Polaris: GCN新一代『物超所值』之王

Table of Contents

1. 👉 Introduction
2. 👉 AMD's GPU Architectures: Polaris and Vega
3. 👉 AMD's Plans for Polaris 11 and 10
4. 👉 Understanding Entry-Level and Mainstream GPUs
5. 👉 AMD vs NVIDIA: The Battle for the Mainstream Segment
6. 👉 Polaris 10 Performance Expectations
7. 👉 Factors Affecting Polaris 10's Performance
8. 👉 Improving IPC with Polaris
9. 👉 Enhancing Shader Utilization with Polaris
10. 👉 The Importance of Instruction Prefetching
11. 👉 The Primitive Discard Accelerator: Boosting Performance in Complex Scenes
12. 👉 Verdict and Conclusion

Introduction

In this article, we will delve into the upcoming GPU architectures from AMD, specifically focusing on Polaris. We will explore the basics of these architectures, AMD's plans for their release, and the performance expectations for Polaris 11 and 10. Additionally, we will discuss the improvements in Instruction Per Clock (IPC) and shader utilization that Polaris brings to the table. So, let's dive in and discover what AMD has in store for gamers and enthusiasts!

AMD's GPU Architectures: Polaris and Vega

Before we dive into the details of Polaris, it's essential to understand AMD's GPU architectures. Polaris is one of the latest architectures developed by AMD, alongside Vega. These architectures aim to provide significant improvements in performance, power efficiency, and features compared to their predecessors.

AMD's Plans for Polaris 11 and 10

AMD intends to replace its current 303 series with the Polaris 11 and 10 GPUs. The launch of these GPUs is expected to happen around the middle of this year, with a rumored launch at Computex. AMD has confirmed that they are on track with their plans, which is undoubtedly exciting news for enthusiasts eagerly awaiting the release.

Understanding Entry-Level and Mainstream GPUs

Polaris 11 and 10 are designed to target different market segments. Polaris 11 will focus on the entry or value segment, while Polaris 10 will cater to the mainstream segment. Entry-level GPUs, such as the Nvidia 750 Ti, 950, or AMD 360 and 370, offer low power consumption and affordability, usually priced at $149 or below. These GPUs are often compatible as upgrades for prebuilt PCs with weak power supplies or poor airflow.

When it comes to the mainstream segment, GPUs like the AMD R9 280, 380, or the Nvidia 960 are considered the norm. Both AMD and Nvidia consider prices below $299 as mainstream. It's important to note that the price target for Polaris 10 falls well below the mid-range or performance segments, where the Pascal 1070 and 1080 dominate.

AMD vs NVIDIA: The Battle for the Mainstream Segment

Many are curious about AMD's ability to compete with Nvidia's mid-range Pascal 1070 given the lower price of Polaris 10. While the full details are yet to be unveiled, it is expected that Polaris 10 will have similar performance to AMD's top GPUs, albeit potentially slightly below the Pascal 1070. This could be due to GD r5 bandwidth limitations and the possibility of Polaris 10 having fewer than the full 256-bit memory controllers, similar to the situation with the Nvidia GTX 970.

Polaris 10 Performance Expectations

Considering historical trends, we can make an educated guess about the expected performance of Polaris 10. When AMD transitioned to the Pecan architecture, which is similar in size and power profile to Polaris 10, the performance was impressive. The AMD 7870, which was a small mainstream chip, outperformed AMD's previous top GPU, the 6970, and even came close to rivalling Nvidia's GTX 580.

Based on this, it is reasonable to expect that Polaris 10 will be significantly faster than Hawaii GPUs like the 390X, and potentially match or even surpass the performance of top-tier GPUs like the Fury X or the Pascal 1070. While this may sound too good to be true, the combination of the 14 nanometer FinFET process and the GCN architecture in Polaris 10 has the potential to deliver a massive performance boost.

Factors Affecting Polaris 10's Performance

The key factors that contribute to Polaris 10's performance boost are its production on the 14 nanometer FinFET node at Global Foundries and Samsung and the improvements in IPC. The 14 nanometer FinFET process has been used in the production of highly clocked chips for Apple, indicating its capability to produce defect-free chips, enabling better performance and higher clock speeds.

Additionally, the GCN architecture, which tends to clock slower than the media's note, is expected to see a significant increase in clock speeds. This combination of the 14 nanometer FinFET process and improved clock speeds can potentially result in a 50% increase in clock speeds compared to AMD's current top GPUs.

