Upgrade an Acer 5610Z

Hi, everyone!

I tagged this onto an ol thread about a week ago, but it doesn't seem to be getting any attention, so I hope starting this thread up again isn't too beyond the pale.

I've got an Acer 5610Z, with a T2060 part. Acer documentation says T2060, with 943GML chipset.

CPU-Z (1.53) says Acer Grapevine mobo, with i943/940GML, 82801GHM southbridge, and 945GM graphics subsystem. It shows the CPU to be a T2050. Oddly, v1.51 of CPU-Z identified this part as a T2060 (which is what it is supposed to be). DXDIAG shows the CPU as a T2060.

Other programs have identified the chipset as a 945 variant, but I believe that's erroneous, since I have seen threads talking about where certain drivers will report incorrect (?) data.

I used to pay attention to this stuff, and at one time could do this myself, but I no longer can with the "chipset and processor of the month" product cycle we have evolved into.

I'd like to upgrade this CPU, and I am just hopelessly confused as to what I might get.

I _think_ the FSB is limited to 533 MHz on this puppy, but I am not sure. If it is not, I'd love to get into a higher B/W part.

Can someone please help with upgrade choices and perhaps explain a bit? I would like to get a couple of more years out of this notebook, and a new CPU would be just the prescription.

Thanks, everyone!

 

I do have the same laptop as yours, I have upgraded my memory from 1GB to 2GB last year, I have upgraded my Hard Drive today from 120GB to 250GB.
My laptop works very well with these upgrades and I'm quite pleased with it. Saying that,my requirements doesn't need a CPU upgrade as I am satisfied with 1.60GHz processor. I can feel that my lappy is tad faster with the new drive.

 

MrPink, your processor uses Intel's socket M, which was standard for the Core Duo and the first of the Merom Core 2 Duo processors (before those switched over to socket P on the Santa Rosa platform).
This means you can almost certainly upgrade to a T2080 or T2130 (also Pentium dual-cores), and probably to Core 2 Duos numbered T5x00 or T7x00 (e.g. the T7200, but not the T7250). Just make sure it's socket M. Here are some examples.

In general I agree with ellalan though. I doubt if a cpu upgrade would be cost-effective. You would probably get much more benefit from more RAM if you're at 1gb now.

Happy New Year and good luck!

 

I had a 5610z, I got it from Wal-Mart quite some time back. Anyway, I don't think anything in it except the motherboard and heatsink is original..lol. I've replaced the processor with a Core 2 Duo T5500, the HDD to a 500 GB Hitachi, 4gb of memory, a new Broadcom dual-band N-wireless card, and a new DL-DVD burner because the old one died for some reason.

EDIT.. and it is the i943 Express chipset. And there was a LOT of confusion about the T2060 when it was released. It's a Yonah core "Core Duo" with half of the L2 cache of its equivalent Core Duo model.

 

That's a very interesting upgrade, I am keen to know how did you manage to get 4GB RAM in that model because I thought the maximum memory for this model was 2GB. Could you share the details with us. Thanks.

 

Appreciate everyone's help!

Jacob and Ellalan, I went to 2GB two years ago and replaced the Hitachi slowpoke drive that was in this originally with a W/D 3200BEKT screamer about nine months ago. Big differences, indeed.

I like XGX's idea of changing out the Broadcom wireless card in this thing- it's always just been really poor in the sensitivity department.

Questions:

So, I can drop in virtually ANY 32-bit (not 64-bit) socket M processor? I knew that there were some that I could do this with, but was unsure about others.

Will this chipset support an 800MHz FSB speed? And if not, if I were to get a part that was supposed to run at 800MHz, are they smart enough to step themselves down to 533MHz? I'm not proposing this, but if I happened to run across a used CPU that ordinarily wants to run at 800 for a great price, could I use it?

I would have gone to 4GB, but the docs say a max of two. XG, did you have trouble doing that?

 

I don't think there are any socket M processors that run at 800mhz, but if there were, the GML943 would probably downclock them to 667.

Technically your socket will support any processor listed here. But the question is whether the bios Acer but on your motherboard will support a core2duo. I did some snooping, but cannot find anyone with your exact model who upgraded to a c2d. I think you should probably email acer support and ask if the bios supports core2duo processors.

 

Now that you say that, it might have been 2. lol... I gave it to my mother a while back. I do remember the memory I put in it I got on sale at office max so... it was PC5300... I know that for a fact. But in reply to Mr. Pink, yea pretty much any socket M with a 667mhz FSB. That was the max for that MoBo. But the wireless that came in mine was Atheros... AR5005G. It's mini-pci, not minipci-e. I remember that because I messed around with OSX86 and it ran perfectly on that particular notebook, sound, wireless, even webcam. I ended up going with an intel T5500 because it was cheap on eBay at the time.

