Acer Aspire 5020 Series AMD Turion 64 Notebook Review (pics, specs) - FOLLOW-UP

[Angello]

It could speed up the graphics card. I will try it.

Doesn`t worth - it may accelerate only 2D, which is fast enough.

I thought that you need to program the PLL chip via SMBus using i2c, this should be possible. I don't know, what high IOPL is

That`s right, but for dealing with i2c i`m using direct port i/o as it`s done in the kernel i2c drivers, which is usually not permitted. To get permission under NT/XP I`m using device driver. That's why this proggie must be run as administrator. Under linux it will require root access rights.

I controll the voltage with http://cpu.rightmark.org/products/rmclock.shtml and it works. Just now, I'm using the maximum of 1.4V, I will try to decrease it later. The voltage has a huge impact on stability.

Yes, rmclock is good. They (Dmitry Besedin) are directly programming the processor p-states (i don`t know how yet).
It won`t kill the cpu even with 1.5v.
When downclocked sufficiently it worked stable@0.8v for me.

[fik]

Updated my Acer 5020 (5021 5024..etc) overclocking tool. Now it can deal with the integrated DDR memory controller...
Also you may avoid changing the cas latency - this locks my computer very often.

Thank you Angelo, changing CL locks my computer all the time, it should be avoided. The only important tweaks are 1T command rate (disable 2T) and the DDR frequency.

But be aware, that the DDR frequency does not survive soft-reset, while the FSB frequency does.

[fik]

That`s right, but for dealing with i2c i`m using direct port i/o as it`s done in the kernel i2c drivers, which is usually not permitted. To get permission under NT/XP I`m using device driver. That's why this proggie must be run as administrator. Under linux it will require root access rights.

In linux there is http://www.lm-sensors.org, which deals with i2c, but I'm afraid, there is not the correct driver for our PLL. Anyway, I have found a tutorial on i2c drivers: http://www.linuxjournal.com/article/7136 and http://www.linuxjournal.com/article/7252

Yes, rmclock is good. They (Dmitry Besedin) are directly programming the processor p-states (i don`t know how yet).
It won`t kill the cpu even with 1.5v.
When downclocked sufficiently it worked stable@0.8v for me.

In my rmclock the maximum is set to 1.4V. Maybe this is a security measure. Just now I'm trying FSB 248 with 1.4V and the cpu is quite hot, 82C when doing torture test in gimps. (250 and 249 failed the test in about two hours or so).

Edit: 248 failed also in about 5 hours. Going even lower.
Edit: 240 failed in 16 hours, maybe it is a temperature problem, I'm waiting for my laptop cooler.

The p-states are programmed in processor MSR (machine-specific-registers). Maybe this will help: http://www.tuxamito.com.es/cpupw, but myslef I'm using a small kernel patch, to change the voltages: http://forums.gentoo.org/viewtopic-t-245197.html

[andyxl]

hi everybody, back again,

recently I bought MSI 655X with MT-37, 400 MHz DDRAM. I'm going to try Angello's utility, hope it works with this model.

Have a question about GPU oc, there is x1600 450/450 and I tried 490/500 so far, do you think this is safe? anybody knows, if these cards have sth like temperature sensor, so they wont get burnt?

On 5024 x700 I haven't tried GPU oc, because it's my brother's machine...

Has anybody tried this > http://www.sanx.org/tipShow.asp?index=14 <
I read that it can speed up your computer FPU by 10%, sounds interesting

Finally, I think that rmclock is a great utility, I run it on ACER 5024 MT34 [0.4x - 0.925V ; 0.8x -1.250 I think] as well as on my MSI MT37 [0.4x - 0.775V ; 10x - 1.150]. Never had problems with that tool, it's doing a great work. Even on that ACER full loaded, not undervolted CPU was too hot - 82C and then automaticaly lowered muliplier to 4x for a while till it cools down.

[fik]

hi everybody, back again,

recently I bought MSI 655X with MT-37, 400 MHz DDRAM. I'm going to try Angello's utility, hope it works with this model.

I'm not sure, it has to have the same PLL chip, in case of Aspire 5020 it is IDT CV137, if I'm not mistaken. If it works, you may need to go down with the memory frequency first, before increasing the FSB.

Have a question about GPU oc, there is x1600 450/450 and I tried 490/500 so far, do you think this is safe? anybody knows, if these cards have sth like temperature sensor, so they wont get burnt?

have you tried http://www.techpowerup.com/atitool ? It should find the highest frequencies without artefacts.

Has anybody tried this > http://www.sanx.org/tipShow.asp?index=14 <
I read that it can speed up your computer FPU by 10%, sounds interesting

I did, it comes from windows-NT already, but it has no effect, just placebo

[pifko]

So, how well does this series of laptops deal with Vista then? Drivers alright/available?

I got a free legit copy of Business edition (through my Uni's MSDNAA partnership) and I might give it a quick go, seeing as I'm gonna refromat soon anyway.

[andyxl]

I'm not sure, it has to have the same PLL chip, in case of Aspire 5020 it is IDT CV137, if I'm not mistaken. If it works, you may need to go down with the memory frequency first, before increasing the FSB.

I tried that utility and it stated that clock generator was found, so looks like its working ok. but it detected wrong fsb speed - only 166. doesnt matter.
I think that's good idea to go down with memory frequency first, I'll try sth around 180.

Yes, I am using ATT, I know that the highest frequency this card can handle without artefact is 533/535, but I'm not sure if this is safe?

I did, it comes from windows-NT already, but it has no effect, just placebo

Really? there is other guy that tried it and ... 10% he said.

[paparas64]

hello.can anyone upload the original cd's of acer aspire 5024 wlmi?

[fik]

I tried that utility and it stated that clock generator was found, so looks like its working ok. but it detected wrong fsb speed - only 166. doesnt matter.

166 is wrong, it is not a good sign, but anyway you can try it. What is the chipset of your notebook?

Yes, I am using ATT, I know that the highest frequency this card can handle without artefact is 533/535, but I'm not sure if this is safe?

It also depends how long you are running without artefacts, 12 hours, 24 hours? And what temerature are you reaching? Too high temperature may cause shorter life of your silicon. This has already happened to me with hot P4.

Really? there is other guy that tried it and ... 10% he said.

Well, it depends what exactly he benchmarked. You can do the same and report here, but I don't beleive you will see any increase.

[andyxl]

What is the chipset of your notebook?

There is ATi RX480M + ATi SB450 Chipset

I tried, oc'ed to 2.2 GHz [FSB & MEM @ 220 ] but after few minutes computer hanged, don't know reason. Probably memory, maybe another program. I can't lower memory freq to 183 - that doesn't work either, hangs computer immediately. Will try 166 later on.

It also depends how long you are running without artefacts, 12 hours, 24 hours? And what temerature are you reaching?

I have tried only for a few minutes... because there is no GPU temperature sensor or utility to find out the temperature.
It's the same as in 5024 - gpu and cpu are sharing only one fan using heatpipe I think, CPU temp is good - max 71 when undervolted. This is good start for GPU oc, isn't it?