CPU 3.0 HT vs. Pentium M 1.8 (745)

[arthur]

I have Sony VAIO GRT280ZG with Pentium 4 3.0Gz with HT. Looking for Thinkpad T42 with Pentium 4 1.8Gz (745) 2 Mb cache.
Just wonder, how those computer are related to each other. Will be they at least close in speed, performance? Thanks.

[csv96]

Almost. The Pentium M 755 is a closer rival to the 3.0ghz Pentium 4 though. A rough estimate of performance is to multiply the Pentium M clock speed by 1.5. So the 1.7ghz Pentium M should be about the same as a 2.55ghz Pentium 4. A 2.0ghz Pentium M should be about the same speed as a 3.0ghz Pentium 4.

http://www.intel.com/products/benchmarks/notebook/?

[CantStopNow]

For one thing, the Pentium 4 looks like it is a 39W part running at its slowest speed of 1.6MHz (it burns 90W at full speed!). Dothan is a much lower power part (7.5W @ 600MHz / 21W @ full speed). The P4-M is more of a desktop replacement part. The Pentium M parts are more for the light & long lasting battery models. At 40W for the processor alone, you'd go through the T42's battery in under 1.5 hrs. To get longer battery life for the P4, you have to carry around a larger & heavier battery.

On top of that, the difference in clock rates is a little complicated. When the Pentium 4 architecture first came out, P4 parts would perform about roughly on par with P3 parts of 1/2 the clock rate. This is due to the longer pipeline in the P4. I think this has improved a little, especially with HT. Just don't discount the Pentium M because of the slower clock rate - keep in mind that Intel has decided to switch their main processor line over to the Pentium M architecture.

[Conmee]

The other issues to keep in mind are that the chipsets, FSBs, memory clocking, are all optimized to consume less power and generate less heat in a notebook computer. While the sheer processing power of a Pentium M 2.0GHz Dothan may be 'equivalent' to a 3GHz P4 or 2.6GHz P-4M, the real-world performance between notebook and desktop is no contest. The P4 with Hyperthreading, 800MHz FSB, PC3200 RAM, and the ability of the P4 to run apps optimized for iSSE2 and Hyperthreading (which Pentium M Banias/Dothan don't execute as quickly, or support, respectively) all play a huge factor in actual system speed.

So the more accurate comparison/statement might be... IF you put a 3GHz P4 CPU (and to a lesser extent a P4-M) into the same T42p as a 2.0GHz Dothan, THEN you would see 'equivalent' performance results between the two. But put a T42p 2.0GHz Dothan up against a 3GHz or 3.6GHz P4 desktop/workstation, and it's no contest.

Of course, the biggest advantage, as already mentioned, is that the processing power-to-power consumption and heat generation ratio with the Pentium M is clearly superior, without question. That's why an Alienware or other gaming notebook with a 3 GHz CPU and 533/800MHz FSB and multiple cooling fans will run for about 1.5-2 hours on an 8-Cell battery, while you can get pretty close to the same performance for daily use on a T42p Pentium M 2GHz cpu and get closer to or more than 6 hours on a 9-Cell battery, without the fans and heat (and no, that extra 'cell' on the T42p battery doesn't account for the 4-5 hour runtime difference.. lol). Oh, and the Alienware 3GHz systems will weigh in at about 8 lbs. without a battery too... lol... not to pick on Alienware, which are killer machines designed for high gaming or desktop replacement performance, but just to point out that the desktop P4 CPus offer different configuration, performance, and aesthetic characteristics from a Pentium-M. So it's really difficult to get much value out of a simple rule of thumb calculation when comparing different CPUs on different platforms (although I do it all the time...lol).

Daniel

[AtmosMan]

I did some research on this a few days ago. I looked up probably 100+ benchmarks for Pentium M - Pentium 4 notebook systems. I found a very genaral conversion chart:

Pentium M Pentium 4

1.6 GHz = 2.66 GHz
1.7 GHz = 2.8 GHz
1.8 GHz = 3.0 GHz

IMO I find multiplying the Pentium M speed by 1.65 gives a fairly accurate equivalent to (non-HT) Pentium 4 speed. You really can't compare a Pentium M to a HT Pentium 4 because the Pentium M isn't built with HT as a main feature of the processor.

Remember that the processor is only one part of a computer. Other components like memory, hard drive and GPU will affect the performance greatly.

Here is one example I found on CNet that shows how hard it is to compare processor performance:

A Dell Latitude D800 with 2GHz P-m, 512Mb 266MHZ RAM, 80Gb 5400rpm HDD, and 128MB Nvidia GeForce FX Go5650 scored a 212 on BAPCo MobileMark 2002. An IBM Thinkpad T42 with 1.7GHz P-m, 512 Mb 333Mhz RAM, 40Gb HDD, and 32MB ATI Mobility Radeon 7500 scored a 208 on BAPCo MobileMark 2002. With a slower processor, and a less powerful GPU, the IBM is able to rival the Dell computer with the fastest processor available.

Intel's benchmarks is a good place to start looking, but try to find more benchmarks to give you a broader view of the world of notebook computer performance.

[akerman]

How did you come to that AtmosMan? I have done my own calculations, based on extensive benchmarks by Anandtech, and a week ago TechReport's benchmarks - I came to a conversion factor of 1.46 for the anandtech benchmarks, and 1.47 for the tech report benchmarks.

[AtmosMan]

I guess I should say "system performance" instead of just processor performance. Sorry 'bout that. Just look up on Cnet some of the BAPCo MobileMark2002 performance ratings (MobileMark is used only when on battery power):

Pentium 4 Systems:

http://reviews.cnet.com/Dell_Inspiron_5 ... ml?tag=top

http://reviews.cnet.com/Toshiba_Satelli ... ml?tag=top

Pentium M Systems (Dothan):

http://reviews.cnet.com/Dell_Latitude_D ... ml?tag=top

Just by looking at the general mobile performance of the 1.7GHz - 2.0GHz Pentium M systems, they beat out the 2.8GHz - 3.06GHz. There are several more examples like this on CNet. I keep finding that the 1.8GHz P-M systems are performing just as well, or better than the 3.06GHz P-4 systems. Maybe I'm just doing this completely wrong! Well, that's why this is a forum of lots of people giving advice!

[arthur]

sorry, i did not specify VAIO GRT280ZG is a notebook. so all comparison more or less relevant. thanks a lot to all guys.

[akerman]

Yeah AtmosMan, those figures can't really be used for cross-platform comparison. For one, they are as you say, system benchmarks - and a system is made up of a lot more than just the cpu. Also, the mobilemark performance depends on how the manufacturer wants the notebook to perform - ie. they can choose between long batterylife, but low performance, or short batterylife, and high performance - most users would prefer the low performance + long batterylife combination, when running on batteries - this is also the formula IBM uses.

Look at the benchmarks for the pentium-m systems - it's not surprising, that the best performers, are also the ones with the shortest batterylife.