Acer TimelineX 3820TG Owners Lounge. Info, benchmarks, tweaks, mods and upgrades

It's almost true, except in my experience the weakest part of the chain is always the USB controller chip on the case. I have seen about 10+ USB external cases or docks die on me or on my friends after about a years use. First they start needing more power, then becaming more unreliable, then some of them even corrupts huge files. While the HDD inside is still very good. BTW 2.5 drives tend to need more power when they became older. As strange as both sounds, I have experienced these issues with more than 10 USB HDD things.

 

Just wait a little, Hungary was late as usual, and finally we got the 5 series in the more expensive shops now. I am waiting for the cheaper shops to start selling.

 

Oh, sounds better and better. Do we have some common knowledge about overcloking possibility, except for "random"? If I can try a laptop in a shop before buying, is there any way to check it? Can we understand some information from the SetFSB diagnostic page? Do same HEX values mean same IC?

 

There are some which does not require 2 usb ports. 2 usb's will give more power and improve file-transferring. 1 usb will usually be problematic and may cause the harddrive to shut off. You have to google the item for comments to find out. My case has also 2 usb-connecters so I'm guessing I need two to get it working. Will be confirming this later today

Mine: SweBurn.com :: Accessories :: USB :: Sumvision E-Gem 2.5" SATA to USB Aluminum Enclosure Black/Grey
Other case which I'm pretty sure only requires 1 usb:
IcyBox IB-221STU-B SATA 2,5" IcyBox IB-221STU-B SATA 2,5 - Dator - Lagring - Extern hårddisk - Hårddiskchassi

 

Looks like Acer refuses to send out parts, regardless if you're under warranty or not......

Looks like I'm going to have to order a US Keyboard from Ebay.

 

Just wanted to say that the 2nd USB cable in a y-cable is only for transferring power, not files..
And even if your external HDD ships with a y-cable, it doesn't mean it is required. Mostly it is just a precaution by the vendor/manufacturer in order to avoid complaints when people use it at bad USB ports and don't get it working.

 

Sorry I can't find the post now. But someone posted how to create shortcuts to setfsb to make it overclock on start up? So I can make it boot with the fsb set at 168Mhz. Anyone now how to do this.

 

found this via google, looks like it'll work.

[SetFSB Auto Tool] - Automatic CPU overclocking utility - v3.1 (Page 1) - N10 Overclock - Asus N10 Series Forum | N10E | N10J | N10Jc | N10Jb | N10Jh |
and this if that one doesn't work
http://forum.notebookreview.com/asus/373672-when-setfsb-autotool-doesnt-work.html

Good to hear that they were willing to cover you under ITW, bronsky. If it's not US bound, I'm going to import one of these in a month.

 

http://forum.notebookreview.com/6434213-post1164.html

 

Will this ever be released in the US?

 

I expect you 3820TG owners to get much better

This is with 4820TG i5-450m version..

 

Wiezok, with what resolution?

 

default, that is I didn't change anything in any settings except oc'd the gpu

 

'Bout time somebody broke 10k. I knew it could be done. Congratulations! No oC on the CPU? Nice to hear. I'll have to start running 3DMark to see what I can do.

Bronsky

 

My 3820TG's arrived only 2.5 days from the time I ordered it.

 

Congrats PlatinuM195. Nice to see another Kiwi on the forum and a 3820TG owner. Doubt there will be many in NZ.

 

what game fps increases are you getting, hendrickson with your OC'ed lappy?

 

Nice what GPU clocks are those? Its strange there are such differences, when the hardware is basically the same?

My best so far:



Doesn't seem to turbo much, but sits with both cores over 3Ghz

CPU-Z Validator 3.1

Haven't really looked to hard but I would say 5-6 fps less in BFBC2 (Need to confirm via fraps) I need to do more testing to see that 166.7 Mhz is stable after a few hours of gaming.
Why would this be? Is strange 2.6Ghz to 3.2Ghz and its slower??!? This just seems to be in BFBBC2 so far.

Losing FPS would seem my CPU isn't stable. But it did handle over an hour of prime95 stress testing with no errors. Odd... I will try lowering the FSB I guess.

