Welcome to MilkyWay@home

Posts by Thamir Ghaslan

21) Message boards : Number crunching : Problem with tiny cache in MW (Message 31681)
Posted 29 Sep 2009 by Thamir Ghaslan
Post:
WHICH IS PRECISELY why I've completly detached from MW and have made my GPU solely run Collatz. Instead of whining and getting frustrated, the better alternative was collatz. Collatz offers 120 tasks, each of which runs 15 minutes, and the queue seldom runs dry.

In the long run I'm sure my RAC on collatz will be higher than it is on MW.
22) Message boards : Number crunching : 3 way 5870 (Message 31611)
Posted 27 Sep 2009 by Thamir Ghaslan
Post:
Something tells me milkyway would scale better!

http://www.youtube.com/watch?v=81Q0n7lgGZg&feature=related

http://www.hardware.info/en-US/news/ymicl5qZwpuacJY/AMD_Radeon_HD_5870_Triple_CrossFireX/

5870x2

http://www.hardware.info/en-US/news/ymicl5qZwp2aaZY/AMD_Radeon_HD_5870_X2_spotted/
23) Message boards : Number crunching : Lucid HYDRA Multi-GPU Technology Launches - End of SLI and CrossFire? (Message 31437)
Posted 24 Sep 2009 by Thamir Ghaslan
Post:
Don't know if anyone sucesfully mixed ATI and Nvidia drivers in a single OS, I tried once, under XP and 7, and failed.

This might be an interesting read.

http://www.pcper.com/article.php?aid=785
24) Message boards : Number crunching : Credit lowering (Message 31435)
Posted 24 Sep 2009 by Thamir Ghaslan
Post:

I am not saying that it could not be done ... but just as a practical matter you would need 10-20 really good people to make a go of it ... and they would have to be really into it ... it is not just that BOINC is so big, but that there is so much that is messed up ...


Maybe its time to outsource boinc to India. :P
25) Message boards : Number crunching : Problem ATI v0.2 and Win7? (Message 31060)
Posted 16 Sep 2009 by Thamir Ghaslan
Post:

Have you tried the W tag? Try increasing it above the default "1" to 1.1 or 1.2.


Thanks for the tip, I've resolved the issue by matching the F tag with my monitor's refresh rate, at 85 hz.

My TV runs at 30hz, now the gui is fluid smooth and so is video playback!

I've reset w and n to their defaults since they have little effect compared with F.
26) Message boards : Number crunching : Problem ATI v0.2 and Win7? (Message 31050)
Posted 16 Sep 2009 by Thamir Ghaslan
Post:
Cat 9.9, win 7 32 bit, 4870x2, boinc 6.10, and ati 0.2 ..

Main problem is choppy video graphics playback and GUI compared with the very smooth video and GUI which was completly not noticable slowdowns in previous setups.

Edited app_info when it comes to the F tag and N tag, 1 concurrent work unit or higher/lower frequency rates had little to no effect on video performance!

I heavily use my computer as a media center and don't wish to pause the application while the computer is in use!

Rollback? Is it expected due to the code rewrite which emphasizes app performance over everything else? Anyone else experience this?


27) Message boards : Number crunching : Probability of burning a 4870x2 ? (Message 30624)
Posted 12 Sep 2009 by Thamir Ghaslan
Post:
Today I've upgraded the drivers to 9.9, and went for the .19G application!

MASSIVE temprature drops!

Temp went down from 80c to 70c at full fan throttle with mem downclocked to 500 mhz.

On auto mode the fan would throttle at 40%, but the chip never exceeds 90c.

I prefer manual full fan throttle.

Bottom line, I like the new driver update, and gipsel's .19g updated app cut down gpu computation by an average of 20 seconds, before and after the driver update!
28) Message boards : Number crunching : Credit lowering (Message 30577)
Posted 12 Sep 2009 by Thamir Ghaslan
Post:
Lowering it ~30% will certainly run users off again


It may but there's going to be no mass exodus of Participants. Right now Milkyway has no Competition for the ATI Boys other than Collatz Project which at the moment doesn't compare to Milkyway as far as Credits for the ATI Cards goes.

That may change though soon and if the ATI Credits at Collatz are comparable to Milkyway then you might see a bigger jump of Participants to the Collatze Project since they may pay about the same.


distriubted.net has an ATI app, don't mind running it once in a while and not sure when yoyo@home will hatch on to the ATI app or how much credit it will grant by converting keys to credits!