Improving IPC with Polaris

IPC, or Instructions Per Clock, is a critical performance metric. AMD has made efforts to improve IPC with Polaris, which means that the architecture can deliver better performance per shader at the same clock speeds. AMD achieves this by improving the hardware scheduler or command processor, which has been an issue in previous GCN architectures, particularly when it comes to non-optimal shader utilization in APIs like DirectX 11.

The improved command processor in Polaris allows for higher shader utilization, leading to an overall performance boost, especially in DirectX 11 games. Another aspect of improving IPC involves the hardware prefetch mechanism. This mechanism prefetches frequently occurring instructions or in-cache texels, resulting in higher sustained performance and fewer drops due to CPU or memory bandwidth bottlenecks.

Enhancing Shader Utilization with Polaris

Shader utilization is another area where Polaris brings improvements. By addressing the non-optimal shader utilization in DirectX 11 games, Polaris can deliver better performance. The higher shader utilization is primarily achieved by AMD's hardware scheduler, which takes advantage of all the hardware schedulers on the architecture, including the command processor and asynchronous compute engines (ACEs).

This enhancement in shader utilization is particularly beneficial in complex scenes with lots of objects Hidden from view. AMD's primitive discard accelerator feature further contributes to improved shader utilization and overall performance in such scenes. This feature culls and removes information that is not visible to the current field of view, resulting in a significant performance gain.

The Importance of Instruction Prefetching

Instruction prefetching is a challenging task, but when implemented correctly, it can lead to noticeable performance gains. Polaris introduces an advanced instruction prefetch mechanism that prefetches frequent instructions, reducing wait states and improving overall performance. This feature can prefetch both instructions and in-cache texels, resulting in smoother gameplay and higher sustained performance.

It's worth noting that while the primitive discard accelerator and instruction prefetching features are valuable in enhancing performance, their effectiveness may vary depending on the Game and its complexity. However, in more demanding scenes, these features can lead to a higher minimum frame rate and an overall better gaming experience.

Verdict and Conclusion

In conclusion, Polaris 11 and 10 hold great promise for gamers and enthusiasts. With improvements in IPC, enhanced shader utilization, and features like the primitive discard accelerator, AMD aims to deliver impressive performance gains. Polaris 10, in particular, is expected to compete with top-tier GPUs, like the Pascal 1070, at a more affordable price point.

While we await the official release and comprehensive benchmarks, the transition to the 14 nanometer FinFET process and the advancements in the GCN architecture have set the stage for significant performance improvements in Polaris. It's an exciting time for AMD and its fans, and we can't wait to see how Polaris GPUs revolutionize the gaming industry.

Pros:

• Potential for significantly better performance than previous AMD GPUs
• Improved IPC and shader utilization
• Affordable pricing in the mainstream segment
• Enhanced features like the primitive discard accelerator

Cons:

• Unconfirmed performance benchmarks
• Effectiveness of new features may vary in different games

FAQ

Q: When will AMD release Polaris 11 and 10 GPUs? A: AMD plans to release Polaris 11 and 10 GPUs around the middle of this year. The launch is expected to take place at Computex later this month.

Q: What market segments do Polaris 11 and 10 target? A: Polaris 11 targets the entry or value segment, while Polaris 10 caters to the mainstream segment.

Q: How does Polaris 10 compare to Nvidia's Pascal 1070? A: While full details and benchmarks are not yet available, it is expected that Polaris 10 will have similar performance to the Pascal 1070, albeit potentially slightly below.

Q: What improvements does Polaris bring in terms of IPC and shader utilization? A: Polaris introduces an improved command processor, enhancing IPC and shader utilization. This results in better performance per shader and improved overall performance, especially in DirectX 11 games.

Q: What are the key features of Polaris that contribute to its performance boost? A: Polaris features a 14 nanometer FinFET process, enabling higher clock speeds, and improved IPC. Additionally, hardware prefetching and the primitive discard accelerator contribute to enhanced performance in complex scenes.

Q: Can we expect Polaris 10 to compete with top-tier GPUs? A: Yes, based on historical trends and the improvements in Polaris, we can expect Polaris 10 to be significantly faster than previous AMD GPUs and potentially match or surpass the performance of top-tier GPUs.

Q: What price range can we expect for Polaris 11 and 10? A: Polaris 11 is expected to be priced around $149 or below, targeting the entry-level segment. Polaris 10, on the other hand, is expected to be priced around $299, positioning it in the mainstream segment.

Q: How will Polaris GPUs impact the gaming industry? A: Polaris GPUs have the potential to revolutionize the gaming industry by delivering impressive performance improvements at affordable price points, making high-quality gaming more accessible to a wider audience.