 

Okay, it sounds like an email to Acer is in order. I will get back here and post any useful response (other than, "Is your computer plugged on?") from them. In the meantime, if anyone else has anything to add, please do!

 

Hi Mr Pink,
I have upgraded my T2060 to T2400 and I think you can safely upgrade to T2400,T2500,T2600 and T2700. If you want C2D, you can upgrade to T5600, T7200,T7400 or T7600. I would go for a T7200 and you may need to flash the latest BIOS. I am going to try without flashing as I have v3.33 BIOS to see whether it just drops in when I get hold of one.

 

Update:

Well, Acer tech support is completely useless. It really IS, "Is your computer plugged on, I will be happy to (not) assist you in the most pleasant of ways," nonsense.

If I really want to talk to someone, they want $2.95 a minute, and you have to buy a minimum of $60 of time. You mean, you guys want $60 to answer a simple question? What a joke. God, I remember when I could get the guy WHO DESIGNED THE PRODUCT on the phone at HP... <sigh>

Ellalan: Okay, this is good information. However, if there is newer BIOS for the 5610Z than I have, I am unaware of it. I found one update that solved a nasty audio stutter problem (documented somewhere else on this site, I believe). Everest says BIOS date is 2/13/07, and my notes say version 3.50. The BIOS file was BL50350A.WPH. I can give this file to you, if you want.

I changed out the Broadcom wireless adaptor for an Atheros-based card and am really happy with the results. Much better sensitivity and apparent throughput because of that. Nice.

As far as FSB speed, Intel says this chipset maxes out at 533 MHz. I sure would like it if it would run at 667 MHz, though.

Isn't the 7200 A 64-bit part? This machine is running XP, and I don't want to have to change THAT part of the equation.

Anyway, any and all more better input is always appreciated. The more data points I have, the better my chances!

 

Hi MrPink,
I have upgraded to T7200 and pleased with it, I didn't flash my BIOS and it was just a drop in and the system recognised it. I have noticed a vast improvement in performance but the new temperatures are bit higher, idle between 42 and 46C, and maxes at 58 while my older CPU was much cooler.
I'll wait for some time for the AS5 to settle otherwise I may have to undervolt.
You can use 32 bit OS with 64bit capable CPUs without any problem, I have attached some pics of how it made the difference to my system.

 

The T7200 is a good CPU. I've used it for a year or so after upgrading from a T2500. And it is not that expensive £60 to £70.

For the undervolting mine was stable with 0.9875v @ 2Ghz (multiplier x12)

 

Unless there's some trick I don't know about, this BIOS shields you from anything you could tweak which would actually make any sort of difference. That seems to be de rigeur for notebooks, and I really hate it. There's no options to even change memory timing, much less options for clocking or voltages. So this is strictly a plug-and-pray kind of thing.

Unless there are other BIOS options- are there?

As far as going with a part that supports 64-bit instructions, I'm a bit confused. As I said before, I haven't paid attention to processors in a while because there are just so many flavors, and they change every other week.

As I understand it, there are 32-bit parts like this 2060 and the 2500, and 32-bit parts that support 64-bit instructions (they call it Intel64- what IS this?) like the 7200, and full-blown 64-bit parts. Naton, why did you upgrade to the 7200 from the 2500? Aside from the 4MB cache and the Intel64 function, they look like identical parts. While I can imagine there being a small speed improvement with the larger cache, the power consumption is almost 10% higher. It also looks like I can get a 2600 for less than the 7200, which might be a sweeter spot.

This is fun!

EDIT: I see that there's now a 3.60 BIOS version. Will go and get that. Anyone know what it fixes?

 

Initially the Gateway I had (I sold it two months ago) had a Celeron M. Then it was upgraded to a T2500. I can't remember who did some tests (either Andy or K-tron) and basically they found that a T5300 was faster than the T2500. In addition I didn't spend a lot for the T7200. I bought it for the same price a I sold the T2500

To undervolt you can use RMclock. Just follow this tutorial: http://forum.notebookreview.com/showthread.php?t=235824
If you wan't to go further you can also try this: http://forum.notebookreview.com/showthread.php?t=390696

I like to think that the T7200 consume less power than the T2500 because it has a newer architecture. In addition the cache which is not used is turned off to save power.

 

If you look at the power figure quoted by Intel, they show the 7200 at about 10% more power. But that may be static maximum. In a dynamic situation, it may have better power management than the 2500, I don't know.

 

I hope that you're not talking about DTP beause this is not the same thing as power consumption.

 

I was. Since DTP is the dissipation that the heat sink must be able to draw away from the part, I presumed that was a decent number to compare relative power consumption with. If DTP is not a good number to use for this, then what spec does one use?

