Welcome to MilkyWay@home

Posts by Purple Rabbit

21) Message boards : Number crunching : ps_sgr_*F_2s_1 (Message 26019)
Posted 19 Jun 2009 by ProfilePurple Rabbit
Post:
You can always look for a PCI slot based GPU card to help things along...

Don't think that I haven't tried :-) My elderly veggies are just happy if they can light up. I can relate to this (oldness, not veggie). I'm afraid that data transfer over the PCI bus would take longer than the computations if even I could find a card that would do it. I always give them the best and I think their 5 year old video cards are it...sigh.

They try to work out, but their 250 watt power supplies limit any heavy lifting. They're up to 2 fluffy bunnies on a bamboo stick. I've offered to put them out to pasture at a local school, but I suspect (like my 486) they're not adoptable.
22) Message boards : Number crunching : ps_sgr_*F_2s_1 (Message 25862)
Posted 18 Jun 2009 by ProfilePurple Rabbit
Post:
Poor Cabbage and Lettuce, hosts 37831 and 37830 respectively (you don't need the link, it's boring!) just about had a nervous breakdown with these new tasks :) Both computers are (almost) identical 1.3 GHz Celerons (upgraded from the screaming 600 MHz P3 they came with in 1999). They probably have no business still processing BOINC, but they are. I'm a pack rat :)

They sit in my basement doing what ever I ask, albeit slowly. The previous (normal?) 1.5 to 2 hour tasks didn't phase them. The current longer tasks take about 5 hours. When these tasks came along they politely asked me, "WTF"?

They're still doing fine, but please don't say the word "GPU" in their presence (I think it's a swear word for them although it only has 3 letters). Some of my vegetables (I'm a wabbit and I have a computer garden) are quite sensitive about this. Potato (my main computer) is the only one with a "real" GPU (Old 8600GTS NVIDIA card...sigh), but he keeps ragging on the other vegetables because he can do SETI on his GPU. My other vegetables in the garden make fun of Cabbage and Lettuce just because they are wimps. I've had to chastise Onion several times about this.

I've taken a great deal of space to say that we (me and the vegetables) have no problems with the new tasks, but don't you dare make them any larger!
23) Message boards : Number crunching : MW power outages (Message 22764)
Posted 20 May 2009 by ProfilePurple Rabbit
Post:
I'm turning blue waiting for the power outage! Everything is running as of 0200 UTC. I hope Travis has his servers inoculated against the power outage. There always seem to have gawd awful consequences when this occurs (like stuff disappearing).

I'm still scared that they're still running with the threat of no power. Bad stuff can happen with an ungraceful shutdown...sigh

Oh well, off to have some tea and then bed.
24) Message boards : Number crunching : MW power outages (Message 22758)
Posted 19 May 2009 by ProfilePurple Rabbit
Post:
All still running here at 23:30 UTC. I've reported and received new work.

So if the web pages are operational, I don't understand why the rest of the project back end can't be operational....


I think it's all the same machine. If the power goes then...nada! I suspect the the outage schedule is quite flexible for the power guys. It goes out when it goes...sigh
25) Message boards : Number crunching : Little help for those using opti apps. (Message 10350)
Posted 11 Feb 2009 by ProfilePurple Rabbit
Post:
I think that experience with a project will tell you if a higher SSE version is useful. For MW, I think it is at least to some extent.

The OS reports what it knows. It was written years earlier. CPU-Z (and the like) are more current. BOINC can help with showing the capabilities, but it can't do anything with things OS doesn't report.
26) Message boards : Number crunching : Little help for those using opti apps. (Message 10338)
Posted 11 Feb 2009 by ProfilePurple Rabbit
Post:
What Banditwolf suggested is quite adequate to get people going. There may be some more opimiation to do.

The reporting of CPU capabilities is about the same between Windows and Linux. I think any Bill Gates conspiracies can be ruled out. It's more about the age of the release than the originator. Accept the the deficiencies and adapt!
27) Message boards : Number crunching : Little help for those using opti apps. (Message 10336)
Posted 11 Feb 2009 by ProfilePurple Rabbit
Post:
Yes, what Ageless said :) I just want to add that any 64 bit extensions CPU-Z shows (doesn't apply to the current SSE*.** discussion) requires a 64 bit OS.

