Homogenous Redundancy?
Message boards : Number crunching : Homogenous Redundancy?
| Author | Message | |
|---|---|---|
|
My Athlon XP was paired with a couple of Athlon X2's and its result was deemed invalid. Zero credits for over 6 hours crunching.
|
||
ID:
1518 | Rating: 0
| rate:
/
|
||
Andre Kerstens
Forum moderator Project tester Volunteer tester 
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Thanks, this will help research this issue even better!
My Athlon XP was paired with a couple of Athlon X2's and its result was deemed invalid. Zero credits for over 6 hours crunching. ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1519 | Rating: 0
| rate:
/
|
||
|
Well, my Barton 3200+ has been paired with A64 X2 or even Opteron and my wus were valid. |
||
|
ID:
1526 | Rating: 0
| rate:
/
|
||
|
A question concerning HR classes, if you don't mind. As I understood from what was said here in the forum the following cpu/os classes for HR have been set up:
|
||
|
ID:
2045 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
A question concerning HR classes, if you don't mind. As I understood from what was said here in the forum the following cpu/os classes for HR have been set up: Almost correct. The Windows PII/III machines are still grouped with the Windows AMD machines. Based on the outcome of some of your results, we will have to change this though and have separate groups (as you show above).
That is correct. Intel PII/III and AMD on Linux produce the same results (as can be seen from your example too).
If a system does not fit into any of these classes, it will get classification 'unknown cpu' or 'unknown os' (or both) and never get work, until we find out that fact and create a class for it (or add it to an existing class). We haven't seen such a case yet though. Thanks Andre
____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2050 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester 
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
A question concerning HR classes, if you don't mind. As I understood from what was said here in the forum the following cpu/os classes for HR have been set up: Hi Alex, I'm guessing that you're talking about the CyrixInstead processor. Only Andre or Memo could tell you for sure, but I do know that they were keying off of the x86 family and model codes. I don't know if they are looking for the GenuineIntel or not but I think they also checked for the word Celeron. I have an old Cyrix machine with RedHat 6 or 7 on it, but I didn't think it would be able to run a work unit, so I never hooked it up. I look forward to seeing how your machine does on Docking and how long it takes to run a workunit. My Cyrix machine hasn't been plugged in for over 2 years, but it was working when it was unplugged. On the Linux AMD/P-II/P-III situation, I noticed that the three machines validated against each other but, IIRC, that's not always the case. I think it depended on the Linux runtime library. There's a possibility that it was the K6 I'm remembering, though. I'm not at my most coherent right now. I've had a member of my immediate family in the hospital the last couple of days, so my schedule has been crazy and I'm a bit exhausted at the moment. Fortunately things turned out OK. -- David |
||
|
ID:
2051 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
Well, it looks like Andre was too fast for me and got an answer in while I was editing my reply *grin*
|
||
|
ID:
2052 | Rating: 0
| rate:
/
|
||
|
A question concerning HR classes, if you don't mind. As I understood from what was said here in the forum the following cpu/os classes for HR have been set up: I don't know if we have any cyrix machines attached but it will be interesting to investigate on those; actually I didn't consider these but if anyone has one attached let us know. On the K6 machines is kind of hard, at the moment, to do testing as we don't have any K6 available at the lab to make tests, but as always we are trying to do something to do testing on that architecture. I hope that your family member gets better soon. |
||
|
ID:
2054 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
You're right about the K6 though; that seems to be a stranger in the family who probably needs its own HR group. We don't have many though (I think only 2) so these people will have to wait an eternity for validation...
Well, it looks like Andre was too fast for me and got an answer in while I was editing my reply *grin* ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2056 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
I don't know if we have any cyrix machines attached but it will be interesting to investigate on those; actually I didn't consider these but if anyone has one attached let us know. I think Alex has one attached. It's machine number 1315 . CPU type CyrixInstead x86 Family 6 Model 0 Stepping 0 250MHz Operating System Microsoft Windows 2000 Standard Server Edition, Service Pack 4, (05.00.2195.00) I hope that your family member gets better soon. She's out of the hospital, as of this morning, and recovering nicely. Thanks for asking. I could dig out my old machine and see if it still works. It would be Linux though. I think it's a Cyrix cpu, but it could be a K6. I think it's 333 MHz. -- David |
||
|
ID:
2057 | Rating: 0
| rate:
/
|
||
|
I'm still having problems with my Athlon XP. It completes work units but gets no credit - everything deemed invalid. Have aborted the remaining WU on that computer as it's obviously futile.
|
||
|
ID:
2074 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
I've asked Memo to look into this and see if there are any other similar machines. If necessary, he should come up with a new HR rule.
