Beta testing HCC update for Windows

[sierra_bound]

WCG beta test announcement:

We will soon begin the beta test for a new version of Help Conquer Cancer on Windows. We were able to address the page fault issue in this version, and we're hoping that this will reduce the number of page faults and inconsistent credits granted in order to address members' concerns about the project.

http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=18563

[SparkyJJO]

yay

[123bob]

Good news. I'll put my last remaining "123clan" 32bit OP SYS machines back on HCC and report back here. If it performs even close to the Vista 64 bit, BOINC 5.10.28 it will be a very good thing. Thx Sierra! :toast:

[[XC] Adywebb]

HCC Beta testing starts 22nd February - although Sekerob reports having received one already: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=18567

Its important that folks here who have experienced these issues to take part to ensure all types of rigs and OS, particularly 4 and 8 core with XP, are thoroughly tested.

To take part, ensure you select to participate via My Grid>Beta Testing - and please report your findings in the thread above.

:toast:

Edit: It was a mistake in the thread - the Beta test has started.....

[sierra_bound]

Update

The Help Conquer Cancer Windows application has been updated to fix an issue with a large number of page faults. The new application will be downloaded to members crunching Help Conquer Cancer work automatically. There is no need to abort any currently running Help Conquer Cancer work units.

http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=18593

[123bob]

Wow, that was a quick beta test:clap:

Thx for the update Sierra. Have you tried a clover on it yet? I'm still gathering results here on my quads.

[sierra_bound]

No, have not tried it yet.

[[XC] Jaco]

I don't know but this doesn't look very promising to me :shrug:
Windows XP 32-bit

[[XC] Adywebb]

Jaco - I think the 5.15 WU's are the old version, and 5.20 the new ones so that may be the reason....

Here is a screeny of my Clover, note the massive pagefaulting on the 5.15's but far lower and no PF Delta on the 5.20's:

http://www.upload3r.com/serve/210208/1203622615.jpg

I'm not really up on what the figures mean, so perhaps Sierra Bound can give his opinion on whether this is an improvement on before or not?

2 x X5340's @ 3GHz, Windows Server 2003 64Bit, 4GB RAM.

[123bob]

That a very good improvement. Your other 5.20 WU looks good too. About what percent complete was this taken?

The best I get here on the Vista 64 bit/ BOINC 5.10.28 combo is page faults in the low 100k range. This is lower than the FAAH, DDT, or HPF2 WUs.

Yours is looking very good....:up: :up:

You can click twice on "image name" if you want to sort these in order. That will put all your WCG at the top of the window.

[[XC] Adywebb]

The percentage was about 65% - just checked again now on the last 5.20 still crunching which is at 76% and it is showing 43K page faults :up:

[123bob]

Your looking very good.

Here's what bad looks like on the QX6700.

http://i149.photobucket.com/albums/s48/123bob_bucket/HCCpagefaults2-19-08.png

This is a shot from yesterday.

[123bob]

Look at the difference here!!! :up: :clap: :clap: :clap: :up:

http://i149.photobucket.com/albums/s48/123bob_bucket/HCCpagefaults515vs5202-21-08.png

Same machine, same moment in time. QX6700 XP pro 32 bit. The 5.20 WUs are rocking.....I'm a believer.

Bob

[[XC] Adywebb]

Yep, looks like they have done a good job - these 5.20's rock!! :up:

[[XC] Adywebb]

I may have spoken too soon - the Q6600 is fine, but although the page faults and crunching time for these new WU's is also fine on the Clover, the claimed credit has reduced dramatically for some reason even though the Boinc benchmark is the same, hence the granted credit has gone down from around 20/hr to 12/hr. If this stays the same it would amount to around 16K points a day on the clover rather than the usual 26K :shakes:

Rather than post the whole lot again here, for those that have Clovers or are interested have a look at my updates in the thread over at WCG: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=18567

[[XC] Adywebb]

Quick update (although judging by the lack of response here it seems folks aren't that interested in the HCC issues these days considering all the fuss we made about it, but hey ho....) - updated Boinc on the clover to 5.10.39 64-bit and claims now back to normal :up:

[[XC] 4X4N]

I have been letting it go a few days before saying that all is well, but all 3 of my quads look to be fine. Page faults are way down, and credit claimed/granted is good. Two of them are with XP Pro 64 bit.

EDIT: Here is a comparason of XP64 vs XP32. L26A is on XP64. Notice how it does the wu in less time, but claims, and is granted more points.

