![[H]ardOCP](/images/blank.gif){kind=small}
- Date:
- Wednesday, June 22, 2005
- Author:
- Brent Justice
- Editor:
- Kyle Bennett
- Google +1
NVIDIA GeForce 7800 GTX Preview
NVIDIA is today launching their next generation video card, codenamed “G70.” Inside we cover what this new NVIDIA powerhouse is bringing to the table and how it stacks up against other high-end video card solutions in true gaming experience.
Transistor Madness:
The GeForce 6800 Ultra was large in its own right, but the GeForce 7800 GTX takes the cake. The GeForce 7800 GTX is manufactured on TSMC’s 110nm process. The GeForce 6800 Ultra used IBM’s 130nm process and has around 222 million transistors; the GeForce 7800 GTX has 302 million transistors. As you can imagine, that is a lot of transistors. To put this in perspective, NVIDIA has concocted a comparison chart that adds up the total transistor count in several console systems and even an Athlon64 FX-55, and the total combined count falls short of the number of transistors in the GeForce 7800 GTX! This is one complex chip folks.
Power and Heat:
Because it is such a large and complex chip, one would think the heat and power requirements would be enormous. Amazingly, however, NVIDIA has been able to reduce the power consumption so that it actually requires slightly less power than a GeForce 6800 Ultra. NVIDIA has implemented a very smart and efficient GPU design to manage those 302 million transistors.
Parallelism is a great thing, and graphics GPUs can take advantage of parallelism to great extremes. Increasing ALU width and keeping the same clock speed can result in the higher performance with similar power requirements. With parallelism, the G70 has increased in pipeline width with only a slight core speed increase, plus the 110nm manufacturing process, which results in the GPU actually using slightly less power than the NV40. NVIDIA has also implemented clock-gating technology that shuts down GPU transistors that aren’t in use to reduce power and heat. The GeForce 6800 Ultra draws around 110-120 Watts of power; the GeForce 7800 GTX draws around 100-110 Watts of power.
The power supply requirement for a single GeForce 7800 GTX is 330W, while a GeForce 6800 Ultra requires a 350W PSU. When you compare the performance of the GeForce 6800 Ultra with its power requirements to the GeForce 7800 GTX’s performance with its power requirements, you will find the 7800 GTX delivers much more frames per second per watt of power.
With all the combined power saving features, NVIDIA was able to produce the GeForce 7800 GTX as a single-slot solution.
Design Goal of the 7800 GTX:
The design goal of the GeForce 7800 GTX is based on architectural improvements rather than brute force. Instead of massively increasing clock speed and memory to ramp up performance, NVIDIA has more elegantly made the proven architecture more efficient with more shader computational power. NVIDIA sees applications taking greater advantage of shaders thereby becoming bound by the performance at which the GPU can compute them. NVIDIA also sees HDR becoming more popular in games with games making use of higher precision targets. NVIDIA also believes that transparency textures in games are on the rise, and there are plenty of games now using transparencies.
With the GeForce 7800 GTX, NVIDIA has chosen to increase performance by utilizing parallelism to improve the width and depth of the pipelines. They have increased the computational math power per pipeline and per clock so that the 7800 GTX can, on average, maintain a 2x performance advantage in pure shader performance over the GeForce 6800 Ultra clock for clock. NVIDIA has also improved HDR performance and is introducing a couple of new Transparency Antialiasing modes that we will discuss later.