So the online rumor mill is saying that these cards could be out as early as July and that they will be based on a new graphics core that is code named GT200, that is supposedly much faster than G92. They are also saying that they will be on a piece of PCB called P651.

Does anyone have any idea what P651 is?

NVIDIA is stumbling right now so I really don't have high hopes for anything from them. It's great if it is true but I just don't buy it.

As of right now, I'm very disappointed that the last release of non-beta drivers for 8800 series cards was in December. Five months!?! WTF? I was forced to use beta drivers just to make Assassin's Creed run at a normal clip. NVIDIA is really disappointing me right now but I don't have any sympathy for AMD either so... Someone better do something major soon before I get bored.

The whole gaming scenario at the moment is really come to a halt....something big needs to happen.

Just seems boring at the moment I think.

I don't think it's gaming because there's a lot of good games out there. It's just the graphic card industry seems to have slowed way down ever since, well, the 9700/9800 series from ATI. Progress has been slow but steady but nothing major like the whole P4 -> Core 2 transition. There needs to be, fundamentally, a rethink of how graphics are done and how it can be done better. Perhaps DX10 and soon, OGL3 were the software side of that transition but the hardware really has failed to pounce on it.

Well, I guess I am waiting for a new decent FPS....Rainbow Six Las Vegas 2? Meh.

Maybe the reason the video card design manufacturing has slowed down so much is because they are up to something we don't know about?

I doubt it, but let's hope so.

I think they're both running into issues with something (perhaps process type) that is preventing them from making substantial steps forward. The whole unified model could also be throwing towels at them that they weren't expecting. We'll see if the 9900GT/GTX happens. Maybe it will be the kick in the rear the industry needs.

I see the main problem as being lack of competition at the high end.

Obviously, all mainstream/ low end cards are based off a high end version. And to be honest, why would nvidia push for a new high end or at least rush it, when they have not had any decent competition at the high end in the best part of 2 years?? (an 8800GTX bought in November 2006 is still pretty much the best card you can buy - apart from the ultra). And I dont mean these dual chip stitched-together solutions.

To see more advances in the GFX industry, there needs to be something to drive it -money. Money slows when there's competition, but there isnt.

Sure, there is competition in the mid range - but most punters buy the mid range based off the permance associated with the high ends SKU's.

FordGT90Concept said:

I hear ya on that one. I would really like to see something that comes out and stomps the 8800GTX like its integrated video. You would think someone would have done this by now considering that its been out for 18 months.

I really hope that the new cards whenever it is that they come out are smaller, or at least no larger than the 8800 GT/GTS... I've got about 3/4 of an inch or so before it becomes physically impossible to put a new card in my case.

Nvidia are such a shower of c**tS, they are as bad as creative labs. I mean at the moment they are trying to CON peeps into upgrading by bringing out 9800 cards which are just glorified 8xxxGTS series cards.....by rebadging these cards with a 9 series code they are conning peeps into thinking these are next gen cards when they are not. They are still based on exsisting tech.

A card costing as much as £75 will just about give the same performance FACT!!!!

Nevermind 9900 series......they are struggling to come up with a new GPU and with the competition they have now they dont even need too.

If the rumours are true that INTEL are indeed getting in on the GPU market (Codename Larabee) I hope nvidia die a f**king slow death.........

a little angry are we?

Jim H said:

Looks like it lol

I agree that the videocard market is rather boring atm. But who am I to speak? I'm still running a 7 Series chip /flex

To me it seems that Nvidia is holding out on us. I think they have the technology to come out with blazing fast new graphics processors, but due to the lack of competition, they have been holding out. The Geforce 9 series would have been better called the 8900 series, the 8800GT should have been marketed as an 8850GT and the 8800GTS G92 could have been the 8850GTS.

I honestly believe that it comes down to four separate things.
1. POWER vs. power. Just how many watts can these monsters pull before you have to have external PSUs and cooling systems.
2. Access to the highest end fabrication facilities. In order to increase the speed, while maintaining a reasonable power (wattage) draw requires Billions of dollars in new fabrication facilities. (this one is related to #1, but is a separate issue)
3. Bus speed limitations. Even PCI-E 2.0 16x2 or 3 or 4 show that there are limitations in how fast the GPU(s) can talk to DRAM and/or the CPU. The incorporatoin of a physics engine on the GPU (I seem to recall this is something that NVidia is working on as I seem to recall either they or AMD just purchased the folks that made a stand alone PEU) would be I right step in this direction, but would require a majority of games to utilize a underlying standard to redirect this type of traffic from the CPU to the GPU/PEU combination ... DX10.2 or 11 perhaps?
4. Which strategic approach do you take to GPU design over the next 5-7 years? There does seem to be some scientific and technological preasures to stop making specific purpose processing units and to focus on True General Purpose, RECODABLE RISC or CELL Processors that are HIGHLY scalable and microcode modifiable in order to assign cores from a general purpose processor to whatever function is necessary. This is to say if you have 4 cores, you can individually, and on the fly (or at least on a reboot for now) reallocate the cores between MPU, CPU, GPU and PEU functionallity, and if you need more horsepower, you can scale in at 8, 16,32, 64, etc. cores, all of which will be assignable via BIOS and OS. This approach, while it seems inevitable, will require multiple regulatory commities as well as OEMs to coordinate on generating the standards necessary. BTW, this type of approach has worked well with very high end Silicon Graphics Mini and Mainframe computers since the 1990s.

