S5R3

Bikeman (Heinz-...

Moderator

Joined: 28 Aug 06

Posts: 3522

Credit: 691450306

RAC: 251192

RE: NB for anyone else

25 Sep 2007 13:11:15 UTC

Message 73242 in response to message 73241

(moderation:

)

Quote:

NB for anyone else wanting to do any testing - have a look at the server status page first. While the "Oldest Unsent Result" is at 8 days or more (creation time of 12:40 on Monday 17 September), you'll probably be getting S5R2 work.

Which is also the reason for the astronomical predicted length of S5R3 on the status page. Most crunching is still done on S5R2, so progress on S5R3 is slow. This will change once the S5R2 pool is completely drained.

CU
Bikeman

archae86

Joined: 6 Dec 05

Posts: 3146

Credit: 7063614931

RAC: 1230786

RE: NB for anyone else

25 Sep 2007 14:54:41 UTC

Message 73243 in response to message 73241

(moderation:

)

Quote:

NB for anyone else wanting to do any testing - have a look at the server status page first. While the "Oldest Unsent Result" is at 8 days or more (creation time of 12:40 on Monday 17 September), you'll probably be getting S5R2 work.

I think it knows what it has sent you. If there is more S5R2 work it would have tried to send you had you done nothing, it just realizes some files are missing and resends them, rather than treating that as a reason to start you a though your were fresh.

Richard Haselgrove

Joined: 10 Dec 05

Posts: 2142

Credit: 2775821103

RAC: 809989

RE: I think it knows what

25 Sep 2007 15:06:25 UTC

Message 73244 in response to message 73243

(moderation:

)

Quote:

I think it knows what it has sent you. If there is more S5R2 work it would have tried to send you had you done nothing, it just realizes some files are missing and resends them, rather than treating that as a reason to start you a though your were fresh.

No, this was a brand new send - I hand-edited client_state.xml to get rid of the old references. I got:

- no trace of a resend.

This box has only ever done 532.xx and 523.xx work using Beta v4.40 previously, newly built and attached to Einstein less than a month ago.

Yin Gang

Joined: 23 Feb 05

Posts: 52

Credit: 120187750

RAC: 0

My first S5R3 WU just

25 Sep 2007 15:21:55 UTC

Message 73245

(moderation:

)

My first S5R3 WU just finished on P4-2.8E (with HT):

name: h1_0539.05_S5R2__158_S5R3a
cpu time: 114,452.47 secs
granted credit: 219.42 pts

so it's about 13.8 pts/hr.

To compared with S5R2...

name: h1_0539.05_S5R2__59_S5R2c
cpu time: 313,573.23 secs
granted credit: 646.69 pts

it's about 14.8 pts/hr.

Welcome To Team China!

picantecomputing

Joined: 1 Jan 07

Posts: 10

Credit: 109837

RAC: 0

Results for PPC Mac G4 1.25

25 Sep 2007 15:57:27 UTC

Message 73246

(moderation:

)

Results for PPC Mac G4 1.25 GHz:

R3: 97,902.35 sec, 219.42 cr = 8.068 cr/hr

R2: 217,336.68 sec, 633.91 cr = 10.500 cr/hr
R2: 159,106.92 sec, 472.32 cr = 10.686 cr/hr
R2 avg: 10.593 cr/hr

So about a 25% drop in credit for R3 vs. R2. Ouch.

archae86

Joined: 6 Dec 05

Posts: 3146

Credit: 7063614931

RAC: 1230786

RE: - no trace of a

25 Sep 2007 15:59:05 UTC

Message 73247 in response to message 73244

(moderation:

)

Quote:

- no trace of a resend.

I stand corrected. Thanks.

I assumed the server as well as client_state.xml had a record of what you were "supposed" to have. Long ago I deleted "just about" everything Einstein in an effort to get away from a batch of results that for over a month had kept one of my machines tied up processing work it was particularly ill suited to. That time I got the same work right back--but I'm sure I did not edit client_state.xml (I only got brave enough to do that more recently), and not sure that I deleted it either.

