I think I will arbitrarily pick a goal number and once it is met, pull the plug on boinc in general unless things overall get better.

Been there, done that, got the shirt ... :)

Things got so bad for me that I left for almost 2 1/2 years ... when I came back it was the same muddle because you have the same basic faces in the same basic places doing the same basic things expecting a different outcome ...

Please be patient, folks...Travis is still a student and has classwork to do as well.

As of 4:30pm the servers are running.

---

While I am not a 'credit whore' when a project constantly cuts back on allotted credit vs run time it does put me off a bit.

...and those of us CPU-only folks are the people bearing the brunt of that credit cut, not those of you with GPUs. You can check that out with your own hosts, particularly the Pentium D. In fact, it appears as though the GPU users credit rates stayed constant or went up some. Your RAC "dropping" is merely due to work fetch issues. If you had a steady stream of work, your RAC would actually be higher now, while those of us with CPU-only systems would still be lower.

I'm sorry, my GPU credits have been going down hill since the last server crash.
My peak with two 4850's on this machine was a little over 90K and still had a very steep climbing slope. Then the crash hit and it is now leveling off at mid 70K.
Do not tell me about credit hits... I should be averaging some 220 to 250K (or more) a day on MW according to the old rules with two 4850 and a 4870. I am barely breaking 100 to 120K most days if that. Work fetch issues means I am in the 60 to 80K range.
My single 4870 was doing 120K+ on its own long ago.

---

I'm sorry, my GPU credits have been going down hill since the last server crash.
My peak with two 4850's on this machine was a little over 90K and still had a very steep climbing slope. Then the crash hit and it is now leveling off at mid 70K.

You started Collatz on September 19, 2009. Other GPU users are notating problems with the two projects playing together.

According to the graph at BOINCStats for your host, looking back across 60 days it becomes plainly clear that your credit per day stayed relatively stable or went up on that host post-cut (09/11 was the date of the cut). The average amount from 09/05/2009 to 09/19/2009 looks to be in the range of 94k - 104k, which includes post-cut rates. Once you started mingling Collatz into the mix, the rates went down. Additionally there have been work availability issues and some bad tasks here and there.

To state that your credit production has dropped within this project is accurate, but doesn't tell the whole story. Had you been getting the same stream of work as before, not mixing another project in, it would've been higher, just like it was during the 09/11/2009 - 09/19/2009 time period, which was after the cut was implemented.

Meanwhile, CPU users generally only come to within 80% of pre-cut rates.

The stats are the stats, and the stats tell the same general story time and time again... CPU users took the brunt of the cut. Period.

I'm sorry, my GPU credits have been going down hill since the last server crash.
My peak with two 4850's on this machine was a little over 90K and still had a very steep climbing slope. Then the crash hit and it is now leveling off at mid 70K.

You started Collatz on September 19, 2009. Other GPU users are notating problems with the two projects playing together.

According to the graph at BOINCStats for your host, looking back across 60 days it becomes plainly clear that your credit per day stayed relatively stable or went up on that host post-cut (09/11 was the date of the cut). The average amount from 09/05/2009 to 09/19/2009 looks to be in the range of 94k - 104k, which includes post-cut rates. Once you started mingling Collatz into the mix, the rates went down. Additionally there have been work availability issues and some bad tasks here and there.

To state that your credit production has dropped within this project is accurate, but doesn't tell the whole story. Had you been getting the same stream of work as before, not mixing another project in, it would've been higher, just like it was during the 09/11/2009 - 09/19/2009 time period, which was after the cut was implemented.

Meanwhile, CPU users generally only come to within 80% of pre-cut rates.

The stats are the stats, and the stats tell the same general story time and time again... CPU users took the brunt of the cut. Period.

I have only added collatz when MW is down. And I was only running it when MW was down. Right now starting today I am running them side by side on same machine. As of Oct/1/09 I took my 4870 completely off MW all together. So any averages or number I refer to are before that happening.

I can look at my boinc client and see very clearly what is going on. You can look at what ever you wish, I have what is going on right from my client. Before Oct/1/09 I had only run Collatz for a total of 2 days on my two machines.

My client MW numbers say this...
Sept/2/09 it was a little under 70K
Sept/8/09 it had climbed to 85K or so but with a sharp incline.
Sept/9/09 server crash drops below 80K
Sept/16/09 it had slowly grown to 92-93K
Sept/22/09 server crash drops below 70K
Sept/29/09 slowly leveled off to 74-76K
Oct/1/09 server crash drops to near 60K

This is the client that has two 4850's installed.

So given this, at best it came within around 20K of the high points. And I will debate this because before the server crash of Sept/8/09 my level was rising sharply. By the time the crash of Sept/21/09 rolled around the changes were already made and the credits were lowered the first time.

---

I'm sorry, my GPU credits have been going down hill since the last server crash.
My peak with two 4850's on this machine was a little over 90K and still had a very steep climbing slope. Then the crash hit and it is now leveling off at mid 70K.

You started Collatz on September 19, 2009. Other GPU users are notating problems with the two projects playing together.

