Also, it depends on what you have running in the background....given that you obviously know so much about CPU architectures, you must know that the number of context switches the processor has to do will severly effect your performance. Further mroe, there are multiple version of the P4, with differing pipieline lengths. Depending on your code, the newer P4 with its longer pipelines could be mis predicting too foten, and having to clear the pipeline, damaging your performance.
And many argue that the PPC archiecture is not inherently better than the P4.
Message Edited by mattcowger on 05-20-2004 11:29 PM
First of all, its a mobile CPU, as you mentioned the architecture is slightly different and the clockspeeds don't match up.
Secondly, it can depend on you other specs. What speed hard drive do you have on the laptop, how much RAM and what speed. Compare this to your desktop. In those tests they won't make too much difference but it will be some.
I would agree with the post above in that you should check what is running in the background, to gain speed. I have only 3 things starting up out of 14 and my sysytem runs fast. also check the services too. you might want to check out a little program called speedswitchxp witch lets YOU manually determine what the CPU does, for example I have it set to Max Power when im plugged in witch runs it at 3.06GHz all the time, and it is set to max Battery when on battery so it runs at 1.6GHZ all the time. This might narrow the performance gap for ya.
Yeah, I guess that's the case, due to the different architectures you can't compare the Pentium-4 to the Mobile Pentium-4 in terms of clock speeds. And you can't really put a desktop processor inside a notebook since the samller and more compact machine wouldn't be able to handle the high power consumption and heat radiation.
The ideal would be to have a processor for notebooks with the same performance as a desktop processor. But there's no chance of that with Intel, since over the years all they have been doing for desktop CPUs is work on a single architecture and push the megahertz or gigahertz as far as possible. And what you get then is fast processors that radiate and heat up like crazy making them unsuitable for laptop computers (that's where the weaker Mobile Pentium and Pentium-M processors come in).
I think it's time for Intel to start working on some new, more efficient processor designs rather than showing off their high clock speeds.
I think it's time for Intel to start working on some new, more efficient processor designs rather than showing off their high clock speeds.
You mean a new processor like Banias or Dothan (Pentium M)?
Also, based on some information I've read at anandtech.com and tom's, Intel is currently developing a new processor designed to be used in both Desktops and Notebooks. Think of a Pentium M with a desktop version.
I was talking about new processor architectures for desktops that would allow high performance but be also suitable to be put on a notebook. I know that Intel has developped new processors like Pentium-M for mobile computing, but that's only because they had to ... desktop pentiums aren't suitable for laptops, and so far Pentium-M or Mobile Pentium 4 just don't do the job (are much weaker then the desktop Pentium processors).
You say that Intel is making a Pentium-M version for desktops:
If this means that they want to produce desktop processors that will radiate less and be more power efficient by compromising performance, I think that is completely crazy ...
If it means that they want to come up with a processor that is as powerfull as a desktop Pentium 4, but whose archotecture allows it to be installed on notebooks, then we're talking!!!
Thats the whole point of the Pentium M! Its not less powerful than the P4, and doesn't have significantly less performance, but does use a quarter of the power and produces far less heat. You say Intel should stop pushing the MHz myth, and thats exactly what they did with Banias/Dothan - they are reletivly low MHz, but still ahve equiavlent performance.
Maybe for MATLAB thats true, but I regularly processlarge matrices for my research (though mine tend to be 3D) with code written in Java and C, and on my PM 1.4, I get about the same speeds as a P4 2.6. Are you matrices FP, or integer based - mine are FP....
When you perform raw and long computations then comparing the performance of a Pentium 4 to Pentium-M is like night and day. In my case I have to run long Matlab routines for my work, where up to 100x100 matrices need to be multiplied and divided inside repeated loops. So it's very wrong to say that the Pentium 4 performs only slightly better than the Pentium M: computing times are up to 10 times slower.
Whatever their architectural differences, simply speaking I don't believe that a desktop P4 2.4GHz can be much faster in number crunching than a mobile P4 3.2GHz when running on otherwise similarly configured boxes.
First, we still don't know basic specs of dinamo02's desktop and notebook: RAM sizes and speeds, HDD speeds, OS and Matlab versions, services and programs running in the background, etc. (if their hardware or software differ significantly, we may be comparing apples to oranges).
I would first check if there isn't something basically wrong with the performance of dinamo02's notebook. So dinamo02, if you provided your specs and some benchmark results (e.g., from Sandra), maybe other I5150 users could give comparable results. At present, your conclusions about the superiority of desktop P4 vs mobile P4 may be premature.