[123bob]

Adywebb;2794526']Quick update (although judging by the lack of response here it seems folks aren't that interested in the HCC issues these days considering all the fuss we made about it, but hey ho....) - updated Boinc on the clover to 5.10.39 64-bit and claims now back to normal :up:

Interesting situation on your clover. (I'm interested! HCC is now the only project I'm crunching....:D ) I can't speak much for clovers since I don't have one, but, I noticed when I started the Vista 64 bit / Boinc 5.10.28 combination a few months ago, they also claimed lower on most WUs. However, they were very consistent and I was producing more of them in a day. Therefore, my score went up on those machines. This indicates to me that the machine was running more efficiently since it wasn't page faulting so often.

What BOINC version were you running before on the clover? EDIT: Disregard the question. I just read through the WCG thread and got this from your post over there...."Low claim issue solved, must have been Boinc - upgraded from 5.10.30 Windows 64bit to 5.10.39 (Beta) and claims are now fine". Looks like clovers have a combination to work with now too. :up:

I also notice that with the 5.20 WUs, on my QX6700 XP pro 32 bit machine, the WUs have gotten very consistent, like they are on Vista. Below are some WUs from this machine.

QX6700 XP Pro 32 bit/ BOINC 5.10.13 / HCC 5.20 WUs
X0000041520862200411011034_ 2-- el-machino-2 Valid 02/24/2008 15:49:14 02/24/2008 21:05:42 3.53 71.2 / 72.6
X0000042331135200411100828_ 2-- el-machino-2 Valid 02/24/2008 14:46:22 02/24/2008 18:33:20 3.60 72.6 / 72.6
X0000042210624200412072001_ 2-- el-machino-2 Valid 02/24/2008 11:26:30 02/24/2008 15:49:13 3.46 70.0 / 64.8
X0000041620489200410260850_ 2-- el-machino-2 Valid 02/24/2008 03:49:04 02/24/2008 08:27:40 3.63 73.2 / 78.1
X0000041521489200411011021_ 1-- el-machino-2 Valid 02/23/2008 21:37:37 02/25/2008 04:08:25 3.53 71.6 / 73.1
X0000041521169200411011027_ 1-- el-machino-2 Valid 02/23/2008 21:24:57 02/25/2008 04:08:25 3.50 71.0 / 71.0
X0000041511245200410251056_ 0-- el-machino-2 Valid 02/23/2008 20:29:14 02/25/2008 00:44:36 3.58 72.9 / 69.5
X0000041511337200410251054_ 1-- el-machino-2 Valid 02/23/2008 20:28:20 02/25/2008 00:44:36 3.61 73.4 / 69.6
X0000041510452200410251109_ 0-- el-machino-2 Valid 02/23/2008 19:59:26 02/25/2008 00:44:36 3.62 73.7 / 86.3
X0000043121457200411230906_ 0-- el-machino-2 Valid 02/23/2008 19:36:27 02/24/2008 22:09:09 3.55 71.6 / 71.9
X0000043081324200412071406_ 1-- el-machino-2 Valid 02/23/2008 17:40:51 02/24/2008 14:48:17 3.50 70.9 / 70.6

3.5 hours CPU time, claiming tight around 71.5 BOINC points.


Just for yucks, here is my old faithful P4 Northy crunching on 123sarah's account. It also had problems with the older HCC WUs

P4 Northwood / Win XP 32 bit / BOINC 5.10.13 / HCC 5.20 WUs.
X0000049111388200505231305_ 0-- el-machino Valid 02/22/2008 09:03:19 02/24/2008 22:07:10 8.95 81.9 / 89.5
X0000048771238200504281133_ 1-- el-machino Valid 02/22/2008 03:34:54 02/24/2008 04:28:55 9.08 83.1 / 77.9
X0000042470778200412062015_ 1-- el-machino Valid 02/21/2008 13:26:38 02/23/2008 10:53:27 8.91 81.4 / 81.9
X0000042350451200411100931_ 0-- el-machino Valid 02/21/2008 05:33:36 02/23/2008 02:18:41 9.19 83.7 / 76.4

5.20 WUs cleaned this machine up too.....:clap:

With this fix, HCC is all I'm running here. 11 quads and a duo lappy......DEATH TO THAT F%$^&*@ CANCER!! :up:

Regards,
Bob

[[XC] Adywebb]

4X4N;2794549']I have been letting it go a few days before saying that all is well, but all 3 of my quads look to be fine. Page faults are way down, and credit claimed/granted is good. Two of them are with XP Pro 64 bit.

