NVIDIA’s launch of their new GTX 1070 Ti is both senseless and completely sensible depending on which way you tend to look at things. The emotional among you are going to wonder why NVIDIA is even bothering to introduce a new product into a lineup that’s more than a year old. On the other hand people who look at things in an analytical fashion will see a yawning gap between the GTX 1080 and GTX 1070 that’s just begging to be filled by a GeForce offering. In my view at least, the actual story behind the GTX 1070 Ti is a bit more complicated than either of those simple explanations.
Unless you’ve been living under a rock for the better part of a year, it should be more than obvious that today’s graphics card market is in a state of flux. Cryptocurrency mining has driven up pricing to stratospheric levels and even at those inflated costs, actually finding a new video card is becoming increasingly difficult. Coupled with the influx of demand propagated by the upcoming Christmas shopping season and Black Friday, this situation is creating a perfect storm that risks emptying retailers’ shelves completely. From a business perspective, NVIDIA couldn’t pass up the opportunity to introduce some fresh blood into a segment that people are ready and more than willing to buy into.
The other point I wanted to raise focuses on competition, or what AMD has that can be sort of considered a competing solution: the RX Vega 56. I won’t dwell too long on Vega’s failures since that particular dead horse has already been beaten to a bloody pulp and it’s already headed for the knackers. But despite being overly power hungry, available in laughably short supply and marked up by retailers to insane heights, the RX Vega 56 was a faint glimmer of hope in an otherwise drab AMD launch. Its in-game performance slotted in perfectly between the GTX 1080 and GTX 1070 and a price of $399 was more than fair. This was a clarion call for NVIDIA to respond and respond they did…..more than two months later with the GTX 1070 Ti.
At the heart of this particular card lies a very familiar face: the GP104 core which was previously seen on the GTX 1080 and GTX 1070. Yet unlike the GTX 1070 which had five Streaming Multiprocessors (or a whole GPC) disabled, the GTX 1070 Ti only has a single SM cut off. That’s the only real change from a core layout perspective. There’s still a 256-bit memory bus spread across eight controllers and 64 ROPs alongside a completely intact L2 cache structure.
Right about now I’m sure you’re thinking exactly the same thing my brain was when I was first told about the GTX 1070 Ti’s specifications: how in the world is NVIDIA going to insure there isn’t a confusing amount of overlap between it and the more expensive GTX 1080? Well that’s a bit convoluted so hold on while I try to explain things.
Despite core layouts that are very similar, there are still some major points of differentiation between the 1080 and 1070 Ti. First and foremost among those is the GTX 1070 Ti’s truncated Boost Clock of 1683MHz (a number that can fluctuate upwards or downwards depending upon GeForce Boost’s algorithms) but perhaps the most significant shift is on the memory front. While the GTX 1080 utilizes a GDDR5X backbone that has since been updated to 11Gbps, the GTX 1070 Ti has 8GB of GDDR5 operating at 8Gbps or precisely the same speed as NVIDIA’s GTX 1070. Will that be enough to differentiate one from the other? Perhaps in bandwidth-limited scenarios but not in too many other situations.
In an effort to insure there’s less chance of the GTX 1070 Ti stepping on the toes of its firmly entrenched siblings, NVIDIA is also limiting –if not outright banning- their board partners from releasing pre-overclocked cards. This means every 1070 Ti will have a homogenous Base Clock of 1607MHz and a Boost Clock of 1683MHz. Some partners will be able to get around this by offering preset overclocks in their software suites but utilizing hardwired BIOS modifications is off the table.
End users can also overclock to their heart’s content but as with all graphics cards these days, there are TDP limits to contend with. Some AIBs like EVGA will be offering GTX 1070 Ti’s with higher TDP levels in an effort to enhance overclocking headroom on their higher tier SKUs.
Due to its expected performance, the $449 price for the GTX 1070 Ti trends closer towards the $499 GTX 1080 than the $399 GTX 1070. More importantly that’s $50 further afield than AMD’s RX Vega 56, a card that was supposed to be the Ti’s initial target. Then again, finding an RX Vega 56 for anything under $450 has been a challenge until yesterday when a few cards mysteriously started turning up at retailers for $420.