Second, since the issue of performance of P4 vs Pentium M has been raised, you may take a look at some benchmarks in
I've got a P4 2.53 Dell tower and have had the chance to run comparison tests on a compute-bound program on two Toshiba laptops, with 2.4 and 2.8 mobile P4 similar to the processor used in the Inspiron 5150. As you would expect, the 2.4 laptop is a little slower than the Dell 2.53 tower and the 2.8 laptop is a little faster. This is on tasks that are compute-intensive - for anything that uses the hard drive or video graphics much, the Dell tower is faster because it has a faster hard drive and graphics board.
The 5150 laptop with 3.2 chip should definitely be faster or at least the same speed as any ordinary 2.4 4550 desktop for compute-intensive jobs. Maybe the test is not long enough since it only takes one second on the 2.4 desktop. Possibly you are seeing a lot of time being used just to load the program from the hard drive.
A useful thing is to do Control-Alt-Delete to bring up the Windows Task Manager - click the Processes tab and then click the CPU on the CPU column to sort that information and see which processes are using time and how much they are working. The System Idle Process is what XP runs to kill time when it has nothing else to do.
What we are really discussing is not whether the application on compute, disk of bus bound, but what kind of compute bound it is. The FP unit on the Pentium M is wicked fast, but the integer ALU is a bit slower....all of that will affect applications differently.
Whatever their architectural differences, simply speaking I don't believe that a desktop P4 2.4GHz can be much faster in number crunching than a mobile P4 3.2GHz when running on otherwise similarly configured boxes.
First, we still don't know basic specs of dinamo02's desktop and notebook: RAM sizes and speeds, HDD speeds, OS and Matlab versions, services and programs running in the background, etc. (if their hardware or software differ significantly, we may be comparing apples to oranges).
I would first check if there isn't something basically wrong with the performance of dinamo02's notebook. So dinamo02, if you provided your specs and some benchmark results (e.g., from Sandra), maybe other I5150 users could give comparable results. At present, your conclusions about the superiority of desktop P4 vs mobile P4 may be premature.
Second, since the issue of performance of P4 vs Pentium M has been raised, you may take a look at some benchmarks in
Well I'm pretty much convinced by now that desktop processors outperform laptop processors in general (not only in my case with the P4 2.4GHz and the Mobile P4 3.2GHz). But hey if you could prove me otherwise and show that there's something wrong with the performance of my notebook than more power to you. Here are the specs for both systems:
Desktop Dimension 4550
Processor: P4 2.4GHz, 533MHz bus speed, 8k level 1 cache, 512K level 2 cache
RAM: DDR SDRAM 512MB at 333MHz
HDD: 120GB, 7200rpm
OS: Windows XP home edition
Running background processes:
Notebook Inspirion 5150
Processor: Mobile P4 with HT 3.2GHz, 533MHz bus speed, 512K level 2 cache
RAM: DDR SDRAM 512MB at 333MHz
HDD: 40GB, 4200rpm
OS: Windows XP home edition
Running background processes:
(the Power Scheme on the Inspirion is set to "Always On" so that the CPU runs always at the max 3.19GHz)
I'm running the same version of Matlab on both systems (Matlab R12 installed from the same disk).
I ran many different routines and got roughly the same results. Here's one of them:
One thousand AxB multiplications where A and B are both 100x100 floating point matrices generated randomly at each of the 1000 iterations.
I ran this code many times and it runs on the Desktop on average in 5.2 seconds and in 22.4 seconds on the Laptop.
mattcowger
2.6K Posts
0
May 21st, 2004 05:00
And many argue that the PPC archiecture is not inherently better than the P4.
Message Edited by mattcowger on 05-20-2004 11:29 PM
sakor1
2.2K Posts
0
May 21st, 2004 05:00
First of all, its a mobile CPU, as you mentioned the architecture is slightly different and the clockspeeds don't match up.
Secondly, it can depend on you other specs. What speed hard drive do you have on the laptop, how much RAM and what speed. Compare this to your desktop. In those tests they won't make too much difference but it will be some.
stu
connorc04
36 Posts
0
May 21st, 2004 13:00
dinamo02
5 Posts
0
May 21st, 2004 13:00
The ideal would be to have a processor for notebooks with the same performance as a desktop processor. But there's no chance of that with Intel, since over the years all they have been doing for desktop CPUs is work on a single architecture and push the megahertz or gigahertz as far as possible. And what you get then is fast processors that radiate and heat up like crazy making them unsuitable for laptop computers (that's where the weaker Mobile Pentium and Pentium-M processors come in).
I think it's time for Intel to start working on some new, more efficient processor designs rather than showing off their high clock speeds.
kwest2
123 Posts
0
May 21st, 2004 14:00
You mean a new processor like Banias or Dothan (Pentium M)?
