NVIDIA Corporation today announces its new GPUs code-named NV36 and NV38 in an attempt to fight its arch-rival ATI Technologies for the performance lead in high-end 3D gaming market. The Santa Clara, California-based graphics company installs GDDR2 memory, a type of RAM the firm had to reject from its products earlier this year, on its new high-end and performance mainstream graphics cards.
GDDR2 memory, a version of DDR-II for graphics cards, was claimed to be a revolution at last years’ Comdex Fall, where NVIDIA unveiled its GeForce FX GPU code-named NV30. The memory was able to function at astonishing rates, but produced a lot of heat, was not enough cost-effective amid the relatively low yields of the actual GDDR2-powered products. In addition, the GeForce FX 5800 Ultra, the first graphics card ever with 128MB of GDDR2, was too late to market and was not a good competitor for ATI’s RADEON 9700 PRO as well as RADEON 9800 PRO. As a result, NVIDIA discontinued the GeForce FX 5800 Ultra in favour of the GeForce FX 5900 Ultra that came out in mid-June, had some minor architecture improvements and conventional DDR memory.
Today high-speed GDDR2 memory is back on NVIDIA GeForce FX graphics cards, as the company announces its GeForce FX 5950 Ultra and GeForce FX 5700 Ultra graphics processing units.
The new GeForce FX 5950 Ultra is a follow-up for the GeForce FX 5900 Ultra architecture that improves the overall performance by using higher-clocked GPU and up to 256MB of GDDR2 SDRAM memory. All architectural innovations of the original NV35 processor are inherited by the NV38 GPU with no significant tweaks. Engine and GDDR2 clocks are said to be 475MHz and 950MHz respectively.
The NVIDIA GeForce FX 5700 Ultra GPU is the first chip produced by NVIDIA in 0.13 micron at the IBM foundry in
Leading graphics cards makers, such as Gainward, ASUSTeK, Albatron and Sparkle, have already announced graphics cards based on the new GPUs from NVIDIA Corporation. Actual products will be available in November, pricing may vary.