How 4810T can become very slow

If your Acer Aspire Timeline 4810T has been manufactured prior to July 2009, it can become unresponsive to the Windows Vista user interface and show processor load at 100% without any user software running in the computer. The immediate recovery means unplugging all sources of power, including the battery. To resolve the problem by upgrading drivers, BIOS, and Vista into SP2 failed because apparently the VGA driver and LAN driver did not get updated. If you have this problem you want to update the VGA and LAN drivers with drivers on Acer download site.

Alex Kidman of cnet.com.au writes in his review of Acer Aspire Timeline 4810t notebook that 'the machine just keeps going and going and going, but it's a pity that it's otherwise underpowered and hampered by Windows Vista'. What he accredits to an operating system may actually have been problems in hardware performance of the machine and hardware device drivers.

I have had the opportunity to install and test two identical 4810T computers, and what I originally thought was an individual problematic machine turned into a rather more sinister view when the second machine I was most kindly swapped with started showing the same behavior.

Don't let 4810T go to the sleep mode

In brief, 4810T should not be allowed to go into the sleep mode.

After recovering from the sleep state, 4810T may show the processor load at 100% and be unresponsive to the user interface. To normally shut down Windows in this state is agony because the normal shut down process of about one minute is going to take about an hour to complete.

The machine may recover by rebooting. It may be necessary to remove all power, including the battery, and then reboot the machine in order to recover the original speed.

Don't burden your machine with intensive software

What a joke; this is what computers are for.

When the machine is loaded with a heavy processing burden while you leave the machine alone on its own for some hours, 4810T may get stuck with the processor again wheeling off at a 100% loading continuously, but with no actual user processes loading the processor core. Heat is generated. The user interface gets sticky to the point of the mouse not responding to movements any more, let alone able to control software. And the process you wanted to run gets hardly any processing time.

Again, there does not seem to be any other way out of this but to eventually shut down, and to reboot the machine. You may have to unplug all supplies of electricity for a moment before restarting.

4810T has a modern low power consumption Intel SU3500 microprocessor. It can dynamically vary its operating frequency, saving power when possible but able to push forward at full blast immediately when there is the need. This mechanism seems to fail in 4810T unless the new drivers have been installed.

You can measure the processor load with Windows Reliability and Performance Monitor (a part of Windows Vista). You see the load level and clocking frequency level. When 4810T gets stuck, whatever the processor frequency level is (I have seen anything from 56% to 100%) the load is 100% and stays solidly there even if you stop all user processes. All this processing time is devoted to something else than the software you are running, and this includes the Windows operating system. In fact, I have seen Windows coming up with notices saying that the hardware is responding too sluggishly.

Look out when you burn CDs or DVDs

Burning the recovery DVDs has been an ordeal for both machines, and exactly the same way. The burning process slows down after the first 10 minutes, and the last 10% of the burning takes 50% percent more time for the first DVD. The second DVD never finishes.

When you look at the Reliability and Performance Monitor you see that simultaneously the processor load peaks to 100% and the disk activity drops close to zero and virtually no data is being read and written. The user processes (the software you are running; the DVD burning software) get practically no processing time. The average processing time is close to 0% with the highest load being 3% percent.

How to resolve

It is very awkward to have to rip off the power cable and battery to keep the computer working. And you do not know when the disaster is going to hit; once the computer has slowed down it is too late to try and save your work as the machine has already become hopelessly slow.

In the course of testing I have re-installed Windows Vista from the hard drive shadow copy, upgraded to Vista SP2, and upgraded the BIOS from the original 1.10 to the newest 1.20. These steps did not help. Upgrading all the device drivers seemed to improve performance but so far the permanent stalls are not gone.

Pulling the steps together, you can try to resolve this issue take the following steps

• Upgrade to Vista SP2 using Windows Update
  
• Upgrade all Device Drivers on the Acer web site
  
• Upgrade to the newest BIOS (version 1.20) on the Acer web site
  

To upgrade to Vista SP2 use the Windows Update software in Vista. Keep installing updates until SP2 update comes up.

To upgrade Device Drivers and the BIOS, go to the Acer driver download site. Then, navigate to "Notebook | Aspire | Aspire 4810T" and select the tabs 'Driver' and 'BIOS'. Check the dates in files and compare them to the dates in your system and the computer manufacturing date. If the files are newer, download them and install following the instructions in the files. Notice that there are more drivers when you pull down the scroll bar to the right. Pay attention to the Vendor field; you have to use the drivers that are published by the company who manufactured the hardware that is actually on your motherboard.

For more information, see section "Acer attempts to fix it". See also the section 'Going inside'.

This is Acer drivers download site. Note that "Driver" tab has been selected and the scroll bar to the right has been pulled down to reveal more drivers. The date field is the last to the right.

What about the processor?

If the supply voltage for the processor is not sufficient the processor can become unstable and we can experience the kind of stall I have been seeing. SU3500 is a modern low-power processor type. It is very special in the sense that the supply voltage has been individually selected for each unit and this information can be found inside the chip.

Intel's home page says the VID voltage range is 1.050V-1.150V, and the initial start-up voltage for the processor should be 1.20 V.

Intel's data sheet for this processor gives ranges for the legal supply voltages (pages 41 and 45) for the '5.5 W ULF single-core SFF processor' as they call the SU3500 (min -- nominal -- max)...

