I'm running 7 P4 HT Computers Here at the Einstein Site, My slowest one a P4 3.06 will do 2 WU's every 10:25 & my fastest one a P4 3.4 will do 2 WU's every 9:05 ...
I also look, like Honza, which computer is the "best" for a certain project. For this a take another "important variable" with the question:" how often this Host was the canonical result in the resultstable?"
By this way I noticed a remarkeable difference on the PAH-project with two of my .puters.
For example: my P4 (M) 1.5 GHz (2MB L2 Cache)[a Acer Notebook, takes about 58 minutes to crunch a PAH-WU] is near twice as fast as my P4 3.4GHz HT [a Dell XPS, takes about 102 minutes/WU.
all other projects is near the same the Notebbook is a little faster then the XPS (per WU), only CPDN there the XPS is faster.
As you can see : Even with 2 hosts it isn't easy to decide....
EINSTEIN:
Notebook about 10 hours 12 minutes per WU(avg.)
Dell XPS about 11 hours 15 minutes per WU(avg.)
@Honza and Poorboy
Thanks for discussion and advises (or statements).
That's very interesting...
It is a know fact that Pentium-M on a quite low CPU frequency has pretty good performance. Still, processing times you have on PAH is impressive comparing to high-end P4/3.4.
I'm promissed to have a machine based on Pentium-M next week for testing. It suppose to be a regular desktop (not a notebook) with low power consumption, low-noise or even noiseless computer. If i have time, i'll give it a try on BOINC.
> I also look, like Honza, which computer is the "best" for a certain project.
> For this a take another "important variable" with the question:" how often
> this Host was the canonical result in the resultstable?"
>
> By this way I noticed a remarkeable difference on the PAH-project with two of
> my .puters.
> For example: my P4 (M) 1.5 GHz (2MB L2 Cache)[a Acer Notebook, takes about 58
> minutes to crunch a PAH-WU] is near twice as fast as my P4 3.4GHz HT [a Dell
> XPS, takes about 102 minutes/WU.
>
> all other projects is near the same the Notebbook is a little faster then the
> XPS (per WU), only CPDN there the XPS is faster.
> As you can see : Even with 2 hosts it isn't easy to decide....
>
> EINSTEIN:
> Notebook about 10 hours 12 minutes per WU(avg.)
> Dell XPS about 11 hours 15 minutes per WU(avg.)
>
> @Honza and Poorboy
> Thanks for discussion and advises (or statements).
>
> greetz from Switzerland
> littleBouncer
>
>
> That's very interesting...
> It is a know fact that Pentium-M on a quite low CPU frequency has pretty good
> performance. Still, processing times you have on PAH is impressive comparing
> to high-end P4/3.4.
and it gives about 95% "canonical results" (what meam: "good results")
> I'm promissed to have a machine based on Pentium-M next week for testing. It
> suppose to be a regular desktop (not a notebook) with low power consumption,
> low-noise or even noiseless computer. If i have time, i'll give it a try on
> BOINC.
>
> > greetz from Switzerland
> > littleBouncer
> >
=====
important is with Pentium-M : 2L Cache of 2MB!!!
=====
One thing I recall from when I built my last machine (back when I had time to do such things), was that the competition between Athlon and the P3s was really hot. The P4s were just starting to roll out. The thing that caught my eye was that Athlon architecture (at that point in time - this is no commentary on the present) generally performed better in floating point calculations than the P3 chip. It seems to me that in a situation like that, where two chips may perform at the same frequency but one can do a lot more of the high-load calculations like floating point in a shorter amount of time, it would perform better in computationally intensive projects like this.
@ The Ox, I'll give the AMD their Credit when it comes to running a AMD CPU against a comparable P4 CPU with the HT turned off, the AMD wins hands down no doubt about it. But once you turn that HT Feature on in the P4 CPU's that are capable of it then the P4 kicks ass & takes names over a comparable speed AMD CPU ...
Let me give you an example of that, one of my computers I have is capable of running the BOINC Site WU's in 2 hr's flat with HT turned off, not really all that slow but not that impressive either. Now when I turn the HT Feature on in that same computer it can run 2 WU's in 2 1/2 hr's, thats a 37 1/2% improvement over running 2 WU's with HT turned off & a comparable speed AMD CPU simply can't match that ... :)
I can't argue with you on that point. I have a HT P4 in my primary machine now and it just screams. I'll be interested to see if AMD catches up in this category. I think I heard somewhere that they were working on a similar design for their own chips. If they could essentially hyper-thread their own architecture (which imho is superior to Intel's) I think they could give Intel an run for their money again.
Hmm.. so maybe there's still some life left in my dual p2-400 Xeon rig for this project since each of those processors has 2meg of L2 cache as well. :P
Can't hurt to try! :)
"Chance is irrelevant. We will succeed."
- Seven of Nine
I'm running 7 P4 HT Computers
)
I'm running 7 P4 HT Computers Here at the Einstein Site, My slowest one a P4 3.06 will do 2 WU's every 10:25 & my fastest one a P4 3.4 will do 2 WU's every 9:05 ...
I also look, like Honza,
)
I also look, like Honza, which computer is the "best" for a certain project. For this a take another "important variable" with the question:" how often this Host was the canonical result in the resultstable?"