 

Basically you need to know the power consumption in ampers and you multply this by the power consumption in voltes for the T2500 and the T7200.

I think the DTP of the T7200 is 34w and not 31w like the T2500 because of the additional 2MB of cache.

 

For the same architecture and size process, it makes some sense, but if you change architectures or size process, then the numbers don't compare very well, as a change in architecture or a change in the size of the production process will change the energy efficiency of the chip. Thus if 2 chips have the same TDP, but one has an architecture/size process twice as efficient as the other, that one will draw half the power (the difference shouldn't be _that_ big, but it's an easy number for demonstration purposes). Also, Intel TDPs are rarely exact, and can even vary by the individual chip. So while on average, a 35 watt TDP chip will run hotter at full load than a 30 watt TDP at full load, you can have an individual "35 watt" TDP chip that runs cooler than a "30 watt" TDP chip.

 

Errrrammmmm, uh... (brain goes BOING) Wait a minute.

If one architecture or process is more efficient than another, that means it will do the same amount of work while using less power. In the case of a topology that is twice as efficient (to use your example) as another, the power consumption to do the same amount of work (say, execute 1M instructions) would be half.

Dissipation is power consumption. It's voltage times current- that's what a watt is. If part A needs to throw away 31 watts (as in the case of the 2500), and part B (7200) needs to throw away 34 watts, then all things being equal, part A is consuming less power than part B. But if say, Part B can do half again as much work with the same amount of power as part A, then the comparison becomes unfair. Presumably at that level, part B would be half again faster than part A. Part B gets its work done in 3/4 of the time, thus using less power on any given task.

So it looks to be confusing. But again, if Intel says, "Hey, you need to be able to wick 31 watts from this part (ignoring individual variance for this discussion), and 34 watts from that part," that means that the power consumption for the bigger part can be presumed to be larger, at least under some unknown condition.

I think what I would like to see is power consumption data for CPUs doing the same set of tasks. A laptop manufacturer would need to know this sort of thing in order to make design decisions about speed, battery life, etc. Anyone seen anything like this?

The decision for me here is 2600 vs 7200, really- they are both close in price, and either would spiff up this lil workhorse of mine. If one is substantially faster but uses a squiq more power, I can live with that. But if there isn't much difference, then I would opt for lower consumption.

 

Here's an interesting tidbit.

Running the CPU stress test, the hottest I can get this 2060 is 51 degrees C. Not very hot.

But here's the really weird thing. Core 0 is about half as fast as Core 1. Core one completes the tests about twice as fast.

Anyone seen anything like this before?

 

If you want to have both CPU at 100% for your test then use Orthos or some similar application. Acer in genaral have good cooling system, and the T2060 (I had one for a while that I selvaged from a faulty Acer 5610 ) is quite cool as a CPU.

How can you tell that core 1 is faster than 0 if you didn't disable one of them?

DTP is not the same as power consumption. This is a comparative I did a while ago for a T7300: max DTP at full load 35W / max Power consumption at full load between 42.58 and 53.30W.

I calculated the power consumption of the T7300 based on the formula W = V x A. From intel's datasheet the max power consumption in Ampers of a T7300 is 41A, and its operation voltage is beween 1.0375 and 1.30V (varies depending on the CPU). Hence:
1.0375 x 41 = 42.58W to 1.300 x 41 = 53.30W

 

Those results _are_ from using Orthos. I am particulalry befuddled by core 0's half-speed results.

And you're right- consumption isn't the same as dissipation, because we're talking about the heat being thrown away by the part- that's the power NOT being used to perform work. And as already said, if a core is more efficient than another, then that comparison becomes unfair. But dissipation is a useful guideline.

Something that is not well known about electronics in general is that current consumption goes up as voltage goes down. So the CPU may draw 41 amps at some voltage (presumably the lowest voltage), but will not draw that amount at a higher supply voltage. So it is likely that the 41 amps is for the lowest voltage value.

At any rate, it seems to me that a newer part will probably be better. The 2600 is a core duo, and the 7200 is a core2 duo.

I am confused about something else, though- and that's the 7XXX C2Ds with part numbers that end in 50. If I can use a 7200, why can't I use a 7250? Is it because of the bus multiplier? Same question with the 7200 v the 7300. The 7300 is significantly cheaper than a 7200, but its multiplier is 10, not 12.

Sigh. This used to be SO simple... pick a socket, pick a bus speed, and pick a clock rate... heh

 

This is really weired

That's good to know. this means that a T7300 would consume a maximum of 42W regardeless of its operating voltage

yes newer architecture is better. A lot of people say that a Core Duo is two Pentium M glued together.

The T7200, T7400, and T7600 are socket M while the T7250, T7300, T7500, T7700, and T7800 are socket P. Socket P and M are phisicaly the same but their pin out is different. So a T7200 won't [work on a Socket P and a T7250 won't work on a Socket M.

