GPUGRID statistics graphs

Message boards : News : GPUGRID statistics graphs
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile MJH
Project administrator
Project developer
Project scientist

Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 38531 - Posted: 16 Oct 2014, 1:17:30 UTC
Last modified: 16 Oct 2014, 1:20:25 UTC

Following up on the announcement to discontinue support for old hardware and the oldest application, here are two graphs showing the normalised throughput of GPUGRID broken down by compute capability and application version.

http://www.gpugrid.net/graphs/

The legends are messed up for some reason - the first graph is cc, {1.3, 2.0, 2.1, 3.0, 3.5, 5.0, 5.2 }, the second app version {42, 60, 65}

These will update daily.

Matt
ID: 38531 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jacob Klein

Send message
Joined: 11 Oct 08
Posts: 1127
Credit: 1,901,927,545
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38536 - Posted: 16 Oct 2014, 11:32:51 UTC - in response to Message 38531.  

For your description of the legends... does it go top-to-bottom in the order you list, or does it go bottom-to-top?
ID: 38536 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile MJH
Project administrator
Project developer
Project scientist

Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 38538 - Posted: 16 Oct 2014, 11:39:10 UTC - in response to Message 38536.  

bottom to top. 1.3 and 42 are the lowest on their respective graphs
ID: 38538 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jacob Klein

Send message
Joined: 11 Oct 08
Posts: 1127
Credit: 1,901,927,545
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38540 - Posted: 16 Oct 2014, 12:15:46 UTC - in response to Message 38538.  

Thank you for the prompt response! I like these data graphs :)
ID: 38540 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile MJH
Project administrator
Project developer
Project scientist

Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 38541 - Posted: 16 Oct 2014, 12:17:38 UTC - in response to Message 38540.  

That blue needs to be gone by the new year!
ID: 38541 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile skgiven
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38542 - Posted: 16 Oct 2014, 12:53:19 UTC - in response to Message 38541.  

The time scale is too short to construe meaning; compute capability distribution vs time should be over two or three months rather than 11 days.
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help
ID: 38542 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile MJH
Project administrator
Project developer
Project scientist

Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 38543 - Posted: 16 Oct 2014, 13:21:14 UTC - in response to Message 38542.  

skgiven- This data collection only started a fortnight ago.
ID: 38543 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TJ

Send message
Joined: 26 Jun 09
Posts: 815
Credit: 1,470,385,294
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38546 - Posted: 16 Oct 2014, 14:58:04 UTC

Matt, Gerard,

I like those graphs.
Is it possible that you place them on the top of the GPUGRID website?
Perhaps a new tab "Statistics", or under "Performance". That way it is easy to access and check regularly. Some of us do, you know :) ?
Greetings from TJ
ID: 38546 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Gerard

Send message
Joined: 26 Mar 14
Posts: 101
Credit: 0
RAC: 0
Level

Scientific publications
wat
Message 38744 - Posted: 29 Oct 2014, 14:45:03 UTC - in response to Message 38546.  

Done! Control+shift+r if you see the menu a bit distorted and you will clean cached stylesheet
ID: 38744 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile nate

Send message
Joined: 6 Jun 11
Posts: 124
Credit: 2,928,865
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 38747 - Posted: 29 Oct 2014, 15:23:11 UTC

Nice! Looks really cool.
ID: 38747 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TJ

Send message
Joined: 26 Jun 09
Posts: 815
Credit: 1,470,385,294
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38748 - Posted: 29 Oct 2014, 15:44:35 UTC - in response to Message 38744.  

Done! Control+shift+r if you see the menu a bit distorted and you will clean cached stylesheet

Thank you Gerard, now it is very easy to check the stats.
Greetings from TJ
ID: 38748 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile skgiven
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38753 - Posted: 29 Oct 2014, 19:09:40 UTC - in response to Message 38748.  
Last modified: 29 Oct 2014, 19:10:16 UTC

The graphs show 7 generations of compute capability but the key shows 3.
Wishful thinking!
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help
ID: 38753 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile MJH
Project administrator
Project developer
Project scientist

Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 38755 - Posted: 29 Oct 2014, 19:48:35 UTC - in response to Message 38753.  

The graphs show 7 generations of compute capability but the key shows 3.
Wishful thinking!


from the bottom 1.3, 2.0, 2.1, 3.0, 3.5, 5.0, 5.2

The legend is broken and I dont have the effort to fix it.

Matt
ID: 38755 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile skgiven
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38760 - Posted: 29 Oct 2014, 21:33:31 UTC - in response to Message 38755.  
Last modified: 29 Oct 2014, 21:36:04 UTC

Key

Compute Capability (CC), GTX series-Scientist, GPU codename, [Maxwell models]

CC5.2 GTX900-Maxwell (GM204) [GTX980, GTX970]...
CC5.0 GTX700-Maxwell (GM107) [GTX750, GTX750Ti]
CC3.5 GTX700-Kepler (GK110)
CC3.0 GTX600&700-Kepler (GK104, GK106, GK107, GK208)
CC2.1 GTX400&500-Fermi (GF104, GF106, GF108, GF114, GF116, GF118)
CC2.0 GTX400&500-Fermi (GF100, GF110)
CC1.3 GTX200 (GT200)

I've not included non-GTX series cards or laptop GPU's because I'm lazy too.
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help
ID: 38760 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TJ

Send message
Joined: 26 Jun 09
Posts: 815
Credit: 1,470,385,294
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38889 - Posted: 10 Nov 2014, 16:50:46 UTC

Now with so many different types of WU, it is difficult in the bar graph to see which bar belongs to which WU.
Perhaps it is possible to show the name when hovering over with the mouse?
When now hovering over, some data is shown already, so would not be to hard to make?
Greetings from TJ
ID: 38889 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : News : GPUGRID statistics graphs

©2025 Universitat Pompeu Fabra