Also, based on some information I've read at anandtech.com and tom's, Intel is currently developing a new processor designed to be used in both Desktops and Notebooks. Think of a Pentium M with a desktop version.
Message Edited by kwest2 on 05-21-2004 10:40 AM
dinamo02
5 Posts
0
May 21st, 2004 16:00
I was talking about new processor architectures for desktops that would allow high performance but be also suitable to be put on a notebook. I know that Intel has developped new processors like Pentium-M for mobile computing, but that's only because they had to ... desktop pentiums aren't suitable for laptops, and so far Pentium-M or Mobile Pentium 4 just don't do the job (are much weaker then the desktop Pentium processors).
You say that Intel is making a Pentium-M version for desktops:
If this means that they want to produce desktop processors that will radiate less and be more power efficient by compromising performance, I think that is completely crazy ...
If it means that they want to come up with a processor that is as powerfull as a desktop Pentium 4, but whose archotecture allows it to be installed on notebooks, then we're talking!!!
mattcowger
2.6K Posts
0
May 21st, 2004 16:00
mattcowger
2.6K Posts
0
May 21st, 2004 17:00
dinamo02
5 Posts
0
May 21st, 2004 17:00
Kiwiel
101 Posts
0
May 22nd, 2004 00:00
Whatever their architectural differences, simply speaking I don't believe that a desktop P4 2.4GHz can be much faster in number crunching than a mobile P4 3.2GHz when running on otherwise similarly configured boxes.
First, we still don't know basic specs of dinamo02's desktop and notebook: RAM sizes and speeds, HDD speeds, OS and Matlab versions, services and programs running in the background, etc. (if their hardware or software differ significantly, we may be comparing apples to oranges).
I would first check if there isn't something basically wrong with the performance of dinamo02's notebook. So dinamo02, if you provided your specs and some benchmark results (e.g., from Sandra), maybe other I5150 users could give comparable results. At present, your conclusions about the superiority of desktop P4 vs mobile P4 may be premature.
Second, since the issue of performance of P4 vs Pentium M has been raised, you may take a look at some benchmarks in
http://www.x86-secret.com/popups/printarticle.php?id=104
datsun240z
93 Posts
0
May 22nd, 2004 23:00
I've got a P4 2.53 Dell tower and have had the chance to run comparison tests on a compute-bound program on two Toshiba laptops, with 2.4 and 2.8 mobile P4 similar to the processor used in the Inspiron 5150. As you would expect, the 2.4 laptop is a little slower than the Dell 2.53 tower and the 2.8 laptop is a little faster. This is on tasks that are compute-intensive - for anything that uses the hard drive or video graphics much, the Dell tower is faster because it has a faster hard drive and graphics board.
The 5150 laptop with 3.2 chip should definitely be faster or at least the same speed as any ordinary 2.4 4550 desktop for compute-intensive jobs. Maybe the test is not long enough since it only takes one second on the 2.4 desktop. Possibly you are seeing a lot of time being used just to load the program from the hard drive.
A useful thing is to do Control-Alt-Delete to bring up the Windows Task Manager - click the Processes tab and then click the CPU on the CPU column to sort that information and see which processes are using time and how much they are working. The System Idle Process is what XP runs to kill time when it has nothing else to do.
mattcowger
2.6K Posts
0
May 23rd, 2004 03:00
dinamo02
5 Posts
0
May 23rd, 2004 04:00
Well I'm pretty much convinced by now that desktop processors outperform laptop processors in general (not only in my case with the P4 2.4GHz and the Mobile P4 3.2GHz). But hey if you could prove me otherwise and show that there's something wrong with the performance of my notebook than more power to you. Here are the specs for both systems:
Desktop Dimension 4550
Processor: P4 2.4GHz, 533MHz bus speed, 8k level 1 cache, 512K level 2 cache
RAM: DDR SDRAM 512MB at 333MHz
HDD: 120GB, 7200rpm
OS: Windows XP home edition
Running background processes:
Notebook Inspirion 5150
Processor: Mobile P4 with HT 3.2GHz, 533MHz bus speed, 512K level 2 cache
RAM: DDR SDRAM 512MB at 333MHz
HDD: 40GB, 4200rpm
OS: Windows XP home edition
Running background processes:
(the Power Scheme on the Inspirion is set to "Always On" so that the CPU runs always at the max 3.19GHz)
I'm running the same version of Matlab on both systems (Matlab R12 installed from the same disk).
I ran many different routines and got roughly the same results. Here's one of them:
One thousand AxB multiplications where A and B are both 100x100 floating point matrices generated randomly at each of the 1000 iterations.
I ran this code many times and it runs on the Desktop on average in 5.2 seconds and in 22.4 seconds on the Laptop.
Any comments or suggestions are welcome.