Compared to the current flagship of the NVIDIA GeForce range, the GTX 980, the performance gain of the GTX Titan X is significant.
The graphics card GTX Titan X, presented to the public at the GDC (Game Developers Conference) 2015, should embody NVIDIA’s know-how regarding video acceleration and thereby become the manufacturer’s technology showcase at a time when direct competition, specifically AMD, is not far behind.
The GTX Titan X is based on the GM200 (Maxwell architecture), with 3072 CUDA cores, 192 TMU and a maximum of 12 GB of 384-bit GDDR5 memory. This card features two power connectors, like the GK-110 PGPU accelerators, which would place it in an maximum TDP of 250 Watts. It could even be less, given the improved energy efficiency of Maxwell over Kepler.
All these conjectures aside, the card tested by VideoCardz (www.videocardz.com) and ChipHell (www.chiphell.com) achieves outstanding results. Our colleagues note a 3DMark 11 Performance at 22 903 points, 35% better than a GTX 980, which is the current top of the line GeForce. Other benchmarks (Extreme, Extreme Firestrike and Firestrike) give similar speedups, always above 30% – of what, again, is probably a pre-production sample with drivers still developing.
As soon as AMD announced the upcoming release of the liquid-cooled Radeon X R9 390, which is believed to be the first to use HBM (High Bandwidth Memory) technology, NVIDIA has set the performance bar pretty high – which remains to be seen once production boards of both models become readily available. This generational leap is important, if only because this architecture is shared with the Quadro M6000.
© HPC Today 2020 - All rights reserved.
Thank you for reading HPC Today.