> I am interested in how a well a fast Celeron will do in E@H. I have looked
> around the results database & found a 1400MHz Celeron that does a unit in
> 42,000 sec & a 1700MHz Celeron that does 49000 sec. [These results look
> somewhat inconsistent, of course.]
>
> Is anyone processing E@H with a Celeron at 2200 MHz or higher?
>
> Thanks,
> ADDMP.
>
Celerons and Distributing Computing, Two worlds a get together, a Celeron is not any good Idea for Crunching, try a P4 3.0 with HT or somethink...else
> Celerons and Distributing Computing, Two worlds a get together, a Celeron is
> not any good Idea for Crunching, try a P4 3.0 with HT or somethink...else
The only difference between a full P4 and a Celeron based on one, is the amount of L2 cache that the CPU has. Otherwise, it still sports the full speed that the same CPU has if it is a full P4.
I think the same thing goes for AMD Athlons and the Durons.
I don't see how DC wouldn't go on a Celeron. I've been running BOINC Projects (up to 7 different ones at the same time) on my Celeron since we were in BOINC Beta. It manages the deadlines if you just set your preferences correctly.
My Celeron lacks L2 cache against the Celeron D. Mine has 128kb cache, so projects like Seti will kill me. But E@H, with its FOPS setting set to a specific amount (Floating point Operations per Second), runs the units correctly, claims even less credit than granted. All on CC 4.30
The Celeron D has 256kb L2 cache, it does crunch Seti units faster. And that's all.
All you who think of Celerons in the old Pentium 2 and 3 days, forget all that. These celerons run at the full FSB speed and at the full CPU speed, just lack some L2 cache, but even then, that L2 cache runs at the CPU speed.
> > Celerons and Distributing Computing, Two worlds a get together, a Celeron
> is
> > not any good Idea for Crunching, try a P4 3.0 with HT or
> somethink...else
>
> The only difference between a full P4 and a Celeron based on one, is the
> amount of L2 cache that the CPU has. Otherwise, it still sports the full speed
> that the same CPU has if it is a full P4.
>
> I think the same thing goes for AMD Athlons and the Durons.
>
> I don't see how DC wouldn't go on a Celeron. I've been running BOINC Projects
> (up to 7 different ones at the same time) on my Celeron since we were in BOINC
> Beta. It manages the deadlines if you just set your preferences correctly.
>
> My Celeron lacks L2 cache against the Celeron D. Mine has 128kb cache, so
> projects like Seti will kill me. But E@H, with its FOPS setting set to a
> specific amount (Floating point Operations per Second), runs the units
> correctly, claims even less credit than granted. All on CC 4.30
>
> The Celeron D has 256kb L2 cache, it does crunch Seti units faster. And that's
> all.
>
> All you who think of Celerons in the old Pentium 2 and 3 days, forget all
> that. These celerons run at the full FSB speed and at the full CPU speed, just
> lack some L2 cache, but even then, that L2 cache runs at the CPU speed.
>
but the Celerons are slow, very slow, what are your times at Seti or Einstein
my P4 3.2 with HT does two WUs at SAH at 2:25...that means a WU is ready to send in 1:13... :) that is fast...
So what? Is DC solely for fast computers, or do you object to waiting 6 more hours to see what credit you get granted? I bet it's more than what you usually claim. Thanks to a slower computer.
If all you want to do is crunch as many units as possible, in as little as possible time, then you have my blessings. But just saying that DC isn't for 'slow computers' only because you had the money to buy a fast processor & according hardware, goes a tad too far.
There isn't even a way to accurately compare timings between projects. I run a total of 6 different projects on my PC at this moment, divided between 2 seperate preferences. If you solely run Seti on your computer, how do you want to compare against one that runs 3 or more projects?
More so, why do we have this conversation on Einstein@Home? What is there to compare between Seti & Einstein? Either project is looking for something completely different.
(If you want to check my timings on Seti or on here, you can do so. The name's the same all over the projects.)
> So what? Is DC solely for fast computers, or do you object to waiting 6 more
> hours to see what credit you get granted? I bet it's more than what you
> usually claim. Thanks to a slower computer.
>
> If all you want to do is crunch as many units as possible, in as little as
> possible time, then you have my blessings. But just saying that DC isn't for
> 'slow computers' only because you had the money to buy a fast processor &
> according hardware, goes a tad too far.
