Ever since Nvidia’s announcement of its 10-series graphics cards with Pascal GPUs onboard, the internet has been abuzz about the potential performance of the GeForce GTX 1080. Today, we got a look at the first independent performance numbers from the new card, but we still didn’t know much about the 1080’s little brother, the GeForce GTX 1070—at least, until now. PCWorld says it confirmed official specifications for the new card with Nvidia after a series of leaks.
The specs PCWorld is reporting for the GTX 1070 pretty much align with our expectations for a cut-down version of the GTX 1080. Here are some highlights. We've included the GTX 970 for comparison's sake since that's the card the GTX 1070 aims to replace.
|GTX 1070||GTX 1080||GTX 970|
|Memory||8GB GDDR5||8GB GDDR5X||4GB GDDR5|
There are two major changes between the GTX 1080 and the GTX 1070. The GP104 GPU inside the GTX 1070 has a quarter of its streaming multiprocessors disabled, leaving it with a total of 15 functional units. Those SMs are good for 1,920 stream processors versus the 1080’s 2,650. While we don’t have stats on the base clock of the 1070, its boost clock is 1600MHz, about equal to the 1607MHz base clock on the 1080.
The 1070 also gives up GDDR5X video memory in favor of GDDR5. We don't know just how much of a performance difference GDDR5X makes on its own, so it's hard to say how big of a deal it is that the new card is sticking with good old GDDR5 right now. GDDR5X only recently entered volume production, and supply of the new chips may be a reason why Nvidia chose to go with regular GDDR5. Overall, these specs do answer some of our questions about where the GeForce GTX 1070 is positioned between the GTX 1080 and the top-end cards from the last generation, but we'll have to see how it performs in the lab to fill in the rest of the blanks.