 

Damn, It seems I cant change language on my new install;<.

 

Why not?

Let me give you a metaphor. You buy a car, it has an engine and transmissions. You'd think you can make the car go faster by making the engine spin faster, but what if the transmission belt starts to slip because it's going so fast so that any increases of spin rate will decrease the actual energy transmitted to the wheels beyond a certain point?

Kind of not a good metaphor for this, but a way to remind you that things aren't exactly as simple as they seem. Here's the purpose of having a clock signal in a digital circuit: to provide synchronization. Each transistor/circuit element has its own time constant as part of its transient response when the input changes, which means output doesn't change instantaneously. The longer the time constant the longer you have to wait from the time the input changes to the time the output stabilizes. Time constant is defined as the amount of time it takes for a output signal to reach within (1/e)% of its final value.

Then you have areas of the chip which has an overall time constant that is a function of all time constants of its constituents. The manufacturer then sets a clock frequency so that each period is long enough so that they are sure the entire circuit has stabilized before the next input change.

When you guys do the overclocking you're playing with that. Not all chips are made the same due to the inherent randomness in manufacturing. You can try to make the circuit go faster, but beyond a certain point one of the following may happen:

- one part of the circuit (a part with a long time constant) no longer functions correctly, but the rest is still fine
- a major part of the circuit no longer functions correctly, and the chip shuts down.

You will most obviously notice the second effect. The first effect is harder to notice. It will cause errors, which may or may not be corrected by the internal logic in a future clock cycle. Regardless, this can very well slow down your performance. Stress testing with just one application might not produce the right combination of switching to cause errors (remember that any transistor switching series is hugely complex).

So, after watching you guys try to push your chips faster and faster for a few weeks, I've finally decided to write something about it. A lot of text, but hope it helps.

source - electrical engineering

 

Hmm well its odd the other overclockers failed to mention this? Unless I have a turbo boost issue? I would liek to here Just a Nick, Mac and the others input on this. I'm sure if they had noticed less FPS in games it would have been mentioned.

SuperPi is quicker, 18 seconds overlocked vs 23 default.
wPrime much quicker too vs stock.
Any cpu benchmark is quicker
3dmark06 scores are up 800+ points
Converting a video is quicker too.

Everything is quicker apart from new games I think that use threads.

I'm running wPrime now non stop over night to see if I can get some errors overclocked and how the temps hold up. A super stress test. Will see how it holds up and if it throttles or not over night.

 

Oh, and also, the question why the chips aren't the same despite carrying the same model number.

Here is why.

You see, the amount of money it costs to make a silicon chip fabrication method goes well into the billions of dollars. You can't expect ATI to actually make ten or twenty of these production lines, can you?

Instead, they use a standardized circuit design to create the circuits. On a slice of silicon wafer, you would have areas, called dies, which contain your actual circuit. The processing done is through chemical and optic means, which inherently has some randomness. This means that some areas of the wafer will produce excellent chips, whereas other areas will not. How do they tell?

They embed test chips at regular intervals. After the wafer is ready, they analyze the test chip. Afterward, they predict that the chips near the test chip are of similar quality. The higher quality chips will receive higher clock frequencies, be paired with more powerful memory, and get a high model number. The lower quality chips will receive a lower clock frequency, and so on.

What does quality mean? You see, depending on how well a piece of semiconductor is made, it may have higher or lower transition time, draw more or less current, and have more or less defective cells. Because of the aforementioned method of determining whether a chip is good or bad, it can very well be case (though not likely) that one chip may not run at all while its neighbor can be clocked to HD 5870 speeds. But, there are variations. So, one HD Mobility 5650 might crash when you bump its speed slightly, whereas another one is very overclockable.

This is why you should never think that just because someone on a forum was able to overclock to some level, you should be able also. That's a good way to break your system.

 

The chip you're dealing with here contains hundreds of millions of transistors. One particular set of input signals can cause the circuit to behave very differently from another set. I only mentioned time constants earlier. There are also many many other issues, like resonance, etc. You know the suspension bridges that have failed? They wouldn't have failed under any circumstance other than when a particular load was placed on them with a particular motion with a particular set of forces from the sides.