>
> There isn't even a way to accurately compare timings between projects. I run a
> total of 6 different projects on my PC at this moment, divided between 2
> seperate preferences. If you solely run Seti on your computer, how do you want
> to compare against one that runs 3 or more projects?
>
> More so, why do we have this conversation on Einstein@Home? What is there to
> compare between Seti & Einstein? Either project is looking for something
> completely different.
>
> (If you want to check my timings on Seti or on here, you can do so. The name's
> the same all over the projects.)
>
what i will say,
a good Cruncher
no celerons, if you had some, be luck with this, ...
Although i don't want to get in the middle of anything here, Ageless is quite right on the Celeron D's. They're not your father's Celeron! LoL.
I don't know that i would specifically recommend a Celeron for crunching, especially if money is not a concern... but hey i'm a grad student here on a Ramen Noodle budget and this was the most bang for my buck. It runs at 533Mhz FSB with 256L2 cache, crunches 24/7, has never once frozen (whatever that's all about), and it does an Einstein unit in 8.5 hours.
Paul's excellent documentation provides some processor comparisons here. My apparently 'unworthy' Celeron crunches both Einstein and SETI in about the same average time as a P4 2.8Mhz that has twice the L2 cache and dual-channel memory to boot.
X-bit Labs did a thorough test of the Celeron D's here and rated them very highly.
All i can say is that i've been extremely happy with mine :)
I'll go for the 2.8Ghz (I take Digger meant that one, not its way lower MegaHertz brother ;)) once I have the change (in dimes...) (and chance ;)). At €96,- it's cheaper than buying a new motherboard and getting a new CPU, plus new memory.
You who win the lottery every week may not notice this, but a complete mobo + CPU + new memory is quite expensive. Try living on a €1099,- monthly income once. You may notice it's hard!
As for Sir Ulli, you don't need to quote all my text each time, then post one line at the bottom. I fricking knew what I posted above. Else I can look it up. Since you think so much of Seti, why not follow the requested Quoting in forums rules there and here?
> Celerons and Distributing Computing, Two worlds a get together, a Celeron is
> not any good Idea for Crunching, try a P4 3.0 with HT or somethink...else
What worries me about comments like "a Celeron is not any good Idea for Crunching" is that it might put people off donating their spare cycles. I am sure the project would rather have a donation from a Celeron than no donation at all.
The very best computer to use for a distributed computing (DC) project is the one you already have.
Personally I have a couple of 700MHz boxes, I got two complete systems second-hand for less than the cost of one cheap printer new -- why would I want to spend more? Some people will no doubt figure E@H/SETI/etc into their buying decision, but I did not, and nor do the vast majority of those donating time. All those donations are welcome to the project.
If you have a Celeron, donate its spare capacity, don't feel excluded. Similarly if you have a VIA C3 or whatever.
If you are buying a new (or second hand!) computer, make your choice according to the primary mission of the box (games? business? study? etc) and whatever you buy on that basis, the box will also be useful to various DC projects. Unless you are very rich, or unless DC counts as your main hobby, please don't go out and buy a computer according to how well it runs DC.
> The very best computer to use for a distributed computing (DC) project is the
> one you already have.
Well said ...
People buy computers for many reasons. One of the primary discriminators is the cost of the machine, and if they are buying it primarily as a DC machine that drives the selection towards higher cost machines.
If, on the other hand, finances are an issue (as with most people), then this is going to drive buyers towards a different class of machine.
As an example, the last computer I bought was a Dell dual-Xeon (3.4 GHz, with 2M Cache) instead of a P4 EE mostly because the same dollars bought a machine that can do 4 Work Units at a time even though the speed was lower than the P4 EE, thusly increasing the throughput.
My next computer will be the next generation Power Mac (or what ever name they give it) and though I hope it will be a dual processor with dual cores, regardless of what it actually comes with, that will be my next workstation.
Like I said, many reasons to buy a specific computer ...
> I am interested in how a
)
> I am interested in how a well a fast Celeron will do in E@H. I have looked
> around the results database & found a 1400MHz Celeron that does a unit in
> 42,000 sec & a 1700MHz Celeron that does 49000 sec. [These results look
> somewhat inconsistent, of course.]
>
> Is anyone processing E@H with a Celeron at 2200 MHz or higher?