I'm still having problems with my Athlon XP. It completes work units but gets no credit - everything deemed invalid. Have aborted the remaining WU on that computer as it's obviously futile. ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2079 | Rating: 0
| rate:
/
|
||
|
Will look into this. According to my previous tests I conclude that this would be no difference from the other AMDs as I tested on an old Athlon (don't know exactly the model but it was like 1800) and I got the same results as a AMD 64. Will look at those tests again and if possible run some again.
|
||
|
ID:
2090 | Rating: 0
| rate:
/
|
||
|
I'm still having problems with my Athlon XP. It completes work units but gets no credit - everything deemed invalid. Have aborted the remaining WU on that computer as it's obviously futile. Looking that most of the time you were trying to validate against AMD 64X2 I compared a test that ran on a AMD 64X2 running windows XP against one that ran on a AMD Athlon XP (I think it is a 3000 but might be a 3200) and here are the results: :~/charmm-test-windows> diff summary-athlon-XP.txt summary-athlon-64X2.txt :~/charmm-test-windows> diff in this case would printed any difference but as you can see both are exactly the same. I will repeat the tests again just to be sure but it will take me a couple of days as I did one at a friends house and the other (Athlon XP) on a computer on my wife's work. Can you think of any other reason why this computer is reporting erroneous results... perhaps it is overclocked or overheating? |
||
|
ID:
2092 | Rating: 0
| rate:
/
|
||
|
This is weird. I have an Athlon XP 2400+ (Thoroughbred, cpu-ID 6.8.1) which often has been validated against K8-class cpus (Athlon 64, X2, Opteron and Turion) and all results of the quorum were valid. I also have an Athlon 64 (Clawhammer, cpu-ID F.4.A) which has been validated against K7-class cpus (Duron, Athlon XP,...) and results of the quorum were valid, too. If you would like to compare, here's the CPU-Z output of my Athlon XP. Does your XP 3000+ support SSE2? As far as I remember, some of the latest socket A cpus support SSE2 (mine does not). This would be the only relevant difference I could think of. However, the K8 do also support SSE2, so it would be strange, if this is the reason... Regards Alex My results during the HR tests |
||
|
ID:
2095 | Rating: 0
| rate:
/
|
||
|
Thank you for your answers, Andre, David and Memo. And sorry for my late reply.
The Windows PII/III machines are still grouped with the Windows AMD machines. No offense intended, but are you sure? If I check my PIII I see a lot of unsent results missing for completed quorums and all results that have been sent out, were sent to other PII/PIII. (With one expection being the "AuthenticAMD x86 Family 6 Model 4 Stepping 2 1333MHz" processor mentioned here , but this might be an error in HR classification.) Memo wrote: I don't know if we have any cyrix machines attached but it will be interesting to investigate on those; actually I didn't consider these but if anyone has one attached let us know. As David already discovered, I have one attached. Yes, I confess. 
Actually, after my K6-III has delivered its result, I exchanged the cpus on the mainboard and gave the old Cyrix a try, out of curiosity.
I have delivered two results with my K6-III up to now, both with 5.03 and one during the recent HR test. Maybe you can make some use of it: http://docking.utep.edu/results.php?hostid=1130 Concerning HR classification of those cpus I would say: If they fit into a larger group of processors, it would be great. But if they don't, I certainly wouldn't mind if the project would set an Intel PII/PIII or AMD Athlon as minimum hardware requirement . Reasons: 1. AMD K6 will continue to disappear from active computers, Cyrix is almost gone by now. 2. Both have a weak FPU, so there are a number of other BOINC projects where they will perform better. 3. Many of these old cpus are used with Windows 9x, which seems to be not supported by Docking@home anyway, right? David wrote: I look forward to seeing how your machine does on Docking and how long it takes to run a workunit. Well, it is now at ~ 50% after 48 hrs. of crunching time, so it should take another two days. And that's with version 5.04. It's working hard, but still feeling fine. Andrew wrote somewhere else in the forum that charmm is mainly doing floating points and this is exactly the weak spot of the Cyrix (and of the K6-x as well). And then in addition the low clock... David wrote: I could dig out my old machine and see if it still works. It would be Linux though. I think it's a Cyrix cpu, but it could be a K6. I think it's 333 MHz. If it's 333 MHz (real clock) then it should be a K6-2. Cyrix maxed out at 300 MHz, which they called PR433. If it is PR333 (pentium rating) than it could be a Cyrix/IBM like mine, usually running at 250 MHz. Frankly, I don't expect to get credit for the Cyrix workunit. I guess it will not be sent out to another computer in the next two decades. 
