core 2 duo and P4 3.2 E

I'v recently upgraded my system form a P 4 3.2 E and a MSI 875 Pe fsir with
512+512M DDR to a core 2 duo 8500 and a Asus P5Q p45 with 2+2M DDR2. I'v
tried to look in the net for some comparison of this tow processors with no
results. I would like to know how must faster the core 2 duo is related to
P4 3.2 E ?

Thanks in advance.

Manuel.

MS wrote:
I'v recently upgraded my system form a P 4 3.2 E and a MSI 875 Pe fsir
with 512+512M DDR to a core 2 duo 8500 and a Asus P5Q p45 with 2+2M
DDR2. I'v tried to look in the net for some comparison of this tow
processors with no results. I would like to know how must faster the
core 2 duo is related to P4 3.2 E ?

Thanks in advance.

Manuel.

You can look for benchmark results on http://www.hwbot.org/
but the site is almost impossible to navigate. They have
SuperPI results for example.

My old P4 running at 3.1GHz, would do SuperPI 1 million digits
in about 45 to 50 seconds. If I run that benchmark on a Core2
overclocked to 3.4GHz, the number drops to about 18 seconds.
So that kind of benchmark is the most impressive comparison.
HWbot has those kind of data, but it would take me at least
30 minutes of clicking on the web page, to offer a
comparison.

SuperPI for download.

http://www.xtremesystems.com/pi/

Paul

Somewhere on teh intarwebs "Paul" typed:
MS wrote:
I'v recently upgraded my system form a P 4 3.2 E and a MSI 875 Pe
fsir with 512+512M DDR to a core 2 duo 8500 and a Asus P5Q p45 with
2+2M DDR2. I'v tried to look in the net for some comparison of this
tow processors with no results. I would like to know how must faster
the core 2 duo is related to P4 3.2 E ?

Thanks in advance.

Manuel.

You can look for benchmark results on http://www.hwbot.org/
but the site is almost impossible to navigate. They have
SuperPI results for example.

My old P4 running at 3.1GHz, would do SuperPI 1 million digits
in about 45 to 50 seconds. If I run that benchmark on a Core2
overclocked to 3.4GHz, the number drops to about 18 seconds.
So that kind of benchmark is the most impressive comparison.
HWbot has those kind of data, but it would take me at least
30 minutes of clicking on the web page, to offer a
comparison.

SuperPI for download.

http://www.xtremesystems.com/pi/

Bear in mind that, unless SuperPI has changed since I used it, it only uses
one of the available cores.

Therefore, if you're looking for comparisons of CPU power, you can
essentially halve the C2D score to compare it with a single core CPU.
--
Shaun.

"~misfit~" wrote in message
...
Somewhere on teh intarwebs "Paul" typed:
MS wrote:
I'v recently upgraded my system form a P 4 3.2 E and a MSI 875 Pe
fsir with 512+512M DDR to a core 2 duo 8500 and a Asus P5Q p45 with
2+2M DDR2. I'v tried to look in the net for some comparison of this
tow processors with no results. I would like to know how must faster
the core 2 duo is related to P4 3.2 E ?

Thanks in advance.

Manuel.

You can look for benchmark results on http://www.hwbot.org/
but the site is almost impossible to navigate. They have
SuperPI results for example.

My old P4 running at 3.1GHz, would do SuperPI 1 million digits
in about 45 to 50 seconds. If I run that benchmark on a Core2
overclocked to 3.4GHz, the number drops to about 18 seconds.
So that kind of benchmark is the most impressive comparison.
HWbot has those kind of data, but it would take me at least
30 minutes of clicking on the web page, to offer a
comparison.

SuperPI for download.

http://www.xtremesystems.com/pi/

Bear in mind that, unless SuperPI has changed since I used it, it only
uses one of the available cores.

Therefore, if you're looking for comparisons of CPU power, you can
essentially halve the C2D score to compare it with a single core CPU.
--
Shaun.
SuperPi 1.1, the newest version shows it is using all 4 cores on my i7 920
Shaun. At least in CoreTemp, all the cores heat up evenly which to me points
to it using all of them.....:-)..... My results at 3.7Ghz for 1m is
10secs...... for 32m is 10min 12secs.