But let’s be honest here for a second; other than the preorder cost for the initial GTX 1070 Ti offerings (more on that later), all of these prices are completely fictitious. Limited supply and sky-high demand have led to every single graphics card on the market going for a premium over its initial SRP. Where the 1070 Ti will eventually land in the grand scheme of things remains to be seen.
From a physical perspective, there really isn’t much to separate the GTX 1070 Ti Founders Edition from every other NVIDIA “reference” design launched in the last few years. There’s a metallic shroud that surrounds an acrylic window that gives you a peak at the headsink contained within the blower-style cooler and an illuminated GeForce logo on the outer edge.
Around back there’s the usual full coverage black backplate with some stylized imprints and NVIDIA’s logo. This whole setup remains an iconic piece of industrial design and it’s good to see NVIDIA didn’t mess with a good thing.
Power input is handled by a single 8-pin PCIe power connector which is par for the course on most sub-200W graphics cards. With that being said, many board partners will likely expand on this layout if they start offering custom cards with higher TDP values for overclocking.
The I/O area is pretty ubiquitous as well with a trio of DisplayPort 1.4 connectors, a single HDMI 2.0 output and a DVI-D connector.
At this point, every single thing I’ve said in this introduction has already been covered in one way or another since NVIDIA’s strategy for this launch is a bit different. There have been leaks aplenty, NVIDIA themselves decided to release pricing, specification details and even opened up pre orders last week. Unboxings -such as they are called- were allowed a little while later. So as they say: this load has already been blown. Today simply marks the day when we can talk about performance (spoiler alert: it’s better than the GTX 1070 but a bit slower than a GTX 1080) and show overclocking results.
Other than all of the challenges I’ve highlighted about current GPU pricing structures, actually getting your hands on one of these GTX 1070 Ti’s will be challenging as well. This will certainly be a hard launch with preorders shipping out immediately but according to our retailer contacts, pre order quantities vastly exceed Day One inventory in many cases. Basically that means get your order in now or be prepared to either wait or pay an inflated price later on. It may not be a great time to be in the market for a new graphics card but the GTX 1070 Ti already looks like a hit on all fronts.
Unless you’ve been living under a rock for the better part of a year, it should be more than obvious that today’s graphics card market is in a state of flux. Cryptocurrency mining has driven up pricing to stratospheric levels and even at those inflated costs, actually finding a new video card is becoming increasingly difficult. Coupled with the influx of demand propagated by the upcoming Christmas shopping season and Black Friday, this situation is creating a perfect storm that risks emptying retailers’ shelves completely. From a business perspective, NVIDIA couldn’t pass up the opportunity to introduce some fresh blood into a segment that people are ready and more than willing to buy into.
The other point I wanted to raise focuses on competition, or what AMD has that can be sort of considered a competing solution: the RX Vega 56. I won’t dwell too long on Vega’s failures since that particular dead horse has already been beaten to a bloody pulp and it’s already headed for the knackers. But despite being overly power hungry, available in laughably short supply and marked up by retailers to insane heights, the RX Vega 56 was a faint glimmer of hope in an otherwise drab AMD launch. Its in-game performance slotted in perfectly between the GTX 1080 and GTX 1070 and a price of $399 was more than fair. This was a clarion call for NVIDIA to respond and respond they did…..more than two months later with the GTX 1070 Ti.
At the heart of this particular card lies a very familiar face: the GP104 core which was previously seen on the GTX 1080 and GTX 1070. Yet unlike the GTX 1070 which had five Streaming Multiprocessors (or a whole GPC) disabled, the GTX 1070 Ti only has a single SM cut off. That’s the only real change from a core layout perspective. There’s still a 256-bit memory bus spread across eight controllers and 64 ROPs alongside a completely intact L2 cache structure.
