X40 processor speed any good? (M 738 1.4 Ghz) - advice neede

[ahammad]

How good or bad is this processor? I've never used anything close to it so I wouldn't know.

The system has a 4200 RPM drive (not so great) and 1024 mb RAM (pretty good). I know it depends on the purpose, so here's what I'll probably be doing: reading pdf files, everything using Microsoft Office, some C++ or perl/php programming and Matlab (and internet, of course).

I think it can handle it but I'm not sure about Matlab. Also, would multitasking be incredibly horrible? What do you guys think of it? What AMD or Pentium processor would it be comparable to?

Cheers.

[weepy]

it compares +- to an athlon 64 at the same frequency, but it is noticeably weaker in floating point (at least everest says so). It should fit your needs, but for matlab there might be better possibilities. Maybe the screen is a bit small for coding... but the keyboard rocks.

[mfbernstein]

How good or bad is this processor? I've never used anything close to it so I wouldn't know.

The system has a 4200 RPM drive (not so great) and 1024 mb RAM (pretty good). I know it depends on the purpose, so here's what I'll probably be doing: reading pdf files, everything using Microsoft Office, some C++ or perl/php programming and Matlab (and internet, of course).

I think it can handle it but I'm not sure about Matlab. Also, would multitasking be incredibly horrible? What do you guys think of it? What AMD or Pentium processor would it be comparable to?

Cheers.

For your purposes, the processor is probably the least significant issue (for light Matlab and coding, it's fine, for heavy Matlab it will likely disappoint). You'll want to max out the RAM (1.25 or 1.5GB). Your main bottleneck will be the slow hard drive, and there's no good solution to that yet.

Unless you need the ultimate lightweight notebook, why not go for an X31/32? You'll be able to fit in more RAM, and upgrade the HD to something faster/bigger if you need to.

[synchromesh]

it compares +- to an athlon 64 at the same frequency, but it is noticeably weaker in floating point (at least everest says so). It should fit your needs, but for matlab there might be better possibilities. Maybe the screen is a bit small for coding... but the keyboard rocks.

The slowest Athlon 64 was 1.8GHz, iirc. They never made a 1.4GHz model at least for desktops. And I'm pretty sure that even comparable speed A64 will trounce a Pentium M in most performance tests, not to mention it can run a 64-bit OS unlike P-M which is 32-bit only. P-M has lower power consumption. My main rig is running a dual core Opteron which is basically a fancy Athlon 64 with more overclocking headroom and more cache than regular A64s.

Aside from that, for most everyday tasks a P-M 1.4GHz is sufficient.

[weepy]

You may be right but i never said otherwise... I merely compared the performance clock for clock based on one benchmark .. I would be very careful with the word "trounce", thats very task-dependent.
In Super-Pi Pentium M is slightly better than equally clocked A64, in integer math (PassMark), Pentium M 1.73GHz has same score as A64 2.0Ghz... that it is weaker in fpu, we agreed already.

http://www.cpubenchmark.net/common_cpus.html

[synchromesh]

That's why I said in *most* performance tests. Anyway, those synthetic benchmarks are not to be trusted for accurate real life representation. And it also depends on whether you are referring to desktop or mobile Athlon 64.

[weepy]

Could you then specify "*most* performance tests"? Link or something? I think this is a very interesting debate, but i don't have the time to do in-depth googling. I opened again everests benchmark library and found pentium M not lacking in FPU power so much, even surpassing (desktop) athlon 64 in Julia, clk4clk. Please note, that Super PI and relevant Everest benches (zlib, julia, mandel...) are not synthetic but quite exact pattern of what a user, matlabber would execute.

I disagree that mobile vs desktop athlon differ in performance, theyre architecturally identical, having different voltage only. Even the power saving is identical to desktop.

[pxa270]

Weepy is right on this one. According to Anandtech, the Dothan P-M is clock for clock about equal to the A64 in most test. Of course, the P-M in the X40 is clocked pretty low, but a 2Ghz Dothan can match a 2Ghz A64 pretty well.

[synchromesh]

I looked at those CPU tests. First of all those are from 2004. Second, here: http://www.anandtech.com/cpuchipsets/sh ... =2342&p=20 they still admit that while it's close A64 is an overall winner even if by a few percent.

Another thing to keep in mind is that few things have changed since the time these reviews were written. Dual cores aside, there have been various socket changes including the socket 939 I'm running on now which is already obsolete anyway. There have also been some driver updates that increase the performance of A64 CPUs slightly.

