OK, Who Has the Biggest OGR Stub?

[JHutch]

Checkpoints are ALWAYS on my machines. I got burned too many times in the beginning!

Right now I'd be happy if I could just get a dot to appear on the progress meter!

Maybe I should start a betting pool on just how big this stub will get ...

Later,
JHutch

 

[biohazard]

Just got an OGR-25 stub through (6-node start):

[Aug 04 11:28:13 UTC] Running again after pause... ('unrealtournament.exe' inactive)
[Aug 04 13:50:59 UTC] Completed OGR stub 25/1-2-10-11-22-9 (178,379,644,537 nodes)
0.02:59:21.73 - [3,140,947.80 nodes/sec]

Took three hours to complete the last 60%

Of course, when those 4 or 5-node stubs come through I'm going to get whumped

 

[BoberFett]

OGR 25 is going to be a HUGE project, isn't it. :Q

I'm not sure I can even begin to comprehend OGR 26 or beyond.

 

[amok]

Yes, OGR-25 is going to be huge. According to my calculations, the ogr-25 stub that I'm currently working is going to be 959GNodes! I believe that they are doing a bit of testing with stub sizes though, because my stub is labelled 1-5-7-40-4, and there are supposed to be 6 differences there instead of 5. That should cut them down to a decent size.

 

[Verdoux]

You know you are allowed to trash all your 4 or 5 marks OGR-25 stubs ?
They were only sent for testing but now they are re-distributed in a 6 marks format.

 

[Viztech]

Verdoux-

That's what I have done except for a couple of machines. I just want to see how big they are. Any stub that is not completed will eventually be reissued, so don't worry about it.
My proxy has all 25/6 nodes in it now, so after the clients finish their OGR 24 work, it's on to 25!

viz

 

[rancherlee]

Here is a huge one
[Aug 04 14:26:36 UTC] Summary: 1 OGR packet (279.27 Gnodes)
0.13:43:15.26 - [5.65 Mnodes/s]

 

[GSmith]

How about this one?

[Aug 05 05:11:40 UTC] Completed OGR stub 25/1-4-6-13-7-22 (262,518,703,588 0.10:56:11.97 - [6,667,654.26 nodes/sec]
[Aug 05 05:11:40 UTC] Summary: 1 OGR packet (262.51 Gnodes)


Took my Athlon 840 almost 11 hours.
[Aug 04 18:15:33 UTC] Loaded OGR stub 25/1-4-6-13-7-22

 

[RC]

Don't believe this is the biggest OGR stub, but considering it's my first experience with OGR I must say it certainly is intimidating.

[Aug 05 06:04:21 UTC] Loaded OGR stub 25/1-10-4-13-7-31

[Aug 05 06:39:41 UTC] Benchmark for OGR core #0 (GARSP 5.13)
0.00:00:16.07 [8,584,948.30 nodes/sec]

[Aug 05 07:37:46 UTC] Loaded OGR stub 25/1-10-4-13-7-31 (8.40% done)

8.4% complete after 93 minutes at a rate close to the benchmark rate of 8,584,948.30 nodes/sec. OGR is not for the person that desires quick results. I like the fact that OGR 25 has only recently started.

 

[rancherlee]

[Aug 05 12:57:37 UTC] Loaded OGR stub 25/1-5-14-2-15-24 (26.30% done)
This one has been going for 6 hours now at my 5.4 average Mnodes/s speed.

 

[Joe O]

I just completed my first OGR-25 stub.
Completed OGR stub 25/1-7-20-16-17-2 (119,936,685,028 nodes)

 

[crystal]

Well, I got this.

Completed OGR stub 25/1-5-15-10-8-4 (210,334,722,144 nodes)

then there is this

Completed OGR stub 25/1-4-46-3-15 (411,682,965,096 nodes)

for the test one.

 

[RC]

First day with OGR:

[Aug 05 14:23:21 UTC] Completed OGR stub 25/1-10-4-13-7-31 (264,831,597,766 nodes) @ [8,932,813.71 nodes/sec]

Took ([Aug 05 14:23:21 UTC] - [Aug 05 06:04:21 UTC]) or about 8 hr 19 min for 264.83Gnodes on my Athlon K7.

[Aug 05 18:05:53 UTC] Completed OGR stub 25/1-9-12-2-26-15 (259,216,364,914 nodes)
0.08:11:58.34 - [6,083,013.21 nodes/sec]

523 Gnodes today from my two PCs at home.

 

[Hawkeye_(BEL)]

This is my biggest so far :

[Aug 05 13:04:23 UTC] Completed OGR stub 25/1-2-16-23-15-6 (134,194,955,715 ...
0.15:15:36.22 - [2,442,740.62 nodes/sec]
[Aug 05 13:04:23 UTC] Loaded OGR stub 25/1-2-16-23-15-7
[Aug 05 13:04:23 UTC] Summary: 1 OGR packet (134.19 Gnodes)
0.15:15:36.22 - [2.44 Mnodes/s]

If I understand it correctly, WU times will go up as we move to higher stubs ?