Regards Alex My results during the HR tests |
||
|
ID:
2096 | Rating: 0
| rate:
/
|
||
|
That's good to hear. My best wishes. Alex |
||
|
ID:
2097 | Rating: 0
| rate:
/
|
||
|
Can you think of any other reason why this computer is reporting erroneous results... perhaps it is overclocked or overheating? It was mildly overclocked (running at 2.26Ghz instead of 2.17) with RAM underclocked (DDR400 running at "DDR348" speed). It's not overheating and works well on other projects. It stil has problems at Docking at stock speed so I don't think the overclock is causing the problems. RAM has been memtested (even at 433MHz) and had no errors. It has had some hard-disk issues (which caused it to reboot regularly) but that was sorted out before I ran the Docking work units. It's been running ABC@Home for the last 2 days (at even higher speed) and worked flawlessly. So it's probably just my computer that has a problem with Docking. Don't waste any more time on it - I will run it on projects where it does work. ____________ 
Join the #1 Aussie Alliance on Docking@Home |
||
|
ID:
2098 | Rating: 0
| rate:
/
|
||
|
Here is another weird invalid result (with Linux application 5.02):
|
||
|
ID:
2100 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
Here is another weird invalid result (with Linux application 5.02): I had a look at machine #1 and it has had a mixture of valid and invalid results over the last few days. Since it's Linux, they were all running version 5.02. Aside from that machine possibly having problems, what I'm wondering is how the minimum quorum of 3 machines was met when one of them failed to match the other 2. -- David |
||
|
ID:
2104 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
David,
____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2105 | Rating: 0
| rate:
/
|
||
|
Here is another weird invalid result (with Linux application 5.02): #1 is mine.....Athlon64 3700+ OC'd 8%, Ubuntu 6.10 all updates, no tweaks. Now you have called my attention to this one, I see lots of zeros from this computer. I'll check it out and see what the zeros match up to for a common factor. |
||
|
ID:
2107 | Rating: 0
| rate:
/
|
||
|
Here is another weird invalid result (with Linux application 5.02): I have not found any common factor yet in the results data. Computers this system has been paired with and not validated have also been paired when successful. I've set clock to standard and once pendings are processed, I can see if it is better, worse or the same. |
||
|
ID:
2108 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
I've set clock to standard and once pendings are processed, I can see if it is better, worse or the same. I just wanted to propose the same thing: do some tests with standard clockspeed and see what happens. We're definitely interested in those results. Thanks! Andre ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2118 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
The Windows PII/III machines are still grouped with the Windows AMD machines. You're right, I was thinking of something else :-) PII/III and AMD have their own groups on windows.
Alex, please give me a note when you get some credit (or not) for results crunched by the cyrix machine; We have to find out if it fits in any existing HR groups or will need its own (in which case we'll need some more cyrixes :-) Thanks Andre ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2122 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
David wrote: I dragged it out and tried to get it working. The bios says it's a K6-III/333MHz. I tried to get it working, but it won't recognize the Ethernet card anymore and the CPU fan isn't working. It seemed flaky when I tried it so I just scavenged the hard drive and added it to the junk pile. I've now got 3 dead machines that I have to figure out how to get rid of, legally. The trash pickup service won't take them. -- David |
||
|
ID:
2123 | Rating: 0
| rate:
/
|
||
|
I've now got 3 dead machines that I have to figure out how to get rid of, legally. The trash pickup service won't take them. Spring last year a new law was introduced here, which made sure that you can get rid of any electronic or electric device at no cost by bringing it to the local recycling service. On the other hand it's now forbidden to dispose it to your own garbage can. Of course, the costs for recycling will sooner or later be paid when buying a new electronic device by some small additional price increase. ;-) After introduction of this law I brought four old PCs (386-586), two old monitors, one terminal, an old fridge and some smaller stuff (such as keyboards) to the recycling service... :-) And since my last CRT monitor died a few weeks ago, I will be visiting them soon, again. Maybe you can find a company, which will take the stuff for free and dispose/recycle it. Regards Alex |
||
|
ID:
2124 | Rating: 0
| rate:
/
|
||
|
Yep. The other two results of the quorum have been sent in the meantime, too. It's interesting that they have been distributed to two Intel cpus: GenuineIntel x86 Family 6 Model 5 Stepping 0 232MHz GenuineIntel x86 Family 6 Model 8 Stepping 3 698MHz The first should be an old Klamath Pentium II, the second looks like a Coppermine Pentium III or maybe Celeron. Could it be that the system currently disregards the first (vendor) part of processor name for the HR classification into this HR class? That would explain why "CyrixInstead x86 Family 6 Model 0 Stepping 0 250MHz" and "AuthenticAMD x86 Family 6 Model 4 Stepping 2 1333MHz" are grouped together with some "GenuineIntel x86 Family 6 Model x Stepping x xxxMHz". Only a guess, though. Anyway, I'm curious if we will get at least two valid results here, let's see... Regards Alex My results during the HR tests |
||
|
ID:
2130 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). This includes your Cyrix machine as well. We'll have to find out if the Cyrix result matches up with an Intel or an AMD result (or not with anything). This workunit will give us an idea if it fits in the PreP4 group.
