15" T60 guys... For those who already own one

[own6volvos]

1st question: If you have the unit on a flat hard table, and use one hand to open the screen, with the other holding the units down, can you see some slight movement of the hinge connection piece (the shiney metal thing) on the left side of the laptop (near the power buttons). If you move the screen up and down, pay attention to the position of the hinge base compared to the bezel around the keyboard. The right hinge on mine is 100% solid when doing this, the left hinge has just a wee bit of movement.

2nd question: If you press down mildly hard above the "mute" button, can you hear the "nipping" sound of the fan making contact with something around it. This one caught me off gaurd a bit. I am guessing the fan is directly under this area.

These 2 things are making me wonder if something might be loose in that area, and some input from other guys/gals with the same size screen might be able let me know if this isn't just on my model.

EDIT: Oh yea, if you lift the screen up from either of the corners on my laptop, the left side hinge squeezes a good amount.

[fullauto]

Hmmm, the answer to both questions is no. My T60 w/ 15" Flexview is absolutely solid around the hinge and "mute" button.

Dude, what do you think about the portion of the palm rest that is over the PCMCIA slot?

Scary, no?

[own6volvos]

The palmrest over the pcmcia slots are solid on mine, the HD section has some flex though. Looks like my Flexview 15" is getting returned as lemon... this really sucks. I think I might just buy an american made welder instead of another laptop

[kaotic504]

Hmmm, the answer to both questions is no. My T60 w/ 15" Flexview is absolutely solid around the hinge and "mute" button.

Dude, what do you think about the portion of the palm rest that is over the PCMCIA slot?

Scary, no?

i've expressed the same issue in my post. it's just too thin compared to my T22, the plastic on this doesn't feel nearly as thick but now we have a "roll cage"

[fullauto]

kaotic: actually is was your thread that brought this issue to my attention. Thanks.

Wow, I don't understand how the palm rest could be different on another T60 with a 15" display.

Oh well, just going to have to be very careful.

After 7 days, the T60 is looking good...except the core-duo is no where near as fast as I was expecting. I am developing a multi-threaded, realtime app; and I thought the new Intel CPUs was going to scream.

Not by a long shot... We need to get from an execution frequency of 60s down to 10s. I am inverting a very large matrix at every execution and the core-duo just ain't up to it...

[christopher_wolf]

kaotic: actually is was your thread that brought this issue to my attention. Thanks.

Wow, I don't understand how the palm rest could be different on another T60 with a 15" display.

Oh well, just going to have to be very careful.

After 7 days, the T60 is looking good...except the core-duo is no where near as fast as I was expecting. I am developing a multi-threaded, realtime app; and I thought the new Intel CPUs was going to scream.

Not by a long shot... We need to get from an execution frequency of 60s down to 10s. I am inverting a very large matrix at every execution and the core-duo just ain't up to it...

Inverting a huge matrix at every execution? Depending on how large that is, I am not sure a even a workstation could handle a load like that. Sounds like it could be optimized.

[brendel95]

It should depends much more on your matrix solver algorithm and size of matrix. To take advantage of the dual core or any multi cpu cluster, you should parallelize code and run it.

[christopher_wolf]

It should depends much more on your matrix solver algorithm and size of matrix. To take advantage of the dual core or any multi cpu cluster, you should parallelize code and run it.

In other words, more optimization.

I have spent many a night optimizing code for very nearly the same reason; I would think "Oh, is this Opteron setup hot; I bet it will blaze through this" then get suprised when it didn't go through nearly as fast as I had expected. I had to optimize my stuff quite a bit to get it to go faster. I had previously gotten a TI83 to do fairly large FFTs in under 1 minute beforehand, so I just slogged away at it. Lo and behold, I got far better results than I expected using the correct complier and flags. It also helps to see what can benefit from parallelization and what won't benefit as much from it. Amdahl's law applies here.

[brendel95]

I was talking about the algorithm like order O(N^2) or O(N^3) even O(N) (in some special case) algorithm. Then I believe, complier optimizing process has to come to speed up or you can parallelize if the algorithm is suitable to do that.

[fullauto]

Lo and behold, I got far better results than I expected using the correct complier and flags.

I am new to multiprocessing, and I suspect (hope) there are switches buried in the documentation on the compiler for taking advantage of a dual core.

I am still perplexed though, I am only seeing a total of 50% CPU loading (I spawn a separate thread to crunch through the calcs), and total elapsed time is approx. the same between the 1.8 core-duo system and a 1.8 single core Centrino M system per execution.

Multiprocessing 101: how can I clearly see the advantage of a dual-core CPU system running XP (i.e. can XP really take advantage of multi-cores?)

[sidd]

Multiprocessing 101: how can I clearly see the advantage of a dual-core CPU system running XP (i.e. can XP really take advantage of multi-cores?)

You can try to search for multithreaded algorithms for matrix inversion. Or, if possible have two separate threads work on two different matrices at the same time. Basically, you need two separate *computation* threads running in parallel to take the full advantage of the dual core system.

[christopher_wolf]

Intel's compilers have some nifty abilities on their own chipsets; my first introduction with it was with FORTRAN parallelization and the Intel compiler on Linux. Going through some of the techniques they listed in the Numerical Recipes Books; after that, I noticed a huge improvement in how the programs operated and executed. I don't know what the deal is with other compilers on other systems...but they should have something very similar to it or at least have an easy way to integerate into the program (preferably via a language that supports it).