According to the graph at BOINCStats for your host, looking back across 60 days it becomes plainly clear that your credit per day stayed relatively stable or went up on that host post-cut (09/11 was the date of the cut). The average amount from 09/05/2009 to 09/19/2009 looks to be in the range of 94k - 104k, which includes post-cut rates. Once you started mingling Collatz into the mix, the rates went down. Additionally there have been work availability issues and some bad tasks here and there.

To state that your credit production has dropped within this project is accurate, but doesn't tell the whole story. Had you been getting the same stream of work as before, not mixing another project in, it would've been higher, just like it was during the 09/11/2009 - 09/19/2009 time period, which was after the cut was implemented.

Meanwhile, CPU users generally only come to within 80% of pre-cut rates.

The stats are the stats, and the stats tell the same general story time and time again... CPU users took the brunt of the cut. Period.

I have only added collatz when MW is down. And I was only running it when MW was down. Right now starting today I am running them side by side on same machine. As of Oct/1/09 I took my 4870 completely off MW all together. So any averages or number I refer to are before that happening.

I can look at my boinc client and see very clearly what is going on. You can look at what ever you wish, I have what is going on right from my client. Before Oct/1/09 I had only run Collatz for a total of 2 days on my two machines.

My client MW numbers say this...
Sept/2/09 it was a little under 70K
Sept/8/09 it had climbed to 85K or so but with a sharp incline.
Sept/9/09 server crash drops below 80K
Sept/16/09 it had slowly grown to 92-93K
Sept/22/09 server crash drops below 70K
Sept/29/09 slowly leveled off to 74-76K
Oct/1/09 server crash drops to near 60K

This is the client that has two 4850's installed.

So given this, at best it came within around 20K of the high points. And I will debate this because before the server crash of Sept/8/09 my level was rising sharply. By the time the crash of Sept/21/09 rolled around the changes were already made and the credits were lowered the first time.

Something changed on your host on 09/04/2009 because at that point it spiked up and stayed up. This is typically due to a change in resource allocation or new hardware. Your daily totals from 09/04 - 09/11 are what BOINCStats sees as your top 5 days, EVER, and they have a max of 116k, so I don't know where the 120k+ figure is coming from and it certainly was not "long ago", unless 1 month is the definition of "long ago". As for the drop post-09/11, another factor are some bad work units that made it out, and any work shortage during that time also would've also caused a drop in production.

As I said, if you had the same workflow as you did from 09/12-09/19, your credits would still be in that same range. You are complaining about post-outage and post-Collatz being added in. Instead of looking at RAC, try looking at the raw credit per unit time. If you did that, you'll note what I'm trying to tell you, which is that your system is +/- 5% of where it was pre-09/11 cut. I'd be highly surprised if it is any more than 8% lower (daily production of 105k), where as your CPU systems, just like all of the other CPU systems out there, are -20% or more lower.

There will always be server problems. The beauty of BOINC is that when that happens you do something else. No one is forcing anyone to put BOINC on their computers. This is all about volunteer distributed computing. You don't like it then go elsewhere. Get over your perceived self importance to any single project.

And as I said, When this exact same system had that single 4870 installed it would get 100 - 120K a day.
When I added the twin 4850's a little over two months ago now the credit should have gotten to at least 160K a day.

You can look at the 60 day averages all you want. I have been looking at them for longer than 60 days. I have been looking at the for months on end.
I know what they were, and what they should be.
I do not care what my 5 highest days were, or are. I know the numbers that people were getting from these cards before they started playing tiddly winks with the credit system.
4870's were good for around 90-100K.
4850's were good for 70-80K. Given that I have two that of course doubles that.
So at the peak times when both were operating and there were no server issues I should have been pushing 200K easy at MW. That is again what was starting to happen (Sept/9/09 with that peak starting to sharply rise) before the servers constantly played hookie, and they fiddled with the credit amounts because some idiot was bitching about them being more than his project.

Those numbers also show very clearly that there was at minimum a 20K difference between all the credit changes running every day, all day same work allotment on the same machine.

You tend to ignore that I am only speaking about a single host. A single host that has only ran collatz for two days while MW was down. So that it ran Collatz at all is completely mute and meaningless to the numbers that I am seeing and speaking of. You seem to think that it matters, MW was down, and I disconnected from collatz after it came back up.
I stopped citing number after I move the host to both projects.
I cannot be more clear than that.

I am not going to keep arguing with what I can plainly see on my host average tab in the boinc client. What you say has little meaning when what I can see is what the host is reporting on average work. When I can look through the logs and see the last time it got work from the server. When I see that it has been constantly fed work there that point is mute. The average is true.

I am not saying that I am all that important to this project, I never did or assumed that I was. The fact is that this project is letting down the participants with the constant changing of the rules for credit. I said that some time ago and very much stand behind it.
I have also stated that I have moved one client totally off MW because of this.
I have also stated that I will be moving my other client completely off unless this project gets a backbone and sticks to their credit granting system and stops lowering credit for work done.
It is not a threat, it is a clear and concise point for my reasoning. And by the looks of it, I am not the only one that feels this way. I am just going further into detail about it than others.

