Is there any compal notebooks that I can purchase with a 512mb video card?

I can't seem to find a compal notebook that can be customized to include a 512mb video card. Do they exist?

 

Dedicated or with TC/Hypermemory? Though I don't think either HEL80 or HGL30 offers either of those (the Go7600 is just 256, which might be a good thing). If you mean 512 dedicated, the only ones I can think of are the Asus A8jm's 512 mb Go7600 or a 7900GTX on a Sager or Dell...

 

More systems are being released with 512 MB go 7600 or x1600 cards. It's been said though that neither card can actually utilize more than 256 MB of VRAM.

The ASUS A8Jm & F3Jm come to mind. There are also a number of Zepto Znotes with 512MB.

Both Compal models being sold as Compals (HGL30 and HEL80) go with 256 MB.

 

asus is the manufacturer of the 7600

so it seems that they are the first ones to use the 512 mb version of it. And for a very long time in computer time, the only ones.

the only other manufacturer with it still is zepto. I have no idea why zepto.

zepto and asus are the only ones in the world.

 

Like people said above, memory bandwidth limits the usefulness of 512MB with a Go7600. It can only effectively use up to 256MB of video memory. But go ahead and spend extra money on something useless if you really think you need it. It's not my money

 

I do not agree with you.

 

Ok, how's this:
Exhibit A
Exhibit B
Exhibit A is a 512MB Go7600 running at 450MHz, with 700MHz memory. Scores a 3509 in 3DMark 05. Exhibit B is a 256MB Go7600 running at 480MHz, with only 369MHz memory, and it scores a 3683 in 3DMark 05. If you do the math, per clock, the 512MB 7600 gets 7.80 3DMarks per clock (MHz), and the 256MB version gets 7.67 marks per clock. I'll let you draw your own conclusions as to whether the 512MB is worth it on the 7600, especially considering the disparity between the memory speeds.

 

If I have time I'll explain. In the future I'll show you links.

3dmark is a marketing tool for the gpu manufacturers.

3dmark 06 on the same cards will show the 512 mb higher.

3dmark 07, 08,09 will show more and more the higher score for more ram.

So you can say, in 3dmark 05 the 7600 doesnt use the 256 mb extra ram. But what does that mean? Are you seeking to run 3dmark 05 higher? what does 3dmark 05 trying to simulate?

3dmark o5 is trying to simulate software for sale in 05. Software for sale in 06, 07,08 may use 512 mb more often. In the future all software will be designed with the thought that the consumer has a 512 mb video card.

64 mb ti 4800 doesnt run fear, 128 mb ati 9200 does. Software compatibility in the future usually eliminates first by dx compatibility, then ram. Almost never gpu speed.


Trust me if you are new to the computer industry. When the 128 mb 9600 was on the market there were the same little blurbs all over saying, a gpu will never need 128 mb. It doesnt even have a use. Or an even better example is like, g4mx 440 with 128 mb.

128 mb card is necessary for todays software. faster 64 mb card wont run it.

3d marks of the future will just include a test that needs 512 mb. the 256 mb version of the 7600 will run through hypermemory or whatever. Its score will be a fraction of the same card with 512 mb.

The test itself is what requires the ram.

You make your own conclusion. Nothing is written in stone for the computer industry, they will make something obsolete in the way that generates the most sales. But previous patterns suggest 512 mb card is going to be better in the future than a 256 mb card. 3dmark o5 gives you no clue to this, it deliberately misleads you as the consumer.

 

And we have a winner in our "No Duh Statement of the Week Award."

Again, this is not the point that's being disputed. We're saying that the GPU itself is not powerful enough to fully utilize the onboard memory it is given. A GeForce 7900 GPU works better with 512 MB of memory because it has more transistors, a 256-bit interface, and additional shader/pipeline power.

 

In my opinion, the extra 256MB does not really matter. It probably will matter for later generation games as they probably will consume much memory space for textures and etc.