 

[RC]

What is the estimated time completion for OGR 25?

I have yet to choose a team for my OGR work. What team do you recommend? Is everyone in this thread a member of Team Anandtech?

Hawkeye_(BEL): Yes, I would expect WU times to increase when we begin to work on higher stubs.

 

[JonB]

RC, to answer your questions, OGR 25 will probably take a full month or two. The WU times will probably average only slightly higher though, because they are (now) using stub lengths of 6. That means the client is calculating up to 19 more values (out the the 25th spot). OGR 24 used a stub length of 5, so it also calculates up to 19 more values. The kicker is, there are a LOT MORE work units to issue for OGR 25, even though they should average the same time per work unit.

Remember, also, that the client only continues working on a stub until it proves it to LONGER than the shortest known OGR 24 or 25 value. As soon as it is longer, it closes the WU and opens the next.

So, RC, don't sit on the fence. Join the Team. Put yourself on our map. http://teamanandtech.dhs.org/map/ Become part of our international team of computer nuts.

 

[crystal]

hehehe...

this is pretty big for 6 length stub.

Completed OGR stub 25/1-5-15-10-8-9 (320,389,472,578 nodes)

 

[RC]

JonB: Thanks for the reply.

My second PC just completed its first WU, and my first PC just completed a second WU, now up to 702 Gnodes for today.

I have joined TA. My reason for joining is that TA seems to have a lot more activity compared to most teams.

 

[RC]

JonB: - "the client only continues working on a stub until it proves it to LONGER than the shortest known OGR 24 or 25 value"

This would also explain why the progress bar is not accurate. I started a WU at 1:00 am and at the 8% point I calculated it would require 17 hours, however the work was completed in 8 hours.

Also I have noticed some inconsistencies with OGR benchmarking. The highest rate I can obtain with the OGR Long benchmark is 8,584,948.30 nodes/sec, but yet I completed a stub at a rate of 8,932,813.71 nodes/sec. The benchmark value should never be lower than the actual rate. Especially when you consider that typically other activities usually occur while the stub is being processed.

I see a lot of references to Mika's server. As a TA OGR member do I need to be aware of Mika's server?

Please excuse the large number of questions - I'm still learning about OGR and Team Anandtech.

I am aware that TA has a FAQ document but FAQ docs usually only answer very general questions.

 

[TwoFace]

RC: Mikas is a pproxy that the team flushes to so we can compare stats... there's also a roundrobin to help relieve it with several pproxies that all flush to mikas... the address to mikas is: teamanandtech.dhs.org and the round robin is: proxy.teamanandtech.com

hope that helps

... 7

 

[RC]

TwoFace: From the configuration you have described it seems that if I want to share my OGR stats with other TA members I need to use proxy.teamanandtech.com rather than the default of distributed.net. Since the pproxies within proxy.teamanandtech.com all flush to Mika's, I see no reason to use teamanandtech.dhs.org.

Thanks for the help.

 

[TwoFace]

RC: The only reason to flush directly to Mikas would be so you could track cows you don't see every day... using the hostnames working for this participant feature in the weeklies

... 6

 

[Viztech]

Hi RC and welcome to TA!

Many of us use the hostnames feature to keep track of remote cows as Two Face has mentioned. The round robin was put into place to spread the load and to lessen the chances of our clients running out of work in case of failure of Mika's proxy. It has become a 'take your pick' kind of situation.

There may be a possibility that in the future the logs from the baby bovines in the round robin will be combined into one stats database, but that may take a little more doing to get accomplished. Meanwhile, you can use jator.2y.net to relieve Mika a bit, and still get the hostnames feature.

BTW, here are my biggest stubs to date.
24/4-1-12-7-11,93682026518,1,1,8010
24/1-5-7-4-14,99512318779,1,1,8010
OGR-25 6 Mark
25/1-2-11-4-27-9,240065012176,1,1,8010
25/1-2-11-9-25-7,240597647666,1,1,8010

I have 2 Athlon 600s doing those blessed OGR-25 4 mark stubs, and a P3 500 doing a OGR-25 5 mark.

viz

 

[RC]

viztech: Fine, I will use jator.2y.net port: 2064

Glad to report that I finished in fifth place today in the TA OGR 25 group. My first day was only 18 hours.

I'm not familiar enough yet with the dnet client to know what happens to completed WUs if the specified pproxy is down. Any risk of losing completed work if jator is down?

I use an Athlon 1000@1100, and an Athlon 700@750. Monday I will reconfigure a K62 400 from RC5 to OGR.

 

[Jator]

Welcome Rc. Feel free to use my proxy or one of the other rr proxies. If you haven't had a chance, check out my web page in my sig as I have links to some other proxies being hosted under the round robin.

Jay