To be perfectly honest the single and only reason why I joined this project was because it was the only project that used the ATI cards. Now there is another project and hopefully more will follow. So in essence this project got the ATI's by default. That is simply no longer the case and people will move to other projects as more support the ATI, and as project admins do things to piss of volunteers.

---

And as I said, When this exact same system had that single 4870 installed it would get 100 - 120K a day.
When I added the twin 4850's a little over two months ago now the credit should have gotten to at least 160K a day.

The stats sites simply do not agree with what you're saying.

You tend to ignore that I am only speaking about a single host. A single host that has only ran collatz for two days while MW was down.

Nope, I didn't look at your X2 host until you claimed 120k+ and so I went searching to see if the X2 had gotten that, which it hasn't...

So that it ran Collatz at all is completely mute and meaningless to the numbers that I am seeing and speaking of. You seem to think that it matters, MW was down, and I disconnected from collatz after it came back up.
I stopped citing number after I move the host to both projects.
I cannot be more clear than that.

...and again I stated that the stat sites reflect relatively stable numbers between 09/12 (post-cut) and 09/19 (pre-Collatz). The drop had something to do with bad tasks. Due to how short of a time it is before tasks are purged, unless you're either tracking them through an alternative means or you check your results on the web page REAL OFTEN, you simply won't see them being declared invalid.

The fact is that this project is letting down the participants with the constant changing of the rules for credit. I said that some time ago and very much stand behind it.

On that point you and I agree, however GPUs didn't take a hit on the credit cut alone. If, and that is a big IF, they did, it was very minor in comparison to the 20-30% cut that those of us with CPUs have had...

Bah... this is pointless...just like it was several months ago. Hopefully the project will work another Scotty-like "miracle" to make people happy that won't involve shoving me or thousands of other people who aren't just in this for the most points right out the door...

A 4870 gets 80-90k.
A 4850 gets 70-80k.
A 3850 gets 25-28k.

My Q9450 with a 4870 and a 4850 in it gets 150k to 180k per day. Was when using 0.19 and now is with 0.20. Now I just need to work out why it locks up every other day and hope that when it hasn't locked up MW has work.

A 4870 gets 80-90k.
A 4850 gets 70-80k.
A 3850 gets 25-28k.

My Q9450 with a 4870 and a 4850 in it gets 150k to 180k per day. Was when using 0.19 and now is with 0.20. Now I just need to work out why it locks up every other day and hope that when it hasn't locked up MW has work.

Your Q9450 is what I looked at to verify your claim from several days ago about not taking a cut. What I think is muddying the waters is the bad tasks from around the 09/11 timeframe as well as work fetch problems. I also suspect that there are tasks that come out invalid that people don't notice. Also there has been a lot of tinkering with driver versions, which may make a difference, as well as overclocking, which could cause errors, which again due to the short retention time on the server, unless you keep an eye on things you just won't notice.

CPU users on the other hand are consistently lower now, by large percentages. We took the majority of the actual credit cut... All one has to do is look at stats of CPU-only systems that have remained active during the past 60 days...

CPU users on the other hand are consistently lower now, by large percentages. We took the majority of the actual credit cut... All one has to do is look at stats of CPU-only systems that have remained active during the past 60 days...

Which is one of the reasons I have been slowly moving off CPUs attached to this project and only letting my ATI cards run work here and now one pair of CUDA GPUs...

Lord I wish UCB would fix the credit problems instead of pretending that they are not an issue...

What was the name of the project that said that they were going to stop awarding credit? As I recall they didn't while I was still attached several months later ... I wonder if they ever did follow through and how that worked out for them?

The funniest thing is that there have been several published papers that talk about how important and central the credit stuff is to the design of BOINC ... it is too weep ...

CPU users on the other hand are consistently lower now, by large percentages. We took the majority of the actual credit cut... All one has to do is look at stats of CPU-only systems that have remained active during the past 60 days...

Which is one of the reasons I have been slowly moving off CPUs attached to this project and only letting my ATI cards run work here and now one pair of CUDA GPUs...

The credit per task even for CPUs is better than I could get at nearly any project. My P4 is still getting 30-35/hr, which is better than Einstein and likely better than an optimized SETI app. My AMD gets close to 50/hr sometimes... The payout is still better, even for CPUs. However, for GPUs there has been only minor fluctuations if one has a stable system, and does not have TDS (Time Dilation Syndrome) where 28-30 days is considered to be "a little over 2 months".

Please be patient, folks...Travis is still a student and has classwork to do as well.

As of 4:30pm the servers are running.

I will support that statement from Blurf. If Travis was to decide that his studies were more important than the hassle he sometimes gets here, where would that leave us?

---

Don't drink water, that's the stuff that rusts pipes

I expect you also support this trend (the bit at the end);






It looks like some are supporting this trend;

---

Now those are pictures worth a thousand words....