____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2132 | Rating: 0
| rate:
/
|
||
|
Coppermine Pentium III is correct. Might take a while. It's sharing CPU with CPDN and SETI/AstroPulse Beta |
||
|
ID:
2135 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). I have a Celeron 1200MHz. Which, if my system info is correct is an x86 family 6 model 11 stepping 1 processor. But it is still placed in the same class with the P4 CPUs. It completes the WUs successfully but is granted no credit. I guess its the HR problem. I run Windows and Linux both on this computer. Linux Computer ID 1058 Windows CID 1371 I will let it continue to complete WUs if that helps in any way. If not let me know. |
||
|
ID:
2250 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). Is it happening on both Linux and Windows, or just on one of them? ____________ The views expressed are my own. Facts are subject to memory error :-) Have you read a good science fiction novel lately? |
||
|
ID:
2254 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). I checked...it is both. 1058 (Linux) hasn't hit the server since 01/04. |
||
|
ID:
2256 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). Were you running this dual boot? |
||
|
ID:
2257 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). For the last two weeks or so I've only had Windows installed. With Linux I only had 7 out of 35 successfully completed WUs granted credit. So far with Windows 6 completed WUs with 5 zero credit and one pending. |
||
|
ID:
2258 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). Interesting, your computer just changed from 1371 to 1374. Did you reinstall BOINC? |
||
|
ID:
2260 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). For the heck of it, install the recommended client instead of 5.8.0 and see if that works. |
||
|
ID:
2261 | Rating: 0
| rate:
/
|
||
|
All 'Family 6' type cpu's currently are put in the IntelPreP4 HR class (meaning PII's and PIII's). This appears to be a 'Tualatin' Celeron, so it is based on the latest version of the Pentium III. I agree, that it should be paired with Pentium II/Pentium III instead. I haven't seen another Tualatin until now, but it's more likely to produce valid results when validated against the Intel 'Family 6' processors: If you look on this workunit from early december where your Celeron (running Linux) was grouped together with a P4 and a PIII, you can see that it has been validated ok against the PIII (and the P4 received zero credits). I think the problem is the processor name string here, which is provided by BOINC. On the first glimpse, you can not tell from the string "GenuineIntel Intel(R) Celeron(TM) CPU 1200MHz" what architecture it is. It could be Pentium III, Pentium 4 or even Pentium M based. Only the frequency gives a hint, but when it comes to mobile computers, the frequency could be anything. IMHO it would be much easier for homogenous redundancy classification, if BOINC would provide the vendor string (like "GenuineIntel") together with the CPU ID (something like "6.11.1") instead of the processor name string ("Intel(R) Celeron(TM) CPU 1200MHz"), because the combination of vendor and CPU ID is completely conclusive. In addition, the BIOS version and operating system should not have any effect on the CPU ID. Regards Alex My results during the HR tests |
||
|
ID:
2262 | Rating: 0
| rate:
/
|
||
|
LOL. And I thought where are all those results I have seen just a few minutes ago... |
||
|
ID:
2263 | Rating: 0
| rate:
/
|
||
|
Yep, I was in the middle of looking at them and poof...... |
||
|
ID:
2264 | Rating: 0
| rate:
/
|
||
|
I reinstalled Windows and then Boinc. That accounts for the change in CIDs. |
||
|
ID:
2265 | Rating: 0
| rate:
/
|
||
|
Let's see what Andre and the Docking team are saying about this. But in case there is no easy solution on their side, it might be quite easy to solve the problem under Windows with a registry hack. It seems BOINC is reading the CPU name on each start from the Windows registry and writes it then to the client_state.xml. I have just changed the name stored there to complete nonsense, stopped BOINC and restarted it again. And now I have an "AMD la la la Processor 50 GHz" instead of my usual "AMD Athlon(tm) 64 Processor 3200+": Screenshot The cpu name is send to the server apparently on each connect. I had a SIMAP workunit already uploaded (before I changed the name in the registry), then I issued the update command for SIMAP and ... uhm ... here's the result: SIMAP computer summary Disclaimer: Changes to the registry are risky. You are completely on your own doing it. I take no responsibility at all. :-) And please do not mess with it, if you don't have a reason at all. ;-) I will change it back later, to avoid further confusion. I hope you are all serious enough not to misuse this information. :-) My point is: If you change your cpu name in the registry to something like "x86 Family 6 Model 11 Stepping 1 1200MHz" then your computer should be grouped together with the other PII / PIII. I don't know, however, if Windows is keeping the registry entry on a reboot. And it could effect other software, which is reading this registry datum. But my recommendation: Let's wait for Andre's opinion about that, maybe the registry hack can be avoided. Regards Alex |
