Folding Forum

Posted: **Wed May 26, 2010 4:30 pm**

Just got on the 6th attempt after restarting the client. Restarting the client is unrtelated to this problem, except for resetting the retry count (Attempt #) so the client is making attempts more frequently. A client that was left alone would also succeed eventually, it just would have taken a little longer.

This is not so much about WU shortage for this particular event, there's somewthing going on witht he servers getiing overloaded or a code issue in the servers. If the servers are working correctly and not overloaded, you should always be able to connect (Couldn't send HTTP request to server) to them.

[16:14:23] Connecting to http://171.67.108.20:8080/
[16:14:23] - Couldn't send HTTP request to server
[16:14:23] + Could not connect to Work Server

Code: Select all

--- Opening Log file [May 26 16:14:03 UTC] 


# Windows GPU Systray Edition #################################################
###############################################################################

                       Folding@Home Client Version 6.30r2

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Users\MikHexa\AppData\Roaming\Folding@home-gpu
Arguments: -verbosity 9 

[16:14:04] - Ask before connecting: No
[16:14:04] - User name: [EV]NissanLeaf (Team 104636)
[16:14:04] - User ID: xxxxxxxxxxxxxxxxx
[16:14:04] - Machine ID: 2
[16:14:04] 
[16:14:04] Gpu type=3 species=30.
[16:14:04] Loaded queue successfully.
[16:14:04] Initialization complete
[16:14:04] - Preparing to get new work unit...
[16:14:04] Cleaning up work directory
[16:14:04] - Autosending finished units... [May 26 16:14:04 UTC]
[16:14:04] Trying to send all finished work units
[16:14:04] + No unsent completed units remaining.
[16:14:04] - Autosend completed
[16:14:04] + Attempting to get work packet
[16:14:04] - Will indicate memory of 12279 MB
[16:14:04] Gpu type=3 species=30.
[16:14:04] - Detect CPU. Vendor: GenuineIntel, Family: 6, Model: 12, Stepping: 2
[16:14:04] - Connecting to assignment server
[16:14:04] Connecting to http://assign-GPU.stanford.edu:8080/
[16:14:04] Posted data.
[16:14:04] Initial: 43AB; - Successful: assigned to (171.67.108.20).
[16:14:04] + News From Folding@Home: Welcome to Folding@Home
[16:14:04] Loaded queue successfully.
[16:14:04] Gpu type=3 species=30.
[16:14:04] Empty passkey
[16:14:04] Connecting to http://171.67.108.20:8080/
[16:14:04] - Couldn't send HTTP request to server
[16:14:04] + Could not connect to Work Server
[16:14:04] - Attempt #1  to get work failed, and no other work to do.
Waiting before retry.
[16:14:23] + Attempting to get work packet
[16:14:23] - Will indicate memory of 12279 MB
[16:14:23] Gpu type=3 species=30.
[16:14:23] - Connecting to assignment server
[16:14:23] Connecting to http://assign-GPU.stanford.edu:8080/
[16:14:23] Posted data.
[16:14:23] Initial: 43AB; - Successful: assigned to (171.67.108.20).
[16:14:23] + News From Folding@Home: Welcome to Folding@Home
[16:14:23] Loaded queue successfully.
[16:14:23] Gpu type=3 species=30.
[16:14:23] Empty passkey
[16:14:23] Connecting to http://171.67.108.20:8080/
[16:14:23] - Couldn't send HTTP request to server
[16:14:23] + Could not connect to Work Server
[16:14:23] - Attempt #2  to get work failed, and no other work to do.
Waiting before retry.
[16:14:40] + Attempting to get work packet
[16:14:40] - Will indicate memory of 12279 MB
[16:14:40] Gpu type=3 species=30.
[16:14:40] - Connecting to assignment server
[16:14:40] Connecting to http://assign-GPU.stanford.edu:8080/
[16:14:40] Posted data.
[16:14:40] Initial: 43AB; - Successful: assigned to (171.67.108.20).
[16:14:40] + News From Folding@Home: Welcome to Folding@Home
[16:14:40] Loaded queue successfully.
[16:14:40] Gpu type=3 species=30.
[16:14:40] Empty passkey
[16:14:40] Connecting to http://171.67.108.20:8080/
[16:14:40] - Couldn't send HTTP request to server
