I also agree

that there is a significance in all our efforts.

So I don't understand why you think the 2 and 4 socket contribution was so small.

However, we may never know the answer to this debate: transparency has never been a hallmark of the project.

[PantherX]

Hey DonMarkoni, I have a suggestion, (hope you don't mind ) While the Official response may or may not come, you can make it easier my making a Poll. You can list these options: (feel free to elaborate but you can have a maximum of 10 options)

1) 2p System
2) 4p System
3) i7 System

Make it a never ending one and set the user selection to the maximum number of choices that you have displayed. I am sure everyone would fine it easier to click a few options rather than to type their specs. This way, it will be useful as you get accurate data easily. It isn't the most accurate but at least some valid data is better than none. (or making guess ) This does have some limitations like if you multiple systems of the same configurations, you can select only one unless you keep duplicate entries which will allow a user to select multiple systems. You can edit your first post to make the poll. look to the bottom left and you will see a tab "Poll creation"

[Grandpa_01]

I am pretty sure you can remove PDC and Anonymous from the stats and be fairly close.

I'm not following you. PDC was the donor that originated the -bigadv WU.

viewtopic.php?f=38&t=12058&p=117859&hilit=pdc#p117860

The question of the OP is how much of the WU of -bigadv were done by I7 verse other 2 and 4 socket machines.

excluding PDC would be pointless.

It would be like asking what percentage of the total fund someone donated to the Melinda Gates foundation;( excluding Bill Gates and Warren Buffet.)

PDC is Panda group they are the owner not a donor Don was wanting to know how many DONORS were running 1P systems.

[mdk777]

PDC is Panda group they are the owner not a donor Don was wanting to know how many DONORS were running 1P systems.

From this source:

viewtopic.php?f=24&t=10697&start=0

Does this have any relation to the large-points value work units and recent high-scoring users?
Yes. The initial projects are 2681 and 2682, valued at ~25K points base. Although these point values seem high, the work units are correspondingly larger, so the base PPD (points per day) value is roughly comparable to standard SMP.

A collaborator has donated a large amount of compute time to this project; those clients were initially running under username Anonymous/team 1. To give proper credit for the donation, we have changed the username to PDC, team 1. During the period of this donation, there are at any time between 100 and 400 8-core clients running under this username (800-3200 cores total).

[PantherX]

This is confusing me, if PG states that they never re-credit the points if you have already submitted them (i.e if a user made a typo, even a genuine one, they won't change their stats to the correct one) then in this case did they make an exception? (I would understand as that's a lot of dedication) They changed the username to PDC and left it as team 1?

[orion]

Hey DonMarkoni, I have a suggestion, (hope you don't mind ) While the Official response may or may not come, you can make it easier my making a Poll. You can list these options: (feel free to elaborate but you can have a maximum of 10 options)

1) 2p System
2) 4p System
3) i7 System

Make it a never ending one and set the user selection to the maximum number of choices that you have displayed. I am sure everyone would fine it easier to click a few options rather than to type their specs. This way, it will be useful as you get accurate data easily. It isn't the most accurate but at least some valid data is better than none. (or making guess ) This does have some limitations like if you multiple systems of the same configurations, you can select only one unless you keep duplicate entries which will allow a user to select multiple systems. You can edit your first post to make the poll. look to the bottom left and you will see a tab "Poll creation"

A poll is just like a survey, it's not scientific. It only reflects what the people who vote in it may or may not be using. They don't even have to be running -bigadv and can still vote in it. I wouldn't try to debate with that kind of info.

To get back OT, Don Just PM Dr.Pande or someone else in the PG with your question.

[7im]

PDC is not Pande Group. Purdue Data Center...

And I didn't read that post as saying that PG edited the stats, just that Purdue had been running as Anonymous, and are now [then] reconfigured and running as PDC.

As for the OP question, I'm not sure what significance the answer would have, even if that data were tracked accurately (i.e. able to distinguish between single sockets with HT or multiple sockets, or multiple sockets with HT, etc.)

The hardware mix is always a moving target, and PG does what it can to best use that mix of hardware.

[DonMarkoni]

I was very busy at work and there has been many valuable comments/suggestions here...

Don you now that starting a thread like this is going to bring in people with different opinions than yours. If you wanted the info from PG than you should have sent one them a PM for it, not asked the question in open forum. It maybe some super secret info they don’t what out in the wild.

You've been running -bigadv since February? That's great and thank you for your contribution to F@H!

I've been running them since November...well until last month when the LINUX client crashed. All bought and paid for by yours truly too

Different opinions are good, if people keep respect to each other and stay civilized. And we all have been so far.
Info in question in the open forum should be useful IMO. I was interested in it and maybe some people would find it interesting too. I hope it's not a secret, don't see any reason to be.
Running -bigadv since November, just like you, (what I posted was just an example)

My question in the OP was intended for officials.

And I agreed with you.
It would be interesting to know.

My point was that it was presumptuous to base your 80% estimate on the number of people who posted their stats in a thread, When the entire project was only opened to donors as a second thought.

I also agree

that there is a significance in all our efforts.

So I don't understand why you think the 2 and 4 socket contribution was so small.

However, we may never know the answer to this debate: transparency has never been a hallmark of the project.

I wasn't underestimating any type of system, I'm sorry if I wasn't clear enough, or sounded like it. I was just saying that overclocked i7's can be on par with more powerful rigs.
And I think 2 and 4 socket systems contribution (from private donors) is small because of very high price.