One good example of a "current" game that can POSSIBLY use all the 512MB is probably Obvlion, or FEAR.

 

Ok, explain this then:
Exhibit C
Notice how the card in the top post is a 256Mb GeForce Go 7600. And it scores 2260 on 3DMark '06. Exhibit A shows a 512MB GeForce Go7600 as scoring 2200 on 3DMark '06 (which is really showcasing the effects of 2007's games. Thanks for trying, play again. 3DMark attempts to show the direction future games will take.)

Face it, some of us do know what we're talking about. 512MB will do very little for you because of the limitations of the hardware in addressing large amounts of memory quickly, which saturnotaku outlined above. The 7600 only has a 128bit memory interface, and runs at a lower memory speed. You need a wider pipeline to efficiently access a larger amount of memory, and that's all there really is to it.

Oh, and for your Ti 4800 vs Radeon 9200 comparison, you should realize that the GeForce 4000 series was in competition with the Radeon 8000 series. It's because the 4800 doesn't have the same capabilities, it's not the amount of memory in it, though that does have some difference. It also has no bearing on the argument you're trying to make in here.

And to further clarify, the TurboCache technology (bottom of the page), NVIDIA makes is only for the GeForce Go 6200 and 7200-7400. Nothing else, especially not the 7600. Only dedicated memory on the 7600 and higher.

And new the the computer industry? Hehe. I may only be 25, but I've got a Bachelor's in Mathematics with a Computer Science emphasis from the Colorado School of Mines, and have worked as a programmer for the past 4 years, 3 of which for a 3D software company. I really do know what I'm talking about. Sorry if I sound like an ass, but I don't take it well when people tell me I'm wrong or question me, especially when I know I'm right and am trying to help them out.

 

My question is, why bother? If you want 512 mb, just get a better video card! Like a 7900 gs or something.

 

i'm thinking of getting a notebook with intel's dual core 2.16 processor and a 512mb video card as it might future-proof my laptop to the next 4-6 years.

 

There's nothing that will future-proof it for that long 256MB will be enough for 3 years or so, and then it'll start being middle of the road. For what it's worth, the Mobile Radeon 9800's were top-of-the-line 3 years ago. So gauge your notebook lifetime off that. Would you be ok with a Radeon 9600 today? That's what anything with a 7600 in it will be like in about 3-4 years.

 

can the radeon 9600 run most of the games today?

 

Ok. Im not going to get caught up in a childish nintendo is better than sony style forum flaming type argument. Please be polite.

No doubt. the memory interface for the 7600 does not address the memory of 512 mb. just like in previous generations.

It does not use it in a way that improves the 3dmark score.

In the past there were even cards that were slower with more memory. Im reasonably certain one was just 9600 with 64 mb was faster than 128 mb version.

Dont reference tommorow by 3dmark scores of today. The 3d mark score is showing literally nothing it is not addressing your question.

IF you have software that uses more than 256 mb textures, a geforce 2 with 512 mb ram is still faster than a 7900 with 64 mb. You are analyzing the statistics of improving speed, Ive seen it before a few generations of gpus exactly the same. the ram is still being addressed faster on the card than through the pci bus.

So, your card with more ram, such as i mean perfect example is what i said g4mx 440 128 mb which is a geforce 2, starts up battlefield 2. 64 mb 9600 doesnt even start it goes to slide show. Yet trust me the 3d mark scores of the 64 mb 9600 are going to be double or more of the g4mx 128 mb.

As I said when Ive got the time I will show you links I dont have time to make this an internet research project for your benefit right now but I wont forget.

I think I have described this well enough. I think you will understand if you think about it, if you really are computer literate as you say you should pick it up.
If not, Im going to take the time to show you software compatibility tables that show examples of what I said earlier in this post.

However it may take me some time Im not online for research projects right at the moment.

 

yes its still compatible with all current software.

provided you have a 128 mb 9600. if you have a 64 mb version you cant run todays software.