>
> Thanks,
> ADDMP.
>
Celerons and Distributing Computing, Two worlds a get together, a Celeron is not any good Idea for Crunching, try a P4 3.0 with HT or somethink...else
Greetings from Germany NRW
Ulli
[img]http://boinc.mundayweb.com/one/stats.php?userID=380 [/img]
> Celerons and Distributing
)
> Celerons and Distributing Computing, Two worlds a get together, a Celeron is
> not any good Idea for Crunching, try a P4 3.0 with HT or somethink...else
The only difference between a full P4 and a Celeron based on one, is the amount of L2 cache that the CPU has. Otherwise, it still sports the full speed that the same CPU has if it is a full P4.
I think the same thing goes for AMD Athlons and the Durons.
I don't see how DC wouldn't go on a Celeron. I've been running BOINC Projects (up to 7 different ones at the same time) on my Celeron since we were in BOINC Beta. It manages the deadlines if you just set your preferences correctly.
My Celeron lacks L2 cache against the Celeron D. Mine has 128kb cache, so projects like Seti will kill me. But E@H, with its FOPS setting set to a specific amount (Floating point Operations per Second), runs the units correctly, claims even less credit than granted. All on CC 4.30
The Celeron D has 256kb L2 cache, it does crunch Seti units faster. And that's all.
All you who think of Celerons in the old Pentium 2 and 3 days, forget all that. These celerons run at the full FSB speed and at the full CPU speed, just lack some L2 cache, but even then, that L2 cache runs at the CPU speed.
> > Celerons and Distributing
)
> > Celerons and Distributing Computing, Two worlds a get together, a Celeron
> is
> > not any good Idea for Crunching, try a P4 3.0 with HT or
> somethink...else
>
> The only difference between a full P4 and a Celeron based on one, is the
> amount of L2 cache that the CPU has. Otherwise, it still sports the full speed
> that the same CPU has if it is a full P4.
>
> I think the same thing goes for AMD Athlons and the Durons.
>
> I don't see how DC wouldn't go on a Celeron. I've been running BOINC Projects
> (up to 7 different ones at the same time) on my Celeron since we were in BOINC
> Beta. It manages the deadlines if you just set your preferences correctly.
>
> My Celeron lacks L2 cache against the Celeron D. Mine has 128kb cache, so
> projects like Seti will kill me. But E@H, with its FOPS setting set to a
> specific amount (Floating point Operations per Second), runs the units
> correctly, claims even less credit than granted. All on CC 4.30
>
> The Celeron D has 256kb L2 cache, it does crunch Seti units faster. And that's
> all.
>
> All you who think of Celerons in the old Pentium 2 and 3 days, forget all
> that. These celerons run at the full FSB speed and at the full CPU speed, just
> lack some L2 cache, but even then, that L2 cache runs at the CPU speed.
>
but the Celerons are slow, very slow, what are your times at Seti or Einstein
my P4 3.2 with HT does two WUs at SAH at 2:25...that means a WU is ready to send in 1:13... :) that is fast...
Greetings from Germany NRW
Ulli S@h Berkeley's Staff Friends Club m7 ©
So what? Is DC solely for
)
So what? Is DC solely for fast computers, or do you object to waiting 6 more hours to see what credit you get granted? I bet it's more than what you usually claim. Thanks to a slower computer.
If all you want to do is crunch as many units as possible, in as little as possible time, then you have my blessings. But just saying that DC isn't for 'slow computers' only because you had the money to buy a fast processor & according hardware, goes a tad too far.
There isn't even a way to accurately compare timings between projects. I run a total of 6 different projects on my PC at this moment, divided between 2 seperate preferences. If you solely run Seti on your computer, how do you want to compare against one that runs 3 or more projects?
More so, why do we have this conversation on Einstein@Home? What is there to compare between Seti & Einstein? Either project is looking for something completely different.
(If you want to check my timings on Seti or on here, you can do so. The name's the same all over the projects.)
> So what? Is DC solely for
)
> So what? Is DC solely for fast computers, or do you object to waiting 6 more
> hours to see what credit you get granted? I bet it's more than what you
> usually claim. Thanks to a slower computer.
>
> If all you want to do is crunch as many units as possible, in as little as
> possible time, then you have my blessings. But just saying that DC isn't for
> 'slow computers' only because you had the money to buy a fast processor &
> according hardware, goes a tad too far.