Hey DonMarkoni, I have a suggestion, (hope you don't mind ) While the Official response may or may not come, you can make it easier my making a Poll. You can list these options: (feel free to elaborate but you can have a maximum of 10 options)

1) 2p System
2) 4p System
3) i7 System

Make it a never ending one and set the user selection to the maximum number of choices that you have displayed. I am sure everyone would fine it easier to click a few options rather than to type their specs. This way, it will be useful as you get accurate data easily. It isn't the most accurate but at least some valid data is better than none. (or making guess ) This does have some limitations like if you multiple systems of the same configurations, you can select only one unless you keep duplicate entries which will allow a user to select multiple systems. You can edit your first post to make the poll. look to the bottom left and you will see a tab "Poll creation"

I don't mind, it's a good idea! Can a thread (this one) be changed to poll, so we keep discussion and poll results together?

I am pretty sure you can remove PDC and Anonymous from the stats and be fairly close.

I'm not following you. PDC was the donor that originated the -bigadv WU.

viewtopic.php?f=38&t=12058&p=117859&hilit=pdc#p117860

The question of the OP is how much of the WU of -bigadv were done by I7 verse other 2 and 4 socket machines.

excluding PDC would be pointless.

It would be like asking what percentage of the total fund someone donated to the Melinda Gates foundation;( excluding Bill Gates and Warren Buffet.)

PDC is Panda group they are the owner not a donor Don was wanting to know how many DONORS were running 1P systems.

No matter, it would be good to know stats with and without them. They are making the contribution for sure, so it would not be fair to totally exclude them. They just have more money to spend on their machines.

A poll is just like a survey, it's not scientific. It only reflects what the people who vote in it may or may not be using. They don't even have to be running -bigadv and can still vote in it. I wouldn't try to debate with that kind of info.

To get back OT, Don Just PM Dr.Pande or someone else in the PG with your question.

I agree, it would not be scientific data, nor statistical, or even accurate, but it would be something for starters.
It would be let at people's conscience to state true data, and I don't think there are many liars here. We all here want to help and contribute, and usually people like that don't lie or feel the need to lie.
I don't want to know the (numerical) answer to the original question, if it is not something PG would like to share. And for sure I don't put to the open something shared only with me.

So, if we don't get an accurate answer, I'll be more then satisfied with the abstract one: every machine is significant for the project. We can all agree.

PDC is not Pande Group. Purdue Data Center...

And I didn't read that post as saying that PG edited the stats, just that Purdue had been running as Anonymous, and are now [then] reconfigured and running as PDC.

As for the OP question, I'm not sure what significance the answer would have, even if that data were tracked accurately (i.e. able to distinguish between single sockets with HT or multiple sockets, or multiple sockets with HT, etc.)

The hardware mix is always a moving target, and PG does what it can to best use that mix of hardware.

Thanks for the clarification!

EDIT: And thanks to every one of you that read this gigantic post!

[PantherX]

I don't mind, it's a good idea! Can a thread (this one) be changed to poll, so we keep discussion and poll results together?

As you started the Thread (you have the first post) you can always edit the first post and even the Thread name.

[DonMarkoni]

I don't mind, it's a good idea! Can a thread (this one) be changed to poll, so we keep discussion and poll results together?

As you started the Thread (you have the first post) you can always edit the first post and even the Thread name.

Thanks! Looked a little better and found everything.
If you have any ideas how to change/improve poll, feel free to advise.

[Grandpa_01]

PDC is not Pande Group. Purdue Data Center...

Learn something new every day. I had always assumed that PDC was Panda since they were the single largest contributor to the Panda Lab team. Well that is what assuming will get you.

[Karamiekos]

So if PDC is not PG and they are indeed a donor, where are you going to put the 400 votes for PDC's systems? Are they running Intel or AMD and are they running 2P systems then.....?

[DonMarkoni]

So if PDC is not PG and they are indeed a donor, where are you going to put the 400 votes for PDC's systems? Are they running Intel or AMD and are they running 2P systems then.....?

As I said...

... it would be good to know stats with and without them. They are making the contribution for sure, so it would not be fair to totally exclude them. They just have more money to spend on their machines.

...but, we'll never know accurately.

The thing about them is that they are not individuals' machines.

[Karamiekos]

I understand. I put my vote in honestly, to try and help out. I hope others follow suit. Good Luck.

[bruce]

And I think 2 and 4 socket systems contribution (from private donors) is small because of very high price.

Judging from (a somewhat faulty) memory of things donors have said about their systems, FAH does have both private donors who spend a lot of money on their farms (whether that's exclusively for FAH or for other purposes isn't documented either). FAH also gets contributions from private donors who may own private business with lots of big-iron or from private donors who have permission from their employer to use the employer's hardware. Even if you were able to get good statistics, I'm not sure that the "private donors" distinction is a clear as it may seem to you.

This probably doesn't apply to you, but if you started with an assumed answer and a desire to prove that theory correct, you wouldn't be the first person who unintenonally shades the data. When you start excluding data, you have to either accept all data or at least have a clear distinction between what's going to be excluded and what is going to be included and why.

For years, medical studies were done by men and the test subjects were men. Now researchers are repeating some studies to see if by excluding women, the conclusions were biased.