[16:14:40] + Could not connect to Work Server
[16:14:40] - Attempt #3  to get work failed, and no other work to do.
Waiting before retry.
[16:15:02] + Attempting to get work packet
[16:15:02] - Will indicate memory of 12279 MB
[16:15:02] Gpu type=3 species=30.
[16:15:02] - Connecting to assignment server
[16:15:02] Connecting to http://assign-GPU.stanford.edu:8080/
[16:15:03] Posted data.
[16:15:03] Initial: 43AB; - Successful: assigned to (171.67.108.20).
[16:15:03] + News From Folding@Home: Welcome to Folding@Home
[16:15:03] Loaded queue successfully.
[16:15:03] Gpu type=3 species=30.
[16:15:03] Empty passkey
[16:15:03] Connecting to http://171.67.108.20:8080/
[16:15:03] - Couldn't send HTTP request to server
[16:15:03] + Could not connect to Work Server
[16:15:03] - Attempt #4  to get work failed, and no other work to do.
Waiting before retry.
[16:15:52] + Attempting to get work packet
[16:15:52] - Will indicate memory of 12279 MB
[16:15:52] Gpu type=3 species=30.
[16:15:52] - Connecting to assignment server
[16:15:52] Connecting to http://assign-GPU.stanford.edu:8080/
[16:15:52] Posted data.
[16:15:52] Initial: 43AB; - Successful: assigned to (171.67.108.20).
[16:15:52] + News From Folding@Home: Welcome to Folding@Home
[16:15:52] Loaded queue successfully.
[16:15:52] Gpu type=3 species=30.
[16:15:52] Empty passkey
[16:15:52] Connecting to http://171.67.108.20:8080/
[16:15:53] - Couldn't send HTTP request to server
[16:15:53] + Could not connect to Work Server
[16:15:53] - Attempt #5  to get work failed, and no other work to do.
Waiting before retry.
[16:17:16] + Attempting to get work packet
[16:17:16] - Will indicate memory of 12279 MB
[16:17:16] Gpu type=3 species=30.
[16:17:16] - Connecting to assignment server
[16:17:16] Connecting to http://assign-GPU.stanford.edu:8080/
[16:17:16] Posted data.
[16:17:16] Initial: 43AB; - Successful: assigned to (171.67.108.20).
[16:17:16] + News From Folding@Home: Welcome to Folding@Home
[16:17:16] Loaded queue successfully.
[16:17:16] Gpu type=3 species=30.
[16:17:16] Empty passkey
[16:17:16] Connecting to http://171.67.108.20:8080/
[16:17:16] Posted data.
[16:17:16] Initial: 0000; - Receiving payload (expected size: 44323)
[16:17:17] - Downloaded at ~43 kB/s
[16:17:17] - Averaged speed for that direction ~60 kB/s
[16:17:17] + Received work.
[16:17:17] + Closed connections
[16:17:17] 
[16:17:17] + Processing work unit
[16:17:17] Core required: FahCore_15.exe
[16:17:17] Core found.
[16:17:17] Working on queue slot 00 [May 26 16:17:17 UTC]
[16:17:17] + Working ...
[16:17:17] - Calling '.\FahCore_15.exe -dir work/ -suffix 00 -nice 19 -priority 96 -checkpoint 30 -verbose -lifeline 2860 -version 630'

[16:17:17] 
[16:17:17] *------------------------------*
[16:17:17] Folding@Home GPU Core -- Beta
[16:17:17] Version 2.09 (Thu May 20 11:58:42 PDT 2010)
[16:17:17] 
[16:17:17] Build host: SimbiosNvdWin7
[16:17:17] Board Type: Nvidia
[16:17:17] Core      : 
[16:17:17] Preparing to commence simulation
[16:17:17] - Looking at optimizations...
[16:17:17] DeleteFrameFiles: successfully deleted file=work/wudata_00.ckp
[16:17:17] - Created dyn
[16:17:17] - Files status OK
[16:17:17] sizeof(CORE_PACKET_HDR) = 512 file=<>
[16:17:17] - Expanded 43811 -> 163067 (decompressed 372.2 percent)
[16:17:17] Called DecompressByteArray: compressed_data_size=43811 data_size=163067, decompressed_data_size=163067 diff=0
[16:17:17] - Digital signature verified
[16:17:17] 
[16:17:17] Project: 10626 (Run 64, Clone 24, Gen 1)
[16:17:17] 
[16:17:17] Assembly optimizations on if available.
[16:17:17] Entering M.D.
[16:17:23] Tpr hash work/wudata_00.tpr:  2551314138 2101076186 1163881490 95594515 3752065284
[16:17:23] Working on 582 p2750_N68H_AM03
[16:17:23] Client config found, loading data.
[16:17:24] Starting GUI Server
[16:18:01] Completed 1%
[16:18:38] Completed 2%

