I would look at getting a Nvidia Geforce4 ti4200, seems to be a great bang/buck ratio.
 
Just looked on http://www.pricewatch.com (not the best, but a decent pricing benchmark) and the above card sells for about $120 USD. That's with 128MB of RAM, too.

Radeon 9500 Pro is a bit more expensive (about USD$150 last time I checked) but will leave a Ti4200 choking on its digital dust and its DX9 support will give it a much longer lifespan than a Ti4200. You'll have to be quick though, ATi plan to replace 9500 and 9500 Pro with the R350-based, but still inferior, 9600 and 9600 Pro.

Don't forget to tell him how "wonderful" driver support is for ATI "choke choke"

Quote:Don't forget to tell him how "wonderful" driver support is for ATI "choke choke"

Actually, I think it's a lot better then the nvidia driver support.

Really.? All i hear from people who actually have an ATI is that they wished driver support was better. Nvidia may have stepped down their driver production in the last year. Both are great cards.

Quote:Quote:Don't forget to tell him how "wonderful" driver support is for ATI "choke choke"

Actually, I think it's a lot better then the nvidia driver support.

You must be very, very new to the PC world as ATi has always had poor driver support in Windows, and is just now starting to fix that as well as help out Linux users. nVidia not only has *much* more solid and stable drivers, but has a much better driver team and continues to enhance performance on newer cards, while still passing on improvements to older ones and being backward compatible. And yes, I have been a long time fan of nVidia but I am currently using a Radeon 9800 Pro on this box because I wanted better DVD support.

THANK YOU. I knew I wasn't crazy. I remember having a Viper GL with the old Riva chip on it. Nvidia released some drivers a full 3 years after that cards release that made it last another year. Those drivers pretty much doubled the video quality and increased frames per second by like 30%! It also added way better GL support. I was able to play 2 games that would not work with the previous drivers. Not bad for a card I would've tossed in the trash.
 
 
The Albatron GeForce 4 Ti 4200 128MB 8X kicks some serious @ss for the price.

The early Riva cards used some cheats to boost performance (something about dropping the video quality if I remember correctly) to catch up with the 3DFX crowd. I started getting really heavy into 3D in '98 when Forsaken came out, and first had a Matrox Millenium AGP card. The acceleration and artifacts were awful with it, so I went to an ATi AIW Pro based on the Rage Pro. It was OK with D3D stuff, but had pathetic OpenGL support. Then, there was the promised mini-GL "wrapper" (many probably haven't heard of this, but it would translate OpenGL commands into D3D so the card could accelerate it, sort of) but it took forever to come out, and when it did it was awful. Then, when they were supposed to come out with a full OpenGL ICD for the Rage Pro, they dropped further development in favor of the Rage 128 line. This was the type of behavior that they continued with; dropping support for one chip when the next was released. 3DFX and nVidia would still release drivers for the older chips, while ATi couldn't be bothered. Now, however, it seems that ATi is sticking with a similar architecture and will be able to support these chips for a while to come.

Quote:and is just now starting to fix that as well as help out Linux users.

nVidia could use a few lessons in helping Linux users themselves as well...

Their current Linux nForce2 drivers are a joke, no APU support, no UDMA support and no non-nVidia AGPGART support. UDMA support is promised in the next kernel (2.4.21) and nVidia are supposedly working on AGPGART but who knows when they'll get around to giving us a proper APU driver.

Quote:Quote:Quote:Don't forget to tell him how "wonderful" driver support is for ATI "choke choke"

Actually, I think it's a lot better then the nvidia driver support.

You must be very, very new to the PC world as ATi has always had poor driver support in Windows, and is just now starting to fix that as well as help out Linux users. nVidia not only has *much* more solid and stable drivers, but has a much better driver team and continues to enhance performance on newer cards, while still passing on improvements to older ones and being backward compatible. And yes, I have been a long time fan of nVidia but I am currently using a Radeon 9800 Pro on this box because I wanted better DVD support.

Well maybe it's just me then. I have always found it to be quite a hassle with nvidia. I have only had a Radeon for a couple of months and have not tried to update the linux drivers or anything. I have been pleased with the windows support though.

