2013 Chimp Challenge Race Updates & Musings

[orion456]

I been told this a lot and not once has anyone been able to give me poof. ever!

Some quotes from here: <--- link now fixed


Folding@home differs from many other projects in that it is important for WUs to be completed in a timely manner.


Work Units are serial in nature. When a completed WU is sent back, a new work unit is generated from those results. This must happen many times over within each project (group of work units). A generation 1 work unit must be turned in before a generation 2 work unit is created and sent out.

To keep these generations moving along, we have to set expiration deadlines in the event a work unit is not uploaded in a timely manner (lost, deleted, whatever). These unfinished work units "expire" and are reassigned to new machines. You will still receive credit for all WUs completed and uploaded prior to the Timeout (formerly preferred deadline). However, after the Timeout, your contribution is not as useful scientifically because another copy of that work unit had to be sent out to another contributor.

It is most helpful to the project to return work units as quickly as possible.

-------------------

So as you can see, a dumped WU becomes less scientifically useful, slows the project because the same WU has to wait to be reissued and subsequent WUs depend on previous results.

The whole object is fast science, so dumping WUs isn't compatible with that goal.

 

[EarthDog]

Spot on!! This isn't about the points, or shouldnt be really. This is about helping science help humanity. The PPD thing is just something tangible for those that are doing this type of work.

This is about science and helping people, not a PPD personal race.

 

[sbinh]

Say you don't like a team. And you aren't going to win any way, you figure. Or maybe your team has already finished the race, whatever.

So instead of staying out of the race, and letting the teams finish, you start folding for the team racing against the team you don't like.

This was mentioned in the EVGA forum recently. It has been against the rules, but this year, the Captains were so inert, by and large, that no rules were formalized.

I see. Let me go check their site

 

[Robert17]

Weather dudes were correct, lost power last night, don't know how long, so my contribution will be a bit lower than expected on today's totals.

 

[m715]

A while back before the F@H site all changed around I saw the quote directly myself while reading through lots of info for fun. Just a few days back I was looking so I could link it in a recent post I made - but could not find it as Adak mentioned I now know why.

Non-official Beta testers are semi frowned upon it turns out (was reading up just a few days ago) - but its where we are at now.

A comment from proteneer made a few days back - the guy behind 7662:

If you don't want to do a complete driver uninstall / reinstall for any reason (maybe its faster or you want latest display driver for games still ), in a post I made recently i left instructions on how to swap out the OpenCL AAP SDK (part that does the folding from Catalyst.) The older 12.7 / 12.8 package had a good version for folding older 11292 & 11293 WUs

13.1 = 2K PPD --- each WU takes about 20hours
12.8 = 7.3K PPD --- each WU takes about 5hrs 30mins
(on my 7950 @ 1050)

Link to Post here
(AS Long as you clean out those mentioned files before installing the new OpenCL I have found no restart is actually required in testing the last few days.)

12.8 - failed with NaNs on the early 7661, not tested with 7662 enough to know. (select Finish Folding and hope to get 7662 to be certain, or let us know how you get on)
13.2 - I think I had the odd issue with the older WU's failing.
13.3 Beta is 100% broken for folding - don't even try it.

Recently I have had much less 11292 & 11293 to contend with thankfully.


Folding on!

Hmm, I didn't know CCC 13.3 was supposed to be broken I've had my 2 main rigs folding with it with no issues...no one wants to by my i5-6950 rig so it will continue to fold until I find a buyer
1) [email protected] w/6950
2) [email protected] w/7950x2

Now my 3rd rig I slapped together last night has crashed a couple of time and I don't know why, its P95 & IBT stable, temps are fine. I've had 2 blue screens of death and 1 hang and I don't know why.
3) Intel G550 CPU w/7770

Old 1/2 broken laptop I gave up on, crashed, was hitting 85C, after last night folding @ home won't even load...don't know why and don't have time atm.


Do these PPDs make sense?
7770@1000MHz = [EDIT] seems stable at 7000 ppd
6950@800MHz = 10000-12000 ppd w/0x17
7950@925MHz= 33000-37000 ppd w/0x17 each

[email protected] = ~10000 ppd
[email protected] = ~7000-9000 ppd (but I'm tying and running netflix)

 

[m715]

Ok, something strange was going on with my 7770 GPU & folding, the ppd kept climbing until it reached 81,000 ppd But, it must have figured out what was wrong now 10 mins later it has stabled out at 7,000 ppd which seems more reasonable I think...