So even though I've focused on NVidia, the ATI/AMD merger seems to be making more sense....If they can ride out the slack time.

IBM/Intel are not new entrants to the GPU market, they've always been there. It's just that they've not focused thier considerable talents on the high end GPU market in quite some time. I would be more than happy to see a third entrient into the high end GPU market, as I seem to recall a time when we had many, and the truely massive changes came about when they were all fighting. S3, Diamond MultiMedia, Intel, ATI, NVidia, IBM, Motorola, Silicon Graphics (SGI), Matrox, 3DFX, Cirrus, the list goes on. I still remember a late `80s manufacturer that made a scan doubler that actually started the EGA craze at that time... I just can't remember the name right now.

So we've come a long way, but the next several years are critical, especially if we want our portable gear to give us some of the same capabilities as our NOW behemouth tower cases. Sony is a early adopter with it's use of cell processors in the PS3, but I think we will see moves towards more processors of every type to this infrastructure.

Remember that Linus Torvalds is or was working on a Trancender Processor at Transmeta that shows some of these capabilities a few years back... I've lost track of him since then, but as an inovator he ranks way up there, even maybe above Steve Jobs.

Gerritt

http://www.canadacomputers.com/index.php?do=ShowProduct&cmd=pd&pid=018031&cid=999.243.390

is this the newest card out?

isnt this a very good price for this card or is it just me?
http://www.canadacomputers.com/index.php?do=ShowProduct&cmd=pd&pid=016647&cid=999.243.390

I see you have the 7600GT

Are you thinking of upgrading it? I have one as well and am thinking about an 8800 series card

yeah im wanting to upgrade it. but everyone is saying to wait for the "new" cards. Theres always something better coming out, why not just buy now.

But yeah, the 8800GT looks like a decent performer. I was also thinking about the 9600GT, but i think the 8800GT is the better card right?

So is that a good price for the card?

CPUs are mostly ALU intensive and quite weak in FPU. GPUs, on the other hand, are all about FPU. What they essentially did was increase the number of FPUs because, after all, gaming uses a lot of both but FPU takes a lot longer to calculate. Essentially what I am getting at is that multiple cores in a single processor are going to encounter the very same limitations as current multi-core processors. It doesn't matter how many FPUs you pump in to it, it still won't perform as well as a discreet video card.

For that reason, I still don't get why NVIDIA/AMD haven't aimed for the stars. They need to set up a ridiculous test and make a GPU capable of doing it without breaking a sweat. For instance, make an application that generates 1,000,000 triangles with 1,000 different textures that runs at 85 FPS with 16x AA, 64x AF, very high quality shaders and lighting, with a resolution of 2560x1600x32 @ 85 Hz. Basically, you got an expensive card with a complex GPU (thus, low Hz) that can do pretty much anything without breaking a sweat. As the market matures, they can make the card last a long time by raising the clock speed over time and making the necessary changes to make sure it remains at the very pinnical of display requirements. For cheaper models, they would, for instance, lower the requriment to 500,000 triangles and 500 textures. As long as those numbers aren't exceeded, the card should perform fabulously at the highest possible resolutions.

Because monitors are the limiting factor in graphics, why don't they push the limit of the most ridiculous monitors and software requriements?

Ford,
At least you of one, actually understands where I'm comming from in the last post.
Gerritt

I'm pretty sure AMD or Nvidia would be able to come up with a GPU such as described by Ford - the main probelm being cost.

No sense making a behemoth of a GPU, that will do what you ask, if you're only getting 10 per wafer.

Let alone the power requirements etc etc. You'd be talking $20k gfx cards.

The contemporary designs are inherintly scalable to such a size as you describe. But the power required to supply a circuit that large would be monumental.

Once again, i think the reason we have a stagnant period on GFX development is pretty much down to competition. Why release something new and fantastic, when your current cards still perform the best, and are still selling like hot cakes.

Its all about the $$$$ in the end.........

There are people that would pay for it at that price and, if they have no problem with that price tag, they'll have no problem with having a stand alone 1000w PSU just for the card. XD

The advantage of making a behemoth like that is it could be easily scaled back to meet "high" end and mid-range at little extra cost.