There is no doubt in it that Double precision was one of the reason of selling dual GK110 based NVIDIA GeForce GTX Titan Z. According to reports new CUDA 7 REL upgrade will double precision to a surprising extent. Where the card typically reads as 1/3 DP, after the redesign, it falls to some place around 1/19 DP. For those of you who utilize FP64, this could be a genuine work-breaker and we would desire such users to stick to the current form until we have a lock on this conceivable bug.

The new CUDA 7 REL Update will Disable Double Precision computing on NVIDIA GTX Titan Z

GTX Titan Z full Specifications and Pictures

NVIDIA GeForce GTX Titan Z

The issue is starting yet unconfirmed and we are anticipating to the authority reaction from NVIDIA on the matter. There is a high probability that it could be a one-time bug in the drivers or extremely specific arrangements. Regardless, the issue of FP64 being injured on GK110 based TITAN cards is really genuine. The whole reason “semi-pros” throw TITANs in any case was a result of the double precision benefit. Without that, the TITAN rapidly turns into a celebrated gaming card with no professional advantage.

On the off chance that this issue is not a bug, then it would make for an exceptionally intriguing ideal model transformation. It would imply that that Nvidia has been quietly yet steadily controlling its FP64 customer base to the a great deal more extravagant Tesla lineup. The extent that the renditions go, the CUDA 7 RC variant seems to still have DP empowered. On the off chance that you have introduced CUDA 7 REL and had your DP crippled, just re-introducing the drivers won’t settle the issue. Utilizing a driver clearing system may work, however, your most solid option would in any case be a fresh install.

An alternate point to note here is that while we have an affirmed report of a TITAN-Z client, the bug or conscious disabled person could be appropriate to different models also – not only a TITAN-Z. We are going to cease from forming a hasty opinion on what precisely is occurring till we get word from Nvidia HQ, so stay tuned for a subsequent when that happens. For users in the feature business or logical industry, this is certainly one issue to pay special mind to.