You are a very bright person. You have gone straight to the answer of this problem.

A few of us are quite foolish and spending time blabbering about whats irrelevant.

4 years from now a 512 mb 7600 might run all of the days software. a 256 mb one might not

a 64 mb 9600 and a 128 mb one had the same 3dmark scores in their day.

That is the self evident pattern, its right there. You place your bet where you want to. It MAY be exactly the same.

It MAY be both cards are obsolete at the same time due to dx 10 compatibility.

NO ONE can say for certain. It will be decided by the computer manufacturers based on market factors.

3d mark scores and memory interface and speed will not likely enter into the equation. That is entirely different subject.


Ive been through this before. Im quoting people extremely well acquainted with this subject.
I used to work for Q A in blizzard software, and I asked the person Im certain knows what he is talking about.

 

Not a chance. For that sort of time-span, you'd need a top-end DX10 card - and they're not out yet. A PPU wouldn't hurt either.

I think you may be thinking about the 9600Pro 256MB versus 128MB. The 256MB ones tended to use 200/400Mhz RAM, while the 128MB ones used 300/600Mhz RAM.

Probably, but they'll both be under 1fps anyway (because there's no way a GF2 can render 256MB of data onto the screen at any reasonable framerate). On the other hand, if you've got software which uses only 64MB of RAM with lots of shaders to deal with then the 7900 will be getting a good framerate (50fps+) while the GF2 will still be under 1fps because it doesn't support those shaders.

Are you sure about that? According to the BF2 specs, it won't even start on a GF4 Ti (because that's got DX8.0 support and BF2 requires DX8.1 support). In any case, you'd expect the 9600 to be slower than an older-generation card in BF2 simply because it does more work - while the older card only runs DX8.1 shaders, the 9600 has to deal with full-precision DX9 shaders instead.

I agree with what has been posted before: more than 256MB on a Geforce Go 7600. If we had 512MB Geforce2 cards now, would that make them good gaming cards? Of course not. The only time I can recall that a lack of RAM really made a difference was on the Asus 128MB 6800GT cards - in that case, having 256MB would have helped quite a bit.

Consider cards which are going out of date. Does/did a 256MB 9800Pro have a significantly longer useful lifespan than a 128MB 9800Pro? Not really. There may have been a few situations where the 256MB one could handle marginally higher detail settings in new games, but in general they'd both be limited by the GPU more than the RAM available.

What about a 6600GT? We've seen 128MB and 256MB ones of those. Would a 512MB one have been worthwhile? I doubt it - in modern games the GPU is the bottleneck, and in older games even 128MB was enough RAM. Correct me if I'm wrong, but the Geforce Go 7600 in the HEL80/HGL30 is 350Mhz core, 8 pipelines, and 350/700Mhz RAM. The desktop 6600GT is 500Mhz core, 8 pipelines (marginally less efficient, but pretty close) and 500/1000Mhz RAM (or 450/900Mhz RAM for most AGP models). This suggests that a desktop 6600GT will have no trouble beating the 7600 in these laptops. But everyone's perfectly happy with 256MB on a 6600GT, so why should a slower GPU (the GF Go 7600) have more RAM?

 

In the HGL30, it's 375/400 for the Go7600, and I think the same on the HEL80. Either way, the bandwidth isn't there for massive memory access, and it may as well use system memory by the time it's hitting 512, because PCIe isn't a slouch of a bus like PCI or even AGP.

Update: the HGL30/HEL80 both ship at 350/350, and the published specs from NVIDIA are 400/400 for the GeForce Go 7600.

 

heh ok pitabread. I think what I already wrote registered to you. No actually when the data is running through the pci bus what you have is integrated graphics. If the pci bus is fast enough to functionally game dont get a graphics card.

I didnt dig your message claiming you knew what you were talking about, when the whole message told me you didnt know what you were talking about. But Im certain you read what I wrote, and youve begun to look it up and ask around. Because you want to know as opposed to claim you know.