@Ice Your confusion in SSE levels is understandable. Presumably the increase in SSE may increase processing speed. Not necessarily, it depends upon the project. MW (in my small universe) seems to run better (faster) with the increasing SSE numbers. This all depends upon on what calculations they do. It seems (to me) that higher SSE versions seem to work better (faster)
28) Message boards : Number crunching : Little help for those using opti apps. (Message 10332)
Posted 11 Feb 2009 by ProfilePurple Rabbit
Post:
Yes, CPU-Z is more accurate. Go for it!
29) Message boards : Number crunching : Little help for those using opti apps. (Message 10326)
Posted 11 Feb 2009 by ProfilePurple Rabbit
Post:
This list is a bit conservative in some cases. Certainly if you go by it everything "should" work. My AMD 5600+ X2 (Linux) shows SSE2, but it can actually do SSE3 (and is doing it). My T8300 (Windows) shows SSE3, but will do SSE4.1 (likewise successful). The operating system has a lot to do with what is reported.

As a first cut this is good advice. Start here and use CPU-Z to find out what the processor really does. It may do more.
30) Message boards : Number crunching : We're back (Message 10323)
Posted 11 Feb 2009 by ProfilePurple Rabbit
Post:
A project reset can fix many problems. For my 1.3 GHz Celerons (Linux) it might be the kiss of death (or at least the kiss of hurt) tho. They have a DCF of 6. Any reset will result in an over fetch. For them it's days of extra work. Even my AMD64 5600+ X2 and AMD64 3800+ X2 (Linux) have a DCF of 3. Only my Q6600 and T8300 (Windows) are close to one. All are running an optimized app.

By all means reset the project if nothing is working, but be careful. You may get more than you bargained for :)
31) Message boards : Number crunching : We're back (Message 10239)
Posted 9 Feb 2009 by ProfilePurple Rabbit
Post:
Us wabbits are quite talented so we don't have your finger problem (there's a LOT of us!). A true wabbit would NOT nibble a bead. This is just not done, but we have just begun to use the optimized vermilion beads.

Although a quad abacus is nice, we soldier on with the HT version. The puce beads are proving to be useful. Unfortunately the bit bucket is getting full of puce beads. I think we need to purge the bucket more frequently. New work has been delayed while we argue about where it goes on the overfull abacus.

PS: Travis and Dave--Both of you are doing great work on a shoe string budget. Please take my comments in the spirit they were intended. I'm WAY off topic and deserve to be deleted--Beware of the wabbit angst tho. Besides Travis, you started it!
32) Message boards : Number crunching : We're back (Message 10232)
Posted 9 Feb 2009 by ProfilePurple Rabbit
Post:
I'm still trying to get the hyper threaded abacus to work. Sharing beads is proving to be a bottleneck. I think I'll paint them blue and puce. That may not increase my credits, but it will certainly be entertaining :)
33) Message boards : Number crunching : Database troubles (Message 10226)
Posted 9 Feb 2009 by ProfilePurple Rabbit
Post:
These things are like peeling an Onion. There's always more...sigh

I'm sure that you'll finally get to the end tho.
34) Message boards : Number crunching : Giant Workunits? (Message 9651)
Posted 4 Feb 2009 by ProfilePurple Rabbit
Post:
Travis, you have given us a challenge. This is a good thing tho. Changing the Fpops has made my vegetables (it's a garden and I'm a Wabbit!) really go bonkers, but this will really help in the long run.

My 1.3 GHz Linux Celerons (Cabbage and Lettuce) suddenly went from 1:37 to 0:04 or so for the estimated time and downloaded a bunch (more than they can comfortably eat). Cabbage decided 2:35 was the answer for a couple of s21 tasks (not crunched yet) tho...sigh. My bigger iron went to 0:04 and downloaded the limit too (but a bunch more)...sigh again. Thank you for the limit! Potato and Onion complained, but I told them to chill. This too shall pass. My vegetables agree that we can work it out and in the end everything will be better once BOINC figures out what's going on and the transient times settle out.