Quote:Quote:and is just now starting to fix that as well as help out Linux users.

nVidia could use a few lessons in helping Linux users themselves as well...

Their current Linux nForce2 drivers are a joke, no APU support, no UDMA support and no non-nVidia AGPGART support. UDMA support is promised in the next kernel (2.4.21) and nVidia are supposedly working on AGPGART but who knows when they'll get around to giving us a proper APU driver.

A lot of that goes to the development of the kernel, however, and not the fault of nVidia. Considering that the company is willing to put as much effort as they have into a free OS while trying to retain their intellectual property (3D acceleration is big money people, and they don't want to just "hand out" their stuff just yet) is a testament to their customer support program. In addition, I was able to get an nForce board working rather well under Gentoo, although it is a rather convoluted process. GRUB had seizures with it until I grabbed an unstable version of it, and I was also able to get the onboard LAN working and hdparm returned rather nice results at the time, so I would assume that UDMA was working.

ViolentGreen, you came in just as ATi started hobbling together a good driver program to compete with nVidia. These Catalyst drivers seem to be doing just fine for me as well, and have been stable overall.

do not get a Ti or g4 - out of date - why buy card that will be ueseless - think future
 
ATi 9500 or 9100 - they are Dx9 compatible - yes i know nothing uses it yet, but better to buy something that can use it when it is in palce so in 6 months your not buying yet another card because your ti won't play games.

Quote:do not get a Ti or g4 - out of date - why buy card that will be ueseless - think future

ATi 9500 or 9100 - they are Dx9 compatible - yes i know nothing uses it yet, but better to buy something that can use it when it is in palce so in 6 months your not buying yet another card because your ti won't play games.

While I understand your points, and to a degree, agree with you, I am going to pose this question for no other reason than to be a jerk. Don't take it personally...

What games (currently) REQUIRE a DirectX 8.x card? I know that some games can take ADVANTAGE of such cards (obviously...) but are there any cards that require >GeForce2/3 etc?

Quote:A lot of that goes to the development of the kernel, however, and not the fault of nVidia. Considering that the company is willing to put as much effort as they have into a free OS while trying to retain their intellectual property (3D acceleration is big money people, and they don't want to just "hand out" their stuff just yet) is a testament to their customer support program. In addition, I was able to get an nForce board working rather well under Gentoo, although it is a rather convoluted process. GRUB had seizures with it until I grabbed an unstable version of it, and I was also able to get the onboard LAN working and hdparm returned rather nice results at the time, so I would assume that UDMA was working.

The AGPGART and APU issues are their fault. They put AGPGART code in their own Linux graphics drivers so they're fully capable of protecting their IP in that environment (nVidia's drivers aren't open source) so how hard can it possibly be to include support for non-nVidia chipsets as well? As for the APU, the current driver is merely a bridge between Linux's AC'97 audio support and the APU, the APU isn't used at all. This isn't support, it's lip service.

Nvidia folks aren't making good hardware nowadays, so thier poor driver team is stressed trying to keep up with ATI.
 
If ATI had Nvidia's driver team, Nvidia would be in serious trouble.
 
 
I must say though, My 9700 pro has been stable from the start, and frequent driver updates have been the norm.

Quote:Quote:A lot of that goes to the development of the kernel, however, and not the fault of nVidia. Considering that the company is willing to put as much effort as they have into a free OS while trying to retain their intellectual property (3D acceleration is big money people, and they don't want to just "hand out" their stuff just yet) is a testament to their customer support program. In addition, I was able to get an nForce board working rather well under Gentoo, although it is a rather convoluted process. GRUB had seizures with it until I grabbed an unstable version of it, and I was also able to get the onboard LAN working and hdparm returned rather nice results at the time, so I would assume that UDMA was working.

The AGPGART and APU issues are their fault. They put AGPGART code in their own Linux graphics drivers so they're fully capable of protecting their IP in that environment (nVidia's drivers aren't open source) so how hard can it possibly be to include support for non-nVidia chipsets as well? As for the APU, the current driver is merely a bridge between Linux's AC'97 audio support and the APU, the APU isn't used at all. This isn't support, it's lip service.