If memory serves me right, 30,000,000 keys / second on a quad core, 500,000,000 on a gtx 280, and 1,200,000,000 keys on a 4870x2!
29) Message boards : Number crunching : Probability of burning a 4870x2 ? (Message 30542)
Posted 12 Sep 2009 by Thamir Ghaslan
Post:
I have a quad with an ATI HD4850 for MW.

Crunching away nicely until about 7.00pm tonight, then the PC died. I thought the 4850 had burnt out, despite the fan running at 60% and the temperature being 77c to 81c with 98% load on the GPU.

I tried rebooting the PC, but zip all. So, for the second time - either the PSU has died (last time mid October 2008, and replaced with a Gold 550w jobbie) or the MoBo (the Asus P5D lost capacitors and went away to be replaced by a Gigabyte MoBo.

If either component is at fault I should get it replaced under the 12 month warranty. But my RAC will drop like a ...


Please keep me posted, I have a gigabyte and an antec 850 watt.

30) Message boards : Number crunching : Probability of burning a 4870x2 ? (Message 30469)
Posted 11 Sep 2009 by Thamir Ghaslan
Post:


One thing I found works is to declock your memory - it has very little effect on the MW app so it should produce less heat if you declock it.

GFX cards are now designed to be pelted, so it should run 24x7 for the whole warranty. However I guess they assume it'll be in a regular environment. Your GTX280 sounds like a dud if it burnt out. If you can't RMA it, tried reseating it/changing the cooler?


Thanks for the tip, I've moved the slider all the way down to 500 mhz for mem on both cards, on average I've dropped 90c to 80c on the primary and from 80c to 70c on the secondary card.

GTX 280 was severely burned, I could hear little metallic clicks coming out of the chip! The card can run in 2d mode, but once it hits 3d or cuda, I'd get a screen full of rainbows!
31) Message boards : Number crunching : Probability of burning a 4870x2 ? (Message 30463)
Posted 10 Sep 2009 by Thamir Ghaslan
Post:
I suppose this has been asked time and time again, but here goes:

July 2008, I bought a GTX 280, ran it 24x7 for 6 months on GPU GRID until it burned out.

Could'nt RTM due to geographic reasons.

Bought a 4870x2, felt happier due to its gaming stability and cheaper price performance ratio.

Now I'm happily crunching 24x7 on milky way for almost 2 weeks. AC in my room is busted, so instead of averaging 80c I'm averaging 90c on the card. Don't mind the noise.

Did you, or did anyone you know had their ATI burned? My little web research shows that nVidias have a notoriously higher failure rate than ATIs, could be bad manufacturing or could be the higher market share, which leads to higher failure counts!

You rarely hear of CPUs burning even with nasty over clocks, but GPU burning is all to common, methinks!
32) Message boards : Number crunching : Server Updates and Status (Message 17269)
Posted 1 Apr 2009 by Thamir Ghaslan
Post:

CPUs and GPUs can crunch for each, though logic dictates CPUs for the short, GPUs for the long. This is predicated on the fact that BOINC does not natively detect ATI GPUs.


Who cares about BOINC!

A simple option on the user preference page to tick a box so that the server can send long WU if I have a GPU and short WU if I have a CPU will resolve this!

However, I guess the admins already decided to split the website in 2, one for GPU and one for CPU.
33) Message boards : Number crunching : Server Updates and Status (Message 17267)
Posted 1 Apr 2009 by Thamir Ghaslan
Post:
How are you going to handle the current credit data base? Freeze it and start over with new projects? Or are you going to determine what gets moved to what project? If the later, how are you going to determine what gets moved to what project? Ex, I have approx 25M on CPU, and 25M on GPU.

Continue accumulation here but only for work done on CPUs using CPU applications.

The GPU project will be GPU only like GPU Grid and will start at 0 for all hands. WIth new applications out from the project, theory says that the participant will be able to connect to both the CPU and GPU project and do work on both, but the totals will be separate.

So, you will have 50M and rising here, and what ever on the new project ...

Not sure how they are going to make this work when the one GPU application we do have is not integrated into the BOINC system ... and if the new application is CUDA will exclude most of the current base ... of course if it is released as OpenCL then both Nvidia and ATI owners can join the fray ... and it would be time for me to get a new GPU ... :)


I don't think determining CPU from GPU tasks and then migrating would be difficult to implement. There are many fields and hints to determine what to split!

If they query the database to move any tasks below say, 10 second CPU time then its a GPU, anything higher then its CPU!

