FAHWatch7

[MtM]

Ahh yeah forgot to mention build 45 add an optional delay field which can be used to simulate the average time needed to upload the results file.

In time I'll expand the functionality so that on entering a work unit which has been run before, it will use the known upload time from that project's results file.

I do notice an issue I need to fix with the parser code, but that will not show for most people. In my case it did since I swapped hardware, nvidia for another nvidia, and my parser code didn't notice the dump event of the work unit, resulting in two work units which both got the same wu id and ran on the same slot id to be listed as completed when the latter one was submitted.

[GreyWhiskers]

FYI, I've not been able to get the remote connections to work with v44 or v45. In v44, it seemed the same for both trying to add the remotes from either the installer or from the options menu from the application. In both cases, when I hit the "ADD" button, the only option seems to be for the app to "enumerate computers". I have two remote computers, but it isn't finding either. In the prior versions, there was an option, which I can't find in 44 or 45, to manually search for the remotes.

EDIT: Running on a Win7 Home Premium computer with FAH v7.1.43.

[MtM]

The 'enumerating computers' is displayed while an activedirectory query is being made, if that times out it should display the manual browse option. I'll look into it, maybe I forgot to reenable the timeout when I was trying the best interval for the time out. When started with the 'simple browser' shortcut it should default to the windows network browser dialog, again maybe I forgot to enable the switch in the last two builds.

Edit: the last build and probably the one before that as well don't parse correctly, I'm working on it ( I did fix allot of other smaller stuff already ).

[PantherX]

...I only asked that gpu clients also logged steps completed vs steps total if possible...

AFAIK, this is done on the FahCore level which means that Joe can't do much. It is up to the developers of FahCore to do it. My best guess is that since there are many programmers involved in the development of the FahCore, reaching a standard/consistent output from all FahCores is something that isn't currently possible. I hope that this changes one day but it doesn't seem likely.

[gwildperson]

But, I didn't request that being standardized, I only asked that gpu clients also logged steps completed vs steps total if possible.

It looks to me like they're doing that:

Code: Select all

16:20:06:WU00:FS02:0x15:Setting checkpoint frequency: 500000
16:20:06:WU00:FS02:0x15:Completed         3 out of 50000000 steps (0%).
16:22:04:WU00:FS02:0x15:Completed    500000 out of 50000000 steps (1%).
16:24:02:WU00:FS02:0x15:Completed   1000000 out of 50000000 steps (2%).

[MtM]

...I only asked that gpu clients also logged steps completed vs steps total if possible...

AFAIK, this is done on the FahCore level which means that Joe can't do much. It is up to the developers of FahCore to do it. My best guess is that since there are many programmers involved in the development of the FahCore, reaching a standard/consistent output from all FahCores is something that isn't currently possible. I hope that this changes one day but it doesn't seem likely.

I'm not sure, but I would guess FAHClient itself does not copy the log file from wudata.log to the FAHClient log ( eg a file to file copy with a datetime stamp in front ).

Like I said, I'm not sure right now ( but I could check... ) how the fahcore's work in conjunction with FAHClient. If FAHClient does not use standardout, it is reading wudata.log, adding a timestamp in front of it, formats string data for slot and wu id , and then adds the wudata.log string. If as I expect, it doesn't look at the files but ties into standardout it's doing the same without file io overhead.

Either way, there is already formatting taking place ( and don't get me wrong, I know it's not a priority as I said before I'm quite happy with the current log standard ).

But if the gpu core's do report steps total and steps completed when queried using the remote interface ( again I haven't even checked.. I just threw the idea out there ), it would create a even more uniform log. And that I think is something which is good to strive for even if it has a low or even trivial priority.

But, I didn't request that being standardized, I only asked that gpu clients also logged steps completed vs steps total if possible.

It looks to me like they're doing that:

Code: Select all

16:20:06:WU00:FS02:0x15:Setting checkpoint frequency: 500000
16:20:06:WU00:FS02:0x15:Completed         3 out of 50000000 steps (0%).
16:22:04:WU00:FS02:0x15:Completed    500000 out of 50000000 steps (1%).
16:24:02:WU00:FS02:0x15:Completed   1000000 out of 50000000 steps (2%).

Got your contacts in? Ok...

