Suggestions I posted to the HFM topic on FAH forum

[Tuna Ertemalp]

I found out about this group after I posted my list on FAH. For convenience, here is a copy/paste: https://foldingforum.org/viewtopic.php?f=14&t=9903

"Being one of the newbies among the rest of the 400,000 of us, and running on 6 hosts with 14 slots: Love HFM! Thank you for the new release with the fixes to the few bugs. I especially appreciate the connection problem to the EOC stats in the status bar.

Just wanted to ask/suggest a few things that might have already been covered in the 45 pages of this forum topic that I couldn't possibly read, so my apologies if some of this amounts to spam.

1) Some acronyms used by FAH might be meaningful to oldtimers, like PPD=PointsPerDay, but not to the new lot of 400,000. At least not readily. It would be awesome if the column titles would bring up a little tooltip to show a blurb about what they are for, when the cursor rests on them, and not just for the acronym columns but all. I find "TPF" as confusing as "Completed"/"Failed", the latter two being almost always 0/0 (therefore resulting in a "Huh?") and sometimes 2/0 or 1/0 (therefore resulting in a "What?"), all of which is not useful without knowing what they refer to, and probably very obvious once the user knows that meaning.

2) SlotType/Core columns are "Unknown" if that slot is not currently churning. But if it did churn ever before, HFM already knows what it is. Actually, when I add the client to HFM, and do a TestConnection before adding it, that little window does show the slot type, but then decides to show unknown if the slot is not churning. Is that data not cached? I understand that the hardware config might change between the uses of a slot, e.g. the GPUs might get replaced, but the log file for that slot seems to be still available while it is not churning, therefore its last-known-Type/Core is still available, and maybe could be displayed in italics or grey or just plain regular font, relying on the fact that 0% progress with unknown ETA/Deadline would be the clues that show to the user that the slot is currently not churning but was previously known as the info displayed from the cache.

3) Just like #1 above, but for the status/stats data at the bottom right status bar. Tooltips on these would help. For example, I belong to Team=999, but the stats there show Team=5, even though in the prefs my StandfordTeamID=999 and my churning WUs are shown as belonging to "Tuna_Ertemalp (999)". Without knowing what "Team" in that status bar refers to, I have no idea what that 5 means.

4) I have the ToggleFollowLogfile on. Great feature. The default width of the left section with the slot's tech data is too narrow for some of my CPUs ("AMD Ryzen Threadripper 1950X 16-Core Processor") and GPUs ("GeForce RTX 2080 Ti Rev. A"), so I widened it. But, that width doesn't seem to get stored in the config (like other column widths in the grid do) even though the toggle/on setting itself is, so the next boot, I do get the Log window, but the left window goes back to its default width.

5) Boy, a multiselect of the grid rows would be awesome. The one time I am really missing it is to force the wait time of the slot from a few hours to now or a few minutes via the pause/fold trick. So, currently, I sort the grid by Progress, then rightclick on each 0% row one by one, to first select PauseSlot and then FoldSlot. It would be awesome if I could just select those 0% rows by sort & select & shift+select to select a range or just multiple ctrl+selects to select non-contiguous rows, and then use the context menu to pause/fold.

6) As I am typing this, I am seeing a CPU slot that shows 0% progress with a teal status (not green), Unknown ETA, but a valid Deadline, with a 286 credit, with proper slottype/core, project, user, download time, and C/F=1/0. Furthermore, looking at the same slot via FAHControl, that CPU is READY, not RUNNING. So, there is a bug where HFM thinks that a slot is running but not running. If there is anything I can collect & send the next time I observe this, let me know. Simply rebooting HFM "fixed" the problem.

7) In the left side of the FollowLogFile window, the dropdown at the top that shows one or more entries formed as "N - Pnnnn (Rnn, Cnn, Gnn)": Without a tooltip or background in FAH, a new user wouldn't know what the initial N stands for, as well as why there might be more than one value in that dropdown, as well as what R/C/G REALLY means beyond Run/Clone/Gen. Another tooltip to explain that dropdown might be useful. Also, there is no visual feedback to indicate that that dropdown might have more than one value in it for me to know to switch between them. So, maybe it would be better if it were a textbox when there is just one entry, but a dropdown when there is more than one entry to select. Or some count on the left of the dropdown indicating the number of entries in it.

8) In the left side of the FollowLogFile window, it would be awesome to have the "Driver" version as well as "Compute" level displayed, if it is a GPU slot. I see that data in the cached log file for each host/slot. This would be useful to see if all GPUs of certain kind are running the same latest driver etc.

9) Similarly, the OS version might be a good info to show, as well. In case of Windows, this should be the output of WINVER, like "Windows 10 Version 1903 (OS Build 18362.720)" or some form of it to show the version/build numbers 6.2.1903.18362.720, instead of the generic 6.2 I see in the cached log; this might be impossible if FAH is not collecting this info on the host down to the build number.

10) The lower right status area: instead of saying something like "9 slots", it could say some form of "9 of 14 (64%) slots" to give me a coarse idea of the busy/idle level of my slots.

11) While trying to figure stuff about #7 above, I noticed that one slot might be in SEND/RUNNING/DOWNLOAD type of modes, sometimes all three simultaneously. I am sure that is not a full nor a correct list; just my quick observation. But the Status column in the grid only shows READY/RUNNING. I am wondering if there would be value in having different states & colors for these, with a prioritization of those states, therefore a sort on that column could show what that slot is primarily busy with: if a slot is RUNNING, or not running but SENDing while also waiting to download, or not running/sending but actively DOWNLOADing, or not running nor sending nor downloading but READY to download, all in that order of RUNNING/SENDING/DOWNLOADING/READY. Just a thought... (And, what is the yellow "RunningNoFrameTimes"???)

