With all the recent leaks, rumors, announcements and launches of high end parts in both the red and the green camps over the past few months, low and midrange cards have taken a back seat. That changes today. Although Nvidia has announced its next generation “Fermi” architecture and released the details of its internal architecture, today they are showing off what they have been doing for the past few months.
Let’s say that the cards that launched today serve as testing ground for the bigger brothers coming down to the road to tackle Evergreen family. This means that Nvidia’s upcoming cards based on Fermi architecture will be based on 40nm fab process. Current Nvidia GPUs are manufactured using the 55nm process. Usually when you transition to a new fab process, yields are low in the beginning. Seeing that Nvidia’s next generation parts are made up of more than 3 billion transistors, it would not be wise to manufacture them on 40nm process for two reasons:
- Yields are low at the beginning coupled with the reports of TSMC’s troubles with the 40nm process.
- You do not want to induce errors into your multi-billion dollar chip due to inexperience with a new process.
Thus today’s low-mid range launch cards serve as practice for Nvidia on the 40nm process much like ATI did with the HD 4770 before launching their new architecture on the new process.
Some of the firsts for Nvidia today are:
- First 40nm GPUs from Nvidia
- First GPUs from Nvidia to support DX10.1
We will be comparing the the G210 and the GT220 to the HD 4650, which is the lowest performing card that I have in my video card suite and also checking out the performance in 3DMark benchmarks, games and video playback.