I disagree, and I am willing to bet that your judgement is a little clouded by your general distaste of anything mainstream (Intel, MS, nVidia, etc). It isn't like the Linux/BSD kernel is the same from release to release, and therefore easy to optimize for. Many companies are used to developing against a "blackbox" and therefore having a stable component to work against, while the Linux crowd is having a hard enough time coming up with a standard installation method that all the distributions like, or where to install applications let alone any core components that they can work with and call "normal". I have had X crash just from using a browser in one deployment, and it would crash just from using certain menu styles on websites. Hell, X is a totally separate entity from the Linux and BSD kernels, while the Windows shell is integrated and maintained along with the kernel and not developed with the intention of being ported from one OS/Distro to another. As for sound issues, I had enough issues getting my Audigy to work right in a few distros and it's a mainstream card. The kernel would "claim" support in some cases but I would have to get ALSA installed and running (along with 10 different processes) to support the sound card. And what about the hobbled ACPI support that hasn't worked for me on several laptops and workstations across 6 distributions and 2 kernel revs? It's funny how there's probably 2 motherboards on the planet that work right with it, while all the other boards "don't follow the standard" according to their documention for the patches. And you have to love the wonderful suspend and hibernate functionality of the kernel (oh wait, I mean lacking) where you have to patch it *again* and do all kinds of crazy partitioning only to see it bomb out. After all this, we get to hear more and more promises of how 2.6/3.0 is going to fix it all, and yet the release is drifting further and further away. Not exactly a tempting environment for a company to sink a ton of resources in, now is it?

Most people that use Linux like running it on low-end systems, like it because it's cheap/free, like the lower resources, or all of the above. Many users that buy high-end cards or regularly upgrade their systems (the kind of people that these companies like) play high demand games, do video production, 3D work, etc. While *nix/BSD systems do have a few high demand games running in them natively, along with other apps like those listed previously, they do not come close to those available for Windows, period. That's where the money is, that's where it's easier to offer the newest features, so that's where the attention is going to go.

So in short, nVidia is doing fine, and ATi is moving forward in *nix support. Considering that there isn't a ton of cash in the area right now and in the forseeable future, it's a nice surprise.

You're missing the point. Again.
 
nVidia are the only people who really know how AGP and the APU in nForce2 work and the only people who can write proper Linux drivers for it. It has nothing to do with the "Linux crowd," they don't know how nF2 works if they did, we would have seen drivers ages ago. nVidia already write drivers for Linux and FreeBSD so they have at least a working knowledge of the platform yet the nF2 drivers (particularly the APU which is the only feature I'm really concerned about. I know as well as you the Linux games market is still too immature to really make the lack of 3D acceleration a big deal) are still a joke. I don't hate nVidia (I bought their motherboard chipset didn't I?), but I expect Linux support to go the whole nine yards, support for all their products equally or none of them at all.

Quote:You're missing the point. Again.

nVidia are the only people who really know how AGP and the APU in nForce2 work and the only people who can write proper Linux drivers for it. It has nothing to do with the "Linux crowd," they don't know how nF2 works if they did, we would have seen drivers ages ago. nVidia already write drivers for Linux and FreeBSD so they have at least a working knowledge of the platform yet the nF2 drivers (particularly the APU which is the only feature I'm really concerned about. I know as well as you the Linux games market is still too immature to really make the lack of 3D acceleration a big deal) are still a joke. I don't hate nVidia (I bought their motherboard chipset didn't I?), but I expect Linux support to go the whole nine yards, support for all their products equally or none of them at all.