Code: Select all

17:10:36:WU02:FS02:0x11:Completed 1%
17:11:36:WU02:FS02:0x11:Completed 2% <- GPU ;)
17:12:09:WU00:FS01:0xa4:Completed 77500 out of 250000 steps  (31%) <- SMP ;)
17:12:35:WU02:FS02:0x11:Completed 3%
17:13:36:WU02:FS02:0x11:Completed 4%
17:14:34:WU00:FS01:0xa4:Completed 80000 out of 250000 steps  (32%)
17:14:35:WU02:FS02:0x11:Completed 5%
17:15:35:WU02:FS02:0x11:Completed 6%
17:16:35:WU02:FS02:0x11:Completed 7%
17:17:34:WU02:FS02:0x11:Completed 8%
17:18:23:WU00:FS01:0xa4:Completed 82500 out of 250000 steps  (33%)
17:18:35:WU02:FS02:0x11:Completed 9%
17:19:34:WU02:FS02:0x11:Completed 10%
17:20:35:WU02:FS02:0x11:Completed 11%
17:21:35:WU02:FS02:0x11:Completed 12%
17:22:00:WU03:FS00:0x11:Completed 78%
17:22:35:WU02:FS02:0x11:Completed 13%
17:23:34:WU02:FS02:0x11:Completed 14%
17:24:34:WU02:FS02:0x11:Completed 15%
17:25:34:WU02:FS02:0x11:Completed 16%
17:26:35:WU02:FS02:0x11:Completed 17%
17:27:35:WU02:FS02:0x11:Completed 18%
17:28:34:WU02:FS02:0x11:Completed 19%
17:29:33:WU02:FS02:0x11:Completed 20%
17:30:32:WU02:FS02:0x11:Completed 21%
17:31:31:WU02:FS02:0x11:Completed 22%
17:32:31:WU02:FS02:0x11:Completed 23%
17:32:54:WU00:FS01:0xa4:Completed 85000 out of 250000 steps  (34%)
17:33:30:WU02:FS02:0x11:Completed 24%
17:34:29:WU02:FS02:0x11:Completed 25%
17:35:19:WU00:FS01:0xa4:Completed 87500 out of 250000 steps  (35%)
17:35:28:WU02:FS02:0x11:Completed 26%
17:36:27:WU02:FS02:0x11:Completed 27%
17:37:26:WU02:FS02:0x11:Completed 28%
17:38:26:WU02:FS02:0x11:Completed 29%
17:39:28:WU02:FS02:0x11:Completed 30%
17:40:27:WU02:FS02:0x11:Completed 31%
17:41:26:WU02:FS02:0x11:Completed 32%
17:42:28:WU02:FS02:0x11:Completed 33%
17:42:54:WU00:FS01:0xa4:Completed 90000 out of 250000 steps  (36%)
17:43:27:WU02:FS02:0x11:Completed 34%
17:43:46:WU03:FS00:0x11:Completed 79%
17:44:27:WU02:FS02:0x11:Completed 35%
17:45:27:WU02:FS02:0x11:Completed 36%
17:46:27:WU02:FS02:0x11:Completed 37%
17:47:27:WU02:FS02:0x11:Completed 38%
17:48:26:WU02:FS02:0x11:Completed 39%

But, you do have a point there, core11 is quite old both nvidia wise and ati wise. So it might be that they are no capable of reporting steps completed, now that I'm thinking about it that's actually something I been told before

Sometimes you can be wrong and right at the same time, congrats and thanks

[bruce]

FahCore_11 = old = Steps not reported
FahCore_15 = new = Steps reported

It looks like the developers agree with you but I doubt they're going to retrofit old cores.

[MtM]

No updates but maybe nice to show some of the things I been working on..







*Problem with graph labels when there are many bar items.


*Solved with some manual pan and zoom. I'm working on code which does this when you select the RCG from the combobox.


Clicking on a label will select the workunit in the history view and show the tpf graph and work unit details.

The parser code is working, but I'm waiting to release an update until I completed the graphs and supporting functions. The last blocking calls in the parser have been replaced with async events so the gui always stays responsive ( last change was async accessing of the remote log files ). Running into some problems with error notification, the systray icon's events will not trigger, not sure on what's up with that.

