NVidia’s much-rumored GTX 1070 Ti will launch on November 2, 2017, with initial information disseminated today. The 1070 Ti uses a GP104-300 GPU, slotted between the GP104-400 and GP104-200 of the 1080 and 1070 (respectively), and therefore uses the same silicon as we’ve seen before. This is likely the final Pascal launch before leading into Volta, and is seemingly the response to AMD’s Vega 56 challenger of the GTX 1070 non-Ti.
The 1070 Ti is slightly cut-down from the 1080, the former of which runs 19 SMs for 2432 CUDA cores (at 128 shaders per SM), with the latter running 20 SMs. The result is what will likely amount to clock differences, primarily, as the 1070 Ti operates 1607/1683MHz for its clock speeds, and AIB partners are not permitted to offer pre-overclocked versions. For all intents and purposes, outside of the usual cooling, VRM, and silicon quality differences (random, at best), all AIB partner cards will perform identically in out-of-box states. Silicon quality will amount to the biggest differences, with cooler quality – anything with an exceptionally bad cooler, primarily – differentiating the rest.
As we understand it now, users will be able to manually overclock the 1070 Ti with software. See the specs below:
GTX 1070 Ti Specs vs. 1080, 1070
| NVIDIA Pascal Specs Comparison | |||||
| Tesla P100 | GTX 1080 Ti | GTX 1080 | GTX 1070 Ti | GTX 1070 | |
| GPU | GP100 Cut-Down Pascal | GP102 Pascal | GP104-400 Pascal | GP104-300 Pascal | GP104-200 Pascal |
| Transistor Count | 15.3B | 12B | 7.2B | 7.2B | 7.2B |
| Fab Process | 16nm FinFET | 16nm FinFET | 16nm FinFET | 16nm FinFET | 16nm FinFET |
| CUDA Cores | 3584 | 3584 | 2560 | 2432 | 1920 |
| GPCs | 6 | 6 | 4 | 4 | 3 |
| SMs | 56 | 28 | 20 | 19 | 15 |
| TPCs | 28 TPCs | 20 TPCs | 19 TPCs | 15 | |
| TMUs | 224 | 224 | 160 | 152 | 120 |
| ROPs | 96 (?) | 88 | 64 | 64 | 64 |
| Core Clock | 1328MHz | - | 1607MHz | 1607MHz | 1506MHz |
| Boost Clock | 1480MHz | 1600MHz | 1733MHz | 1683MHz | 1683MHz |
| FP32 TFLOPs | 10.6TFLOPs | ~11.4TFLOPs | 9TFLOPs | 6.5TFLOPs | |
| Memory Type | HBM2 | GDDR5X | GDDR5X | GDDR5 | GDDR5 |
| Memory Capacity | 16GB | 11GB | 8GB | 8GB | 8GB |
| Memory Clock | ? | 11Gbps | 10Gbps GDDR5X | 8Gbps | 8Gbps |
| Memory Interface | 4096-bit | 352-bit | 256-bit | 256-bit | 256-bit |
| Memory Bandwidth | ? | ~484GBs | 320.32GB/s | 256GB/s | 256GB/s |
| Total Power Budget ("TDP") | 300W | 250W | 180W | 180W | 150W |
| Power Connectors | ? | 1x 8-pin 1x 6-pin | 1x 8-pin | 1x 8-pin | 1x 8-pin |
| Release Date | 4Q16-1Q17 | TBD | 5/27/2016 | 11/2/2017 | 6/10/2016 |
| Release Price | - | $700 | Reference: $700 MSRP: $600 Now: $500 | Reference: $450 | Reference: $450 MSRP: $380 |
It seems that the 1070 Ti will effectively kill the GTX 1080, with its MSRP of $450, and will directly challenge the AMD RX Vega 56 card that we previously noted as a strong alternative to the 1070. Initial Newegg listings for the 1070 Ti (these are for pre-orders – as always, don’t do that; wait for the reviews before buying) have the card listed at ~$470 for the cheapest options, with GTX 1080 options listed at $520. We’re not counting blower 1080s, as they aren’t very good, but those start at $510.
If the cards are nearly the same in performance, an overclock by the user should make that up. We’ll see. It seems as if the primary factor in overclocking will be power limit, where some boards and VBIOS configurations will permit higher current/power to the card. That’d be worth noting, and is something we’ll look at in reviews.
This is a move by nVidia to combat the Vega 56 card. We’d wager that a 1070 Ti wouldn’t exist without Vega 56, as nVidia would rest easily with its 1070 and 1080 gapped by a $100 bill. Given V56 availability and pricing, same-cost 1070 Ti cards might take some of the wind out of those sails – but we’ll see if V56 cards see price drops, or if AMD instantiates some sort of bundle deal to try and keep eyes on Vega.
We’ll have a review on this for launch.
- Steve Burke