It was the same message I have seen for years. Almost word for word. Because computer magazines print the same articles for the same reason.

What you're going to find again and again is that a computer magazine such as maximum pc, pc magazine, etc is going to give you some very consumer biased information.
Its not data for you to consider, its something to convince you you need to buy something expensive.

A 7900 gets a score twice as good as a 7600 to convince you you need to spend the cash for that one. Only in next years test, they are almost the same score. And definitely in 3 years their score difference is pointless. This is the marketing scheme.

the 3dmark scheme is not data for you to compare cards. The score is made to market the cards. It has to be designed in a way to justify the cost of each option for sale. Its not the other way around. Its not an unbiased measurement.

The consumer of this is also someone who enjoys thinking he understands the data.

 

You're just going on against marketing? Of course marketing is bull-honkey. That's how it's always been. I'm looking at the published, verifiable specifications of the cards in my arguments, not just the 3DMark scores or marketing slogans.

The "whole message" told you that I didn't know what I was talking about? This is coming from the same guy who claimed that a GeForce2 of ANY speed with 512MB of RAM would go faster than a Geforce 7900 with 64mb of RAM? How about this:
The GeForce 2 Ultra (the fastest chip, desktop only) had at most 7.3GB/s memory bandwidth. The MX400? 2.7GB/s. GeForce 2 Go version? About 1.35GB/s. A PCIe x1 interface has 2.5GB/s of bandwidth. A 7900 will run on an PCIe x16 slot. It would have 40GB/s of bandwidth. To the system memory. Even giving them the benefit of the doubt and hamstringing the card with only an x4 PCIe connection to the video card, that's 10GB/s of bandwidth. A 7900 with NO onboard memory will go faster than ANY GeForce 2 with ANY AMOUNT of onboard memory.

Now, just because you think you're clever and have "figured out" the marketing crap, it doesn't mean that you aren't full of bull-feces yourself. Listen to us. We know stuff. It would do you well to learn to listen to people that know more than you obviously do. Feel free to get a card with 512MB of RAM. It's a waste of money, because the graphics will be going too slow if it's trying to access all of your 512MB of VRAM over the 128bit, 400MHz bus. You'd be close to playable framerates, but not quite there. Therefore, no reason to have it.

 

Slatye, I noticed you're quoting the memory speed at one-number/that-number-doubled (i.e. 350/700). Could you fill me in on why you're doing that? When Zepto says their machines are clocked at 450core/700mem, does that mean the same as 450core/350mem?

Also, I'm wondering if anyone knows how large an increase in frames per second in a typical game you would see in a machine clocked at 350core/350mem versus one clocked at 450core/450mem, both using a 7600 card. Thanks,

Eric

 

At least someone gets the whole picture.

 

Certainly.

Pretty much all RAM now is based on DDR. This means that it transfers data twice in every clock cycle (while SDR would only transfer once in every clock cycle). As a result, if the RAM has an actual clock speed of 350Mhz (350,000,000 cycles per second), the data transfer rate is twice that (700Mhz or 700,000,000 transfers per second).

When I say "350/700Mhz RAM" I mean that it's actually running at 350Mhz, but the data transfer rate is 700Mhz. Most PC shops will tend to list the data transfer rate, simply because it looks better.

On a somewhat related topic, if you want to convert that to RAM bandwidth just multiply the data transfer rate by the bus width and divide by 8 (to convert bits into bytes). For the Geforce Go 7600, bus width is 128 bits, so RAM bandwidth = 700Mhz * 128 / 8 = 11200MB/s = 11.2GB/s.

That depends on the game and settings. In something like FEAR which really does stress the video card, at 1680x1050 resolution with maximum details, the video card would be the bottleneck almost all the time. As a result, doing the 28.5% overclock you suggested could actually give a ~25% increase in performance.