Richard Haselgrove

Joined: 10 Dec 05

Posts: 2142

Credit: 2775821103

RAC: 809989

Since Bruce asked for

25 Sep 2007 21:07:43 UTC

Message 73248

(moderation:

)

Since Bruce asked for observations on the 'early release' S5R3, to help fine-tune the bulk run later, here's another one to fill in the time while we're waiting for the last of the S5R2s to be issued.

Host 1001562 is working it's way down through an almost consecutive run from the 522.10 set - I've reported 168 thru 162 so far, and I've got most of 161 thru 150 in stock. There's been a very steady decrease in crunching time as I've gone through them, averaging about 1% per WU, and since the credit claim has been static (218.66 every time), a corresponding steady increase in the rate of return. Could that account for the wide variations people have been reporting?

Edit - I've got the same wingman (989567) for all of them, and his run is even more complete than mine. Results are only just starting to come in, but early indications are that it's the same pattern. He's running Linux on AMD against my WIntel, so might be a useful pairing to keep an eye on.

Udo

Joined: 19 May 05

Posts: 203

Credit: 8945570

RAC: 0

RE: Since Bruce asked for

25 Sep 2007 21:39:35 UTC

Message 73249 in response to message 73248

(moderation:

)

Quote:

Since Bruce asked for observations on the 'early release' S5R3, to help fine-tune the bulk run later, here's another one to fill in the time while we're waiting for the last of the S5R2s to be issued.

Host 1001562 is working it's way down through an almost consecutive run from the 522.10 set - I've reported 168 thru 162 so far, and I've got most of 161 thru 150 in stock. There's been a very steady decrease in crunching time as I've gone through them, averaging about 1% per WU, and since the credit claim has been static (218.66 every time), a corresponding steady increase in the rate of return. Could that account for the wide variations people have been reporting?

Edit - I've got the same wingman (989567) for all of them, and his run is even more complete than mine. Results are only just starting to come in, but early indications are that it's the same pattern. He's running Linux on AMD against my WIntel, so might be a useful pairing to keep an eye on.

You wingman is processing
h1_0522.05_S5R2__35_S5R2c in 87,063.38 Sec. and gets 461.04 credits but
h1_0522.10_S5R2__171_S5R3a in 71,280.50 Sec. gives him only 218.66 credits!
That's 188,84 sec./credit against 325,99 sec./credit.

Udo

Donald A. Tevault

Joined: 17 Feb 06

Posts: 439

Credit: 73516529

RAC: 0

Well folks, It looks like

25 Sep 2007 23:04:08 UTC

Message 73250

(moderation:

)

Well folks,

It looks like my AMD 6000+ machine is taking a bit of a performance hit with the S5R3's. They're taking just a bit over half as long to do as it would a monster S5R2, but only giving me about one-third of the credit.

This box is running Kubuntu 7.04 Linux. I would be interesting if I could have a Windows setup with the same hardware to make a comparison.

Ciao,
Donnie

archae86

Joined: 6 Dec 05

Posts: 3146

Credit: 7063614931

RAC: 1230786

RE: It looks like my AMD

25 Sep 2007 23:34:02 UTC

Message 73251 in response to message 73250

(moderation:

)

Quote:

It looks like my AMD 6000+ machine is taking a bit of a performance hit with the S5R3's. They're taking just a bit over half as long to do as it would a monster S5R2, but only giving me about one-third of the credit.

This box is running Kubuntu 7.04 Linux.

Two comments on your experience.

1. Of your three S5R3 results, one has successful quorum partner--a Core 2 Duo running XP. That person saw about a 3% degradation from their S5R2 resuls on your worst (slowest) result, much less degradation than you report.

2. Your experience is another example of strong CPU time variation for results in the same series with the same credit:

all three results are of the form h1_0545.10_S5R2__rrr_S5R3a_n

Res CPU_______credit
185 57,125.42 220.28
178 52,348.26 220.28
174 50,046.86 220.28

S5R3

Forums › Cruncher's Corner

Comment viewing options

Forums › Cruncher's Corner