| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: GO Fight Against Malaria
Thread: Results on AMD-FX Inconclusive? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 62
|
|
| Author |
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
I don't expect you're going to hear anything until they announce a beta to address the issue. That's the typical manner of operation. They also have the option of forcing results to go to similar machines, but again, you won't hear about it until it's done or nearly so.
---------------------------------------- Distributed computing volunteer since September 27, 2000 |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
We are looking at this pretty closely. There definitely is a problem. The AMD-FX cpus are validating against Intel cpus and not AMD cpus. This suggest a more generalized issue with the assignment of homogeneous redundancy classes. We are trying to determine what the proper change should be to how the hr class is assigned. This will actually be a server change and not a application binary change.
You can read a bit about hr here: http://boinc.berkeley.edu/trac/wiki/HomogeneousRedundancy |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
We just deployed a change to the server code. For DSFL and GFAM we have changed the hr classes so that AMD processors will only be checked against AMD processors and same for Intel. The exception is that based on the information above, AMD FX is validating against Intel so AMD FX will be compared against intel. We will monitor (and ask people to report their observations) and see if this resolves the issue.
Please note that this only applies to work distributed from now going forward. Work already assigned was assigned under the previous rules. |
||
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
I just downloaded a few GFAM and pushed them to the front of the queue. Should have an answer for you before Friday ends.
----------------------------------------Edit: Suppose I should mention this is on my FX8150 Distributed computing volunteer since September 27, 2000 [Edit 1 times, last edit by KWSN - A Shrubbery at Jan 6, 2012 7:01:18 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
 Having fun waiting for more info.  Boy, am I weird!At times like this I regret letting my subscription to IEEE Micro lapse. I actually enjoy these finicky little details.  Lawrence |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Am interested about the Bulldozer myself. The right compiler should bring out the strengths of the chip. Yeah, I know it takes work and to be weighted against other works in the WCG ToDo's list. (The WCG T-shirt seems more exciting at the more casual level of things...)
P.S: CA lawrencehardin, as they say, the devil is in the details -- so that smiley with two horns that you used is just about 'right on the  ; |
||
|
|
kateiacy
Veteran Cruncher USA Joined: Jan 23, 2010 Post Count: 1027 Status: Offline Project Badges: 
|
Having fun waiting for more info. Boy, am I weird!At times like this I regret letting my subscription to IEEE Micro lapse. I actually enjoy these finicky little details. Lawrence So glad to know I'm not the only one like that!   |
||
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
Well, for all of you waiting with bated breath.....
----------------------------------------The first results are in and the validator has fallen down again. As for general impressions, this thing is blazingly fast on CMD2 and C4CW. HCC runs slower than most of my other systems so there is something in the type of computation in these results that plays to the shared hardware weaknesses. I'll have to play around with some other sciences to see which ones are the most efficient. GFAM finished in under six hours. I haven't been running those lately because I'm working on the ending sciences. Let the speculation begin... Distributed computing volunteer since September 27, 2000 |
||
|
|
kateiacy
Veteran Cruncher USA Joined: Jan 23, 2010 Post Count: 1027 Status: Offline Project Badges:
|
As for general impressions, this thing is blazingly fast on CMD2 and C4CW. HCC runs slower than most of my other systems so there is something in the type of computation in these results that plays to the shared hardware weaknesses. I'll have to play around with some other sciences to see which ones are the most efficient. GFAM finished in under six hours. I haven't been running those lately because I'm working on the ending sciences. Are you running WUs on all 8 cores? Are you running under Windows or Linux? C4CW work units are consistent enough in size that I can see how you can determine that the FX 8150 is blazingly fast on them (how long are the runtimes?). Do I recall correctly that C4CW is the only 64-bit WCG application? Keep the information coming! She who waits with bated breath |
||
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
Ubuntu 11.10. I run a mix of HCC, HCMD2, and C4CW. I've got 16MB of ram, no overclock. (edit: yes 8 cores)
----------------------------------------The C4CW appear to be averaging about 1.75 hours, the HCMD2 haven't exceeded 4 hours yet, even on parents. This may be due to the complexity of the current parents, but I'll keep an eye on it over time. Back to this thread, the first GFAM has run through the validator and... Project Name: GO Fight Against Malaria Created: 01/04/2012 17:14:49 Name: GFAM_x1j3iWt_NDP_2WATs_0002542_0731 Minimum Quorum: 2 Replication: 3 Result Name App Version Number Status Sent Time Time Due / Return Time CPU Time (hours) Claimed/ Granted BOINC Credit GFAM_ x1j3iWt_ NDP_ 2WATs_ 0002542_ 0731_ 2-- - In Progress 1/6/12 19:20:11 1/10/12 19:20:11 0.00 0.0 / 0.0 GFAM_ x1j3iWt_ NDP_ 2WATs_ 0002542_ 0731_ 1-- 611 Inconclusive 1/6/12 06:41:53 1/6/12 13:43:36 5.26 148.2 / 0.0 <-- me GFAM_ x1j3iWt_ NDP_ 2WATs_ 0002542_ 0731_ 0-- 611 Inconclusive 1/6/12 06:41:50 1/6/12 18:27:43 5.08 66.0 / 0.0 Distributed computing volunteer since September 27, 2000 [Edit 1 times, last edit by KWSN - A Shrubbery at Jan 6, 2012 7:31:36 PM] |
||
|
|
|