In older games (like CS), you'd see virtually no difference - even at stock speeds, the video card can easily run CS as fast as the CPU can supply the data, and it'd be drawing frames far faster than the LCD can actually display them.

I'd probably expect a noticeable performance gain in most new games - 10% or 15% sounds reasonable.

 

So in other words, it is *highly* unlikely that the Zepto's memory is running at 700/1400. Rather, it is running at 350/700, which seems to jive better with its 3dMark score of 3509 (and its core is running at 450MHz.)

 

Yes, this is correct. RAM tend not to overclock particularly well anyway; a 100% overclock (from 350/700 to 700/1400) would be almost impossible.

Running the RAM at 700/1400 in that laptop would be a bit pointless anyway - the GPU simply doesn't need that much RAM bandwidth to perform well. A desktop 7600GT needs RAM that fast, but it's got something like 80% more processing power (560Mhz core clock, and 12 pipelines instead of 8).

 

is there a huge difference between the prices of a 256mb top of the line card and a 512mb top of the line card?

the more i think about it it's probably not justifyable spending say $200 more for a 512mb card cuz i forgot to consider directx 10 heh

 

It's basically a function of which machine you want, archaic. Compal's only have 256MB cards. The Asus A8JM has a 512MB card in it.
Our point is that you shouldn't worry about the amount of memory for a GeForce Go 7600, as long as it's at least 256MB. Look at the other features of the laptop to decide what you want.

 

what other ones have a 512mb video card with at least a 15" screen?

 

For desktops, it's virtually impossible to find a 256MB top-end card - all the 7900GTXs and X1950XTXs are 512MB, and of course the 7950GX2s are 1GB.

For laptops it's really impossible to tell. If one laptop has a 256MB card and costs $1200 and another has a 512MB but costs $1800, is the $600 difference due to the different cards? Or is it just because they're completely different laptops?

I agree. Really, 256MB will be fine for games that the 7600 can actually run pretty well. After a few years you'll need 512MB to run new games, but by then the GPU won't be able to keep up anyway.

You could consider things like the Asus S96J and Z96J if you need 512MB of RAM - but they're not 'real' 512MB systems. They've got 256MB of RAM and 256MB through HyperMemory (which just borrows system RAM and takes CPU time too).

 

You will have to concentrate.

I will be polite.


If you want to see the speed difference between a 256 mb card and a 512 mb card, the same card even, it would make no sense to reference 3d mark 05. That is inherently poor idea. This shows me basic ignorance of the concept of 3d mark test. I may be wrong. I explain this for your benefit specifically.
Because no 512 mb software was available in 05 the test would not factor in textures over 256 mb. 128 mb even. The score SHOULD show the cards are exactly the same.

So this test would show the same speed of a card with more ram.

In the future, if software uses 500 mb of textures, a 512 mb card will be significantly faster than a 256 mb card.

It will not be as fast as the cards of the day. The memory bandwidth is not made to run 500 fast.

But it will be light years ahead of the same card with 256 mb.

( I will leave a blank space for you to refute this, you are refuting this? You have not refuted this but you made a claim its possible this is not the case. )




In the future, tests will be incorporated into 3dmark that deal with larger textures. In those future 3d mark tests, the 512 mb 7600 would have a higher score. It will coincide with the need to convince consumers they need a 512 mb video card, 07 sounds good.

It is not a waste of money to get a 512 mb 7600. It would certainly depend on the amount of money.

The pattern has been repeated several times in the past, the same way. If software requires 512 in the future the 512 mb 7600 will run it better than the 256 mb verion.
Im not saying its fps will be impressive compared to the cards current at that time. I will not even estimate for time contraints, but something like todays 9600 or 9200 running oblivian makes sense.
7600 256 will not boot it up possibly.

these are the factors to consider. I gave these points as clearly and directly to whoever needs them as I could.

 

If by chance, the geforce 2 with 512 mb is actually slower than integrated graphics I stand corrected.

