New NVIDIA Titan X GPU, based on the Pascal architecture, announced

NVIDIA has announced a new Titan X graphics card that will be based on the Pascal architecture. The new NVIDIA Titan X features 3584 CUDA cores and its core will be clocked at 1.42GHz (with a boost clock at 1.53GHz). The GPU will come with 12GB of GDDR5X VRAM, and will offer 480GB/s memory bandwidth.

The NVIDIA Titan X will feature 12 billion transistors and will sport 11 TFLOPS. The GPU will require one eight-pin and one six-pin PCI-E power connectors, and will come with 3x DisplayPort, HDMI 2.0b and a dual-link DVI port.

This new NVIDIA Titan X GPU will be priced at $1200 and is schedule for release on August 2nd.

Enjoy!

NVIDIA TITAN X: The Ultimate Graphics Card, Powered by Pascal

56 thoughts on “New NVIDIA Titan X GPU, based on the Pascal architecture, announced”

    1. “Stop_AMD_Propaganda”

      Oh yeah, this guy’s definitely not totally, completely, utterly biased……….

  1. Okay i’m ready. One question though. WHY nvidia doesn’t partner directly with h20 cooler companies to make a “hybrid” version themselves? Wouldn’t the markup be higher so, potentially, more money for them ?. I’m tired to wait 3 months after vanilla release date to get silent yet efficient cards.

    Painful.

    1. Titan cards usually dont see AIB models as in you wont be seeing MSI Titan x gaming z or evga titan classified, the titans always stay at the reference model, its almost like those are exclusively held out for only NVIDIA directly. Try finding a Titan Hybrid or Seahawk, they dont exist unless custom done by the one who bought it. You can find the evga reference or pny reference but you wont see custom pcbs or custom coolers on titan products.

      1. Yea i was refering mainly to custom coolers. I know pcbs stay the same etc. It’s just that demand for these products tend to be higher than what they can manufacture early on. So we always end up f5’ing store page every 15mins for weeks. Annoying.

      1. Amd can’t charge this much with their actual market position. They have to be aggressive on the price war. We haven’t seen benches yet but it’s safe to say that dx12 positively benefits amd much more than nvidia (when properly done “time spy benchmark fkup”). I am not sure a Fury “vega” would be able to get as much raw performance as this card but the price would be considerably lower hence why it would be a better buy. Also considering that dx12 will be the main api for years to come (sadly i’d prefer vulkan but ms are dixs soooo yea) amd would be a better choice. But i’ll end up getting that titan anyways personally, price isn’t a concern and i just prefer Ngreedia over amd for now (580/670/690/Tx).

        But nothing says i won’t be switching someday. Recently amd got their act together. Both on the driver and hardware side. Plus they have a good advantage over NV and their prices are more “people friendly”. Cause you know not everybody can buy a 1200$us card. I’m letting amd do their moves but as of now i still think nv is better imo.

          1. Yea true that. Not everybody has that kind of money to invest in a gpu. When you think about it, it can be the price of a whole pc… But they’ll sell like hotcakes as usual! I like hotcakes.

          2. I will :). Only to be pissed 3 months later @1080ti being as powerful and 300usd less…

          3. regular gamer? probably not many. but for people needing compute but does not have the money to get full fledged tesla then this kind of card will be a “steal” for them. even titan x maxwell was quite popular as a compute card according to anandtech.

        1. vega will be HBM2, when titan is only g5x. though vega will be much smaller chip (there should be 2 chips and we dont know much about any of them), but easier to manufacture and under DX12 could be close to GP102 performance. Really depends on what libraries they use and how far the fregvency will get. Save to say price will be lower though.

          1. Oh and when will vega be released ? I have seen no announcements yet. I’m quite curious to see the performance. This might be the year i jump ship. If they happen to show their top of the line card before i end up getting the new titan, i’ll consider. Otherwise, i feel amd is always counter attacking nvidia and not throwing blows first on gpu releases. It also feels like nvidia could do more and they always keep hidden stuff until they get beaten..

          2. Well vega should come out in early 2017, there should be 2 chips and there are some info online but not much. We know one chip will have 4096 stream processors (it should sucessor of Fury) but we dont know if its the smaller or bigger chip. We dont even have confirmed that vega will be 2 chips. We know though it will high end from AMD with HBM2 and as you can see today Fury on vulkan is blazingly fast. And since polaris is 100-300USD segment, Vega will be 300USD+. Still very far from 1200USD xP.
            Still in new API vega with 4096 cores could be faster than Titan if AMD manage to clock it similarly.

          3. AMD’s live event, Capsaicin, coming to SIGGRAPH 2016 on the 25th of this month. I think they will show Vega there, but we’ll see.

          4. Nice. Reaaaalllly curious about this. Meanwhile i expect nvidia to post benchmarks of the new Tx

          5. I think a few months ago the announced that they had finished Vega. So we shall see. It might also be the 490(2x 480’s on one card).