>
> There isn't even a way to accurately compare timings between projects. I run a
> total of 6 different projects on my PC at this moment, divided between 2
> seperate preferences. If you solely run Seti on your computer, how do you want
> to compare against one that runs 3 or more projects?
>
> More so, why do we have this conversation on Einstein@Home? What is there to
> compare between Seti & Einstein? Either project is looking for something
> completely different.
>
> (If you want to check my timings on Seti or on here, you can do so. The name's
> the same all over the projects.)
>
what i will say,
a good Cruncher
no celerons, if you had some, be luck with this, ...
Greetings from Germany NRW
Ulli
[img]http://boinc.mundayweb.com/one/stats.php?userID=380 [/img]
Although i don't want to get
)
Although i don't want to get in the middle of anything here, Ageless is quite right on the Celeron D's. They're not your father's Celeron! LoL.
I don't know that i would specifically recommend a Celeron for crunching, especially if money is not a concern... but hey i'm a grad student here on a Ramen Noodle budget and this was the most bang for my buck. It runs at 533Mhz FSB with 256L2 cache, crunches 24/7, has never once frozen (whatever that's all about), and it does an Einstein unit in 8.5 hours.
Paul's excellent documentation provides some processor comparisons here. My apparently 'unworthy' Celeron crunches both Einstein and SETI in about the same average time as a P4 2.8Mhz that has twice the L2 cache and dual-channel memory to boot.
X-bit Labs did a thorough test of the Celeron D's here and rated them very highly.
All i can say is that i've been extremely happy with mine :)
Happy Crunching!
Dig
I'll go for the 2.8Ghz (I
)
I'll go for the 2.8Ghz (I take Digger meant that one, not its way lower MegaHertz brother ;)) once I have the change (in dimes...) (and chance ;)). At €96,- it's cheaper than buying a new motherboard and getting a new CPU, plus new memory.
You who win the lottery every week may not notice this, but a complete mobo + CPU + new memory is quite expensive. Try living on a €1099,- monthly income once. You may notice it's hard!
As for Sir Ulli, you don't need to quote all my text each time, then post one line at the bottom. I fricking knew what I posted above. Else I can look it up. Since you think so much of Seti, why not follow the requested Quoting in forums rules there and here?
> Celerons and Distributing
)
> Celerons and Distributing Computing, Two worlds a get together, a Celeron is
> not any good Idea for Crunching, try a P4 3.0 with HT or somethink...else
What worries me about comments like "a Celeron is not any good Idea for Crunching" is that it might put people off donating their spare cycles. I am sure the project would rather have a donation from a Celeron than no donation at all.
The very best computer to use for a distributed computing (DC) project is the one you already have.
Personally I have a couple of 700MHz boxes, I got two complete systems second-hand for less than the cost of one cheap printer new -- why would I want to spend more? Some people will no doubt figure E@H/SETI/etc into their buying decision, but I did not, and nor do the vast majority of those donating time. All those donations are welcome to the project.
If you have a Celeron, donate its spare capacity, don't feel excluded. Similarly if you have a VIA C3 or whatever.
If you are buying a new (or second hand!) computer, make your choice according to the primary mission of the box (games? business? study? etc) and whatever you buy on that basis, the box will also be useful to various DC projects. Unless you are very rich, or unless DC counts as your main hobby, please don't go out and buy a computer according to how well it runs DC.
~~gravywavy
> The very best computer to
)
> The very best computer to use for a distributed computing (DC) project is the
> one you already have.
Well said ...
People buy computers for many reasons. One of the primary discriminators is the cost of the machine, and if they are buying it primarily as a DC machine that drives the selection towards higher cost machines.
If, on the other hand, finances are an issue (as with most people), then this is going to drive buyers towards a different class of machine.
As an example, the last computer I bought was a Dell dual-Xeon (3.4 GHz, with 2M Cache) instead of a P4 EE mostly because the same dollars bought a machine that can do 4 Work Units at a time even though the speed was lower than the P4 EE, thusly increasing the throughput.
My next computer will be the next generation Power Mac (or what ever name they give it) and though I hope it will be a dual processor with dual cores, regardless of what it actually comes with, that will be my next workstation.
Like I said, many reasons to buy a specific computer ...