By this way I noticed a remarkeable difference on the PAH-project with two of my .puters.
For example: my P4 (M) 1.5 GHz (2MB L2 Cache)[a Acer Notebook, takes about 58 minutes to crunch a PAH-WU] is near twice as fast as my P4 3.4GHz HT [a Dell XPS, takes about 102 minutes/WU.
all other projects is near the same the Notebbook is a little faster then the XPS (per WU), only CPDN there the XPS is faster.
As you can see : Even with 2 hosts it isn't easy to decide....
EINSTEIN:
Notebook about 10 hours 12 minutes per WU(avg.)
Dell XPS about 11 hours 15 minutes per WU(avg.)
@Honza and Poorboy
Thanks for discussion and advises (or statements).
greetz from Switzerland
littleBouncer
That's very interesting... It
)
That's very interesting...
It is a know fact that Pentium-M on a quite low CPU frequency has pretty good performance. Still, processing times you have on PAH is impressive comparing to high-end P4/3.4.
I'm promissed to have a machine based on Pentium-M next week for testing. It suppose to be a regular desktop (not a notebook) with low power consumption, low-noise or even noiseless computer. If i have time, i'll give it a try on BOINC.
> I also look, like Honza, which computer is the "best" for a certain project.
> For this a take another "important variable" with the question:" how often
> this Host was the canonical result in the resultstable?"
>
> By this way I noticed a remarkeable difference on the PAH-project with two of
> my .puters.
> For example: my P4 (M) 1.5 GHz (2MB L2 Cache)[a Acer Notebook, takes about 58
> minutes to crunch a PAH-WU] is near twice as fast as my P4 3.4GHz HT [a Dell
> XPS, takes about 102 minutes/WU.
>
> all other projects is near the same the Notebbook is a little faster then the
> XPS (per WU), only CPDN there the XPS is faster.
> As you can see : Even with 2 hosts it isn't easy to decide....
>
> EINSTEIN:
> Notebook about 10 hours 12 minutes per WU(avg.)
> Dell XPS about 11 hours 15 minutes per WU(avg.)
>
> @Honza and Poorboy
> Thanks for discussion and advises (or statements).
>
> greetz from Switzerland
> littleBouncer
>
>
> That's very
)
> That's very interesting...
> It is a know fact that Pentium-M on a quite low CPU frequency has pretty good
> performance. Still, processing times you have on PAH is impressive comparing
> to high-end P4/3.4.
and it gives about 95% "canonical results" (what meam: "good results")
> I'm promissed to have a machine based on Pentium-M next week for testing. It
> suppose to be a regular desktop (not a notebook) with low power consumption,
> low-noise or even noiseless computer. If i have time, i'll give it a try on
> BOINC.
>
> > greetz from Switzerland
> > littleBouncer
> >
=====
important is with Pentium-M : 2L Cache of 2MB!!!
=====
littleBouncer
@littleBouncer - thanks for
)
@littleBouncer - thanks for additional info on Pentium-M; a large cache may be a important issue.
> @littleBouncer - thanks for
)
> @littleBouncer - thanks for additional info on Pentium-M; a large cache may be
> a important issue.
>
You're wellcome.
happy crunching littleBouncer
Honza, LittleBouncer,
)
Honza, LittleBouncer, Poorboy,
One thing I recall from when I built my last machine (back when I had time to do such things), was that the competition between Athlon and the P3s was really hot. The P4s were just starting to roll out. The thing that caught my eye was that Athlon architecture (at that point in time - this is no commentary on the present) generally performed better in floating point calculations than the P3 chip. It seems to me that in a situation like that, where two chips may perform at the same frequency but one can do a lot more of the high-load calculations like floating point in a shorter amount of time, it would perform better in computationally intensive projects like this.
Just my two bits worth...
www.clintcollins.org - spouting off at the speed of site
@ The Ox, I'll give the AMD
)
@ The Ox, I'll give the AMD their Credit when it comes to running a AMD CPU against a comparable P4 CPU with the HT turned off, the AMD wins hands down no doubt about it. But once you turn that HT Feature on in the P4 CPU's that are capable of it then the P4 kicks ass & takes names over a comparable speed AMD CPU ...
Let me give you an example of that, one of my computers I have is capable of running the BOINC Site WU's in 2 hr's flat with HT turned off, not really all that slow but not that impressive either. Now when I turn the HT Feature on in that same computer it can run 2 WU's in 2 1/2 hr's, thats a 37 1/2% improvement over running 2 WU's with HT turned off & a comparable speed AMD CPU simply can't match that ... :)
@PoorBoy, I can't argue
)
@PoorBoy,
I can't argue with you on that point. I have a HT P4 in my primary machine now and it just screams. I'll be interested to see if AMD catches up in this category. I think I heard somewhere that they were working on a similar design for their own chips. If they could essentially hyper-thread their own architecture (which imho is superior to Intel's) I think they could give Intel an run for their money again.
www.clintcollins.org - spouting off at the speed of site
Hmm.. so maybe there's still
)
Hmm.. so maybe there's still some life left in my dual p2-400 Xeon rig for this project since each of those processors has 2meg of L2 cache as well. :P
Can't hurt to try! :)
"Chance is irrelevant. We will succeed."
- Seven of Nine