The new parser code is much better at updating ( eg from a previously parsed state, up till now it parsed the whole log file just skipping already known work units ).

I'm also considering changing the remote clients setup, replace the shared client data folder with just the \logs folder and retrieving the active log through the remote interface.

[jimerickson]

really like it MtM! can't wait for the update.

[PantherX]

The mini statistics at the bottom left is very intriguing!

[MtM]

I'm in a bit of a dilemma, Harlam has been so nice to give me a FAHClient interface which isn't on the svn yet ( let's call it 'beta testing ' ) so I have what I need to finish the project. I had some problems I thought, since I had some work unit's with those strange frametime graphs, thought it was my code but it was not. The problem is really in the log files themselves:

Code: Select all

23:19:51:WU00:FS01:0x11:Completed 56%
23:20:07:WU01:FS02:0x11:Completed 54%
23:21:17:WU00:FS01:0x11:Completed 57%
23:21:35:WU01:FS02:0x11:Completed 55%
23:22:42:WU00:FS01:0x11:Completed 58%
23:23:01:WU01:FS02:0x11:Completed 56%
23:24:08:WU00:FS01:0x11:Completed 59%
23:24:28:WU01:FS02:0x11:Completed 57%
23:25:34:WU00:FS01:0x11:Completed 60%
23:25:55:WU01:FS02:0x11:Completed 58%
23:26:59:WU00:FS01:0x11:Completed 61%
23:27:22:WU01:FS02:0x11:Completed 59%
23:28:25:WU00:FS01:0x11:Completed 62%
23:28:49:WU01:FS02:0x11:Completed 60%
23:29:51:WU00:FS01:0x11:Completed 63%
23:30:16:WU01:FS02:0x11:Completed 61%
23:30:40:WU02:FS00:0xa3:Completed 160000 out of 1000000 steps  (16%)
23:31:16:WU00:FS01:0x11:Completed 64%
23:31:43:WU01:FS02:0x11:Completed 62%
23:32:47:WU00:FS01:0x11:Completed 65%
23:33:15:WU01:FS02:0x11:Completed 63%
23:34:13:WU00:FS01:0x11:Completed 66%
23:34:42:WU01:FS02:0x11:Completed 64%
23:35:39:WU00:FS01:0x11:Completed 67%
23:36:09:WU01:FS02:0x11:Completed 65%
23:37:04:WU00:FS01:0x11:Completed 68%
23:37:36:WU01:FS02:0x11:Completed 66%
23:38:30:WU00:FS01:0x11:Completed 69%
23:39:02:WU01:FS02:0x11:Completed 67%
23:39:56:WU00:FS01:0x11:Completed 70%
23:40:30:WU01:FS02:0x11:Completed 68%
23:41:21:WU00:FS01:0x11:Completed 71%
23:41:56:WU01:FS02:0x11:Completed 69%
23:42:47:WU00:FS01:0x11:Completed 72%
23:43:24:WU01:FS02:0x11:Completed 70%
23:44:13:WU00:FS01:0x11:Completed 73%
23:44:18:WU02:FS00:0xa3:Completed 170000 out of 1000000 steps  (17%)
23:44:50:WU01:FS02:0x11:Completed 71%
23:45:38:WU00:FS01:0x11:Completed 74%
23:46:17:WU01:FS02:0x11:Completed 72%
23:47:04:WU00:FS01:0x11:Completed 75%
23:47:44:WU01:FS02:0x11:Completed 73%
23:48:30:WU00:FS01:0x11:Completed 76%
23:49:11:WU01:FS02:0x11:Completed 74%
23:49:55:WU00:FS01:0x11:Completed 77%
23:50:38:WU01:FS02:0x11:Completed 75%
23:51:21:WU00:FS01:0x11:Completed 78%
23:52:05:WU01:FS02:0x11:Completed 76%
23:52:47:WU00:FS01:0x11:Completed 79%
23:53:32:WU01:FS02:0x11:Completed 77%
23:54:13:WU00:FS01:0x11:Completed 80%
23:54:59:WU01:FS02:0x11:Completed 78%
23:55:38:WU00:FS01:0x11:Completed 81%
23:56:26:WU01:FS02:0x11:Completed 79%
23:56:53:WU02:FS00:0xa3:Completed 180000 out of 1000000 steps  (18%)
23:57:04:WU00:FS01:0x11:Completed 82%
23:57:53:WU01:FS02:0x11:Completed 80%
23:58:30:WU00:FS01:0x11:Completed 83%
23:59:20:WU01:FS02:0x11:Completed 81%
23:59:55:WU00:FS01:0x11:Completed 84%
23:40:48:WU01:FS02:0x11:Completed 82%
23:41:22:WU00:FS01:0x11:Completed 85%
23:42:15:WU01:FS02:0x11:Completed 83%
23:42:48:WU00:FS01:0x11:Completed 86%
23:43:41:WU01:FS02:0x11:Completed 84%
23:44:14:WU00:FS01:0x11:Completed 87%
23:45:08:WU01:FS02:0x11:Completed 85%
23:45:39:WU00:FS01:0x11:Completed 88%
23:46:35:WU01:FS02:0x11:Completed 86%
23:47:05:WU00:FS01:0x11:Completed 89%
23:48:01:WU01:FS02:0x11:Completed 87%
23:48:31:WU00:FS01:0x11:Completed 90%
23:49:28:WU01:FS02:0x11:Completed 88%
23:49:36:WU02:FS00:0xa3:Completed 190000 out of 1000000 steps  (19%)
23:49:56:WU00:FS01:0x11:Completed 91%
23:50:54:WU01:FS02:0x11:Completed 89%
23:51:22:WU00:FS01:0x11:Completed 92%
23:52:21:WU01:FS02:0x11:Completed 90%
23:52:48:WU00:FS01:0x11:Completed 93%
23:53:48:WU01:FS02:0x11:Completed 91%
23:54:13:WU00:FS01:0x11:Completed 94%
23:55:14:WU01:FS02:0x11:Completed 92%
23:55:39:WU00:FS01:0x11:Completed 95%
23:56:41:WU01:FS02:0x11:Completed 93%
23:57:05:WU00:FS01:0x11:Completed 96%
23:58:07:WU01:FS02:0x11:Completed 94%
23:58:30:WU00:FS01:0x11:Completed 97%
23:59:34:WU01:FS02:0x11:Completed 95%
23:59:56:WU00:FS01:0x11:Completed 98%