||
|
ID:
2266 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Yeah, the celerons are a big problem for us (and boinc) as you point out in this post. They only thing we can filter on is the string in the p_model field of a host record and for Celeron (pre or post P4) this is always Celeron blabla. I agree that the frequency is not a good thing to filter on as well (although it might be our only possibility for now) since it is not 100% conclusive.
This appears to be a 'Tualatin' Celeron, so it is based on the latest version of the Pentium III. I agree, that it should be paired with Pentium II/Pentium III instead. I haven't seen another Tualatin until now, but it's more likely to produce valid results when validated against the Intel 'Family 6' processors: ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2270 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester 
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
So far I don't have an idea for the server side solution, because as you know the information around CPU is claimed with client_state.xml. One thing I can imagine is, making participants choose which Celeron their hosts are running through docking@home preference or something like that.
Yeah, the celerons are a big problem for us (and boinc) as you point out in this post. They only thing we can filter on is the string in the p_model field of a host record and for Celeron (pre or post P4) this is always Celeron blabla. I agree that the frequency is not a good thing to filter on as well (although it might be our only possibility for now) since it is not 100% conclusive. ____________ 
I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions. |
||
|
ID:
2274 | Rating: 0
| rate:
/
|
||
|
Hello Cold Shot, according to this result you are using BOINC 5.8.3. Would you mind posting the content of the <p_capabilities> tags in the file client_state.xml of your BOINC directory to this thread ? It might help finding a solution to the 'Celeron' problem... Thanks Alex |
||
|
ID:
2298 | Rating: 0
| rate:
/
|
||
|
... is back to 'normal' in the meantime.
Windows overwrites this registry entry on the next reboot. Which makes sense, otherwise the operating system would never notice a cpu upgrade. :-) Alex |
||
|
ID:
2299 | Rating: 0
| rate:
/
|
||
|
Hello Andre, the quorum of the Cyrix M2 workunit is now complete. It has been validated against two Pentium II/III-class processors and was deemed invalid in comparison to those: Workunit ID 20898 Well, if it is not too much of an effort, you could check it against the K6 results in the database, maybe it fits there. I also have a Pentium MMX crunching which hopefully finishes its workunit tomorrow. Maybe this one also fits into the K6 class or is compatible with the Cyrix M2. Currently it shares a quorum with two Pentium 4 D, I do not expect it will be validated ok against them: Workunit ID 23162 Regards Alex My results during the HR tests ____________ |
||
|
ID:
2387 | Rating: 0
| rate:
/
|
||
|
I look forward to seeing how your machine does on Docking and how long it takes to run a workunit. Hello David, it is not very good at crunching for Docking@home even on clock-per-clock basis. The Cyrix M2 @ 250 MHz needed ~ 4 days for the workunit, on the same mainboard the AMD K6-III @ 400 MHz needs only ~ 1,5 days (both with application version 5.04). I switched back to the K6-III for now. Regards Alex |
||
|
ID:
2388 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Hello Andre, I checked in it also doesn't compare to the K6. Sorry... I am actually completely redesigning the HR rules also based on the new boinc client info we get back. It's kind of complex because we have to be able to catch both 'old' client and 'new' client strings. But I can hopefully deploy these new rules on Docking by monday. Maybe we should be the first project that will ask for boinc client 5.8.8 as a minimum :-) Will make life much easier! But only when the linux client is released too of course.