I've had a chat with all the vegetables (8 of them: 4 Linux (optimized), 4 Windows (stock)) and we all agree that we can handle this. It's a one time thing that helps the project. All of us are vigilant about task waste, and we apologize for the abandoned tasks.

It's a PITA, but we can handle it :)

Rick (aka: Strangely colored Wabbit)
35) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 9230)
Posted 26 Jan 2009 by ProfilePurple Rabbit
Post:
Here's apples to apples. SSE_32 and AMD SSE3_64 are both winners. First of all they run! Here's what I've seen so far (remember sample size is 1 for the new app).

SSE_32: Running on two 1.3 GHz Celeron computers running SUSE 10.3. The stock app took about 2:02 hours. The optimized app (both .14 and .16) take about about 1:35 hours.

AMD SSE3_64: My AMD64 5600+X2 running SUSE 10.3 64 bit ran the optimized .07 (SSE2) app in 0:31 hours. The stock apps from 0.12 to 0.16 ran in about 0:53 hours. The new optimized app ran in 0:23 hours.

AMD SSE3_64: My AMD64 3800+X2 OC'd to 2.35 GHz running SUSE 11.1 64 bit ran the 0.16 stock app in 1:01 hours. The new optimized app ran in 0:28 hours.

Thanks Speedimic for the optimized apps. They are working VERY well, but what will you do for an encore? :-)

Rick
36) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 9050)
Posted 25 Jan 2009 by ProfilePurple Rabbit
Post:
The .12 version saw an increase from 2:02 to 1:37 CPU time as compared to the stock version.


I meant decrease ...sigh. Stupid computer sends what I type rather than what I meant!
37) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 9028)
Posted 25 Jan 2009 by ProfilePurple Rabbit
Post:
I'll take that as a yes :-) I just downloaded the SSE version. Unfortunately only my 1.3 GHz Celerons can take advantage of these apps (much to their pleasure I might add). The .12 version saw an increase from 2:02 to 1:37 CPU time as compared to the stock version.

My AMD 3800+ x2 and AMD 5600+ x2 can't handle them (as expected). I only tried the 64 bit SSE3 app tho. Maybe the 32 bit SSE2 will work? It's much too late tonight to embark on what may be a major task. I'll try it tomorrow unless someone knows it's futile.
38) Message boards : Number crunching : Reverting to Stock (Message 8798)
Posted 21 Jan 2009 by ProfilePurple Rabbit
Post:
Yep, that's what I saw. Future WU are .12 apps. Anything left from before seem to run under the old app (unless you delete it (I really need to do this)-I suspect it then has a conniption fit!)

Hopefully Travis will give a bit more of a reprieve. I've only got a few more to go of what I think are OK apps, but have been declared deficient.

Rick
39) Message boards : Number crunching : Reverting to Stock (Message 8796)
Posted 21 Jan 2009 by ProfilePurple Rabbit
Post:
Poor Boy: I just looked at my latest contribution to MW. It still looks like it's using the old "optimized" app (very fast), so I guess I need to delete it as well.

Rick
40) Message boards : Number crunching : Reverting to Stock (Message 8795)
Posted 21 Jan 2009 by ProfilePurple Rabbit
Post:
Travis, you seem to have kept the .07 app on the server. That is a big plus (I think). It's not clear my cache is using it, but believe me, I'm trying. I want to give you useful results.

I didn't detach/reattach. The WU list showed the "new" app (.07 stock) after I did what I did. I stopped BOINC, removed the app.xml file, and restarted. I didn't remove the old optimized app.

My main problem is to assure BOINC that the WU won't run at 5 minutes, but 57 minutes. OK, I enable new work, get some stuff, then disable new work. So far so good.

Of course whether this works awaits validation. I'm turning blue holding my breath waiting for the final decision!

Rick


Previous 20 · Next 20

©2020 Astroinformatics Group