The problem is, it is faster. You are not describing the correct factors involved in running textures through pci-e

However I dont care. Forget I said it if it distracts you. I must have been wrong.

I tried to make an example to explain that the slower card, lower ram speed, lower gpu, bandwidth etc is much faster than pci-e

What I think Im trying to say, is that the 512 mb running though the 128 bit interface at that core speed and ram speed is faster by a huge huge factor over the 512 mb running through pci bus in system ram.

And that is what we are trying to coherently discuss. Direct to this specific instance, these are the facts.

You are incorrect. You have given a consumer incorrect information. I spend the time to try and explain a relatively involved consideration for this forum format.

I do not think its a big deal, the difference between these two peices of computer hardware.
I think there is a difference. I think if someone asks again I will reference this thread and these last two posts only as I dont have time to type essays on it again.

I reread them twice, they are as direct and clear as I get. I certainly have no stock in asus to want to believe their version is better. I dont care what you buy.

I have no bias. I asked someones opinion on this who tests software for a living. I used to have a similar job. He gave a lot of factors to consider, and those were some of them.

I didnt ask him to make a buy recomendation I think he said it might not matter too much. I dont think theres any more calories I can put into this subject.

 

*sigh* You asked someone who tests software for a living? And you don't take the opinion of someone who's MADE software for a living? As in, 3D software? I'm telling you that it DOESN'T matter because of the limitations of the hardware. If we were talking a 7900 or 7800, then yes, it would be better to go with 512. But most games are barely using 128 now, and doubling that (especially on a machine that runs at a "lower" resolution like the HGL30) would be pointless. There's only so much memory you can use and/or need for a 1280x800 screen, and 256Mb is much more than that.

And for the GeForce 2 being faster than integrated graphics, I still stand by the assertion that you're wrong. I showed you how the memory bandwidth to the GeForce 2 wasn't anywhere near what even a low-end PCIe card will get on system RAM. The GeForce 2 Go couldn't even use more than 64MB of RAM in the first place. I know you were trying to "make an example to explain that the slower card, lower ram speed, lower gpu, bandwidth etc is much faster than pci-e", but the problem is that you are just wrong. No two ways about it, you don't even understand the numbers that I'm showing you and referencing, so what makes you think you're qualified to make such an assertion?

And I know what you're trying to say about the 512MB running through the chip interface to the RAM is faster, but by the time you're running into 512MB of RAM on a 128bit interface, it'll be so slow that it won't matter if it's coming from the system or dedicated RAM (see my first paragraph about resolutions and memory. Higher resolutions require geometrically more memory, but since the HGL30 isn't that high-res, then it doesn't matter)

I don't really have a "bias" so much as I see it not making a difference, even in next-generation games and benchmarks, and it'll be obsolete period in 4 years (you really think that games will more than double, sometimes more than quadruple their RAM usage by next year, and not work with current systems by downgrading gracefully?), by the time you might even think of needing 512MB of RAM, so there's really no real point other than bragging rights or being able to limp along with crappy effects far into the future.

I have NOT given any incorrect information, I have backed up EVERY SINGLE THING I HAVE SAID with numbers. You have said nothing except "I think", which you have then proven yourself to not do by not considering the facts of the matter.

Go ahead and point people to these last two posts. They'll see mine here as well.

 

I appears that Pitabred and SLAYTE have an excellent understanding of this topic. Their explanations are clear and accurate. Thanks

 

Heh lololol

Pitabred,

I worked for blizzard software from 1998-1999

The person I asked was shane incargili (sp) who is one of the heads of QA at blizzard in irvine.

Its not that I dont understand. I understand all of it.

Memory bandwidth, size of screen, texture size. A texture of 500 mb, running through a 7600 to a 1200 x 800 screen would be slow

that uses the advanced mathamatics called einsteins theory of LONG DIVISION.
The entire process takes perhaps algebra one level concepts.