Right about now I’m sure you’re thinking exactly the same thing my brain was when I was first told about the GTX 1070 Ti’s specifications: how in the world is NVIDIA going to insure there isn’t a confusing amount of overlap between it and the more expensive GTX 1080? Well that’s a bit convoluted so hold on while I try to explain things.
Despite core layouts that are very similar, there are still some major points of differentiation between the 1080 and 1070 Ti. First and foremost among those is the GTX 1070 Ti’s truncated Boost Clock of 1683MHz (a number that can fluctuate upwards or downwards depending upon GeForce Boost’s algorithms) but perhaps the most significant shift is on the memory front. While the GTX 1080 utilizes a GDDR5X backbone that has since been updated to 11Gbps, the GTX 1070 Ti has 8GB of GDDR5 operating at 8Gbps or precisely the same speed as NVIDIA’s GTX 1070. Will that be enough to differentiate one from the other? Perhaps in bandwidth-limited scenarios but not in too many other situations.
In an effort to insure there’s less chance of the GTX 1070 Ti stepping on the toes of its firmly entrenched siblings, NVIDIA is also limiting –if not outright banning- their board partners from releasing pre-overclocked cards. This means every 1070 Ti will have a homogenous Base Clock of 1607MHz and a Boost Clock of 1683MHz. Some partners will be able to get around this by offering preset overclocks in their software suites but utilizing hardwired BIOS modifications is off the table.
End users can also overclock to their heart’s content but as with all graphics cards these days, there are TDP limits to contend with. Some AIBs like EVGA will be offering GTX 1070 Ti’s with higher TDP levels in an effort to enhance overclocking headroom on their higher tier SKUs.
Due to its expected performance, the $449 price for the GTX 1070 Ti trends closer towards the $499 GTX 1080 than the $399 GTX 1070. More importantly that’s $50 further afield than AMD’s RX Vega 56, a card that was supposed to be the Ti’s initial target. Then again, finding an RX Vega 56 for anything under $450 has been a challenge until yesterday when a few cards mysteriously started turning up at retailers for $420.
But let’s be honest here for a second; other than the preorder cost for the initial GTX 1070 Ti offerings (more on that later), all of these prices are completely fictitious. Limited supply and sky-high demand have led to every single graphics card on the market going for a premium over its initial SRP. Where the 1070 Ti will eventually land in the grand scheme of things remains to be seen.
From a physical perspective, there really isn’t much to separate the GTX 1070 Ti Founders Edition from every other NVIDIA “reference” design launched in the last few years. There’s a metallic shroud that surrounds an acrylic window that gives you a peak at the headsink contained within the blower-style cooler and an illuminated GeForce logo on the outer edge.
Around back there’s the usual full coverage black backplate with some stylized imprints and NVIDIA’s logo. This whole setup remains an iconic piece of industrial design and it’s good to see NVIDIA didn’t mess with a good thing.
Power input is handled by a single 8-pin PCIe power connector which is par for the course on most sub-200W graphics cards. With that being said, many board partners will likely expand on this layout if they start offering custom cards with higher TDP values for overclocking.
The I/O area is pretty ubiquitous as well with a trio of DisplayPort 1.4 connectors, a single HDMI 2.0 output and a DVI-D connector.
At this point, every single thing I’ve said in this introduction has already been covered in one way or another since NVIDIA’s strategy for this launch is a bit different. There have been leaks aplenty, NVIDIA themselves decided to release pricing, specification details and even opened up pre orders last week. Unboxings -such as they are called- were allowed a little while later. So as they say: this load has already been blown. Today simply marks the day when we can talk about performance (spoiler alert: it’s better than the GTX 1070 but a bit slower than a GTX 1080) and show overclocking results.
Other than all of the challenges I’ve highlighted about current GPU pricing structures, actually getting your hands on one of these GTX 1070 Ti’s will be challenging as well. This will certainly be a hard launch with preorders shipping out immediately but according to our retailer contacts, pre order quantities vastly exceed Day One inventory in many cases. Basically that means get your order in now or be prepared to either wait or pay an inflated price later on. It may not be a great time to be in the market for a new graphics card but the GTX 1070 Ti already looks like a hit on all fronts.