Thank you for listening! And thank you for the great tool to get such nice overview...

Tuna"

[harlam357]

Hi Tuna!  Wow... I'm sooo sorry I haven't responded to your GREAT ideas and observations here.  A lot of information here... so let me address what I've absorbed.

1) I've had several others give me similar feedback regarding tooltips to explain things better.  I concur.

2) I think you'll find the "Unknown" issue is much better in the latest 0.9.17 release posted today.

3) Basically same answer as 1).  These are the same labels used by extremeoverclocking.com.  So if you look at your user stats page there, the labels line up 100%.  The reason they're so terse is just due to horizontal real estate in the status bar.  My idea is to move this data to somewhere else in the UI where things will be more clear.

4) Some folks are asking for the GPU or CPU hardware info to be included in the primary grid.  So I think this data will be leaving this view of the work unit queue... and that's what it is...the queue of work units for the slot.  This queue area in the UI was much more useful back in the FAH v6 days and it just hasn't been as useful for FAHClient v7.

Going a little deeper on history.  HFM was developed originally to work with a COMPLETELY different Folding@Home client and I adapted the best I could when version 7 came out... and supported BOTH clients in the same app.  However v6 is dead and I just recently removed a bunch of the old v6 client stuff from the app... which is allowing me to make some changes where I don't have to work around concerns for different FAH clients.

Regarding the width not "sticking".  This is another v6 hangover... there was really no need for it to be wider so I never implemented saving the position of the divider.  Low priority but I agree would made a ton more sense if it did maintain the position.

5) Gotcha, select multiple slots and act on several at once.  Let me just say this... the UI layer needs work and that's all easier said than done until I get in there and deal with some decisions from 10 years ago.  Again, back in the v6 days selecting multiple clients (slots wasn't a thing) didn't make any sense.  HFM was just a monitor, you couldn't effect the actual FAHClient (Fold, Pause, Finish).  Anyway, UI rework is on my todo list.  In fact, I want to get HFM to the point where I can define slots and configure the entire client... a complete alternative for FAHControl.  That's a ways down the road, don't get too excited.

6) I've seen this on a few occasions, not many, and I think I have it addressed in 0.9.17.  Please let me know if you continue to see this issue.

7) I like the idea of an indication when there are multiple items in the slot queue.  Great suggestion!

8) Driver and/or Compute version.  If I were to add a CPU / GPU column I think this version might be better there.  "Geforce RTX 2060 (432.0)" for example.  Can you expand on why Compute is a desirable value?  I can see driver version, but I'm not even sure what the Compute version represents.

9) I'm bound to the OS version reported by the FAHClient.  So perhaps better detection is something I can push on their developer.  I agree.  It would be better to have the actual version rather than 6.2.  I'm somewhat familiar with how/why that is returned in certain cases by Windows.  Don't know why microsoft decided to go down that road, but I'm sure they have their reasons.

10) That represents the number of slots online.  If I move the EOC stats data from the status bar then I'd have more real estate to do something like your suggest.  If you look at HFM's system tray icon it does show "working" and "non-working" (perhaps "idle" would be a better word) slots in its tooltip.

11) There are some hang overs from the v6 days with the Status in the grid.  But you'll notice that the FAHClient reported status is echoed in the Queue Viewer (left of what you call FollowLogFile, I call it the Log Viewer).

RunningNoFrameTimes - LOL... yeah, most folks just kinda "know" what this means.  It all has to do with how many frames (percentages of progress) that have been observed.  The default for the app is to calculate frame times based on an average of the last three "frames".  You can adjust this option by going to Edit > Preferences > Options > Calcaulate PPD based on:.  So when a new WU starts and three frames haven't been completed yet HFM can't calculate a three frame time average.  So it's "Running" with "No Frame Times".  In this situation HFM estimates TPF, PPD, and Credit based on the data available in the Benchmarks (Tools > Benchmarks Viewer).

Better tooltips for Status would be good... especially for RunningNoFrameTime.  Thinking something like "Running (Benchmark TPF)".

So, you're obviously passionate about the HFM, which I really, really dig... nothing makes a dev happier than to have engaged users.  So if you're serious about helping drive some of these changes, here's where you can help me... open issues on GitHub and detail these asks there.  ONE issue for ask please.

https://github.com/harlam357/hfm-net/issues

Once these things are logged there we can continue to conversation and refine the scope.  Also, please understand that I develop and support HFM on my own time and for my love of the Folding@Home project.  I have to prioritize issue based on the time I have available.

[Tuna Ertemalp]

Thanks for your detailed response! And, yes, #2 now is much better. And, I will create issues shortly, for more focused ongoing conversation.

Thank you for the new version!

Tuna

[Tuna Ertemalp]

And, I think I am done with all the issue opening. I also added a bunch of new suggestions/observations I noticed during my use of HFM since my first message in this thread.

Have fun with them! :) 

Tuna

[harlam357]

I see that... you've been busy. :-)  So, to be super clear... I'm looking for engagement when evaluating these requests.  If I get no engagement on an issue, I won't put time into it.  Also, I may not address every issue that comes in.  But I will certainly explain why before closing it.

For example, I see quite a few issues around the static generated HTML.  While this is not something that is going to go away I also don't want to invest a lot of time in it.  It's based on very old web standards and I'd rather put my efforts into a completely new and modern web view that's more interactive.

So, I'll take a look at these new issues as time permits and we can continue the conversation and refine scope.  As I said before, you've got some great ideas and observations, so I'm sure some of these will come for fruition.

Thanks again Tuna!  Talk to you soon.