Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 23
|
![]() |
Author |
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
One of my teammates is having an issue with his new 8 Core 2.8 Xeon Mac Pro. It's running a clean copy of Leopard 10.5.4. HCC runs at 85% on all cores. The HPF2 project on the same machine runs at 100%
Here's some more information - http://www.dslreports.com/forum/r21119217-Re-...runching-at-full-capacity Is this a known issue with HCC? |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Hi,
----------------------------------------It's a first report for sure and all 8 cores and HCC only doing is more than obscure. Given it's rather constant on all processes, it is not reflective of a BOINC throttle which is applied to all sciences. And operates in a run/pause type arrangement as in either 0 or 100%. See nothing in that thread that gives me an idea. 10.5.4 is that the very latest OS-X, straight from the shop?
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi, it's me that's having the problem... It's been driving me nuts for over 2 weeks now.
Today I did an absolute clean install of 10.5.4 on a spare drive. I tried both the 5.10.45 client from WCG site as well as the latest 6.2.18 from berkeley. Both did the exact same thing. 10.5.4 is not the latest, 10.5.5 is. However it behaves the same just after an update to 10.5.5. It's not that much of a biggy now, since I can crunch HPF2 on this machine as our team does both HCC & HPF2. However if I can be of any assistance in helping to figure this out, I'm all for it. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello HiVolt,
My first thought is that there might be some memory limitation - second thoughts indicate that such a problem should not manifest the way you say it does. But since nothing else pops in my mind (empty air with a very distant bird wheeling against fluffy clouds) I might as well ask how much memory your system has. If you change the device profile to only run 6 threads instead of 8, does that change the CPU utilization per core? Lawrence |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The machine has 6GB of ram, so I don't think that should be the issue. Everything else is running super smooth on it.
----------------------------------------I have changed it to run 4 cores, and then each core was running at approximately 93%, still not maxing out. However, when I reset the project and it flushes all the work units, the first WU it receives runs at 100%, until more download and start running, then it gets throttled down. If you look at the screenshot in the linked thread, there's that strange issue of the System % going up as soon as the User % load drops from 99% to 85%. When running HPF2, the System % does not go up and the user % stays maxed out. User % - Green System % - Red HCC http://www.dslreports.com/r0/download/1350522...f9a51d722/Picture%204.png HPF2 http://www.dslreports.com/r0/download/1352704...3e9e4f425/Picture%201.png [Edit 2 times, last edit by Former Member at Sep 23, 2008 10:31:23 PM] |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
There used to be an issue with a science to run ultra-low (HPF2 and HCMD on the UD agent), "lower than low" so to speak. Maybe HCC has been compiled with extra nice and gives way to regular nice?
----------------------------------------As for system use, HCC is after RICE the smallest by far so resource with HPF2 being a WCG heavy I'd not consider an issue. cheers, and good night.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Sep 23, 2008 10:32:34 PM] |
||
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7665 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Not knowing very much about the MAC architecture, but thinking about how the system manifests its symptoms, I would guess there is a bottleneck in some part of the information flow. This would be a question for the HCC programmers in conjunction with more knowledge of the I/O architecture of the Xeon running Leopard. The question would be : Does HCC push more information around than HPF2 does ? And how does it do it ? It could be register level starvation or inefficient use of cache. Hopefully brains better than mine can diagnose this.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Not knowing very much about the MAC architecture, but thinking about how the system manifests its symptoms, I would guess there is a bottleneck in some part of the information flow. This would be a question for the HCC programmers in conjunction with more knowledge of the I/O architecture of the Xeon running Leopard. The question would be : Does HCC push more information around than HPF2 does ? And how does it do it ? It could be register level starvation or inefficient use of cache. Hopefully brains better than mine can diagnose this. Cheers The thing is, I have been running WCG on another Mac Pro (4-core) for over a year with no issues with HCC WU's. If it wasn't saturating the other machine, why is it screwing around with this one. The other Mac Pro also has 6GB ram. I also run WCG on other Intel Macs which are Dual Core, and never had any issues... It seems the most powerful somehow gets choked with HCC. I think it has to be some kind of bug in the HCC modules. As for the architecture, the current generation Mac Pro uses Intel's 5400 server chipset along with dual 45nm Quad-Core Xeons, using FB-DIMM memory. BTW, Since it's an Intel Mac, I also have Windows XP installed on another drive. I installed BOINC on it and HCC's crunched flawlessly on it. |
||
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
From the TD thread I understand that your teammate ajeb has the same machine as yours and has not experienced your problem (yet?). What I have not found explicitly said is if he has tried to run HCC exclusively on the eight cores as you do. If he could say it and confirm that he is still running at about 100 % on all cores that would help to know if the problem is between HCC and 8-core Macs in general or if it is between HCC and your 8-core Mac and its setup.
----------------------------------------Cheers. Jean. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
From the TD thread I understand that your teammate ajeb has the same machine as yours and has not experienced your problem (yet?). What I have not found explicitly said is if he has tried to run HCC exclusively on the eight cores as you do. If he could say it and confirm that he is still running at about 100 % on all cores that would help to know if the problem is between HCC and 8-core Macs in general or if it is between HCC and your 8-core Mac and its setup. Cheers. Jean. Ah, good idea. He never did specify. I'll ask him. |
||
|
|
![]() |