What i'm getting at is, just because wprime doesn't crash, and other tasks run well, doesn't mean that the circuit is free of errors regardless of what input you give it.

Oh, and last thing. Just because the temperature is low, does not mean the circuit can't fail.

High temperatures will cause permanent damage, whereas simply causing calculation errors will not.
Let me give you another small example. Suppose my method of logic is the following. I have a bucket sitting in front of me. I can communicate a 1 or a 0 to you through:

- if the bucket is empty, I mean to say 0
- if the bucket is full, I mean to say 1

so if I were to try to tell you something like 10011110101010100101 very quickly, it means that I have to fill and empty the bucket very quickly, meaning I have to do a lot of work, which produces heat (like sweat, lol), but in circuits, this increase in current causes more heat.

But just because the circuit isn't getting hot doesn't mean that you haven't pushed me to the point where sometimes I haven't been able to fully fill the bucket after a 0->1 transition before you told me to empty it again. That's when you start to get glitches.

 

To overclock my cpu I would need the pay version of setfsb right? at least the version I tried didn't seem to have specific model numbers like the asus listed for hendrickson..

 

Your unit from Germany was delivered?

@wiezok - if the setfsb does not recognize your chip it probably will not OC. Try Pll diagnostics. See if you get an error.

 

The good news, my replacement US keyboard for the German 3820TG arrived today.....

The bad news, I don't have a laptop since my Computeruniverse order was refused once it could even load into the plane...

I can sit here at the table and pretend I'm typing on my new TimelineX

 

Bastages!

Bronsky

 

I wrote a letter to SetFSB guy asking about what should we inspect to get some information about the overclockability of our laptops.

Here replied:

If someone could or could not overclock his 3820TG, could you share your PLL control registers with the result of your overclocking? Maybe Acer did not use that many ICs, just a few, which we could collect from SetFSB diagnostic page.

You can even get the registers from the shareware version, when it asks for code, just press enter and it has a Capture button just for this purpose. If we could collect all the captures in this forum then we might understand what model changes happened.

 

Does anyone know of a program similar to Vaio Care's battery management for Sony laptops? That allowed me to stop charging the battery once it reached 80% to prolong battery health.

 

Finally finished formatting and re-installing and everything seems fine, thought I'd post a couple of quick pics for potential buyers.

An overall shot


The metal casing

 

Nice, so it shipped with a english keyboard?

I found a thread that has a package download with the battery care software in it:
http://forum.notebookreview.com/son...y-care-function-after-clean-installation.html

maybe it can work for other brands? no idea.

 

Yup with an english keyboard with some chinese character strokes on it but otherwise pretty good.

One thing I've noticed is that I can't seem to mute/unmute while in a full screen game but I can still adjust the volume fine.

 

Damn you beat my scores hendrikson! Maybe I should try with different RAM to see if overclockability changes, although I doubt it. I guess it's just my CPU that won't do higher. Also I've not found any way to increse FSB in small increments with the freeware FSB version, it's straight from 160.4 to 166.8 FSB.

 

Overclocked and 6+ hours of Prime95 torture test and no error. Temps maxed at 92 degrees. Not bad at all.
Here's the proof:



OC-Freak are you seeing less FPS in games with overclock?

 

Hey so you ordered it from world.pchome.com.tw?

Which model did you order? The i3-350m? "ACER Aspire 3820TG-352G50n. (I3-350/1G independent display)"
or the one that i think is on sale "ACER Aspire 3820TG (i3 +1 G alone were sale one product) (Product Number: DHAA5R-148942717)"
for $26900?

The google translate is confusing the hell out of me. I'm guessing independent display means discrete graphics?

I may try getting this to the US since you had luck with them. Who did you ship with?

Thanks for your help btw.

 

What are the odds for THIS:



Must be like an 1000:1 chance to get exactly 10k, oh well.

Settings: i5-450m @ 2934MHz (162.7 FSB)
HD5650 @ 740/1130MHz
1280 x 768 resolution.