Ed

Ed Medlin wrote:

"~misfit~" wrote in message
...
Somewhere on teh intarwebs "Paul" typed:
MS wrote:
I'v recently upgraded my system form a P 4 3.2 E and a MSI 875 Pe
fsir with 512+512M DDR to a core 2 duo 8500 and a Asus P5Q p45 with
2+2M DDR2. I'v tried to look in the net for some comparison of this
tow processors with no results. I would like to know how must faster
the core 2 duo is related to P4 3.2 E ?

Thanks in advance.

Manuel.

You can look for benchmark results on http://www.hwbot.org/
but the site is almost impossible to navigate. They have
SuperPI results for example.

My old P4 running at 3.1GHz, would do SuperPI 1 million digits
in about 45 to 50 seconds. If I run that benchmark on a Core2
overclocked to 3.4GHz, the number drops to about 18 seconds.
So that kind of benchmark is the most impressive comparison.
HWbot has those kind of data, but it would take me at least
30 minutes of clicking on the web page, to offer a
comparison.

SuperPI for download.

http://www.xtremesystems.com/pi/

Bear in mind that, unless SuperPI has changed since I used it, it only
uses one of the available cores.

Therefore, if you're looking for comparisons of CPU power, you can
essentially halve the C2D score to compare it with a single core CPU.
--
Shaun.
SuperPi 1.1, the newest version shows it is using all 4 cores on my i7
920 Shaun. At least in CoreTemp, all the cores heat up evenly which to
me points to it using all of them.....:-)..... My results at 3.7Ghz for
1m is 10secs...... for 32m is 10min 12secs.


Ed

What do you see in Task Manager ? I thought the one in the above
link (xtremesystems), is single threaded. One of the reasons I like
it as a bench, is it is a conservative means of predicting the speedup.
In other words, if all my software is single threaded, my benchmark won't
be predicting speedup factors I'll never see in practice. For example,
the only thing I have on the computer right now that uses two
cores, is Windows Movie Maker. So what I can count on, is ~45/18
as a speedup factor. (I don't currently leave it overclocked,
so my speedup is a bit less than that.)

That particular program apparently has no publicly available source.
The above executable, has been maintained by editing the binary
with a hex editor. The change to improve the timing resolution,
the anti-cheat features, were all added in machine code.

http://en.wikipedia.org/wiki/Superpi

For a multithreaded test program, you could try the Prime95 Torture
Test. Since the directory on the original server tends to be closed
now, you can get a copy here. If this doesn't trip the power limiter
on your board, nothing will.

http://majorgeeks.com/Prime95_d4363.html

Paul

Somewhere on teh intarwebs "Ed Medlin" typed:
"~misfit~" wrote in message
...
Somewhere on teh intarwebs "Paul" typed:
MS wrote:
I'v recently upgraded my system form a P 4 3.2 E and a MSI 875 Pe
fsir with 512+512M DDR to a core 2 duo 8500 and a Asus P5Q p45 with
2+2M DDR2. I'v tried to look in the net for some comparison of this
tow processors with no results. I would like to know how must
faster the core 2 duo is related to P4 3.2 E ?

Thanks in advance.

Manuel.

You can look for benchmark results on http://www.hwbot.org/
but the site is almost impossible to navigate. They have
SuperPI results for example.

My old P4 running at 3.1GHz, would do SuperPI 1 million digits
in about 45 to 50 seconds. If I run that benchmark on a Core2
overclocked to 3.4GHz, the number drops to about 18 seconds.
So that kind of benchmark is the most impressive comparison.
HWbot has those kind of data, but it would take me at least
30 minutes of clicking on the web page, to offer a
comparison.

SuperPI for download.

http://www.xtremesystems.com/pi/

Bear in mind that, unless SuperPI has changed since I used it, it
only uses one of the available cores.

Therefore, if you're looking for comparisons of CPU power, you can
essentially halve the C2D score to compare it with a single core CPU.

