Hi,
I am a brand new user, because seti is short of work units. Normaly my PC does a Seti Unit in about four hours. After downloading the first climate work unit I got a prdiction of more than 2100 hours to complete the workunit. Well, my PC normally runs about two hours a day, what makes it abot five years to complete.
I guess, that can\'t be true. What could possibly be wrong ? It\'s a 2Ghz AMD processor with 512MB RAM.

---

Nothing wrong.
Welcome to the toughest DC project there is. And the one with the longest work units.

The problem with 5 years, is this project isn\'t open ended like SETI and the others. It may well have achieved it\'s aim before then.
Click on the Climate Science link to the left of here to see what it\'s all about.

Well, so I believe that makes no sense for me to take part and I will cancel my membership. I\'d never be able to finish a single Work Unit. Sorry for that and my best wishes to that fascinating project. Maybe I\'ll come back, after the power of the PCs have tripled....
Thanks and good luck, Joerg

---

running a P4D and 1.024 megs of RAM, although the estimates, and I stress \"estimates\", are of around 1500 hours, I have experienced a rality of 1000 to 1100 hours.......

So don\'t let the estimates scare you, and whatever portions you finish are of value to the researchers....

---

I have a P4 2.8GHz, and my time to completion is another 2720 hours (already done 124 hours!!)
If I open the graphic though, I can see my \"crunching speed\" of 3.11 seconds/time-step.
What are you getting with yours (hopefully about 1 sec/TS)? When you say a \"P4D\", do you mean a dual-core Pentium 800 Series-D processor (820, 830....which one?)
I\'m upgrading soon to a dual-core and would be interested to know about the performance advantage.

Regards,

Neil.

running a P4D and 1.024 megs of RAM, although the estimates, and I stress \"estimates\", are of around 1500 hours, I have experienced a rality of 1000 to 1100 hours.......

So don\'t let the estimates scare you, and whatever portions you finish are of value to the researchers....

The \'time to completion\' figure is misleading, best to calculate it from the % done.

Dual core is excellent for the project, but won\'t help the times per model - it lets you run two models at once.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

I have a P4 2.8GHz, and my time to completion is another 2720 hours (already done 124 hours!!)
If I open the graphic though, I can see my \"crunching speed\" of 3.11 seconds/time-step.
What are you getting with yours (hopefully about 1 sec/TS)? When you say a \"P4D\", do you mean a dual-core Pentium 800 Series-D processor (820, 830....which one?)
I\'m upgrading soon to a dual-core and would be interested to know about the performance advantage.

Regards,

Neil.

running a P4D and 1.024 megs of RAM, although the estimates, and I stress \"estimates\", are of around 1500 hours, I have experienced a rality of 1000 to 1100 hours.......

So don\'t let the estimates scare you, and whatever portions you finish are of value to the researchers....

Neil,

A bit OT but, what the heck, it relates to calculation times...

Had five P4 HT boxes, from 2.8 to 3.4; three remain. I was an Intel-bigot for hadware because I thought they crunched floating-point better/faster than AMD. It was probably true once upon a time. Now? AMD might have a slight edge. Also, I think the AMDs use less power, hence a bit quieter, than Intel boxes of equivalent power. So, I now also have three A64 X2 4400+ boxes designed to crunch CPDN (good MoBo, low-end Graphics card, good low-latence memory), have Win32 XP on one, Win64 XP on the other two -- and SuSE 10.0 64-bit across the board. (I\'m very pleased with the X2s; two currently running TCM in XP and one running Seasonal in Linux.)

Lots of info on the net. Here is a good place to start: http://www.tomshardware.com/
(Don\'t let the \"busy\" home page turn you off; there\'s a lot of good stuff buried in there.)

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

---

And with your 4400+ PC\'s what sort of secs/TS are you getting per core?

Neil.


A bit OT but, what the heck, it relates to calculation times...

Had five P4 HT boxes, from 2.8 to 3.4; three remain. I was an Intel-bigot for hadware because I thought they crunched floating-point better/faster than AMD. It was probably true once upon a time. Now? AMD might have a slight edge. Also, I think the AMDs use less power, hence a bit quieter, than Intel boxes of equivalent power. So, I now also have three A64 X2 4400+ boxes designed to crunch CPDN (good MoBo, low-end Graphics card, good low-latence memory), have Win32 XP on one, Win64 XP on the other two -- and SuSE 10.0 64-bit across the board. (I\'m very pleased with the X2s; two currently running TCM in XP and one running Seasonal in Linux.)

