Nvidia rolls out latest laptop graphics chip
GeForce 200Ms offer more performance, less power hungry than 100Ms
Nvidia has introduced a quintet of 40nm mobile graphics processors to fill out its GeForce 200M series.
Spreading across almost the full range of market segments, with parts targeting Mainstream, Performance and Enthusiast buyers, all replacing older, 55nm GeForce 100M chippery.
The Mainstream part is the G210M, a 16-core GPU that runs at 625MHz. Its memory - up to 512MB of GDDR 3 - runs at 800MHz and connects over a 64-bit bus.
The remain four GPUs have 128-bit memory buses and hook up to a maximum of 1GB of VRam. In the case of the GT230M, it's GDDR 3 - the GT240M, GTS250M and GTS260M can all use GDDR 5. The two GT chips' memory is clocked to 800MHz. The 250M's Ram runs at 1.6GHz, while the 260M's is set at 1.8GHz.
The GTs contain 48 processor cores, while the GTS parts have 86. The GT230M and GTS250M run at 500MHz, but the GT 240M and GTS 260M are clocked at 550MHz.
But there's a clear increase in graphical horsepower moving up range, with the G210M, GT230M, GT240M, GTS250M and GTS260M notching up Gigaflop scores of 72, 158, 174, 360 and 396, respectively, Nvidia claimed. They all do DirectX 10, but not 10.1.
Nvidia said those figures represent big performance jumps over previous-generation mobile GPUs but not at the cost of a higher power consumption. Quite the reverse: the 200Ms use less energy overall than 100Ms. In the case of the G210M, 50 per cent less.
We have to stress that that's idle power consumption - the power used when the GPU isn't actually doing anything, but the shift from a 55nm fabrication process down to 40nm should help reduce flat-out power consumption too. We'll tell you how much when the review the chips in due course.
Meantime, Asus and Acer will be the first laptop makers with GeForce 200M-equipped machines on the market, but Nvidia said more than a hundred machines from "all the top vendors" will be released during the second half of 2009 with its new chips on board. ®
Sponsored: Hyper-scale data management