No I don't expect that one to validate based on the current and previous experience. If you want to suspend that one for now, that would be fine by me. Pentium MMX is a Family 5 model and those do not exist in the current HR rules, but will be there in the new set. Cheers Andre ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2390 | Rating: 0
| rate:
/
|
||
|
Ok. I'd say if it does not fit into a larger group, you shouldn't bother about it. It's very rare.
No problem. I try to let it finish. It looks as if it would take ~ 3 days per workunit, which is not that bad at only 166 MHz core clock (most RAM is clocked higher today). Even without credit for me you will get a result which you can compare if necessary. Since far more Pentium MMX have been sold than Cyrixes for example, I guess there's a chance that a few more will show up later, even if they are quite slow. At least there have been boards for dual Pentium MMX with up to 512 MB RAM, those could still do some crunching. Regards and have a nice sunday. Alex |
||
|
ID:
2392 | Rating: 0
| rate:
/
|
||
|
It's ready now, took a bit longer than first indicated. As expected it was deemed invalid against the two Pentium 4 D. Regards Alex |
||
|
ID:
2416 | Rating: 0
| rate:
/
|
||
|
Just out of interest where does a Celeron M 1.4Ghz fall in the HR classes? its on Linux and i was considering throwing a portion of it Docking's way if it falls under the Linux Intel banner as i think you have all the Linux AMD's you need correct?
|
||
|
ID:
2672 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Celeron on Linux would be HR class 258 (Linux on Intel).
Just out of interest where does a Celeron M 1.4Ghz fall in the HR classes? its on Linux and i was considering throwing a portion of it Docking's way if it falls under the Linux Intel banner as i think you have all the Linux AMD's you need correct? ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2673 | Rating: 0
| rate:
/
|
||
|
I just recieved your email regarding update to the new BOINC-version because of too less info in the older versions. I don't quite understand that, what's missing in
this:
AuthenticAMD It's been created with my 5.8.8. I don't like to run after every new issue, I'd change once something really new arrived, (and the step from 5.4.5 to 5.8.8 was something like it, but 5.8.8 -> 5.8.11 hardly seems worth the hassle). Is it really necessary? |
||
|
ID:
2687 | Rating: 0
| rate:
/
|
||
|
I have 5.8.7 running and dont need new update because this is stable for me.
|
||
|
ID:
2688 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
The reason is that only from 5.8.17 the linux client provides the Family/Model/Stepping information (which is missing in your info below). Your model string would look something like:
I just recieved your email regarding update to the new BOINC-version because of too less info in the older versions. I don't quite understand that, what's missing in this: ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2690 | Rating: 0
| rate:
/
|
||
|
For now please do not upgrade, because 5.8.17 seems to have problems that we did not notice initially. We'll ask people to upgrade again when the latest stable Linux client is released. What is the problem with 5.8.17? The change to glibc 2.4 is likely to be permanent in the official versions. ____________ |
||
|
ID:
2694 | Rating: 0
| rate:
/
|
||
|
For now please do not upgrade, because 5.8.17 seems to have problems that we did not notice initially. We'll ask people to upgrade again when the latest stable Linux client is released. Right. It's not so much a problem, as a change. The glibc change is the fix to the benchmark problem for linux clients. Linux users need to upgrade to 2.4. ____________ Dublin, CA Team SETI.USA |
||
|
ID:
2695 | Rating: 0
| rate:
/
|
||
|
I have changed to 5.8.15 at my remote site, (as I happen to be here today), as all machines there are running XP. I'll try at home, but last time, I had trouble with my NT4 system, it seems less stable with the 5.8.x versions.
|
||
|
ID:
2697 | Rating: 0
| rate:
/
|
||
|
The glibc change is the fix to the benchmark problem for linux clients. Linux users need to upgrade to 2.4. I'm curious about this. Do you have more information about it? TIA ____________ |
||
|
ID:
2699 | Rating: 0
| rate:
/
|
||
|
The glibc change is the fix to the benchmark problem for linux clients. Linux users need to upgrade to 2.4. Just what I've been reading on the alpha mailing list. Most of it is over my head, but I was able to understand the reason behind the change. ____________ Dublin, CA Team SETI.USA |
||
|
ID:
2701 | Rating: 0
| rate:
/
|
||
|
What is the problem with 5.8.17? Linux 5.8.17 is not reporting the full processor information. Look at the list of top hosts and there are a bunch just being reported as 'AuthenticAMD' or 'GenunineIntel' without all the other capabilities. |
||
|
ID:
2707 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
What is the problem with 5.8.17? Don't the other versions report it, too, or just this version? suguruhirahara ____________
I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions. |
||
|
ID:
2711 | Rating: 0
| rate:
/
|
||
|
Seems it is something of an
ongoing issue
.
