Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 822
Posts: 822   Pages: 83   [ Previous Page | 2 3 4 5 6 7 8 9 10 11 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 1952240 times and has 821 replies Next Thread
GunZi
Senior Cruncher
Joined: Apr 1, 2020
Post Count: 160
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

I just got 15 GPU work units, exciting! (first finished in 00:01:31 minutes)
----------------------------------------
[Edit 2 times, last edit by GunZi at Apr 7, 2021 8:04:44 AM]
[Apr 7, 2021 8:03:32 AM]   Link   Report threatening or abusive post: please login first  Go to top 
BladeD
Ace Cruncher
USA
Joined: Nov 17, 2004
Post Count: 28976
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

I just got 15 GPU work units, exciting! (first finished in 00:01:31 minutes)

...running on what?
----------------------------------------
[Apr 7, 2021 8:08:57 AM]   Link   Report threatening or abusive post: please login first  Go to top 
goben_2003
Advanced Cruncher
Joined: Jun 16, 2006
Post Count: 146
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

If patterns continue, we will see 81,600 GPU tasks a day, and 1,038,000 CPU tasks a day. Hopefully this improves - what's the purpose of waiting 9 years for a new GPU project & the programming effort so a task runs 500 times faster if the vast majority of batches are still reserved for CPUs?

Keep in mind that new projects start out slowly(on purpose). Also - 81,600 GPU work units is actually several times as much work as 1,038,000 CPU work units. CPU work units have 1-3 jobs(ok, some really small jobs have 4 or 5) in one work unit. The GPU units have a range of approximately 30-70.

Keep in mind also that CPUs outnumber GPUs here a gazzillion to1. wink

True. However, my rtx 2060m does in ~3min what would take ~330min on my 6c/12t i7 using all 12t. wink
It is part of the reason that even though there are many CPUs, there is no shortage for CPUs, but the GPUs are doing more science and have a shortage.

It will be interesting to see how much more science will happen on the GPUs vs the CPUs once the pipeline is proven and there is not a shortage of work units.

Edit: the above numbers are from running only 1 work unit at a time on the rtx 2060m. That only uses ~26% of the card, so I actually run 4 work units simultaneously when enough of them are available.
----------------------------------------

----------------------------------------
[Edit 1 times, last edit by goben_2003 at Apr 7, 2021 8:28:22 AM]
[Apr 7, 2021 8:25:48 AM]   Link   Report threatening or abusive post: please login first  Go to top 
bozz4science
Advanced Cruncher
Germany
Joined: May 3, 2020
Post Count: 104
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

All is running smoothly on my end, and the WU shortage was to be expected as well tbh. However, the largest sustained load in the last beta run, was 182 WUs on 2 GPUs, each running 3 concurrently. Hardly ever, either one of those cards ran short of the defined app_config setup of running 3 at a time, as the cache immediately filled back up within the regular 2 min interval. The 'flow' kept my cards busy for ~2hrs. So far I have gotten ~10 jobs in nearly a day.

Roughly assuming that each OPNG job in beta had 30 jobs packed each, that amounted to 5,460 jobs done by two mid-tier NVIDIA cards in 2 hrs. According to several boinc stats sites, that 2 hr-beta run had me jump into the top 5% RAC on WCG. And that by far outnumberd my CPU contribution up until that point. Now, I am nowhere near that, what I take as a good sign that work is picked up and returned rather swiftly. However, I don't like seeing that one task pops up every now and then just blocks off other tasks and projects (all defined to compute on 100% of the card vs. 33% on OPNG), and then just finish off within minutes. I'll probably set WCG to compute on just 50% of the GPU, restrict it to runnning at most 2 tasks and if it only works on 1 job due to a low cache, I'll set up a less demanding project to also use 50% of the card's compute power, so the 'unused' capacity is avoided.
----------------------------------------

AMD Ryzen 3700X @ 4.0 GHz / GTX1660S
Intel i5-4278U CPU @ 2.60GHz
----------------------------------------
[Edit 2 times, last edit by bozz4science at Apr 7, 2021 9:39:34 AM]
[Apr 7, 2021 9:38:19 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Ulrich Metzner
Cruncher
Joined: Apr 8, 2020
Post Count: 7
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

Until now i haven't got a single GPU WU, although i got some BETA WUs.

Simply bad luck or did you finally lock out the GT-Series of NVidias on purpose, although some are capable of OpenCL 1.2?
----------------------------------------
Aloha, Uli


----------------------------------------
[Edit 1 times, last edit by Ulrich Metzner at Apr 7, 2021 12:31:55 PM]
[Apr 7, 2021 12:07:43 PM]   Link   Report threatening or abusive post: please login first  Go to top 
kittyman
Advanced Cruncher
Joined: May 14, 2020
Post Count: 140
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

Kinda like the odds of snagging Astropulse work in the latter days of Seti, eh? LOL.

Meow.
----------------------------------------

[Apr 7, 2021 12:29:25 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Jim1348
Veteran Cruncher
USA
Joined: Jul 13, 2009
Post Count: 1066
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

I finally have enough for a good average over the last 10 work units.
On my GTX 1650 Super (Win10), they are running 2 minutes 47 seconds.
[Apr 7, 2021 1:15:16 PM]   Link   Report threatening or abusive post: please login first  Go to top 
uplinger
Former World Community Grid Tech
Joined: May 23, 2005
Post Count: 3952
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

Is there a way how to quickly show/filter OPNG tasks in the Results Status page? My results have 262 pages and looking thru them all to check just the GPU tasks is not very convenient.


I will ask if there is a way to optimize this filter, I'm not sure the ease of it though.

Thanks,
-Uplinger
[Apr 7, 2021 1:27:14 PM]   Link   Report threatening or abusive post: please login first  Go to top 
uplinger
Former World Community Grid Tech
Joined: May 23, 2005
Post Count: 3952
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

Until now i haven't got a single GPU WU, although i got some BETA WUs.

Simply bad luck or did you finally lock out the GT-Series of NVidias on purpose, although some are capable of OpenCL 1.2?


Simply bad luck unfortunately. The filter at the moment is on OpenCL 1.2 at the moment.

Thanks,
-Uplinger
[Apr 7, 2021 1:33:31 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Grumpy Swede
Master Cruncher
Svíþjóð
Joined: Apr 10, 2020
Post Count: 2575
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Work unit availability

Well, the last WU's I got here. was 2 iGPU tasks, more than 9 hours ago. The last WU for Nvidia was finished almost 18 hours ago. I'll let the dust settle before I decide if I continue to run my GPU cruncher, or just shut it down.
[Apr 7, 2021 1:33:42 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 822   Pages: 83   [ Previous Page | 2 3 4 5 6 7 8 9 10 11 | Next Page ]
[ Jump to Last Post ]
Post new Thread