EVERY SINGLE PERSON WHO HAS MADE IT INTO HIGHSCHOOL FROM JUNIOR HIGH UNDERSTANDS HOW TO COMPARE THE SPEED FROM 3d mark scores.
Im not arguing with that thats coming from the voices in your head.

Ive asted my time analyzing the relative pipe lines memory interface, overclocking etc etc etc many many hours in the past. Its more an exercise for desktop seperate gpu buyers.
Im telling you that doesnt matter, relevant to to the person buying a notebook right now and what gpu to invest in for the long term.

the 256 mb card and the 512 mb card will be similar for running the software for the first year maybe two. Its possible that down the road the 512mb one might run software better.
Based on previous generations of cards.

Based on the example, ati 9200 64 mb, ati 9200 128 mb. This is based on previous gpus and previous patterns. Based on any previous example of: same card, more ram, speed to slow to process textures that large.

A notebook gpu is a long term investment so I point out the amount of time it is functionally compatible with software, not just the amount of time it is running the top line games.

Im saying quite simply dont base your prediction of the cards software compatibility time based on its speed using the software of today. Software of tommorow could be different.

If anyone, I mean anyone actually read through this and doesnt understand Im at a loss. I think I made it clear.


I did a real life test of the theory of geforce 2 vs pci recently.

I tested software on a 32 mb g4mx 420 which is about 5 years old with an amd 1800.

It ran recent software that it was compatible with much faster than a dell e1705 with gma 950, core duo etc fairly recent.

dx compatibility with the geforece 2 series is much lower but the speed is much higher.

 

What software did you test? What are the exact system specifications of each system? What exactly are your numbers?

If you are so confident of your assertions, you will have no problem providing this exact data.

 

Im talking apples and oranges with someone who is like, needing to repeat oranges oranges etc.

Im going to say apples lol. The 512 mb 7600 will most likely be compatible with gaming software for a longer period than the 256 mb version.

Within a short period of time really, I dont know how long, all gpus will be 512 mb.

Even ones slower than the 7600. I suppose those must be the really insane ones. Something along the lines of a 7300, or lets call it the next card, the 8300 will have 512 mb.

This is the form the cards ill take because thats the pattern they have been sold for at least 8 years now.

Once the gpu cards for sale are entirely 512 mb, the gpu manufacturers will work with software and hardware manufacturers to make software that takes advantage of this hardware better.

Just a pattern weve all seen before. The factors of software compatibility in gpus, as Im repeating are


1. dx compatibility. That means get the higher number first

2. amount of ram. that means the 128 mb 9600 is compatible with more software than the 64 mb 9700

3. Speed of the card. That means the pipelines, the speed of the core and the ram. it almost never reaches this. software that is compatible with the dx version, and the ram will always run on even the very lowest end example. for instance 9200 runs oblivian. Its in the low detail functional arena.

gma 9500, ati 200,, nvidia 6150, gpu too little ram: none of those run oblivian.

Everything Ive said is entirely correct lol. Its the relevant information to consider in this case. If you dont want to consider it dont.

 

sure

I will do that for you.

This is just software running on my older laptop, and then running it on a dell e1705 with gma 950 that I borrowed.

this is sony pcg nvr 23 spec amd 1800 512 mb ddr 32 mb g4mx 420
circa 2001
core duo 1.66 512 mb ddr2 gma 950

all games tested actually. but such as battle for middle earth was the goal.

vampire bloodlines.

Grand theft auto san antonio. Basically older software that was dx compatible with the g4mx 420, however the most advanced and recent examples of it.


Heh Im going to bookmark this page and one day see if I can come back and visit with the 256 mb 7600 cult.

theres a pretty good chance that nvidia knows what its doing better than joe poster. They have a pretty good idea about whats going to run the software better longer. They designed the 512 mb 7600. in fact in a short amount of time they wont even sell a 256 mb version.