 

[Humanoid1]

Hmm, I didn't know CCC 13.3 was supposed to be broken I've had my 2 main rigs folding with it with no issues...no one wants to by my i5-6950 rig so it will continue to fold until I find a buyer
1) [email protected] w/6950
2) [email protected] w/7950x2

Now my 3rd rig I slapped together last night has crashed a couple of time and I don't know why, its P95 & IBT stable, temps are fine. I've had 2 blue screens of death and 1 hang and I don't know why.
3) Intel G550 CPU w/7770

Old 1/2 broken laptop I gave up on, crashed, was hitting 85C, after last night folding @ home won't even load...don't know why and don't have time atm.


Do these PPDs make sense?
7770@1000MHz = [EDIT] seems stable at 7000 ppd
6950@800MHz = 10000-12000 ppd w/0x17
7950@925MHz= 33000-37000 ppd w/0x17 each

[email protected] = ~10000 ppd
[email protected] = ~7000-9000 ppd (but I'm tying and running netflix)

I tried testing 13.3 the other day on my 7950. It would go into a loop of download and insta-fail like:

23:23:29:WU00:FS00:0x17:ERROR:Win32: 0xc0000005: Exception access violation
23:23:29:WU00:FS00:0x17:Saving result file logfile_01.txt
23:23:29:WU00:FS00:0x17:Saving result file log.txt
23:23:29:WU00:FS00:0x17:Folding@home Core Shutdown: BAD_WORK_UNIT
23:23:30:WARNING:WU00:FS00:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
23:23:30:WU00:FS00:Sending unit results: id:00 state:SEND error:FAULTY project:7662 run:32 clone:28 gen:41 core:0x17 unit:0x00000038ff3d48355139278df9d99d6b

on all available AMD type work units.
I gave it a good (install) try and a few restarts etc. Just would not play ball for me!
But if on Your machine things are smooth with good PPD, carry on

 

[Mandrake4565]

Do these PPDs make sense?
7770@1000MHz = [EDIT] seems stable at 7000 ppd
6950@800MHz = 10000-12000 ppd w/0x17
7950@925MHz= 33000-37000 ppd w/0x17 each
[email protected] = ~7000-9000 ppd

Here's what I'm getting.
5850 @ 775/1125 8-9k ppd
450 gts @ 900/1850 14k ppd
650 gtx @ 1150/2600 14k ppd
I5-2500k @ 4.5 18-22k ppd depending on WU I also have the 2 nvidia cards running on it.



Im running CC driver 12.3 on the 5850 I wonder if I'd pull more pages with 12.8? It also seems as if the Nvidia cards do better then the AMD considering my 5850 is much more powerful then either the 450 gts or the 650 gtx.

 

[EarthDog]

13.3's did the same thing for me.. I went back to 13.1 WHQL and all was well (until the beta units started sucking).

 

[BrotherJohn]

Weather dudes were correct, lost power last night, don't know how long, so my contribution will be a bit lower than expected on today's totals.

I woke up to dead comps also.
So I lost a few hours in dead time.
I really should get new batteries for my UPS's but at 48 volts per UPS it adds up to quite a few $$.
6.5 hours til my next dump.

 

[Macaholic]

I been told this a lot and not once has anyone been able to give me poof. ever! this is like the 10th time. The closest I gotten for proof was a link to someone else saying the same thing.
you can also block the IP of servers that give out harder WU's, but I never go that far. more then half of the time I just fold what I get.
But if my cpu client (don't care about gpu's) goes over 4min tpf I kill it and just get a new WU. if it's under 5%, they run my system to hard and use up to much power compared to the smaller ones.
not to mention that you can specify what WU's you get(does not guarantee you will get them)

Check Dr. Pande's blog. Still holds true today.

 

[Angry]

I remoted in and turned off a couple of my systems, AEP finally showed up to replace the pole outside my house. After finding out the gas line had been ran right next to the pole.... I left a couple on that shouldn't mind if the power goes out or flickers.

 

[Humanoid1]

Here's what I'm getting.
5850 @ 775/1125 8-9k ppd
450 gts @ 900/1850 14k ppd
650 gtx @ 1150/2600 14k ppd
I5-2500k @ 4.5 18-22k ppd depending on WU I also have the 2 nvidia cards running on it.



Im running CC driver 12.3 on the 5850 I wonder if I'd pull more pages with 12.8? It also seems as if the Nvidia cards do better then the AMD considering my 5850 is much more powerful then either the 450 gts or the 650 gtx.