|
||
|
ID:
2716 | Rating: 0
| rate:
/
|
||
|
Just checked my inbox and stumbled (11 days to late) on the emails that were send regarding the use of the latest boinc client.
|
||
|
ID:
2721 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Hi Rene,
Just checked my inbox and stumbled (11 days to late) on the emails that were send regarding the use of the latest boinc client. ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
2722 | Rating: 0
| rate:
/
|
||
|
I've tried 5.8.15 and 5.8.17 and ran into a random freeze of the x-server.
Hi Rene, ____________ 
|
||
|
ID:
2723 | Rating: 0
| rate:
/
|
||
|
Ok to be on the safe side (I hope...) just installed 5.8.16.
|
||
|
ID:
2724 | Rating: 0
| rate:
/
|
||
|
From another thread at another project, I have noticed that practically identical hardware can show different capabilities lists.
|
||
|
ID:
2725 | Rating: 0
| rate:
/
|
||
|
From another thread at another project, I have noticed that practically identical hardware can show different capabilities lists. You haven't mentioned the OS for each of those. For example, SSE will not be reported by a client running on NT4. ____________ |
||
|
ID:
2726 | Rating: 0
| rate:
/
|
||
|
I posted this in another thread but I'll post it here so it's understood. It's the *compiler* not the c library that helps the benchmarks. GCC 4.1.2 doesn't care which version of glibc that you link against. However, whoever built 5.8.17, built it with a distro that happens to have glibc 2.4. You *can* build against glibc 2.3 and still get the benchmark increase from GCC 4.1 as long as they use a distro that happens to have GCC 4.1 and glibc 2.3. I'm trying to impress this on the BOINC build team but there hasn't been much activity as of late on the list.
|
||
|
ID:
2804 | Rating: 0
| rate:
/
|
||
|
Ok to be on the safe side (I hope...) just installed 5.8.16. 5.8.16 has done the trick for my Athlon... it's running without any problems. System remains stable under Docking, Malaria and Seti-beta. ;-) ____________
|
||
|
ID:
2807 | Rating: 0
| rate:
/
|
||
|
:-) quite a difference in the badges shown for 'nix vs 'doze my palimino (266 Mhz bus) amd's show 26 vs 5 badges respectively.
|
||
|
ID:
2832 | Rating: 0
| rate:
/
|
||
|
No credit for
this wu
and flagged as invalid. Another HR issue?
|
||
|
ID:
3221 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
No credit for this wu and flagged as invalid. Another HR issue? It could also be a checkpointing issue since the result shows it restarted from a checkpoint twice. The other machines were Core 2 machines. Has anyone noticed a problem when matching a Pentium-M against a Core 2 ? Does 5.07 use any of the advanced features of the Core 2 that aren't on the Pentium M 1.86GHz ? -- David ____________ The views expressed are my own. Facts are subject to memory error :-) Have you read a good science fiction novel lately? |
||
|
ID:
3222 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Must be something else; I did some database queries on andrianxw's results and he has been matches with Core2's all the time and validated. For example
this wu
. He restarted from a checkpoint multiple times there as well. The new checkpointing is also pretty robust and we have not seen many problems with it up to now.
No credit for this wu and flagged as invalid. Another HR issue? ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
3223 | Rating: 0
| rate:
/
|
||
|
I have a invalid WU paired against 2 P3 class systems
here
.
|
||
|
ID:
3250 | Rating: 0
| rate:
/
|
||
|
I have a invalid WU paired against 2 P3 class systems here . You are running boinc 5.8.15, which doesn't include the Family/Model/Stepping info. Try installing version 5.8.17 , as that includes the Family/Model/Stepping info for HR. |
||
|
ID:
3251 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Could you send the cpuinfo of your machine to dockingadmin@utep.edu or post it here? That will give us the family/model/stepping info; we might have to adjust our HR rules. To get that info enter: cat /proc/cpuinfo in a terminal and copy the output.
