NVIDIA IS STILL not talking to us, and competition in the graphics world means you the reader win. They have five new 40nm mobile parts launching tomorrow, luckily we can bring you the specs now at discount prices.
The short story is that there is little of interest here, the top part is 40% of a GT200, slightly downclocked. The rest is pretty standard, nothing to write home about. Three use GDDR3, two use GDDR5, and all are 40nm.
Here is all you need to know
Won’t it be fun to watch and see if Nvidia has figured out their defective GPU problems on the 40nm node? The early adopters will be fun to watch. If you look at the numbers though, the 230 and 240 have the same TDPs, it looks like there are still a lot of leakage problems and NV is using creative binning to get around it.
Same with the 250 and 260, a 10W TDP difference for a 10% speed bump? Ouch. Looks like NV is barely scraping out 40nm parts with terrible bin splits and severe clock limits. If asked, they will say everything is perfect though, there still isn’t an official admission about which chips are bad, so what is a little whoopsie like yields in the toilet?
The poor sods who sit through the official presentations will be awash in how cuda will bring greatness to the starving children, and how the one game with physx is worth playing. *YAWN*. There will be also be lots of one sided slide shows about how the parts are faster than light, and if we are lucky, only two journalists will commit suicide before the presentation is over, they are honestly that boring.
That is why we bring it to you in short form. You can get all the info early, without handcuffs, and without having to sit through the mind-numbing presentations. Instead of reading a 24 page writeup, you an sleep in for an extra 30 minutes tomorrow. You will be glad you did.S|A
Latest posts by Charlie Demerjian (see all)
- AMD puts massive SSDs on GPUs and calls it SSG - Jul 25, 2016
- AMD revamps workstation lines and adds three cards - Jul 25, 2016
- AMD unveils Loom and ProRender software - Jul 25, 2016
- Emulex puts NVMe over FC with NVMf - Jul 20, 2016
- ARM needed Softbank, Softbank wanted ARM - Jul 19, 2016