Posted: **Wed May 26, 2010 4:32 pm**

As bruce noted in another thread, PG is ramping up the GPU3 release slowly so they can do several things along the way, and make course corrections as/if needed, like tuning the settings on the Assignment and Work Servers.

It does no good to send out 1,000,000 work units only to find that after about 5,000 WUs are completed that there is a fatal error in the code somewhere. I'm sure they're actively checking the returned WUs for data accuracy and consistency before opening the flood gates.

Posted: **Wed May 26, 2010 4:36 pm**

Hehe, you keep us informed about what your privvy to 7im, and we'll keep you informed about the glitches along the way. This situation is another hour of science opportuntiy that got missed, and some clients can complete a whole WU in under an hour. So this isn't complaining, just letting people know what's up, including other folders who may not be sure if it's their client, network or the WU servers.

Posted: **Wed May 26, 2010 4:56 pm**

Just paraphrasing what bruce said, nothing privvy there.

Agreed, good to keep the communications flowing. To rephrase, PG is weighing that opportunity missed against the potential for opportunity wasted. They tend to err on the side of caution. Some would say it's better to have the GPU potentially sit idle, than to potentially run a lot of bad WUs and also waste even more donor electricity.

And as you said, there is that unfortunate consequence of added uncertainty because of the slower roll out, but all beta testing comes with some degree of uncertainty, woulnd't you agree? So keep it flowing.

Posted: **Wed May 26, 2010 5:09 pm**

Well, it sounds like I have all the correct settings at least, so I'm happy to let me card keep chugging. Hopefully I'll see a GPU3 WU come down the pipes soon!

Posted: **Wed May 26, 2010 6:54 pm**

I think the server is setting up a wait list que when it has to many request at once. This I think is because there is only one server that is distributing the WUs perhaps setting up a second server to distribute WUs as well would be a good idea.

Posted: **Wed May 26, 2010 11:54 pm**

CadBane wrote:I think the server is setting up a wait list que when it has to many request at once. This I think is because there is only one server that is distributing the WUs perhaps setting up a second server to distribute WUs as well would be a good idea.

I have no doubt that that will happen in time, but setting up a new server doesn't ever happen overnight. (It probably takes a lot of work, including taking some other projects off the hardware to free it up for whatever is different about FahCore_15.) It's probably also part of the start slowly philosophy that bruce said and 7im mentioned.

The PandeGroup wants to use your hardware but they also want to be sure the results are correct and folks are not having too many strange problems. That's why they call it a beta test, not a full-production release. There will be plenty of WUs soon© and the full-release will happen soon.©

Posted: **Thu May 27, 2010 1:13 am**

It was just an observation. Nothing else. Its also an idea for later not exactly this moment.

Posted: **Thu May 27, 2010 2:00 am**

Earlier today someone posted a thread that was quickly deleted for reasons unkown, but it basically said that he would get core_15 WUs if he didn't have -advmethods on, and would only get core_11's if he did have -advmethods set. I removed -advmethods from my GTX 260 when it was in the middle of a core_11, and it had gotten nothing but core_11's for the past 10 WUs (as far back as the queue went). The first time it went to get a new WU, it got a core_15. Coincidence, or a strange quirk of the new client/server code/etc ?

Posted: **Thu May 27, 2010 5:16 am**

Removing the -advmethods flag in the middle of a WU won't change anything. (Once you're assigned a WU you are expected to complete it.) That may change the NEXT assignment. I don't see that as a problem. With the flag, you can get projects that are not yet mature enough to called released (also known as "late beta"). Without the flag you'll only projects that have been released. Of course the AS will only assign projects that will work on your hardware, whether it's advmethods or released fah.

