The Pricing of GPUs: Understanding the Differences Between a GTX 1080 and a 1070
Introduction to GPU Pricing and Binning
Have you ever wondered why a GTX 1080 and a GTX 1070 from the same architecture can cost significantly different prices? The answer lies in the subtle differences between the chips and how they are manufactured and sorted.
Manufacturing and Binning
Almost all CPU and GPU production involves a process where not every chip works perfectly. This is known as binning. During the manufacturing process, NVIDIA, for example, produces a batch of GP104 chips, which are the powerhouses behind the GTX 1080 and GTX 1070. The ones that pass quality assurance (QA) tests are designated for the GTX 1080. However, some chips may not perform as well, with as many as a dozen out of sync CUDA cores or not stable at the target voltage/clock speed. These chips typically end up being downgraded to a GTX 1070, with non-functional parts disabled and sold at a lower price point.
Even chips that meet the requirements for a GTX 1080 are not all created equal. The subtle variances in performance are the result of the silicon lottery, where some chips can be overclocked slightly higher than the standard specification. This variation leads to price differences among the same GPU models.
Clock Speeds and Overclocking
Let’s take a closer look at how binning affects the pricing. Manufacturers like EVGA sort their GP104 chips to find the best-performing ones. An EVGA GTX 1070 with a higher clock speed can be sold at a slight premium. The example given shows that one EVGA GTX 1070 has a base clock that’s 88MHz higher and a boost clock that’s 101MHz higher, justifying the extra $10 cost. This is a result of better binning, where these chips have slightly higher performance and stability.
Better-performing chips are not only overclocked but also get better treatments in terms of cooling and packaging. Gigabyte’s GTX 1080 example is a perfect illustration. Two different GTX 1080 models from Gigabyte have significantly different clock speeds, with one featuring a premium copper water block and a custom PCB for improved performance. The more expensive model is expected to cost $100-150 more than its cheaper counterpart.
Mini-ITX GPUs and Market Demands
In addition to standard GPUs, there’s a market for compact mini-ITX GPUs. These GPUs, while physically smaller, often come from chips that have been sorted for lower power consumption and less heat production. For instance, a mini-ITX GTX 1080 might have a similar cooling system to a full-size GTX 1080 but run cooler because of better binning. This means it can use a smaller, more efficient cooler. Despite having a lower bill of materials, these mini-ITX GPUs can still command a premium due to their expected performance and smaller form factor, which is highly valued by consumers.
Conclusion
The pricing differences between a GTX 1080 and a 1070, or the variation among different models of the same GPU, are a result of meticulous sorting and overclocking processes. Binning, clock speeds, and even form factors play crucial roles in determining the final price. Understanding these nuances can help consumers make informed decisions when purchasing a graphics card that meets their specific needs and budget.