 

Yup that's the exact model I got, although I paid 27,900 for it and it went on special the day after I got it.

I realised that the core i3 throttles at 90c whereas the core i5 throttles at 104c. Everything seems ok so far though but then again it's winter down here plus the core i5 was about a US$186 upgrade which I didn't think was worth it as it didn't come with a RAM or HD upgrade too.

For google translate, I find that it's much easier using the latest version of chrome. If you simply surf to the pchome page using chrome then a notification will come up asking Do you want to translate this page?

It then translates the page without reloading it and it's really useful. Hopefully that helps . I think independent display means discrete graphics, just make sure it says 5650 in the specs.

As for shipping, it came to NZ via EMS. Is that the information you were after?

 

Any particular games that is good for testing it? I'm mostly playing Strategy/MMO/rpg games, so not much of the typical fps type games.

 

This place in Thailand sells 3820TG. Doesn't look like they ship internationally though.

$1063 Canadian for i5-450M 2GB RAM
$1400 Canadian for i5-520M 4GB RAM

Search | invadeIT - Online Computer Store Thailand: 3820tg

 

I so want one of these...too bad the conversion is so high in Thailand - otherwise I'd have one picked up in a few weeks.

 

Did you realize that futuremark incorrectly reports what GPU that is used?

HD5650 is the most powerful GPU found in the timelineX models.

My guess is that he ran at 1024 x 768 resolution and not 1280 x 768.

 

(delete please)

 

FPS games, new games. Overclock = less fps

 

Hmm, but I don't feel much like waiting... especially since the performance difference between the 4- and even the 5- series does not seem worthy of the premium. Already in Germany, the model with the 450m sells for 60 EUR more than the one with the 430m, I think this is quite a lot for a 3% hike in performance.

Anyway, is my understanding correct that all 'i' mobile CPUs use the same socket? So basically, by buying a 430M today, I have the option to replace it later on all the way to a i7-840QM (nothwithstanding heat considerations)?

 

Almost two hours at 167Mhz FSB @ 3.36Ghz prime95 + Furmark stress test she reaches a hotter 95 degrees. No errors or crashes. Remember my ambient temps at the moment are hot 30 degrees. For some reason GPU temps would not display? My guess the GPU was pretty damn hot.



And my overnight 6+ hours of Prime95 torture test and no errors. Temps maxed at 92 degrees.





I think I know my issue with the games, seems the CPU is not going over the multiplier of 18 or its even less stuck at 10? Its odd, but i think it explains my much lower FPS and temps as the CPU is not clocking up to turbo or x 19.

Is there a way to make my CPU turbo boost all time?

 

use ThrottleStop?

 

So did anyone try that 1inc adapter from amazon? Im realy thinking about it, it may fit into my ZeroshockIII with my 3820. When I first saw the timeline's adapter I knew I had to change it to something more reasonably sized ASAP...

 

How do I know if my ATI card is running at the correct PCI-E speed?
At amd gpu clock tool under PCI-E STATUS it says Link Width x16. Is that correct?

EDIT: The clocks are also wierd. Engine shows 99 and Memory clock shows 149 ;<.

 

Hi guys!

Just got my laptop from computeruniverse yesterday. Seems like the same one Bronsky's got!
LX.PV102.172

3dMark06 yields 9107 with 680/1050 GPU overclock, test crashed twice before it ran to the end though...GPU temp at 72 degrees. Don't quite get it since it's not all that hot yet it still crashed.

Cannot clock the CPU with SetFSB freeware 2.2 at all! Freeze the screen (crash) and had to hard reset every time.

Any chance somebody managed to overclock the CPU in this particular machine?

Same here liath! Until I just put in 680/1050 anyway...

 

i have the same problem if i overclock anything higher then 645 1045 there is a chance that 3dmark06 crashes i tested with 670 1070 and sometimes it crashes and sometimes not, and temp is not even close to hot and some people seems to overclock up to 740 1100. im wondering if 740 1100 was with thermal paste applied on the cpu/gpu?