EggXpert

I am in the market for a SLI system. I have used ATI in the past but have had issues with the drivers they make for the last couple or years. Also I hate hate hate the Catylist control center that it trys to install, I do know that you can get the driver without it but I still hate ATI's driver packedge including the Old freaking splash screen of some kind of red digital blob that they refuse to let die.

 I am running Windows 7 X64 Pro on an i7 920 system with 6 gigs of ram and can play Fallout 3 pretty well on Ultra settings with AF set to 8X. I am looking into playing more games that will stress the card that I have currently (Geforce 8600 GTS) and the future of DX11.

I have been looking at getting 2 GTX 285's and SLIng them together but I don't want to drop 500+ on non DX11 cards if it is going to be the way games are going.

I guess the real question is in you honest opinion should I wait for Nvidia's Fermi chip sometime in December or next year. I know info on this is sketchy at best and no firm release date has been set. I am not sure if what I have to spend will even be able to buy one GTX300 series card when they get here anyway.

I am getting really ansy and want to make a move, but again I do not want an ATI card for my own personal reasons. I am not an Nvidia Fanboy just like the driver support and how easy it is to get ahold of them. Unified drivers are the way to go in my opinion.

Thanks for any help.

Ok then you will have to wait for the GT 300 series cards to release next year if you want DX11 or go with ATI's 5800 series. And they are not going to be cheap from what I understand.

The high end GT300s ($400+ per card) will be released in a month, the low end GT300s are set for Feb of 2010..

If you're an nvidia fan like me, I'm sure you'll just wait the month or so..

I'm still torn between getting a GTX 360 in december or waiting until the end of feb. for a GTS 350..  Might have to rebuild my gaming PC, yet again.. I've rebuilt it like 3 times in the past 2 months..

You can expect the GTX360s to perform the same at a 5870 give or take 5%. The GTX 380 of course, should take the crown, at least unless ATI rolls out a 5870 X2 edition in the $600+ range per card which they likely will before the end of the year.

Curious where you are reading the month for the first GTX300 cards? I see a bunch of speculation but no real firm dates anywere. I will most likely get one of the high end cards and just see what type of performance that gets me then get a second if I need (want) to later.

http://www.techarp.com/article/Desktop_GPU_Comparison/nvidia_4_big.png

Feb is the middle of Q1 2010..

 

I have also heard that you can leave in a older Video card for PhyX but I am new to this concept is it worth doind and what model Card is required if this is true?

Thanks

More then likely the 300 series will have PhyX I don't see why Nvidia would take that away when they are the ones that bought out Ageia in the first place.

 

http://www.brightsideofnews.com/news/2009/9/30/nvidia-gt300s-fermi-architecture-unveiled-512-cores2c-up-to-6gb-gddr5.aspx

http://www.brightsideofnews.com/news/2009/4/22/nvidias-gt300-specifications-revealed---its-a-cgpu!.aspx 

http://www.overclockersclub.com/news/25640/

http://www.engadget.com/2009/04/26/nvidias-gt300-specs-outed-is-this-the-cgpu-weve-been-waitin/ 

Yes, you can throw in any 8000 series or later card and use it for PhysX, even on motherboards that don't support SLI or Crossfire.. 

 However, keep in mind that recent nvidia drivers have disabled using an 8000 series graphics card for PhysX IF you are rendering on an ATI card..  It's kind of an under-handed thing to do, but there may be very good reasons for it related to possible compatibility problems with ATI's new 5000 series that needs to get worked out in later drivers.

But yea, if you keep your cards nvidia, then any 8000 series card or later can be turned into a dedicated PhysX card, and it works quite well.. Problem is there are stlil very few games out there that use PhysX for their physics (some use Havok, some use their own proprietary engine to get around having to pay PhysX Royalties).. So it's not worth doing it unless you do have spare cards around and play PhysX games

List of PhysX games is here:

http://en.wikipedia.org/wiki/PhysX

If I were in your shoes, I will wait for awhile at least after Nvidia rolls out its new card and read some benchmarks.

Nvidia must be having a lot of trouble with the 300 series if they can't release it before the busiest shopping season of the year.

Frankly I'm not expecting any of the 3xx series to be out untill next year. Between the crazy low gt300 yield rate and the fake "fermi" card they just showed off things arn't looking good.

And this is coming from a big green team fan.

KingofMadCows:

Nvidia must be having a lot of trouble with the 300 series if they can't release it before the busiest shopping season of the year.

nvidia's been busy with Ion, but they also have a bumpy road.. It wasn't that long back that Intel pulled out the "Oh, you have a license to sell chipsets for our old processors, but you don't have a license to sell chipsets for our new chips. If you want to make chipsets for our new i5/i7 series, since you're directly competing against our onboard graphics chipset, we're going to charge you much more money to get the license because you'll be costing us lost sales of our own chipset." ..

