NVIDIA has taken its time when it came to announcing the Titan version for the Maxwell architecture from its previous iteration, the Titan Z. However, this time they have stepped their game up and at an engagement at Stanford University, Jen-Hsun Huang unveiled the new NVIDIA Titan X, for the Pascal architecture.
The specifications have been taken up quite a few notches in comparison to its Maxwell counterpart. Check them out below:
- 11 TFLOPS FP32
- 44 TOPS INT8 (new deep learning inferencing instruction)
- 12B transistors
- 3,584 CUDA cores at 1.53GHz (versus 3,072 cores at 1.08GHz in previous TITAN X)
- Up to 60% faster performance than previous TITAN X
- High performance engineering for maximum overclocking
- 12 GB of GDDR5X memory (480 GB/s)
Almost everything in this list is a major improvement from last year’s Titan X. NVIDIA themselves, in their blog say:
We said our GTX 1080 delivers an “irresponsible amount of performance.” It was a bit reckless. But this is even more reckless.
Right from the record breaking number of CUDA cores to it achieving a mind boggling 11 teraflops of computing performance, NVIDIA surely has won the hearts of the enthusiast market here. However, it just as always does not come at a cheap price as the Titan X is slated to release on 2nd August, 2016 with a hefty price tag of 1200$.
Here is a direct comparison of the various technicalities currently in the GPUs at the enthusiast level: (Source: AnandTech)
Has NVIDIA hit all the right chords here or will we see their current ‘flagship’ model fail to attract customers? Let us know in the comments below!
Check out the YouTube link here: