Pentium II too weak?

Ziran
Ziran
Joined: 26 Nov 04
Posts: 194
Credit: 578,460
RAC: 1,349

Lets se now, 14*24=336. So

Message 20018 in response to message 20017

Lets se now, 14*24=336. So crunching 24/7 you got 336H. My PIII 450 did a result in 40H, and I think it was 4-5 times faster then my P233 doing SETI classic. I think you would be looking at something like 160-200H for an Einstein result. The Albert results I am running now is about 1/5 of an Einstein result, so you might get lucky with 32-40H.

I think 64Mb is too little.

Hmmm, but it sounds interesting. I think I try it on my P233 128Mb at work

Then you're really interested in a subject, there is no way to avoid it. You have to read the Manual.

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1,364
Credit: 3,562,358,667
RAC: 0

RE: Lets se now, 14*24=336.

Message 20019 in response to message 20018

Quote:

Lets se now, 14*24=336. So crunching 24/7 you got 336H. My PIII 450 did a result in 40H, and I think it was 4-5 times faster then my P233 doing SETI classic. I think you would be looking at something like 160-200H for an Einstein result. The Albert results I am running now is about 1/5 of an Einstein result, so you might get lucky with 32-40H.

I think 64Mb is too little.

Hmmm, but it sounds interesting. I think I try it on my P233 128Mb at work

I've yet to get any short alberts, the 3dozenish I've had so far're all around 80-85%.

Out of curiousity, are you using the p1 for anything or's it just gathering dust? I thought the P3-550 I've got as an nt4 testbox was decrepit, and when it most recently had a problem so'd the IS person that answered my help request. Problem was a HD with dieing bearings. Rather than taking a drive from a a newer box and ghosting the old one over, a situation I'd be happy with (would also allow performance testing on a junker). He tried to get nt4 working on a p4-1.8, with a cpu fan as loud as the dieing HD. (louder than the last OCing fan I had,fortunately my need for NT4's <1h/mo). With the IS dept playing musical cubes and consolidating alot of thier operations in a single building his advice was to resubit the request when they were done and let it be someone elses problem.

I was worried about the ram being more of a limit. Nothing I can do there, the mobo won't take more than that.

NikolZ
NikolZ
Joined: 31 Dec 05
Posts: 6
Credit: 4,741
RAC: 0

I use win98se.I got enugh ram

I use win98se.I got enugh ram 190mb to run xp wwindows but it may use cpu for os and whay for i gonna waste time and power such as os.
Athlon xp1600+/kt133 on win 98se
It its posible to run AthlonXp on win 31 ??,wich this os may fully load in half to one second :P
?

Ziran
Ziran
Joined: 26 Nov 04
Posts: 194
Credit: 578,460
RAC: 1,349

The P233 are running win98

Message 20021 in response to message 20019

The P233 are running win98 and is currently working on Predictor 25H/result. I use it at work running Word, Excel and surf the web. I m currently unemployed, so the government have me working a couple of months for a non-profit organisation, running there fair trade shop. They didn’t have a spare computer so I brought one of my old boxes. It is sufficient for the work I use it for.

My computer at home is a celeron 600 384Mb. You take what you can get then you dumpster diving for computer parts. :) It’s got a 15 Gb HD and my satellite receiver got a 200GB. :)

I cant find the ram limit for this project, so you could always attach the host an se what the scheduler says.

Then you're really interested in a subject, there is no way to avoid it. You have to read the Manual.

gravywavy
gravywavy
Joined: 22 Jan 05
Posts: 392
Credit: 68,962
RAC: 0

RE: I have also seen it

Message 20022 in response to message 20013

Quote:
I have also seen it "switch" projects even when it should not.
...
I doubt we will see a change for this either as it would require tracking which CPU needed a change instead of the "simpler" do them all ...

Another approach is to have a separate thread for controlling each cpu.

At present there is one BOINC client thread that controls rpc, benchmarks, and all cpus. Take the cpu control off that thread and put the same code into a new thread, cut down to control only one cpu. Then run as many copies of that thread as you have cpus. A little bit of extra code would be needed to keep track of which processes were already running in other threads.

Switching into edf mode could either happen one cpu at a time, or be handled by an inter-thread message that overruled the usual autonomy of the separate cpu-handlers.

Cost is a little more design separating tasks among threads, saving is no cpu has to know any more about other tasks than 'already running elsewhere'.

Of course, it is even easier to leave it as it is, but with multi-cpu boxes becoming more common the demand for the change will be growing. If we don't see it this year I'm sure it will come soon after.

R~~

~~gravywavy

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.