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

Quick Go ยป
No member browsing this thread
Thread Status: Active
Total posts in this thread: 55
Posts: 55   Pages: 6   [ Previous Page | 1 2 3 4 5 6 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 8427 times and has 54 replies Next Thread
NLY74893
Cruncher
Joined: Mar 5, 2008
Post Count: 5
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

7.14 i will try to download 7.15
OSX runs 12,6 hours and is granted 37 points/hour
Linux runs 17,1 hours and is granted 5 point/hour
Mac is intel i5, Ubuntu runs on i7
Problem is not the time consumption onlybut more the lousy points.
[Oct 10, 2015 8:45:34 PM]   Link   Report threatening or abusive post: please login first  Go to top 
hnapel
Advanced Cruncher
Netherlands
Joined: Nov 17, 2004
Post Count: 82
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
smile Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Seeing significant performance improvement on Linux 64 bit, typically processing a WU is going from about 17 down to 11 hours of runtime on the same PC.
----------------------------------------
[Edit 1 times, last edit by hnapel at Oct 10, 2015 10:10:06 PM]
[Oct 10, 2015 10:09:43 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Time reductions (32-bit Linux):
Pentium Dual Core: 38%
Core 2 Duo: 27%
Core Solo: 21%
----------------------------------------
[Edit 2 times, last edit by Former Member at Oct 11, 2015 4:32:16 AM]
[Oct 11, 2015 4:30:58 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Sgt.Joe
Ace Cruncher
USA
Joined: Jul 4, 2006
Post Count: 7697
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Finally got back the the lone FAH2 unit to compare to the four Beta unit which happened to occur on the same machine and which were all manually started within seconds of each other.

FAH2_ avx38741-ls_ 000063_ 0016_ 002_ 0-- mint Valid 10/8/15 19:04:40 10/11/15 04:24:29 41.24 / 41.26 440.4 / 440.4

BETA_ FAH2_ avx101119_ 000064_ 0056_ 001_ 0-- mint Valid 10/9/15 07:49:21 10/10/15 17:11:13 29.45 / 29.46 447.4 / 447.4
BETA_ FAH2_ avx101119_ 000002_ 0027_ 001_ 0-- mint Valid 10/9/15 05:05:20 10/10/15 17:11:13 29.85 / 29.86 453.5 / 453.5
BETA_ FAH2_ avx101119_ 000086_ 0003_ 001_ 0-- mint Valid 10/9/15 03:40:01 10/10/15 19:47:59 32.55 / 32.56 494.4 / 494.4
BETA_ FAH2_ avx101119_ 000020_ 0021_ 001_ 1-- mint Valid 10/9/15 02:28:53 10/10/15 19:47:59 32.21 / 32.22 489.3 / 489.3
This would be a 24.8 % improvement in the run time of the BETA units (BETA average of 31.02 divided by the FAH2 time of 41.24). Even though this reflects only a single data point for FAH2, I think it is in line with what others have also found. Linux Mint 17 on a Xeon 5320 running off a flash drive.

Cheers
----------------------------------------
Sgt. Joe
*Minnesota Crunchers*
----------------------------------------
[Edit 1 times, last edit by Sgt.Joe at Oct 11, 2015 11:05:27 AM]
[Oct 11, 2015 11:04:38 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Rickjb
Veteran Cruncher
Australia
Joined: Sep 17, 2006
Post Count: 666
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Comparisons between Betas under Linux Virtual Machines vs Windows x64 Native
Linux VMs are Debian 7 or 8 x64, running under VMWare 10
=======================================================================
Based on CPU times, not elapsed times.

Machine: "Callisto" - 3770K @ 4.3GHz, mem 1866MHz, C9:
Running 7 cores Linux - OET + FAHB beta, host 1 core Windows 7 x64 - FAHB.
Linux Betas - average of 7 = 11.6h
Windows production - average of 2 = 10.3h
* Linux takes 13% longer than Windows
---------------------------------------------------------
Machine: "Io" - 3770K @ 4.0GHz, mem 1866MHz, C9:
Running 7 cores Linux - OET + FAHB beta, host 1 core Windows 7 x64 - FAHB.
Linux Betas - average of 7 = 12.8h
Windows production - average of 2 = 11.3h
* Linux takes 13% longer than Windows
---------------------------------------------------------
Machine: "Charon" - 2600K @ 4.0GHz, mem 1066MHz, C7
Running 7 cores Linux - OET + FAHB beta, host 1 core Windows XP x64 - FAHB.
Linux FAHB betas - average of 7 = 14.7h
Windows production - average of 2 = 11.8h
* Linux takes 25% longer than Windows
---------------------------------------------------------
Machine: "Triton" - i7-970 (LGA1366 Gulftown hex-core) @ 3.9GHz, mem 1288MHz, C7 (only 2 channels, 3 supported)
Running:
- Native (host) - 4 cores Win 7 x64 - FAHB
- Linux VM #1 - 4 cores OET + FAHB beta
- Linux VM #2 - 4 cores DENIS@home + FiND@home (VINA)
Linux FAHB betas - average of 4 = 16.6h
Windows production - average of 2 = 14.4h
* Linux takes 15% longer than Windows
---------------------------------------------------------
Note: On all machines, running OET WUs were suspended to run the FAHB betas.
On the 2600K and 3770Ks, 4 or 5 betas were started initially. Then, after the remaining OETs had finished and the suspended tasks had been resumed, the remaining betas were started.
The FAHB betas may have run more slowly when they occupied all cores.
----------------------------------------------------------
Conclusion: The Virtual Linux FAHB betas ran much faster than the previous FAHB production WUs,
but they still run more slowly than FAHB under Windows x64.
Some of the difference might be explained by the overhead of the VM, but on the other hand the VMs may actually run more efficiently than native Linux.

