GTX 690!

I was expecting a 4 GB 680. Ended up reading 4GB GTX 690 at Tom's this morning. Wow....

Would this card do something about the numerical computing deficiency of the GTX 680? My suspicion is nVidia would keep the consumer-end and the professional-end cards apart.
9 answers Last reply
More about tomshardware
  1. Can't say because till now their is no 690 in the market ,therefore,it is really difficult say it can work more efficiently than 680 SLI or equivalent to it but I thing is sure getting 690 in place of 680SLI is really good for the consumers who don't like SLI setup.
  2. aviral said:
    Can't say because till now their is no 690 in the market ,therefore,it is really difficult say it can work more efficiently than 680 SLI or equivalent to it but I thing is sure getting 690 in place of 680SLI is really good for the consumers who don't like SLI setup.


    A detailed shoot-out at Tom's and other expert sites would answer all these questions....Let's see
  3. SSri said:
    I was expecting a 4 GB 680. Ended up reading 4GB GTX 690 at Tom's this morning. Wow....

    Would this card do something about the numerical computing deficiency of the GTX 680? My suspicion is nVidia would keep the consumer-end and the professional-end cards apart.


    Well, it is 2 GB per prosessor, so no extra memery in there... The real computayion card wait untill the end of the year or later to the next year.
  4. SSri said:
    A detailed shoot-out at Tom's and other expert sites would answer all these questions....Let's see

    I am also expecting the same .
  5. Right now the best consumer level cards for compute are the 580 and 7970, with the latter performing much better while using much less power. The only reason I'd see for buying a 580 at this point is if you absolutely need CUDA for programs that aren't built with a cross-platform API like DirectCompute or OpenCL. GCN simply scales better for compute than Kepler at this point but Nvidia is certainly planning something for Big Kepler that will solve this issue to avoid alienating their Quadro and Tesla customers. If you need a card now, I'd grab a 7970 if at all possible (580 if you absolutely need CUDA support), but waiting for Big Kepler will probably be the best decision in the long term although they may only release cards with better compute performance in the Tesla or Quadro brands, costing substantially more. The compute performance of GCN was one of the big reasons I grabbed an AMD GPU. I do OpenCL programming for fun and I wanted to run my programs on some new hardware and really test the architecture. After I read the Kepler review, I knew I had made the right decision, as my 7870 outperforms the 680 in most compute benchmarks (besides media encoding).

    http://www.tomshardware.com/reviews/geforce-gtx-680-review-benchmark,3161-15.html
    http://www.tomshardware.com/reviews/geforce-gtx-680-review-benchmark,3161-16.html
  6. aicom said:
    Right now the best consumer level cards for compute are the 580 and 7970, with the latter performing much better while using much less power. The only reason I'd see for buying a 580 at this point is if you absolutely need CUDA for programs that aren't built with a cross-platform API like DirectCompute or OpenCL. GCN simply scales better for compute than Kepler at this point but Nvidia is certainly planning something for Big Kepler that will solve this issue to avoid alienating their Quadro and Tesla customers. If you need a card now, I'd grab a 7970 if at all possible (580 if you absolutely need CUDA support), but waiting for Big Kepler will probably be the best decision in the long term although they may only release cards with better compute performance in the Tesla or Quadro brands, costing substantially more. The compute performance of GCN was one of the big reasons I grabbed an AMD GPU. I do OpenCL programming for fun and I wanted to run my programs on some new hardware and really test the architecture. After I read the Kepler review, I knew I had made the right decision, as my 7870 outperforms the 680 in most compute benchmarks (besides media encoding).

    http://www.tomshardware.com/reviews/geforce-gtx-680-review-benchmark,3161-15.html
    http://www.tomshardware.com/reviews/geforce-gtx-680-review-benchmark,3161-16.html


    Thanks for the detailed explanation.

    I'm not in hurry. It will be for my new build; I think it is worth waiting for.
Ask a new question

Read More

Graphics Cards Gtx World Of Warcraft Graphics