EDIT: Here is a comparason of XP64 vs XP32. L26A is on XP64. Notice how it does the wu in less time, but claims, and is granted more points.
Thanks for the update - looking good especially with 64bit OS's :up:

........With this fix, HCC is all I'm running here. 11 quads and a duo lappy......DEATH TO THAT F%$^&*@ CANCER!! :up:

Regards,
Bob
I knew you were always interested Bob :up: - was just wondering where all the clover users were considering the problems we had :shrug:

Looks like we can finally say HCC is all fine and dandy for everyone to crunch - and as you say......Lets get Crunching them WU's and Death to Cancer!! :D

[sierra_bound]

Adywebb;2795911']I knew you were always interested Bob :up: - was just wondering where all the clover users were considering the problems
In my case, I have moved on. I crunch DDDT exclusively now. Have been doing so for months. Generally I like to focus on one project at a time. When I started running WCG, I crunched only FAAH. There was a transitional period when I was doing FAAH, DDDT & HCC all at the same time.

To be quite honest, I was very disappointed with WCG's initial response to the page fault issue. Go back to the original thread at WCG. Their attitude was and this is a direct quote, "It isn't a problem." That kind of arrogance left a bad taste in my mouth.

I know many people have personal reasons for crunching HCC and I respect that. Cancer has claimed many lives. My interest in DDDT stems from the fact that it has become a real threat in California. People in my state die each year from West Nile virus. For the past few years, officials have had to do aerial and ground spraying in my area to try to control the mosquito population. My feeling is that Dengue will become the next big viral threat to the US. Dengue has been around for centuries. Some historians believe Alexander the Great died from West Nile virus.

One of the things I enjoy about WCG is that you have a choice regarding which projects you can run. It's my choice to run one project. When that project is completed in about six month's time, I will move on to something else.

[twilyth]

In my case, I have moved on. I crunch DDDT exclusively now. Have been doing so for months. Generally I like to focus on one project at a time. When I started running WCG, I crunched only FAAH. There was a transitional period when I was doing FAAH, DDDT & HCC all at the same time.

To be quite honest, I was very disappointed with WCG's initial response to the page fault issue. Go back to the original thread at WCG. Their attitude was and this is a direct quote, "It isn't a problem." That kind of arrogance left a bad taste in my mouth.


That quote can be found at http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=17244#140206

I often wonder to what extent the opinions of the CA's at WCG reflect the attitudes of the people who actually do the work. My impression is that WCG tolerates some of the krap the CA's toss around simply because they don't have the time or resources to interface directly with members - unless there are critical issues that affect a lot of people. I don't think the CA's represent the opinions of WCG, they express their own opinion and just don't get called on it unless it's truly beyond the pale.

They are little more than gate keepers and I personally wouldn't give their statements any more weight than I would anyone elses - usually less. This is especially true in the case of those who are most prolific - D, S and R. I was happy to see Jean appointed CA since I think he is less interested in stroking his own ego than in actually doing the job he was appointed to do.

I think WCG understands that points are a measure of efficiency and that it is in no one's interest to have resources wasted because of inefficient code. However we all understand that they have to prioritize and that the HCC issue happened to coincide with other arguably more serious technical problems. LH at one point was even recommended that people affected by the problem should de-select HCC. He didn't follow the lead of his peers in trying to gloss over the problem.

People who spend any amount of time on the WCG forums will quickly realize who is a tool and who isn't.

[sierra_bound]

My understanding is that Didactylos is involved in the development of BOINC. He could have said, "Hmmm, let me check that out to see if there is a problem that needs to be fixed." Instead, his attitude was, there is no problem. In the end, the issue was addressed, so that's good. But responses like his don't make me eager to want to report any issues that may come up.

[twilyth]

My understanding is that Didactylos is involved in the development of BOINC. He could have said, "Hmmm, let me check that out if to see if there is a problem that needs to be fixed." Instead, his attitude was, there is no problem. In the end, the issue was addressed, so that's good. But responses like his don't make me eager to want to report any issues that may come up.
Ditto. It's people like him that make me grateful that the vast majority of WCG members never go to the forums.

[[XC] Adywebb]

.....To be quite honest, I was very disappointed with WCG's initial response to the page fault issue. Go back to the original thread at WCG. Their attitude was and this is a direct quote, "It isn't a problem." That kind of arrogance left a bad taste in my mouth.


Couldn't agree more S_B.......sometimes I have to hold my nose when posting over there these days :shakes:


People who spend any amount of time on the WCG forums will quickly realize who is a tool and who isn't.
Couldn't have summarised it better :up:

My understanding is that Didactylos is involved in the development of BOINC. He could have said, "Hmmm, let me check that out to see if there is a problem that needs to be fixed." Instead, his attitude was, there is no problem. In the end, the issue was addressed, so that's good. But responses like his don't make me eager to want to report any issues that may come up.
Didactylos can never be wrong....and even when he is shown to be he just adjusts his stance and will never admit it or ever apologise.

Ditto. It's people like him that make me grateful that the vast majority of WCG members never go to the forums.
Again, couldn't agree more :up:

[SparkyJJO]

Apparently they are sending out the 5.20 HCC now for normal operations. This is fold5.