I mean seriously, usually the next iteration is hailed as the performance part for a while. It doesn't seem that this happened with GDDR2. Am I missing something?
GDDR-2 was featured on the GeForce FX 5800 Ultra. the intent was to use faster memory so Nvidia could stay with a 128-bit bus, and still be competitive with ATI's not 9700/9800 lines.
The memory had some problems:
* Early yields were poor, making it rather expensive. Also, the memory used on the 5800 boards was actually not to GDDR-2 spec: it used the same voltage as original DDR (2.5v, instead of the spec 1.8v). This resulted in vastly higher power consumption over DDR (this is the reason why the leaf blower fan was created).
* With a maximum speed of only 1GHz, the DDR2 memory was unable to make 128-bit cards competitive with 256-bit DDR cards. Improvements in DDR speed and price only widened the gap (at the time, 700 MHz DDR was showing up in graphics boards).
* With GDDR-3 on the horizon claiming much higher performance and lower power consumption, interest in GDDR-2 as a high-end technology was short-lived.
Ultimately, Nvidia junked the 5800 series and created a 256-bit DDR card, the 5900. About a year later, special edition cards featuring GDDR-2 were released, but they were not popular. One year after that, cheap GDDR-2 running at the original spec of 1.8v replaced DDR in the low-end marketplace.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}