GeForce GTX 870M 

I wonder what this means for the Tesla line, and whether we will see a variant of Maxwell with a large boost in 64-bit performance, or if that line will be in an awkward place until Pascal.

The memory, but not the cache, is capable of ECC (error-correction) for enterprises who do not want a stray photon to mess up their computation.

The TMU count is a bit off since the maxwell core architecture holds 8 TMUs per SMX so we were supposed to see 104 TMUs in total instead of 138 so it could be a GPU-Z fault which could mean that these specifications might not be the actual deal.

Last time, I did the analysis of the leaked GeForce GTX 880 PCB which was in engineering state and found out some intriguing details such as the 256-Bit memory bus, the GM204 chip being larger than GK104 while being based on the current 28nm process node.

If the recent GeForce GTX 880 benchmark is taken for reference, we might be looking at a 20-30% performance uplift from Maxwell over GK110 which is insane and makes us wonder the performance uplift with the GM200 core which won’t ship to consumers till next year.

Soon after launch, NVIDIA is expected to launch a third Maxwell graphics card aiming the more budget and gaming crowd, known as the GeForce GTX 860.

The details received only specifying a high end Maxwell GM204 housing a 256 bit wide bus.

This is because even though TSMC has begun volume production on the 20nm node for everyone, the process is facing maturation problems for high performance ASICs like the GM204, but this is expected to start from sometime next year.