SuperPi 1.1, the newest version shows it is using all 4 cores on my
i7 920 Shaun. At least in CoreTemp, all the cores heat up evenly
which to me points to it using all of them.....:-)..... My results at
3.7Ghz for 1m is 10secs...... for 32m is 10min 12secs.

Hi Ed. When I used it last (not sure of version, I'm on my laptop currently)
on my C2D both cores were used, both to ~50% and the result it gave was
about what I would have expected from a single core. I've also used it on a
mate's C2Q and it most certainly didn't use all cores to 100%. His C2Q @
3.2GHz had a slower time than my C2D @ 3.6GHz.

Cheers,
--
Shaun.

Hi Ed. When I used it last (not sure of version, I'm on my laptop
currently) on my C2D both cores were used, both to ~50% and the result it
gave was about what I would have expected from a single core. I've also
used it on a mate's C2Q and it most certainly didn't use all cores to
100%. His C2Q @ 3.2GHz had a slower time than my C2D @ 3.6GHz.

Cheers,
--
Shaun.

I wonder if it has anything to do with all 4 cores being on the same chip on
the i7? After messing around with various programs I found that Lavasys'
Everest is by far the best program to stress all four cores a ton and also
for benchmarking memory and CPU. It goes far beyond what I believe any
processor could be stressed in any real application. It even made me back my
processor down to 3.6Ghz from 3.7Ghz because it started to do some
throttleing when it actually got the processor up to 100C. Yep......100C
which is the Tjunction so that seems to fit. 8-12 instances of Orthos didn't
get anywhere near that. The strange thing is that even at those temps there
were no failures of any kind. It stayed rock stable. Temps during gaming and
other benchmarks like 3DMark 06 are much more in line with the C2Ds and
Quads staying at temps that don't worry me.....:-). At 3.6Ghz it takes much
less of a vcore increase and just might run stable at default vcore. I will
do some testing just to see how high it will go at default. I have only gone
to 3.2Ghz with vcore on 'Auto' with no issues. As I went higher, I used
voltages that others had success with rather than to check myself. I usually
don't do that and it really isn't the correct method since individual
processors, even the same batch, can all be different as far as vcore and
stability goes. Memory perfomance is great with the triple channel Corsair
DDR3 XMS 1333 @ 1440 memory with read being 14896mb/s, write at 12992mb/s
and latency at 36.7ns.


Ed

Ed Medlin wrote:
I wonder if it has anything to do with all 4 cores being on the same chip
on the i7?

I just tested my Q9550 @ 3.2 Ghz (1:1 with 400 clock) with Super-Pi and
Super-Pi mod. It does appear to use all cores, but total CPU utilization was
only about 27 percent. As I recall, memory speed and timing were always
a big factor with this program so it must be memory bandwidth limited. I
tried setting he core affinity to a single core and it shaved 2.5 seconds from
the time (17.7 s / 15.2 s). The result using two cores was comparable to
that using all four. It would seem that it was about at the limitation of the
memory subsystem and the multithreading just added more overhead. I
really must reseat my CPU cooler one of these days so I can crank it up!


Snip

Memory perfomance is great with the triple channel Corsair DDR3 XMS 1333 @ 1440 memory with read being 14896mb/s,
write at
12992mb/s and latency at 36.7ns.

Holy crap! Try setting the core affinity at *two* cores, since your CPU
utilization was at 50 percent, and see what you get...

Memory perfomance is great with the triple channel Corsair DDR3 XMS 1333
@ 1440 memory with read being 14896mb/s, write at
12992mb/s and latency at 36.7ns.

Holy crap! Try setting the core affinity at *two* cores, since your CPU
utilization was at 50 percent, and see what you get...

When set to 2 cores/no HT memory performance went up from 14896mb/s read to
16827mb/s.....write up to 13939mb/s.


Ed

Ed Medlin wrote:

When set to 2 cores/no HT memory performance went up from
14896mb/s read to 16827mb/s.....write up to 13939mb/s.

But- I mean't Super-Pi, since you weren't bandwidth limited as I was
with more than one core. Since I shaved 2.5 s by eliminating the
threading overhead, I wondered if you would see a similar gain by
eliminating two threads, as your CPU utilization was 50% to my 25%.