Ack, my point was that kernel development changes too much for many basic things to be supported let alone having major corporations dumping large amounts of cash into it for items that are still "toys" in the computing world. The server vendors are putting money into development because they are actually selling boxes preloaded with Linux (Compaq, IBM, etc) and they actually see some money in the investment. However, you seemed to have missed my point, again, yourself. Give them a more unified environment, and you will see a lot better results. Obviously by your complaint of the cost of your Radeon 9500 Pro in another thread you realize that money doesn't grow on trees, so where do you think the return on investment is? All of the hardware you are using was designed Windows, a platform that is on over 90% of the desktop PCs worldwide and is on many of the systems shipped out by system integrators. Linux, on the other hand, is a vendor-hostile environment since there is *so* much work to be done, yet there is such a little return on investment right now for many industries in it. I mean, if companies have had financial troubles (anyone remember 3FDX?) or driver and hardware development issues (Matrox, ATi, etc.) for the most popular platform in the world, then why start dividing already thin resources and moving to one that has far less return for sales?

So many people like to think that "Open Source" is the way to go, and that it's so much better than close source development. Well, things like a little trojan in the most popular SSH server app proved that can be a problem. People also claimed that Linux-based distributions are so much more secure than stable than Windows, and yet I have far more alerts coming from RedHat and Gentoo lists than I can recall from MS in the last few years. The concept of "having many eyes" looking at the source of an application to make sure that it's secure is *not* working as advertised, and when things like a simple editor (vi) can be used to take over an installation it does quite a bit to dimish the image of Linux. Do you think that any company would not only want to give out proprietary information on their drivers to make them open source, but then have them used to hold worms or viruses?

I know that you are pissed because you aren't getting what you want, but you have to remember that *you* bought hardware developed for Windows, and *not* for Linux. I am trying to explain to you that there are many reasons for slow development of Linux drivers, since there aren't many people on other forums (or anywhere else) that will. They aren't making hardware/software as a charity or some non-profit organization and it takes a long time to develop a succesful program around the BSD and GNU licensing models (IBM is one of the few that has). If Linux can stabilize long enough to get core concerns out of the way, and bring in *real* features to the kernel without 5,000 extra patches, then they can attract a much stronger (and wealthier) user base and then the vendors will follow. That's why I feel that Linux is to blame, and that is why I will continue to say so. I am a Linux user, and even I can accept this.

I haven't missed your point, it just wasn't relevant. If nVidia weren't already releasing Linux drivers for its graphics cards (or they were as dodgy as the nForce drivers) I might have agreed with you but the very existance of the nVidia Linux graphics drivers (and the ease at which they seem to have been ported to both IA64 Linux and FreeBSD) shoots your argument to pieces. If kernel changes make driver development so difficult why then, up until fairly recently, have there been precompiled binary drivers for virtually every major Linux distribution (all of which use different base kernel versions) and then after that, why did nVidia recently move to streamline the installation process to make it even less distribution dependent? The graphics drivers would also indicate that nVidia are have both the money and the resources to put into the development Linux drivers for its products. Finally, given that the Linux drivers share a fair bit of the same code as their Windows counterparts and that there are already decent drivers for AGPGART and the APU on Windows it shouldn't be all that hard to rig up something similar for nForce2. All this makes the lack of decent Linux support for nForce2 inexcusable and its no one else's fault but nVidia's.

Umm, actually, their existence tends to back up my opinion. If you note, there are a lot more owners of GeForce-based cards than nForce-based motherboards. Also, there has been more time for nVidia to do development work in Linux for these. Next, you will note that something as simple as a bootloader not working for some motherboard/BIOS combinations (GRUB) can be enough to thwart the interest of a developer to bother dedicating a lot of time to that OS. Also, if Intel (a much, much larger company with a *WAY* larger installed user base) had problems with their chipsets not having UDMA enabled (as such with my P4PE), what makes you think that a smaller company is going to feel inspired to dedicate a ton of time to getting theirs working? Again, this goes to resources being used which is something I have repeatedly stated. Larger user base = more money = more support. If this is something that you can't accept, then you're in for a long road ahead. Now, go out and get thousands more people to buy nForce-based boards for use in Linux and you will definately start seeing more support. Oh wait, if they feel the same way as you and think nVidia just doesn't like Linux, (hmmm, they did make some nice drivers for their video cards though so that's thrown out)or maybe does not like you personally and will never release drivers for the nForce chipset then they won't bother getting a board. Sounds like a Catch-22 there, eh? In any case, thank you for proving my "irrelevant" point.