 

Indeed. And if Nvidia know what they're doing, why did they even design the 7600 so it could use 256MB of RAM? Why not limit it to a minimum of 512MB if they think it'll be necessary later? Even the 7600GT (desktop) cards normally still have 256MB of RAM, and that's perfectly adequate for them.

Nvidia don't even make most of the cards - they just make GPUs (and therefore they don't really get to decide how much RAM goes on most cards).

Your examples are flawed, for two reasons:
(1) You're testing completely different video chipsets. If you were testing a 256MB Radeon 9600 versus a 128MB Radeon 9600 then it'd be fair (and performance comparisons would be worthwhile), but testing an MX420 with 32MB of RAM against GMA950 with ~224MB of RAM is just silly. Why don't we also test a 256MB Geforce 7600GT against a 512MB Geforce 6200? I think we'll find that the 256MB card is much quicker.
(2) You said that the tests were done with old software so that the GF4 supports it. That's the whole point! By the time you need the extra RAM (in new games), the card can't run new games anyway (because it has no DX9 support, and the GPU wouldn't stand a chance of running DX9 games reasonably fast anyway).

We do see cheap cards with lots of RAM (like the 512MB 6200 I suggested above), but that's just because slow RAM is cheap and it looks good for marketing. It won't help performance in any way.

I stand by what I said before: extra RAM won't help a card last longer. Are there any games which a 128MB GF4 Ti4200 can play but a 64MB version can't? I don't know of any. The only possible example I can think of is the Radeon 8500 in BF2 (128MB one can play it but the 64MB one can't). That's a single example, and not a very good one (because while a 128MB Radeon 8500 can run BF2, there's a difference between running BF2 and running it well. It'll barely manage to keep a decent framerate at minimum resolutions and detail settings).

In a very small number of games there may be a very slight advantage in having the extra RAM. In the vast majority it'll be irrelevant.

Quite apart from that, a bit of logic:

(1) The quickest 256MB cards at the moment are the X1900XTs. There are also a huge number of users with 256MB 7600GT/GS and 256MB 6600GT cards.

(2) This constitutes a large market. Game developers want this market.

(3) If they're designing a game and they want this market, it means that if the game can run reasonably well on a 7600GT/GS or 6600GT GPU, it must also run well with 256MB of RAM. Therefore they'll make sure that it can do this, even if it means compressing the textures a bit more. They'll eventually stop making games which need less than 512MB of video RAM when the GPU on these cards simply can't keep up anyway (so making a 256MB version would not get them any extra market share).

(4) The Geforce Go 7600 is slower than any of the cards listed above. Therefore, if the Go 7600 GPU is capable of running a game reasonably well, the 7600GT/GS and 6600GT GPUs will also be capable of running it fairly well.

(5) Combining points (3) and (4), any game which the Go 7600 GPU can run fairly well will also be designed to fit into 256MB of RAM, so that the game developers can capture that part of the market. Before games start to require 512MB (when they're reaching the absolute limits of 256MB), the Go 7600 GPU will no longer be quick enough to run them at reasonable speeds (so the RAM size becomes irrelevant).


Essentially, I stand by my old point: while having 512MB of RAM may be handy on some cards like the X1950XTX, it's 99.9% useless on the Geforce Go 7600. The GPU is simply not fast enough to make use of it - by the time games need 512MB of RAM, it'll be struggling to maintain 10fps. Sure, a 512MB card might be able to start a newer game - but personally I don't really see the point in being able to run a game at 3 frames per second.

 

This is getting a bit off-topic. Let's try to keep this on the topic of the orignal post.

Thanks.

 

Sorry about that.

To answer the original question: so far, there are no Compal notebooks which can be customised with a 512MB video card. The closest options would be the Asus S96J and Z96J, both of which have 512MB Hypermemory cards (256MB dedicated + 256MB borrowed from system RAM).

There are also quite a few other Asus non-barebone options which have 512MB cards in them.