I "upgraded" from a 5850 as it happens, did actually know what I was walking into, and the 7950 gets Less PPD even highly overclocked.

Catalyst 12.8 is the latest Stable driver (OpenCL APP SDK) for 11292 & 11293
but 12.7 is reputed to be a Tiny bit faster.

When my 5850 was overclocked to 825MHz I would be a reliable 8,400PPD on avg. Your points sound about as good as they get on that card perhaps.

Soon(ish... still in early Closed Beta testing) now there will be new official GPU WU's that would see an OC'd 7950 getting as much about 74K PPD.
There are a little closer to public release a registered Open Beta 7662 WU that gets 37K PPD.

Good times ahead for us and Pande Labs, as PPD are awarded based on scientific throughput in the form of completed work the scientists can make use of and learn from.

All achieved through finally updated (you can imagine the lengthy test process to make sure no errors are introduced!!!) GPU core and Work Units that support AMD's fantastic Graphics Core Next architecture (my addition the GCN part, but pretty sure that's where the Huge gains came from.) as well as probably improved for older generations

 

[Guns]

Weather dudes were correct, lost power last night, don't know how long, so my contribution will be a bit lower than expected on today's totals.

Lost Power at my house as well due to T Storm and tornado watch. My rigs were down about an hour, got them back up and folding.

 

[Mandrake4565]

When my 5850 was overclocked to 825MHz I would be a reliable 8,400PPD on avg. Your points sound about as good as they get on that card perhaps.

Thank you Humanoid, I appreciate it. I was concerned that I didn't have the correct driver for the 5850 for F@H. I raised the overclock on it to as high as 1000/1145 but my ppd didn't go up much at all so I settled for the 775/1125 I'm running now. I have issues with any driver higher then 12.3 for daily usage. It doesn't crash, but the screen tends to flicker when I'm scrolling a web page or something similar. So I'll leave it as is, it just burns my bottom that I'm getting 14k with two cards that are less powerful then the 5850. I want more power

On another note a friend lent me 2 dell workstation towers to add to the CC. They both have XEON's in them and they are a little old I believe 07/08. I figured even if I could get 10k ppd out of them it would be worth it. Ummm not even under ubuntu they were doing 1k ppd, oh well.

 

[Humanoid1]

@m715

Further thought on 13.3
The most likely reason it is "working" for you is that when you uninstalled the last Cat version from Add/Remove Programs (or just did a straight upgrade (this case also I believe)) is the certain files integral to the OpenCL folding part of the Catalyst package do not get removed from the uninstall process:

C:\Windows\System32

amdocl64.dll
OpenCL.dll
OpenVideo64.dll
OVDecode64.dll
SlotMaximizerAg.dll
SlotMaximizerBe.dll

and

C:\Windows\SysWOW64

amdocl.dll
OpenCL.dll
OpenVideo.dll
OVDecode.dll
SlotMaximizerAg.dll
SlotMaximizerBe.dll

So you probably are not in fact fully folding on 13.3.

I would imagine the Official AMD Uninstaller removes these. (though I've not tested as I was removing them manually)

 

[Guns]

Here's what I'm getting.
5850 @ 775/1125 8-9k ppd
450 gts @ 900/1850 14k ppd
650 gtx @ 1150/2600 14k ppd
I5-2500k @ 4.5 18-22k ppd depending on WU I also have the 2 nvidia cards running on it.



Im running CC driver 12.3 on the 5850 I wonder if I'd pull more pages with 12.8? It also seems as if the Nvidia cards do better then the AMD considering my 5850 is much more powerful then either the 450 gts or the 650 gtx.

I have 4 EVGA GTX 580 hydro coppers and get 31k ppd each.
Just thinking your 650 gtx should do better.

 

[Mandrake4565]

I have 4 EVGA GTX 580 hydro coppers and get 31k ppd each.
Just thinking your 650 gtx should do better.

I would think the same Guns considering the 450 gts is doing the same ppd on the same pc. I have the 650 gtx slightly overclocked but it's not mine so I'm not going to push it. I did read here what some other people are getting on the 650 so I guess I'm doing ok.

 

[kyfire]

I would think the same Guns considering the 450 gts is doing the same ppd on the same pc. I have the 650 gtx slightly overclocked but it's not mine so I'm not going to push it. I did read here what some other people are getting on the 650 so I guess I'm doing ok.

I running two rigs both with GTX 650 at stock. Over the last 11 WUs on them I've averaged 6762 PPD.

 

[Mandrake4565]

at stock

What's that, I don't understand the word "stock" this is Overclockers.