This is the system my 4y old plays minecraft on ect, so he likely changed the timezone information I changed back around the time of the 'anomaly'.

I 'could' work around this by assuming any new frames can not be in the past, and override the hour mark parsed from the log. But I'm afraid that might open up a whole new can of worms

And, tbh, I would rather have the client keeps it's logging function in adherence with the UTC offset which is shown at the start ( eg ignore any changes made when the client is running, only pick them up at start time OR display in the log that the UTC offset has been adjusted to x ). If someone could make a ticket for this I would really enjoy that

<removed>

Edit: I removed some things I probably said because I felt so sicks, so don't worry

[Grandpa_01]

MtM do not be a fool, if you do make it to the couch wrap a blanket around yourself and go to sleep. You do realize that is a self imposed deadline, and can be adjusted as needed. I do believe that being sick is a justifiable reason to move the event.

[MtM]

Ahh I know, and I made it back. I really feel like a dog, but the problem is I have one as well and it needs to go out. And it's cold outside

I did make the parser code ignore malformed or wrong lines without throwing errors. The wurst it can do now is not parse the unit, at least it would work for the rest of the history.

Sorry for the negativity in the previous post, I won't give up. I'll compile a new build tomorrow with the parser update and hopefully better project graphs. I been thinking, I've already added the hw description to each project's tpf and ppd graph, and preceded them with one showing the average values for that project, but I should really make it so it show average, then average for each hardware type, and then individual prcg data. That would make it easier to compare performance for the same projects across different hardware.

Edit: I don't sleep that much btw, and with a cold it's even harder, but I hope to catch some more after walking the dog

[whocrazy]

Perhaps you could submit your flu samples off to stanford and we could fold all the proteins to try and find a cure for your cold. heheh

[MtM]

Hey yeah that might be nice. It's already a great deal better then it was, so I don't think it's actually needed. Anyway, I promised an update and you'll get one but I don't think I'll include any update's to graphs as my head still hurts and I need to rewrite parts which requires more clarity then I'm able to gather right now. So just an update for the log parser and some UI tweaks.