Lots of info on the net. Here is a good place to start: http://www.tomshardware.com/
(Don\'t let the \"busy\" home page turn you off; there\'s a lot of good stuff buried in there.)

[/quote]

---

And with your 4400+ PC\'s what sort of secs/TS are you getting per core?

On one box, two Models @ 1.92 s/TS; on the other, 1.90 and 1.93 s/TS.

Much of the difference has to do with parameter mix. Before the global restart, the first machine had two at 1.91 and the other had two at 1.92.

The machines are similar except for memory. The first has Kingston HyperX 2-3-2-6 1T running to spec. The other has Corsair XMS 2-3-3-6 1T running at 2.5-3-3-6 1T, best I could get without manually resetting voltages (which is more than I care to do). No CPU O/C.

(Oh, different brand but otherwise similar low-end video cards.)

Built a similar box for a friend but with Kingston HyperX 2-2-2-5 1T running to spec. If memory serves, her times were both at 1.90, but I won\'t swear to it. (One Model might have been @ 1.89.) I\'ve been given to understand that A64 X2s, perhaps all A64s, like fast memory.

HTH.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

AstroWX,
Many thanks for this input. Just for clarity - is that 1.9s/TS for each core? So you could say your PC is running at 0.95s/TS overall. If it is, than that is amazing. :-O

Neil.

And with your 4400+ PC\'s what sort of secs/TS are you getting per core?

On one box, two Models @ 1.92 s/TS; on the other, 1.90 and 1.93 s/TS.

Much of the difference has to do with parameter mix. Before the global restart, the first machine had two at 1.91 and the other had two at 1.92.

The machines are similar except for memory. The first has Kingston HyperX 2-3-2-6 1T running to spec. The other has Corsair XMS 2-3-3-6 1T running at 2.5-3-3-6 1T, best I could get without manually resetting voltages (which is more than I care to do). No CPU O/C.

(Oh, different brand but otherwise similar low-end video cards.)

Built a similar box for a friend but with Kingston HyperX 2-2-2-5 1T running to spec. If memory serves, her times were both at 1.90, but I won\'t swear to it. (One Model might have been @ 1.89.) I\'ve been given to understand that A64 X2s, perhaps all A64s, like fast memory.

HTH.

---

AstroWX,
Many thanks for this input. Just for clarity - is that 1.9s/TS for each core? So you could say your PC is running at 0.95s/TS overall. If it is, than that is amazing. :-O

Neil.

Yes, 2*1.9n throughput. And, yes, they are amazing machines.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

AstroWX,
Many thanks for this input. Just for clarity - is that 1.9s/TS for each core? So you could say your PC is running at 0.95s/TS overall. If it is, than that is amazing. :-O

Neil.

Yes, 2*1.9n throughput. And, yes, they are amazing machines.

Astro is good but..... 1.47 and 1.49 for my 2 models on a dual core AMD overclocked to 2.85GHZ and yes it is stable as I finished two coupled spin-ups on this machine

can\'t wait for Conroe, the first Intel offering I have been exited about in years, its going to eat the A64 alive and I am an AMD fanboy

---

\"Conroe\" does look like a step-up in performance - no question. Whether it\'ll beat the FX-60 remains to be seen, but the current benchmarks I have seen look promising. However, AMD aren\'t doing to stand still, and I am sure that the next generation of FX-62/64 CPUs running on the AM2 socket mobos which allow DDR2 RAM to be used for the first time will sing with the AMD chip. I\'m an Intel fan!!!!....but probably moving to AMD for a while!

Neil.

[/quote]
Yes, 2*1.9n throughput. And, yes, they are amazing machines.[/quote]

Astro is good but..... 1.47 and 1.49 for my 2 models on a dual core AMD overclocked to 2.85GHZ and yes it is stable as I finished two coupled spin-ups on this machine

can\'t wait for Conroe, the first Intel offering I have been exited about in years, its going to eat the A64 alive and I am an AMD fanboy[/quote]

---