Or they could look into the GIPSEL string under application version.


34) Message boards : Number crunching : Server Updates and Status (Message 17207)
Posted 31 Mar 2009 by Thamir Ghaslan
Post:
5,000 rocks, easily - each of my 3800s gets about 17,000/day.
5,000 WUs, never. Though I'd get a lot closer than I am at the moment if, when one of my PCs asked for WUs, it actually got some. :)

Hopefully the bigger WU plan will help with this.

Sorry for the triple post. Not only is this board very slow, I just kept clicking on Submit until it was finally accepted, but it doesn't allow me to delete my own posts.


According to: http://boincstats.com/stats/user_graph.php?pr=milkyway&id=3578

these are my best five days:

Date Credit
2009-03-15 15:59:18 38,811
2009-03-16 16:01:47 29,782
2009-03-27 16:02:31 23,771
2009-03-25 16:01:12 17,095
2009-03-14 16:00:58 15,774

Averaging here about low and high credit work units, thats around 1400 to 1500 tasks which took about 6 hours of GPU time.
35) Message boards : Number crunching : Server Updates and Status (Message 17196)
Posted 31 Mar 2009 by Thamir Ghaslan
Post:
Any one got lucky enough in the past to get the full 5,000 quota per day?

Just averaging here and not sure how correct my calcs are, my 4870x2 can do one task in 30 GPU seconds, forgeting CPU time and wall clock time!

5000 * 30 seconds = 150,000 seconds.
150,000 / 60 = 2,500 minutes
2,500 / 60 = 41.6 hours.

Give or take 2 days for 5,000 tasks and possibly one day if the 4870x2 uses two GPUs.

And give or take 130,000 to 150,000 credits for these 5,000 tasks!
36) Message boards : Number crunching : 3rd.in - optimized apps (Message 16742)
Posted 24 Mar 2009 by Thamir Ghaslan
Post:
I changed it to .19 and it seems slower... is that right? I figured with a newer version it would be faster.

It's possible you were crunching slightly shorter WUs with the 0.16 app than with the 0.19 app. A ps_s7x or ps_s8x WU is shorter than a ps_s2x WU, I've noticed.


Correct me if I'm wrong, but I'd give more weights on directx 9, 10, and 11, and catalyst drivers than the actual milkway gpu app!
37) Message boards : Number crunching : 3rd.in - optimized apps (Message 16552)
Posted 22 Mar 2009 by Thamir Ghaslan
Post:
I'm using Cat 9.3 and windows 7 x64 and yes .19e. As far as following directions I downloaded .19e then extracted it, then made all the system32 changes, closed BOINC which asked me if I wanted stop running my applications I said yes, checked the task manager and all that no BOINC or Milkyway. took the files put them into my milkyway project folder and there you go, restarted Boinc and everything went to hell...kind of.


My advice, get 9.1 for vista and run it in compatiblity mode.

9.3 does not allow me to crunch and does not allow me to connect the TV properly!

So by far 9.1 is the best for me and I'm in a wait and see until the native 7 drivers get more improvements and gets out of beta!
38) Message boards : Number crunching : WU abuse (Message 16551)
Posted 22 Mar 2009 by Thamir Ghaslan
Post:


what was illegal was optimized apps that were not true to the project and that all happened on Seti@home a long time ago.



I remember those days and what I remember most was the claims that these apps were not calculating as correctly as the non optimized apps.
39) Message boards : Number crunching : WU abuse (Message 16388)
Posted 21 Mar 2009 by Thamir Ghaslan
Post:
My computers are not hidden. You can view them anytime you like. The only one running the update script is the q6600. It is the only one that runs out of work.
<edit>
My cache is set to .5 days


Would you do me a favor and test something? Increase your cache to 3 days and then let BOINC do its' own thing and see if things are better...


Mine is set at 10 days and it makes no difference!

On other projects I'm overfilled with tasks!
40) Message boards : Number crunching : WU abuse (Message 16260)
Posted 20 Mar 2009 by Thamir Ghaslan
Post:
OIE!

Not that I'm trying to flame here! But I can't believe the rambling here!

What we are dealing with here is simply the project having more crunching power than it needs! And this has been flying off the chart with the GPU clients.

The obvious fix is that this project produces more intensive computation and then everyone will be happy!

In the meantime I have other projects attached and simply don't care if I get milkyway units or not!


Previous 20 · Next 20

©2020 Astroinformatics Group