Some serious bull from Intel that nvidia has to deal with right now.. I mean, this is after nvidia was very open and licensed their SLI technology to Intel so Intel could build chipsets that supported SLI.  Meh, whole bunch of legal problems nvidia has ahead of them..  AMD & ATI are in bed together and Intel is making things hard on nvidia because Intel wants to push their own onboard graphics and chipsets.. Not that Intel chipsets are bad or anything.

Obviously Intel is still sore over nvidia barging in and pretty much kicking Intel's integrated graphics chips right out of Apple's desktops. So now Intel is trying to force nvidia to repurchase ridiculously expensive licensing rights everytime Intel rolls out a new architecture.

serves them right imo for all the bull they pulled these last couple of years

karma came to bite them in the a$$

I am not getting any good feelings from Nvidia.. I think the release is getting pushed back and they are not concerned that people are making a move to ATI in the mean time. I mean I am pretty much fed up with the fact that I have to wait at all.

The funny part is on Hexus ATI and Nvidia execs are fighting back and forth about Nvidia not caring about the gamers anymore and that Direct Compute is all they care about. I would tend to agree.

If direct compute is soo important than put it in Tesla and Quadro only cards and leave the Geforce cards to the gamers.

And the issues with TSCM or TSMC (what ever teh acronym is) is having so many yeild issues why would they have not been able to correct this before now. I mean Intel has been producing 45nm chips in quantity for like a year now.

Imho I don't know why Intel doesn't just purchase Nvidia and dominate the market. I would love to see that paring. Then the lawsuits and licensing issues would be solved.

Also has anyone thought much into RamBus sueing Nvidia over tech that is in Fermi as to why they are delaying it so much?

tuberthechosen314:

If direct compute is soo important than put it in Tesla and Quadro only cards and leave the Geforce cards to the gamers.

You're not getting it...The evolution of DirectX from a fixed pipeline to a programmable shader pipeline to programmable geometry shaders and so on will eventually lead graphics cards down this path anyway. Graphics cards of the past could not compute how light refracts off of water, they couldn't compute meshes that explode into different chunks, they couldn't compute objects that glow so bright they impact the lighting of other meshes..  The fixed pipeline model of graphics cards is the wrong way to do things if you want to get realistic graphics. If you want to get a jist of what a fixed pipeline model looks like, go back and play the original Doom 2.. The only thing the graphics cards were capable of doing in those days was drawing a frame that was already computed, colorized, and rendered using the CPU.  Running modern games on a fixed graphics pipeline is, to put it frankly, impossible.. The CPUs of today just cannot do that much lighting, rendering, and shader computation in addition to handling the more advanced A.I. and physics that modern games have..

 So basically, what I've read your argument as.. That you're not interested in Direct Compute, OpenCL or any of these other architectures, is that you don't care at all about awesome realistic looking graphics.  That may be fine for you, but most people don't feel that way.. These architectures are bringing with them advances in computer graphics that running them in real-time is completely unheard of.. Even today, if you want Ray-Tracing and 100% accurate mirror-to-mirror-to-mirror-to-mirror reflections then you're going to need to sit and wait a few minutes for the CPU to render a single frame.. Fermi and these other architectures can do Ray-Tracing now which means they can do the same frame in seconds..Ray-Tracing light is about as realistic as you can get and we're not even 10 years away from playing Ray-Traced games and the only way that's going to happen is if the graphics pipeline is more flexible.. Which is the whole point..

 

 

tuberthechosen314:

And the issues with TSCM or TSMC (what ever teh acronym is) is having so many yeild issues why would they have not been able to correct this before now. I mean Intel has been producing 45nm chips in quantity for like a year now.

It's all about money.. nVidia has an obligation to stock holders to make as much money as possible.. Sure, they could have had the issues ironed out months ago if they dropped millions of more dollars into it..  But if people are still going to buy their old generation cards because they're good enough, then nvidia makes more money by just waiting for the chip fabrication technology to develop.. In other words, there likely was no financial motivation to rush into the new technology as the GTX 295 was pretty much unrivaled..

 

tuberthechosen314:

Imho I don't know why Intel doesn't just purchase Nvidia and dominate the market. I would love to see that paring. Then the lawsuits and licensing issues would be solved.

More people buy CPUs than GPUs..  I mean, corporations could have thousands of computers, with thousands of CPUs, and no graphics cards..  It's not uncommon for CPUs to lead the way to new chip fabrications tech..

 

tuberthechosen314:

Also has anyone thought much into RamBus sueing Nvidia over tech that is in Fermi as to why they are delaying it so much?

?? What technology are you referring to?  I'm pretty sure nvidia would license whatever it is that they need to use..

All good points. Thanks for educating me.  Didn't realize that the Direct compute impact would be so big for games, gamers, the media on-line makes it sound like a research only type of technology.

Also the ram bus issues may not be big but with anything big or small it could affect the release of the new chip. Here is just one of the links to the story: http://www.pcworld.com/businesscenter/article/148236/rambus_sues_nvidia.html 

Not sure if it is worth consideration but you never know.

 Thanks for the actual knowledgeable responses. I have been out of the cutting edge for some time and have no one knowledgeable to bounce things off of so that is why I love EggXpert so much.

 Thanks