At this point in time, I really doubt that there are any FahCore_15 projects that can be called mature although there should be plenty of FahCore_11 projects.

The system log does show that a certain person created a topic called "GPU3 beta with Flags Reverts to GPU2 Core" and then deleted it. I'm not sure what your concern is. Do we need to ask him why that happened? From the title I'd say it was similar information to what codysluder posted above and what I've added here.

Posted: **Thu May 27, 2010 5:56 am**

Interesting. I just removed "-advmethods" from my additional client parameters, shut down and restarted my client. I finished the WU (number 587 for the card, nice) and checked HFM.net a few minutes later, and I'm now running an OpenMMGPU work unit. Apparently "-advmethods" didn't do what I expected it to do, because I'm running the newer units with it turned off. One newer unit, at least, though I'll certainly check in nine hours when I wake up.

Posted: **Thu May 27, 2010 11:34 am**

My initial speculation was 100% wrong hence I have deleted it to avoid further confusion. Although I did observe that removing -advmethods flag allows me to get FAHCore_15 WUs while using the flag would get me FAHCore_11 WUS. Not sure why but I will be experimenting with it latter when I have some more time.

Posted: **Thu May 27, 2010 4:07 pm**

bruce wrote:Removing the -advmethods flag in the middle of a WU won't change anything. (Once you're assigned a WU you are expected to complete it.) That may change the NEXT assignment. I don't see that as a problem. With the flag, you can get projects that are not yet mature enough to called released (also known as "late beta"). Without the flag you'll only projects that have been released. Of course the AS will only assign projects that will work on your hardware, whether it's advmethods or released fah.

Yes, I said that only affected the next WU I got, not the current one.

bruce wrote:The system log does show that a certain person created a topic called "GPU3 beta with Flags Reverts to GPU2 Core" and then deleted it. I'm not sure what your concern is. Do we need to ask him why that happened? From the title I'd say it was similar information to what codysluder posted above and what I've added here.

I wasn't concerned with the topic being deleted, I just didn't want to take credit for discovering this when someone else did.

Posted: **Thu May 27, 2010 4:26 pm**

PantherX wrote:This is just my speculation about the -advmethods flag so I could be 100% wrong.

As the GPU3 Client is in BETA Stage, I believe that running it without any flags or when using only -verbosity 9 will cause the AS to give you the BETA WUs aka FAHCore_15. This is "hard-coded" in this BETA Client and is invisible to the user and my guess is to keep it simple.

Using the -advmethods flag would mean that the user doesn't want to contribute to the BETA program thus the AS will give you normal WUs aka FAHCore_11. This could be an exception in the code as it would be giving the user a choice to either participate in the BETA program or to simply install the new Client and run the normal WUs.

BTW, I have removed the -advmethods flag and got FAHCore_15 WU. My PPD loss is roughly 3K points but I am happy to contribute in the BETA program. If I can get any official wording, then I can add it to the guide here. If offical wordingisn't possible at this time, a user consolidation will do too.

We've had no indication of hard coding in the GPU3 client. And if hardcoded, an additional flag would NOT override the hard coding, hence the name.

For instance, selecting Small for the WU size in the SMP client setup (SMP client is hard coded for BigWUs) does not override the BigWU hard coding.

And -verbosity 9 only changes the detail level written to the fahlog, it has NOTHING to do with WU selection.

Please take care not to speculate too much. We don't want people to get the wrong idea, and then have that wrong idea perpetuated for the life of that client. An example of such a myth is using the -advmethods switch with the CPU client will earn you more points. It doesn't. It simply changes the mix of WUs you get. It can have no affect, it can make points go up, and it can even make your points go down. It all depends on the mix of WUs in advanced testing at any given time.

Posted: **Thu May 27, 2010 4:31 pm**

Well, mine is not to speculate why, but since removing the -advmethods switch I've run four straight GPU3 work units successfully after doing around 10 normal GPU2 units before when I had -advmethods on. I'll try adding it again this evening and see if there's a switch back to GPU2 units.

Folding Forum

171.67.108.20

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs

Re: Not receiving GPU3 WUs