Sure... that rig isn't standard, but it's also probably a lot more powerful than most people need as well. That being said, the ATI 3800 series video cards have amongst the lowest idle power consumption of any video card available today. Like I said in my original post, the X38 chipset draws a lot of power compared to other modern chipsets.

I gave you an example of a relatively high-end system, and showed that it draws less AC power than my mother's old Athlon system from 2001. I'm not sure what you're getting at, but if a "worst-case" system from now draws less than an "average case" system from 2001 (it was a basic HP model), then I'm not sure how else to convince you of this...

Wrong. Take any modern, 45nm quad-core Xeon CPU and stack it next to an old S601 socket Xeon. The modern CPU draws less power.

Compare an Intel Q9650 and a P4EE 3.4GHz. I don't have the numbers on hand, but I'd be willing to put money on the fact that the P4EE draws more power.

There's no magic about this, nor is there a "golden rule" that says more cores = more power consumption. The 45nm HF based process that Intel has adopted has reduced power consumption over the previous SiO2 process that's been used for the past 10-15 years (if not longer). It's simply a matter of improvements in technology to reduce power consumption in CMOS devices.

Intel has several well-documented papers on the technology used in their 45nm process (as documented as they can go without exposing secrets). A simple web search will help you find them if you're interested.

Dude! I don´t get how you think and where your logic is at but this is really not the smartest thing to say. Listen, it has hardware T&L and I have to correct myself there a bit as it´s DX 9.0b and not 9.0c. Then you come with a link to a guy that is talking bullcrap like:
Err...

And where do you get this from? Refresh rate is response time?...You are totally, but totally wrong there... totally .

I won't bother replying to most points as they've for the most part all been refuted 3 or 4 times already.


Funny how a several years-old chipset/motherboard that uses a not very popular chipset differs from today's modern & popular designs. Who would've thought?


How so? I don't expect requiring more than 10 channels of high definition audio anytime soon, nor is the codec going to break or anything like that.

In a lot of cases, there's no point in buying separate cards for commodity jobs anymore. A while ago, various ports (e.g. IDE, floppy, parallel, serial, etc) weren't onboard (ISA cards rather). Then they were integrated onboard, and that didn't make them suck. The network cards -- every board built in the last few years has at least 100mbit ethernet, most of the new boads now even have gigabit ethernet (sometimes two ports even), and they're not bad either (my onboard Realtek NIC even does TCP checksum offloading and more). Same for audio now. When $50 mATX motherboards come with onboard audio as good as a Audigy 2 ZS or similar (a card that used to cost ~$110), and for the most part have FAR better drivers too (all-around better solution, except perhaps if you care for EAX), it's no wonder Creative's shares have been dropping a LOT over the last few years. Their shares peaked to like $30 circa 2000, and now they're down to like 4$ and still dropping (they were removed from Nasdaq last year). There should be more motherboards with basic video onboard (for non-gamers too), and AMD seems to be making this happen.


And what would having to buy a $50 PCI-e SATA controller bring me over the onboard controllers? What would a $50 PCI-e NIC bring to me? What would wasting $100 on Creative garbage bring me? Maybe I should buy a USB card too? The onboard stuff is quite good in most cases, no need to buy a card separately. All buying cards for this stuff would do in most cases, is fill all of the motherboard's slots very quickly, and cost a whole lot more.

The newer drivers introduced hardware vertex shading, which is a totally different thing from TnL. I linked to a single post in that thread, and could care less what the rest of the thread said.

Taken from OFFICIAL INTEL SPEC SHEET:



http://www.intel.com/products/chipsets/gma950/index.htm

There. Now you'll say that Intel themselves are lying in their specs?



The response time determins the maximum refresh rate, and the maximum FPS rate that the monitor can actually display. Running a lower refresh rate is only going to make things slower.

You can contradict me all you want, i'm off to patch some stuff in a game, see you tomorrow.

1. The Intel GMA950 (940GML, 945G<x> chipsets) can NOT do hardware transform and lighting nor advanced vertex shading *in hardware*. This was not added to the chip hardwre until the GMA X3000/X3100 (GL960 and GM965 chipsets). It can be done in software, but it's not done in hardware on a GMA950 chip.

