Chordonblue's Journal: NVIDIA = Intel Two Years Ago

Was it that long ago when it seemed that Intel just couldn't do anything right? Not only was there the RAMBUS debacle, but then there was the horrible chipset mistakes, slow performing processors, and all sorts of production problems.

They call it 'execution' folks. Intel's got it back, but NVIDIA seems to have lost all semblance of it.

Screwups? Lots. The FX series of cards is underpowered for DX9 games. This didn't seem like that big a deal until the Half Life benchmarks came out. Now it's a VERY big deal. A buddy of mine had just purchased a 5900 Pro right before ATi's 'Shader Day'. Talk about feeling '0wns0r3d'!

Then there's the NFORCE3. What's wrong with that? Well nothing much if you don't mind a loss of about 33% in AGP performance and no SATA...

NVIDIA, what the hell is wrong with you??!! You've had how much time now to get this chipset right? AMD's schedule slipped WAY back, so you should have had mucho time to fix these problems before now. How it is that VIA is able to kick your ass in the Athlon 64 benchmarks is simply shameful. Get it fixed - now.

There's been some talk on the boards that the NV40 will save NVIDIA's butt, well maybe that's true. But don't forget, ATi is still out there with the R400, and I'm sure they won't be playing softball either. It's sure to have killer performance.

The real question is: How did NVIDIA drop the ball in the first place? The answer lies closer to the way NVIDIA and ATi are structured I think. ATi has chip foundries. They can have the tech to make their own chips. True, it's not the most state of the art compared to Intel, Via, AMD, or some of the Taiwanese chip firms, but least they fully understand what they're dealing with.

Then you have NVIDIA. They build advanced chip architectures, but they must farm these out to others to build them. I imagine there can be a lot of miscommunication and trial and error going on here. In the old days when ATi really wasn't concerned about being very competitive in the 3D field, NVIDIA could take time with their partners to get the new process technologies right. Now, they can't.

When .13 micron appeared two years ago NVIDIA jumped on it, even though it really hadn't been proven yet because they felt it would give them the edge on ATi, who was using a .15 micron tech. Had they delivered the FX series on time it would have been very competitive for sure. But the process technologies weren't ready yet which caused delays, the yields were awful, there were heat problems, and there was another problem related to a decision made at a much earlier date - memory choice.

While ATi chose to stick with good 'old' DDR, NVIDIA attempted to jump the gun and go with DDR II which is faster, however has less total bandwidth (128 bits vs. 256). Higher clock speeds meant more heat, and again, more problems as well, hence the 5800 "Dustbuster".

These missteps may have cost NVIDIA a lot of money. The whole FX series will be looked upon as inferior when the new crop of DX9 games arrives. There's simply no other way to see this. When you have a card like the 9600 Pro selling for $200 and kicking total ass at HL2, and then you have a $500 5900 PRO barely able to keep that pace, something is very wrong and no Detonator driver is going to somehow ride in and save the day here folks.

I supposed the biggest problem from NVIDIA's standpoint would be all that lost time and opportunity. They've given up more ground to ATi than ever before, which of course has emboldened ATi to get even more into the game by purchasing better process technologies. What can NVIDIA do now?

PREDICTIONS:

1) NVIDIA goes the way of 3Dfx. Which is kind of fitting since they actually own them now. They can't maintain this kind of rep for long, they will need to rebuild their reputation and that will take a lot of hard work. It may well be that even if NVIDIA survives, they will be a very different company that we know them as now. Which leads me to this, more likely scenario.

2) NVIDIA merges or gets bought by someone with process technologies. I truly believe that in order for NVIDIA to remain competitive, they will need to have more control over the actual chip making process. ATi may kill them if another misstep like this happens. The thing is, when you own your own chip foundry, you truly have your fingers on the pulse of the operation; but when you farm things out, you only have the word of those who own the equipment.

It is entirely possible to imagine IBM or even AMD buying or merging with NVIDIA. Stop laughing back there, I'm very serious! NVIDIA's got good tech, but they have desparate need of foundries now that ATi and even upstart XGi may be in the game for real.

Either way, I don't believe NVIDIA can continue on their current path for much longer. Things will need to change and change quickly. Let's hope they stick around. The competition has been excellent for all of us!