The T7200 is more expensive because, besides the T7400 and the T7600, it is the fast CPU for a socket M. The T7400 and T7600 are very rare and hence very expensive.

The T7300 is relatively cheap because it is not the fastest for socket P.

 

Well, let's go with some (completely hyopthetical) numbers. Let's say we have 2 chips that both take 100 watts, one with a 50% efficiency and one with a 25% efficiency. Thus, even those take the same power load, one will have a TDP of 50 watts, and the other will have a TDP of 75 watts. I do admit to my numbers being wrong in the direction, but the point was more an illustration of how TDP for different achitectures/efficiencies were uncomparable, thus the fact that it's very difficult to use TDP to compare, for example, a Core Duo to a Core 2 Duo.

I had a very similar conversation with lackofcheese over this sort of thing not too long ago, back in this thread ( http://forum.notebookreview.com/showthread.php?t=472199). We were more trying to figure out exactly how much power a processor actually would use compared to TDP (ballpark), but didn't quite manage to seem to come to a final conclusion. Intel seems to pretty much only put out required voltage ranges and maximum current draw, which, truth to tell, is all most manufacturers should need.

It's also worth noting that TDPs are usually measured at near full load, so a faster chip with a higher TDP may still run cooler and draw less power than a slower chip with a lower TDP, simply because it doesn't have to run as hard to match the same instruction load.

Isn't that assuming the same power draw, though, using Joule's law of Power = (Voltage) x (Amperage)? If the power draw changes along with the voltage, the current draw may not necessarily rise.

 

This subject is a little confusing. Every circuit wants to use a particular quantity of energy to operate. While I can't absolutely say this is true in every single example, in general if one lowers the supply voltage to a circuit, current demand will increase, and vice versa. A CPU is a very, very big circuit!

I think I'll be on the lookout for a 7200. When I get one, I'll report before and after. They seem to be a little expensive right now, but history shows a pretty wide price range. I think that will keep me happy with my machine for a couple of more years.

 

Not necessarily. It really depends on the components in the circuit. What I would call a "fixed" circuit, where all the components have a fixed power draw and are always on, will. The problem is when you start adding in components that are not always on, and/or do not always operate at the same level. For example, your house can be considered a very very big circuit as well, but the power drawn will depend on exactly what you have turned on in the house. Don't forget that a CPU, while a very large circuit, is composed of millions of little circuits that are constantly turning on and off (that's what transistors basically are).

Still, for a fixed power draw then yes, if you lower the supply voltage to a circuit, then the current demand will increase. The problem is that very few things operate on a fixed power supply; this is why a combination of Joule's law and Ohm's law (current = voltage / resistance) is a little more common, resulting in power = (voltage * voltage) / resistance. This is a little more common because for a specific circuit, usually the resistance will be the fixed value, and thus, as you can see, raising the voltage will result in increased power draw, because more current is actually being pushed through, not less. The easiest example of this is the old lightbulb and battey circuit; if you double the voltage in that circuit the lightbulb will be brighter because it's drawing more power, not less. To get the lightbulb to not increase in brightness as you raise the voltage, you'd need to insert a power regulator that would limit the current draw as the voltage rises to keep the total power drawn the same.

The T7200 is a generally good processor for most uses. I'm using one myself, although my machine admittedly came with it.

 

We're off the track now, but I must say that my generalization was not intended to be applied to circuits consisting of passive components in a static configuration. It applies to the very thing we were talking about, that being circuits containing active devices that are dynamic, such as a microprocessor. I've designed hundreds of circuits for various applications, from switching power supplies to GPS-disciplined Rubidium time and frequency standards. And in general, when one starves a circuit that contains active devices for voltage, it will want more current. There's no fixed resistance we can use for a calculation for a circuit that contains anything other than passives and is actually doing something.

A light bulb isn't applicable to my generalization because it essentially is a passive component with a twist. The filament is low resistance until voltage exceeds a certain amount, and then resistance goes up and the lamp lights. Once the lamp is lit, resistance will remain about the same. You are exactly correct that one would need to current-limit the lamp to keep it from blowing up with more voltage than it should have. It's static. More voltage across a fixed resistance means more current. More current times more voltage means more power, in watts.

It's convenient that you chose a house for an example, as this will illustrate the point. A home is not a static circuit, as you observe. Brownouts (reduced voltage conditions) are particularly bad for devices and appliances in homes because the stuff that's running or connected suddenly wants more current, can get it, and then can exceed the current capacity of things like transformer primaries, fuses, motor windings and other components.

Anyhow, I think we are on the same train, just perhaps in two different cars.

I will definitely report back with benchmarks and such when I get my hands on a 7200.

Cheers!