I have a invalid WU paired against 2 P3 class systems here . ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
3253 | Rating: 0
| rate:
/
|
||
|
processor : 0
|
||
|
ID:
3257 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Thanks. This will help making our HR rules better!
processor : 0 ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
3258 | Rating: 0
| rate:
/
|
||
|
The other machines were Core 2 machines. Has anyone noticed a problem when matching a Pentium-M against a Core 2 ? Another wu with zero credit, Pentium M 1.86GHz crunched beside Core 2's ____________ Wave upon wave of demented avengers march cheerfully out of obscurity into the dream. 
|
||
|
ID:
3270 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Thanks. It seems that we have to kick or the pentium-M or de Core2 out of the hr group.
The other machines were Core 2 machines. Has anyone noticed a problem when matching a Pentium-M against a Core 2 ? ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
3279 | Rating: 0
| rate:
/
|
||
|
I have some C2D and they have been validating correctly. Now I wonder if there is a difference among the 3 C@D cores: Allendale, Conroe, Merom
|
||
|
ID:
3288 | Rating: 0
| rate:
/
|
||
|
Are we using this feature on Docking ??
|
||
|
ID:
4087 | Rating: 0
| rate:
/
|
||
|
Are we using this feature on Docking ?? Hi Conan, Thanks for your message. We are using homogeneous redundancy. We will check the results and see where the problem is. Thanks Arun |
||
|
ID:
4091 | Rating: 0
| rate:
/
|
||
Message boards : Number crunching : Homogenous Redundancy?
Database Error: The MySQL server is running with the --read-only option so it cannot execute this statement
array(3) {
[0]=>
array(7) {
["file"]=>
string(47) "/boinc/projects/docking/html_v2/inc/db_conn.inc"
["line"]=>
int(97)
["function"]=>
string(8) "do_query"
["class"]=>
string(6) "DbConn"
["object"]=>
object(DbConn)#91 (2) {
["db_conn"]=>
resource(198) of type (mysql link persistent)
["db_name"]=>
string(7) "docking"
}
["type"]=>
string(2) "->"
["args"]=>
array(1) {
[0]=>
&string(51) "update DBNAME.thread set views=views+1 where id=106"
}
}
[1]=>
array(7) {
["file"]=>
string(48) "/boinc/projects/docking/html_v2/inc/forum_db.inc"
["line"]=>
int(60)
["function"]=>
string(6) "update"
["class"]=>
string(6) "DbConn"
["object"]=>
object(DbConn)#91 (2) {
["db_conn"]=>
resource(198) of type (mysql link persistent)
["db_name"]=>
string(7) "docking"
}
["type"]=>
string(2) "->"
["args"]=>
array(3) {
[0]=>
object(BoincThread)#3 (16) {
["id"]=>
string(3) "106"
["forum"]=>
string(1) "2"
["owner"]=>
string(3) "271"
["status"]=>
string(1) "0"
["title"]=>
string(22) "Homogenous Redundancy?"
["timestamp"]=>
string(10) "1214191603"
["views"]=>
string(4) "3251"
["replies"]=>
string(2) "85"
["activity"]=>
string(19) "1.897375257083e-103"
["sufferers"]=>
string(1) "0"
["score"]=>
string(1) "0"
["votes"]=>
string(1) "0"
["create_time"]=>
string(10) "1164029073"
["hidden"]=>
string(1) "0"
["sticky"]=>
string(1) "0"
["locked"]=>
string(1) "0"
}
[1]=>
&string(6) "thread"
[2]=>
&string(13) "views=views+1"
}
}
[2]=>
array(7) {
["file"]=>
string(63) "/boinc/projects/docking/html_v2/user/community/forum/thread.php"
["line"]=>
int(184)
["function"]=>
string(6) "update"
["class"]=>
string(11) "BoincThread"
["object"]=>
object(BoincThread)#3 (16) {
["id"]=>
string(3) "106"
["forum"]=>
string(1) "2"
["owner"]=>
string(3) "271"
["status"]=>
string(1) "0"
["title"]=>
string(22) "Homogenous Redundancy?"
["timestamp"]=>
string(10) "1214191603"
["views"]=>
string(4) "3251"
["replies"]=>
string(2) "85"
["activity"]=>
string(19) "1.897375257083e-103"
["sufferers"]=>
string(1) "0"
["score"]=>
string(1) "0"
["votes"]=>
string(1) "0"
["create_time"]=>
string(10) "1164029073"
["hidden"]=>
string(1) "0"
["sticky"]=>
string(1) "0"
["locked"]=>
string(1) "0"
}
["type"]=>
string(2) "->"
["args"]=>
array(1) {
[0]=>
&string(13) "views=views+1"
}
}
}
query: update docking.thread set views=views+1 where id=106