Progress: 34.8 Elapsed:9 days 10 hours Remaining:66Days 18 hours.

This on one of my tasks is typical of how wildly out estimates are with either 7.12 or 7.14.x running under Wine (Ubuntu.) Last run of Hadcm3 tasks under native Linux BOINC were much more realistic, slightly under the eventual total time from memory.

Because each new batch of tasks tends to be a different region and or resolution, BOINC never learns from the tasks I complete but recent batches have been so far out that it makes me think either the Gflops estimate is wrong or BOINC under WINE misreports cpu details, something I remember suggesting a while ago. Are native windows users experiencing this as well?

Running 7.12 on Win10 and WAH task estimates seem to be OK.

The estimates are OK for me too, running BOINC 7.12.1 on Win7 64-bit. On one machine, I have an app_config.xml:

<app_config> 
<app> 
  <name>wah2</name> 
  <fraction_done_exact/> 
</app> 
</app_config>

On the other machine, I don't. But it does not seem to matter. Maybe it did at first; I have been running them for a while.

Most of mine are a wee bit high: all boxes are Intel, Win10 (I'm ashamed to admit - [all were converted from Win7...]).

There was a small batch more than a month ago that was a bit under in its guesstimate. Later batches, already running, fit the usual mold, so I didn'tmake an issue of it.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

All replies so far from those using windows natively. Would be interesting to see if others using WINE are experiencing the same thing?

Hi Dave

My WINE Windows computers are way out too.
This one, which BOINC thinks is Windows XP, says:

pnw25 (batch 747) 59% after 9 days, 61 days to go
pnw25 (batch 747) 35% after 5 days, 98 days to go
sam25 (batch 742) 74% after 5 days, 17 days to go
sam25 (batch 742) 70% after 4 days, 19 days to go

I just let them get on with it, and don't normally look at this info, but if I do want to know, I use the percentage done, and the elapsed time to work it out roughly.

Mostly I look at which zips they're up to, so I know when to turn off the network so that I can log the final file sizes.

Thanks Les,

Wanted to check it wasn't just me, both my boxes have close to the same installation, so I thought it possibly something in my setup.

I suspect it is something to do with WINE but not as simple as a misreporting of CPU as I checked this computer out and both it has the same CPU speed type,stepping etc listed for both Windows and Linux.

I will do a little more searching to try and identify the problem but suspect that without developer knowledge of/skills with WINE and or BOINC I may not get very far.

Thanks Les,

Wanted to check it wasn't just me, both my boxes have close to the same installation, so I thought it possibly something in my setup.

I suspect it is something to do with WINE but not as simple as a misreporting of CPU as I checked this computer out and both it has the same CPU speed type,stepping etc listed for both Windows and Linux.

I will do a little more searching to try and identify the problem but suspect that without developer knowledge of/skills with WINE and or BOINC I may not get very far.

Sorry I didn't see this earlier. It's all about the floating point boinc benchmarks. For whatever reason, the benchmarks for boinc in WINE are too low, by quite a bit. This is what boinc uses to estimate task time. FWIW, I exit boinc, edit the p_fpops value in client_state.xml for a more realistic floating point benchmark value. For example, my i7 4790K had a WINE fpops benchmark of about 1200. I edited it to 5200 and much better task estimates were the result.

Thanks George,

Looking more closely at the computers' details pages, this one is

Measured floating point speed 1424.39 million ops/sec under WINE.
Measured floating point speed 2707.12 million ops/sec with native LINUX client.

Now you have pointed me to this, I will post on the BOINC boards to see if anyone has an idea about this.

Edit: will adjust value to slightly lower than the figures would suggest, as initially I always get a slightly optimistic estimate with LINUX.