One beta was received & run on another machine under native Linux (no Windows FAHB WUs for comparison), and as is also the case for OET, the ratio of CPU/elapsed time was worse under native Linux than is achieved by the other machines running Linux in VMs.
(The Intel virtualisation feature (VT-x) in the BIOS is set to On in all CPUs.)

Hope this helps ...
----------------------------------------
[Edit 1 times, last edit by Rickjb at Oct 11, 2015 11:39:57 AM]
[Oct 11, 2015 11:30:55 AM]   Link   Report threatening or abusive post: please login first  Go to top 
NorthernRaider
Cruncher
Canada
Joined: Dec 10, 2008
Post Count: 12
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Got 4 To process, everything went very well, even with System Restarts the processing continued good from the last trickle point.
All units processed in similar times AMD FX8350 / 32Gigs on XFire5, with OET1 units loaded up too.
Linux Jessie 64 Bit - Boinc Client 7.15 - Throttled machine to 60% of CPU time.

BETA_ FAH2_ avx101119_ 000083_ 0034_ 001_ 0-- Jabbah Valid 10/9/15 02:34:39 10/11/15 09:33:30 12.17 / 12.45 321.5 / 321.5
BETA_ FAH2_ avx101119_ 000044_ 0047_ 001_ 0-- Jabbah Valid 10/9/15 02:16:29 10/11/15 08:32:56 12.14 / 12.28 317.3 / 317.3
BETA_ FAH2_ avx101119_ 000028_ 0018_ 001_ 0-- Jabbah Valid 10/9/15 02:12:00 10/10/15 23:04:49 12.33 / 12.51 323.1 / 323.1
BETA_ FAH2_ avx101119_ 000001_ 0014_ 001_ 0-- Jabbah Valid 10/9/15 01:14:26 10/10/15 18:46:35 12.12 / 12.33 318.4 / 318.4
----------------------------------------


----------------------------------------
[Edit 2 times, last edit by DutchRaider at Oct 11, 2015 1:34:51 PM]
[Oct 11, 2015 1:13:26 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

All computers are running Linux Mint 17.1

Returns so far.

Desktop Pentium D
BETA_ FAH2_ avx101119_ 000030_ 0021_ 001_ 0-- desktop Valid 10/9/15 05:34:45 10/10/15 21:57:01 33.23 / 36.14 571.9 / 571.9

Laptop Core Duo
BETA_ FAH2_ avx101119_ 000028_ 0042_ 001_ 1-- HP-Laptop Valid 10/9/15 03:16:53 10/10/15 14:26:07 26.94 / 28.58 450.9 / 450.9
BETA_ FAH2_ avx101119_ 000034_ 0028_ 001_ 0-- HP-Laptop Valid 10/9/15 02:14:11 10/10/15 04:29:24 24.44 / 25.87 408.1 / 408.1

Netbook Atom (Still Running)
BETA_ FAH2_ avx101119_ 000068_ 0011_ 001_ 0-- netbook In Progress 10/9/15 03:28:17 10/13/15 03:28:17 43.68 / 0.00 155.4 / 0.0
running for 57+ hrs with 16+ hrs to go.

BETA_ FAH2_ avx101119_ 000004_ 0000_ 001_ 0-- netbook In Progress 10/9/15 01:15:01 10/13/15 01:15:01 47.20 / 0.00 155.4 / 0.0
running for 59+ hrs with 15+ hrs to go.
----------------------------------------
[Edit 2 times, last edit by Former Member at Oct 11, 2015 3:18:16 PM]
[Oct 11, 2015 3:15:28 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Finished on track:

BETA_ FAH2_ avx101119_ 000044_ 0013_ 001_ 0-- starbase3.command Valid 10/9/15 05:45:54 10/11/15 11:49:44 41.33 / 42.97 263.2 / 263.2

Looks like the beta versions are saving around 22% crunching time.
[Oct 11, 2015 7:05:24 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Aperture_Science_Innovators
Advanced Cruncher
United States
Joined: Jul 6, 2009
Post Count: 139
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Is anyone else seeing low CPU usage?
On my dual Xeon E5-4667 v3 setup (32c/64t) I normally see a 15-minute load average (as reported by the 'uptime' command in Linux) of about 68 or 69. Last night while this system was running almost entirely FAAH-Beta WUs, the load average was much lower--about 40 or so. I opened up System Monitor to see indeed that a bunch of the vCPUs were bouncing around, often hitting as low as 10% activity. RAM usage wasn't a problem--I was only at about 4.5GB of 8GB in the system (and BOINC is told that it can use up to 80% of the RAM the computer has).
----------------------------------------

[Oct 11, 2015 7:08:25 PM]   Link   Report threatening or abusive post: please login first  Go to top 
OldChap
Veteran Cruncher
UK
Joined: Jun 5, 2009
Post Count: 978
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Beta Test - Fight Aids at Home - Phase 2 - v7.15 - Oct 8, 2015 [ Issues Thread ]

Not here. According to Boinctasks all 3 rigs running 16c/32t E5's are above 99%. results suggest each thread is losing 3mins over 16 hrs on 1st, about 10 mins in 23 hours on 2nd and less on 3rd

This is running Linux Mint 17.n and with <no_priority_change> set in order to let the kernel do the scheduling
----------------------------------------

----------------------------------------
[Edit 1 times, last edit by OldChap at Oct 11, 2015 7:56:30 PM]
[Oct 11, 2015 7:53:32 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 55   Pages: 6   [ Previous Page | 1 2 3 4 5 6 | Next Page ]
[ Jump to Last Post ]
Post new Thread