2. LCDs do have a refresh rate, but it's not as important as response time. The equivalent response time (from dark to bright and back to darkness) from an LCD to equal a CRT refresh of ~120Hz is 8.33ms. The reason this is not as important is that unlike a CRT, which uses the refresh rate both as a way to image and illuminate the screen, an LCD uses the shutter (no refresh rate) to illuminate a pixel, and thus the response time is what measures this. The reason a CRT needs to refresh (refresh rate) is that otherwise you'd get "flicker", the result of the phosphor decay on the screen after the energy from the electron gun that transferred to the phosphor material behind the screen starts to decay (slowly) until it is refreshed when the electron beam from the gun hits that phosphor location again.

Since LCD monitors do not use phosphors to display on their screen, refresh rate is not of real concern to most people. The LCD transistors used to illuminate the pixels stay open or closed until told to switch. However, the refresh rate of an LCD does matter in some instances, if the hardware supports it. A refresh rate of 60Hz means that you are capped at refreshing a pixel 60 times a second, which is usually fast enough. Saying it doesn't matter at all is not correct, but saying it's just as important as response time is also incorrect, for almost all applications. Since some of you guys are gamers I'd assume you will end up eventually havng an application fall outside that "almost all applications" bucket , but for everyday use 60 or 75Hz (again, check your hardware specs for the optimal refresh rate for the LCD and the card) should be perfect.

I stand corrected; The Intel GMA950 has no TnL hardware onboard, my mistake and sorry (slams head trough the table...).

Are you also drunk? You must be sharing drinks with BenoitRen.

As Cluberti already explained, LCD refresh rates are tied to their response time. If fact you couldn't be more wrong. The fact that the interface is running at 60Hz means that even if your response time was 0ms you'd still be able to update the screen no faster then 60 times a second. The response time only affects how long it takes for the image to change once it receives instructions to change in one of those 1/60th of a second lapses. So the response time has nothing to do with the maximum refresh rate nor does it have anything to do with FPS. Get yourself another beer and go watch TV.

Meh. It doesn't really matter. It was a moot point in the first place.

Like I said, I don't actually need any 3D performance (no games at all), so 3D features are pointless, and the potential lack of them is an absolute non-issue.

It's not like TnL will help when I'm checking email, paying bills, coding something, playing mp3's, burning discs or whatever other everyday tasks.

If anything, I think he made my point: his "OMG GMA sucks" stance seems to be all about the lack of 3D features (which are pointless for me -- it's like complaining a bike has no doors, even though they're clearly not needed), and no actual reasons why it sucks so bad for everyday non-gaming tasks (i.e. why I shouldn't use it). Framerate is also of little concern for things like browsing web pages, typing a word document, viewing por err... I mean photos! Seriously, I'm perfectly happy with the performance of a 10 year old card (except perhaps for video decoding acceleration). Onboard video is plenty fast for a lot of people.

I agree with you - I used an onboard ATI M200 chip for the last few years in my desktop with shared memory because I just write code, do some web browsing/email, and run some VMs. No point in paying for anything but the bare minimum, and onboard is fine with me. My new box doesn't have an onboard video like the last, but I still bought a relatively low-end ATI Radeon video card that runs aero just fine, and while there are some here who would say it's underpowered, I say, "for what?". I do 2D only, like yourself, and a GMA950 would be fine for what I do as well. I was merely pointing out some of the inaccuracies in the threadline.

I actually did look for a box with onboard video too, but couldn't get a motherboard that met my needs with an onboard video, or I'd probably be running an onboard Intel or AMD again.

Excuse me jumping into the middle of a very interesting topic, but as regards this:

what software do you use to turn this into a router?

Monowall will do that nicely for you, as well ClarkConnect (don't know if CC is free anymore...).

Here's a list for you to look at: http://en.wikipedia.org/wiki/List_of_Linux...l_distributions

I use FREESCO. www.freesco.org

Thanks for both replies. I didn't know there was even a Wikipedia list.

I have a P1-133 and a P2-450 in the junk pile as well as a stack of 3Com NICs. Both computers hopelessy underpowered but nice and quiet because the CPU is passively cooled. Both also extremely heavy, incidentally, because of their very solid case + chassis. Another week-end project!