    1. It’s gonna be like 980Ti vs Titan X tbh , like 1080Ti in gaming at least will be similar in terms of performance and almost half the price

      Eh its a niche product , if someone has the money he could buy it i guess. It’s not about value at this point but raw single gpu performance.

      1. yea the Ti v Titan will happen all over again and the Ti will either be 2% slower or even 2% faster in a lot of cases like the 980ti v titan and probably at….im guessing…..800$ for the 1080Ti

          1. Still its 25+% faster card than 1080, for almost double price. 950 would still be quite expensive considering competition.

      1. Damn, I’m stupid. Somehow I thought Pascal was an older architecture. And that was in fact Maxwell.

  2. $200 more than the last one, so in 6 months time expect
    GTX1080Ti
    $849.99 – $899.99

    Nvidia really have whacked up the pricing this time, my budget usually gets me the 80s of the generation but this time I only got a 1070

    1. Nvidia does not really care about there customers because most feed into there price gouging. As long as we start boycotting there products there probably come back to earth. Also AMd have to continue innovate and put the pressure on nvidia but because there in a financial plunge it’s hard to do I think

      1. This. Their near-80% Market share has made them extremely arrogant in light of the lack of any real competition from AMD.

        Now that AMD’s getting better results in the DX12/Vulkan-powered games, they’re going to start gaining traction & market share again (slowly), which means Nvidia’ll start dropping prices soon (read: next year), as they know this is temporary. These over-inflated tags could never compete with AMD’s by far more reasonably priced offerings, & they know it.

        1. Yea, and I’m more than happy to stick with my 980 G1 for the next year until those prices come down.

      2. So why don’t you just buy AMD products? Why doesn’t everyone just buy AMD products? Yes, because NVIDIA still makes better cards.

        Sure, they cost more, but who are you to tell what is a right price for them? How would you know what it costs to manufacture these products? I mean, if they could afford to drop the prices significantly, they could effectively kill AMD straight away. Why wouldn’t they just do it?

        It’s not overpriced if many people consider it being better value than the competition.

        1. You sir are really idiot. It’s people like you who makes it easy for nvidia to hike up there prices whenever they please and buy them. It just does not make sense look at the new Titan x the specs are a bit underwhelming and I wonder how are there going to fit the 1080ti. For this generation of cards nvidia have become worst now they would retail there cards like for example look how the 1080 is set up people are paying more for what the original retail was including that Fe price. Now with the future APIs (Vulcan and Dx12) and the results are coming it’s looks like the way to go is amd which are reasonable in price and competitive. Nvidia cards are not responding well with the new API should customers take the risk of buying there products and not getting the performance they deserve. Yeah dx11 nvidia got that in the bag but dx11 is on its way out. Listen competition brings out the best in companies but having a company almost run a monopoly in a market is a big disadvantage to customers choice and pockets. Greed eventually leads to rebellion from the other side getting bullied.

    2. What we need is the 3rd player on the market and should not be Intel.Like in old good times.
      But its hard to find or appear someone new on the this market.

      Now Nvidia its like Apple or Intel,premium prices for 1.5 years or lower.After 1.5 years you need to buy another premium card because the graphics fashion “evolved”.

  3. FE is worst move for customers nvidia could do because it affects price of non-reference cards, at least for now 🙁

    1. yea it really sucks which is why even as a primarily nvidia customer i want amd to do well and hopefully show nvidia they cant just charge whatever they feel like and get away with it

  4. Fast cash grab before the lack of performance during DX12, Vulkan & asynchronous compute is discovered?

    Remember devs are primarily targeting GCN due to console dominance.

  5. If you got money to burn I can imagine it’s almost unbearable not to succumb to peer pressure of fresh, boasting, ego-boosting owners of the latest and greatest from nvidia.
    Not to mention that if YOU can afford it why would you care about the rest of the poor basterds who can’t.
    The higher the price the more supreme elitist you become.

    1. With the die shrink they get more chips manufactured on the same piece of silicon wafer which should result less cost per product sold.

      So just by maintaining the previous price structure they would have recouped the R&D costs and exceeded the profit margins of Maxwell

      1. That was true in the past but no longer. The die shrinks are getting more expensive rather than cheaper. When Nvidia through TSMC went from 40nm Fermi to 28nm Kepler there was a shortage of chips due to the high rate of semi-functional chips hence we got the mid range GTX 680 being sold for what used to be the high end price of $500. Then the GTX 780 to clear their inventory of partly dysfunctional high end chips and later when the process was refined we got the real high end Kepler GTX 780 Ti. Nvidia buys their chips by the wafer and they don’t always get fully functioning chips that they have paid for already.

        You have to bear in mind that not only do we get a smaller process node but we also get more transistors packed onto the chip.

        I think TSMC is experiencing something similar to what happened when they went to 28nm and there are currently shortages of chips so card manufacturers are having problems getting chips and thus there is some price gouging going on.

Leave a Reply

Your email address will not be published. Required fields are marked *