(power use) VCCHFM VCC at Highest Frequency Mode (HFM): 0.775 — 1.1 V
(light use) VCCLFM VCC at Lowest Frequency Mode (LFM): 0.8 — 0.975 V
(sleeping) VCCSLFM VCC at Super Low Frequency Mode (Super LFM): 0.725 — 0.925 V
(starting) VCC,BOOT Default VCC Voltage for Initial Power Up: — 1.20 — V

All these voltages show a range because the actual voltage is individually selected. Says Intel ", Each processor is programmed with a maximum valid voltage identification value (VID), which is set at manufacturing and cannot be altered. Individual maximum VID values are calibrated during manufacturing such that two processors at the same frequency may have different settings within the VID range." If the processor is loaded the supply voltage should be decreased very slightly according to Intel.

What I have is way below these values. 'CoreTemp 0.99.5' measures the core voltage to be VID = 0.9000...0.9375 V. SiSoftware Sandra reports that the core is asking for 0.938 V power supply voltage and reporting a legal range of 0.713...0.938 V.

The core temperature measaures between 46...60 degrees C in normal operation. The absolute maximum for the circuit is TjMAX = 105 degrees C. Plenty of headroom there!

Going inside

I opened 8410T to see how the air circulation is arranged and how the microprocessor heat sinking is done. There is a fairly thick solid heat-forming heat conducting sheet between the processor and the heat pipe taking the heat to the fan and radiator in the side of the machine. I checked that the heat pipe is good contact with the processor by gently pressing it down on the processor. I checked the two mounting screws keeping the heat piple in position.

There had been no blue screens, just slowing down, but I decided to take off the two memory DIMMs and seat them back in properly.

I test drove the machine with the bottom cover off but otherwise in its normal orientation. Running a SiSoftware Sandra stress test I could see the temperatures increase, but significantly less than with the bottom cover on. The bottom cover has several slotted air inlets. However, the combined area of these inlets is not much and there is a significant air movement resistance because the critical openings are covered with a metal mesh. Providing larger open area in the underside of the machine would decrease system temperature and also make this 'Laminar Wall Jets' system washing the outside of the computer with cool air actually work. I was continuously using the machine while Sandra was pushing it continuously to 100% processor usage. There was no stickiness. The response times were good and short. The system temperature was around 57 C.

I put the covers back on and kept driving the Sandra stress test at full 100% level. The temperature increased to 60 C. All software has been working so far and there has not been a slowing down or stalling of the machine (yet). Now, with Sandra off the temperature is down to 50 C.

The temperature figures I mention do not show anything dramatical. The absolute maximum internal temperature for the processor is 105 C, so in that regard there is plenty of headroom.

What did I change? Nothing. The same problems continued.

Process Explorer

I happened to stumble upton Microsoft Process Explorer (www.sysinternals.com) and the software has proven to be very helpful in understanding what you computer is using all that time for. I had Process Explorer running when 4810T again went into the snail speed mode. Now I had the chance to see who was taking the time. It was no software at all. It was hardware interrupts. Hardware inside the computer is communicating by setting up interrupts in the processor. An interrupt (as the name says) stops the current software processing and makes the processor go away to service the interrupt in what is called an interrupt service routine. All that is normal, and usually takes much less than 1% of the processor time. Well, in my 4810T the processor was now busy servicing interrupts more than 90% of its time. If you consider that the processor clock speed is some 1.4 GHz, there was some serious interrupting and servicing going on. I was getting just a few percent of the processor for my software. No wonder I was feeling that the machine had suddenly almost frozen.

Usually the reason for this type of a fault is a hardware failure or very badly misbehaving software. The hardware is going grazy. The physical interrupt signal line inside the computer is oscillating. Some repair is in order.

After one week it turns out that the reason actually seems to have been faulty software drivers.

Acer attempts to fix it

Finally I wrote all of this into a document, packed it with the computer, and shipped the computer off to Acer service to be fixed. The turn-around time was fast: one day. What was done: Windows recovery was run (erasing all changes to the system), and LAN and VGA drivers were updated.

Does the machine work now? Yes, it seems to work. It is now able to finish the recovery DVD burning and it wakes up from the sleep state normally. The interrupt rate has never exceeded the peak of 15% seen during DVD burning. The machine jammed up to 100% after 12 hours of use, but recovered back to normal after about 20 minutes. I must say I thought "here we go again" when the machine froze and did not come back to its senses after a couple of minutes of waiting. 20 minutes is a long time to wait for a computer to finish its private business and let you carry on with yours.

According to Acer service they updated the VGA and LAN drivers. The drivers that came with the repaired machine are --

• the VGA driver on Acer's download site is actually called "Display Adapter/Mobile Intel 4 Series Chipset Family" in your machine, and the installed driver is dated May 7, 2009, and has the version number 7.15.10.1752, and it is published by Intell.
• the LAN driver in Acre's download site is actually called "Network Adapters/Atheros AR8131 PCI-E Gigabit Ethernet Controller" in your machine, dated April 1, 2009, version 1.0.0.8, publisher Atheros.
  

When you want to check you drivers, go to the Control Panel > System > Device Manager, and open each device (these devices are inside your computer). Using the right mouse button select Properties. Inspect the Driver tab.

Before you send you machine to the service, update these drivers. See also section "How to resolve".

What was it that I failed to do and Acer did? The only difference I can see is the VGA driver. I did not update that. Acer did.

What can we learn from all this?

Acer must have known fully well that 4810T has these stability problems. Yet, they choose to go on the market with the machine. Did they inform the customer of the solution to this problem on their web site or when the customer contacts the service? Of course not.

I free piece of advice to Acer -- you would make much more money per a machine if people do not send machines to be serviced when the resolution is a simple driver update. Put the information on your web site so that it is easy to find. Be proactive. Write a clear notice for the customer. Do not play hide and seek about where the file is available and if someone is clever enough to find it and to guess that this is the information he needs.