Another thing overlooked in such a comparison is the fact that a *system* based on A64 is still going to be considerably faster even at stock speeds. A64 uses regular desktop RAM and regular desktop components which are faster in most cases than their laptop counterparts than P-Ms are restrained too. At the end, the user will use a whole system, not just the processor itself.

Of course, I will not get into overclocking and number of motherboard options that A64s present as opposed to P-M since there is hardly any comparison then but adhering to the original topic ie clock for clock, I still maintain A64 is slightly ahead. This is from somebody who used systems based on either extensively.

[weepy]

I looked at those CPU tests. First of all those are from 2004. Second, here: http://www.anandtech.com/cpuchipsets/sh ... =2342&p=20 they still admit that while it's close A64 is an overall winner even if by a few percent.

So much for trouncing in most benches. Thanks for proving my point . They use ancient 855 agpset and ddr 333 single channel memory which is really not the best they could have used...

Another thing to keep in mind is that few things have changed since the time these reviews were written. Dual cores aside, there have been various socket changes including the socket 939 I'm running on now which is already obsolete anyway. There have also been some driver updates that increase the performance of A64 CPUs slightly.

OT: As another extensive AMD user I would welcome some "driver" for increasing the performance of my desktop X2 3800+. Link appreciated .

Another thing overlooked in such a comparison is the fact that a *system* based on A64 is still going to be considerably faster even at stock speeds. A64 uses regular desktop RAM and regular desktop components which are faster in most cases than their laptop counterparts than P-Ms are restrained too. At the end, the user will use a whole system, not just the processor itself.

Of course, I will not get into overclocking and number of motherboard options that A64s present as opposed to P-M since there is hardly any comparison then but adhering to the original topic ie clock for clock, I still maintain A64 is slightly ahead. This is from somebody who used systems based on either extensively.

Mate, we really get away from topic right now. The question was about processor speed, not platform speed or o/c. We would have to make much broader talk on who is better by platform and though i am a firm amd fan and though their igp rock more and more, Intel is still my choice for mobile and the market confirms my choice (as does your own choice, seeing your signature).

I'm buying an X41t and I'm looking forward to benching it on my own versus my desktop amd *drool*

[synchromesh]

So much for trouncing in most benches. Thanks for proving my point . They use ancient 855 agpset and ddr 333 single channel memory which is really not the best they could have used...

I was very surprised myself. Of course with some chipset and software improvements over the years it's probably a bit different nowadays but I couldn't find any recent reviews.

OT: As another extensive AMD user I would welcome some "driver" for increasing the performance of my desktop X2 3800+. Link appreciated .

Just google for athlon 64 windows driver. You can also get the X2 driver this way, it helps.

Mate, we really get away from topic right now. The question was about processor speed, not platform speed or o/c. We would have to make much broader talk on who is better by platform and though i am a firm amd fan and though their igp rock more and more, Intel is still my choice for mobile and the market confirms my choice (as does your own choice, seeing your signature).

I'm buying an X41t and I'm looking forward to benching it on my own versus my desktop amd *drool*

One of the reasons I have an Intel CPU in my laptop is because IBM/Lenovo didn't make anything based on AMD processors. Altogether, Intel did get the performance crown back with the Core 2 Duo but their P4 CPUs were pathetic. That's the reason I got the Athlon in my desktop.

Also, one last thing I forgot to mention. The 1.8GHz processors were mostly at the bottom of the range for AMD. P-Ms of 1.8GHz and up were top of the line. So logically it would make sense to compare them model for model (ie top models to top models rather than bottom-mid models to top models like in this case). Otherwise it's a silly comparison since they're different architectures and scale differently.

But I suppose if you precisely clock for clock performance is what you are after, yes, they're pretty close, at least according to benchies.

[weepy]

Just google for athlon 64 windows driver. You can also get the X2 driver this way, it helps.

OT: I have these installed. Their purpose is in better thread handling and rdtsc fix. They're actually a windows scheduler update rather than a driver. I didn't measure speed improvement but the m$ scheduler sucks with or without it .

One of the reasons I have an Intel CPU in my laptop is because IBM/Lenovo didn't make anything based on AMD processors.

Good point .

... Otherwise it's a silly comparison since they're different architectures and scale differently.

But I suppose if you precisely clock for clock performance is what you are after, yes, they're pretty close, at least according to benchies.

I believe that clock 4 clock is very relevant, since for the given process/transistor count you should get similar power consumption (or you can do a better comparison) which is quite important considering mobile. Clock 4 clock helps prevent comparing apples and chickpeas.