Welcome to MilkyWay@home

Posts by Dunx

21) Message boards : Number crunching : Benchmark results - times wanted for any hardware, CPU or GPU, old or new! (Message 61130)
Posted 15 Feb 2014 by Dunx
Post:
No probs on it being o/ced :), just for the average 5 they need to be run at the same clock speed, otherwise the result is meaningless.

Only 75% GPU util.? Did MW not have a dedicated CPU core for the 5 averaged WUs?



Yup ! Always feed my babies properly.... unless the GPU really isn't using over half a "core"....

HD 7950 @ 950 GHz - around 120 seconds....

dunx

P.S. If the stupid R9 didn't have sticky-outy heat-pipes it would fit in the damned PC ! ! !
22) Message boards : Number crunching : Benchmark results - times wanted for any hardware, CPU or GPU, old or new! (Message 61107)
Posted 14 Feb 2014 by Dunx
Post:
Sure the PSU is upto it?

Btw I asked you this a few days ago :-

Btw I just had a quick scan of your valid 213.76 cred WUs & some were down to the low 500s.
Was the time you gave with the GPU at nr. 100% load?


Mikey
So any spare C2Qs or 5850s? ;)

Also, you never did confirm if the results you gave for your grx cards were at stock MHz, are they?


Sorry work gets in the way of my fun....

PSU is fine, but the Motherboard isn't.... only a GTX 480 will work ( bulletproof build quality and it's hot anyway ! And very noisy ! )

The GPU is around 75% utilisation, CPU is down at stock clocks for stability ( used to be at 4GHz 24/7 ) variation will be due to SkynetPOGS or MW CPU WU's (IMHO).

dunx

P.S. It's more likely that the first few Wu's were run at stock clocks, but as I'm an OC addict.... ( Sorry ! )
23) Message boards : Number crunching : Benchmark results - times wanted for any hardware, CPU or GPU, old or new! (Message 61105)
Posted 14 Feb 2014 by Dunx
Post:
Thanks, I appreciate your input, just wish my tired, burnt-out old Asus P6T7 WS could hold it's two HD7xxx cards without crashing....

dunx
24) Message boards : Number crunching : 4x HD7990 in one PC? (Message 61086)
Posted 11 Feb 2014 by Dunx
Post:
I think they are just seeing the added latency from a PLX type interface, the native x16 slots are fine ( up to three in use), but add a fourth or use seven single slot GPU's and the interface hardware shows it's delay....

IMHO, not many BOINC projects require "gaming" levels of bandwidth.

Anyway hope to find out within a month or two !

dunx

P.S. In general they saw about a 10% shortfall, hardly PCI-E x1 performance....
25) Message boards : Number crunching : Benchmark results - times wanted for any hardware, CPU or GPU, old or new! (Message 61028)
Posted 9 Feb 2014 by Dunx
Post:
GTX 480 @ 750 MHz = Average of 520 seconds....

HTH

dunx

P.S. Old PC is it set up right ? Too many inconclusive for my liking ? ? ?

http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=561093
26) Message boards : Number crunching : 4x HD7990 in one PC? (Message 61027)
Posted 9 Feb 2014 by Dunx
Post:
I have a pair of Asus X58 boards here - both with burnt out ATX 12V pins.... P6T6 WS and P6T7 WS, neither have additional PCI_E connectors for the GPU's - but the X79 now does.....

dunx

P.S. I just bought a short extension lead and soldered the female half to the board, and the male half replaced the burnt out ATX connector on the PSU ATX lead : - )

P.P.S. GTX 480 and three GTX 460's, VS HD 5870 + HD 7950..... now added a R9 280X !
27) Message boards : Number crunching : Hi I'm back, but n-body tasks run for 6 seconds ? (Message 52168)
Posted 1 Jan 2012 by Dunx
Post:
As the day has passed I have got more WU's but is this a test of the new server ?

CPU useage of around six seconds is a bit silly...

dunx
28) Message boards : News : Moving to a new server (Message 52158)
Posted 1 Jan 2012 by Dunx
Post:
I got a few six second n-body jobs...

dunx
29) Message boards : Number crunching : Hi I'm back, but n-body tasks run for 6 seconds ? (Message 52152)
Posted 1 Jan 2012 by Dunx
Post:
Had a few n-body cpu wu's and they all lasted only six seconds...

dunx


Task 58688677

Name nbody-Plum_Embedded_4027393_0
Workunit 44893073
Created 1 Jan 2012 | 11:40:03 UTC
Sent 1 Jan 2012 | 11:40:26 UTC
Received 1 Jan 2012 | 11:42:03 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 360356
Report deadline 13 Jan 2012 | 11:40:26 UTC
Run time 6.15
CPU time 8.63
Validate state Valid
Credit 0.77
Application version MilkyWay@Home N-Body Simulation v0.84 (mt)
Stderr output

<core_client_version>7.0.2</core_client_version>
<![CDATA[
<stderr_txt>
<search_application> milkyway_nbody 0.84 Windows x86_64 double OpenMP, Crlibm </search_application>
Using OpenMP 7 max threads on a system with 8 processors
<search_likelihood>-1286.982628237925600</search_likelihood>
11:40:26 (3548): called boinc_finish

</stderr_txt>
]]>
30) Message boards : Number crunching : MilkyWay@Home N-Body Simulation arrors... (Message 48442)
Posted 2 May 2011 by Dunx
Post:
name de_nbody_orphan_test_2model_4_11216_1304343747
application MilkyWay@Home N-Body Simulation

All failed !

dunx
31) Message boards : Number crunching : What different tricks do you guys use to optimize your system performance? (Message 47476)
Posted 10 Apr 2011 by Dunx
Post:
I use "Process Lasso" when messing with my system...

You can assign CPU affinity and process priorities to suit.

I.E. I have four GPU WU's serviced by 4 HT cores.

And four MW CPU WU's serviced by 4 different HT cores on my PC.

HTH

dunx

P.S. Also allowing one core to be busy will enable turbo-boost/speedstep on an i7 cpu for example.


Previous 20

©2021 Astroinformatics Group