AMD Ryzen: How a long suffering underdog rose to challenge a titan

I wasn’t sure I’d actually see the day that AMD would once again challenge Intel for the CPU performance crown. As luck would have it, I did. With its Ryzen platform, AMD appears to have changed forever the equation as far as value for money and CPU core count is concerned.

How did AMD do this?

Let’s get some basics out of the way first

All things being equal, the amount of “work” a processor can do depends on the amount of data it can process in a given amount of time. This can be measured by an indicator called IPC or Instructions per Cycle, but this isn’t a perfect indicator of performance because a more efficient instruction set will process more data faster.

For example, if the processor is capable of performing an instruction like 2+3×4 in one step, it’s more efficient than doing 3×4 and then 2+12. Individually, the latter instructions are faster, but overall, the first method processes more data in one step.

Oversimplifying how a CPU works, one can say that instructions are processed on an independent logical unit called a core. One core can independently execute one set of logical instructions at a time. It’s like a production line (called a thread) that can process data in a particular manner. There are tricks to make these cores execute more than one instruction, and we’ll talk about that aspect later.

In theory, the greater the number of cores, the greater the number of instructions that can be executed simultaneously and the faster the data is processed.

A clock-generator makes sure that all actions in a processor are synchronised. This determines your CPU frequency (in Hz) and, after a fashion, your CPU speed. The higher the frequency, the higher the number of instructions that can be processed, but it’s only relevant when comparing CPUs of the same family. A 3.6GHz Intel Pentium 4 pales in comparison to a 1.3GHz Intel Core M3, for example.

The work between multiple cores is handled by things like schedulers and branch prediction, which are way too complicated to get into right now. Suffice it to say that you can think of these as managers delegating tasks to the cores. The better the manager, the more efficiently your cores work. Intel’s advantage, even today, is that it has more experienced managers.

Speaking of managers, a technology called hyper-threading was also introduced by Intel. This essentially allowed a single core to work on two tasks more efficiently. While one task will have priority, every time there’s a lull, say the task is halted because it’s waiting for data, it will switch to the other task. Hyperthreading ensures that a CPU is always working and that no time is wasted. 1 core working on 2 threads is not as fast as 2 cores working on 2 threads, but it is faster than 1 core working on 1 thread.

The first AMD 8-core chip

Around five years ago, AMD claimed to have come up with a challenger to Intel. At the time, Intel was fielding a processor with four independent cores and hyperthreading. AMD’s answer was an ‘8-core’ CPU that was clocked very high.

AMD had a winner, right? Wrong!

AMD’s Bulldozer architecture was a bit of a hack as far as CPU cores are concerned. Each core wasn’t independent, like Intel’s. While two Intel cores can be considered to be two complete, independent production lines, two AMD “cores” are two incomplete production lines that share a third incomplete production line between them.

It’s not that AMD’s design was bad, however. It was a risk, a trade-off between price and performance. The problem was that software didn’t know how to take advantage of it.

READ  Cyprus talks: Erdogan dismisses full Turkish troop withdrawal

Given how operating systems were designed at the time, particularly Windows 7, they couldn’t take advantage of AMD’s design and performance took a hit. There were other issues with AMD’s Bulldozer as well, such as longer pipelines and increased latency to accommodate higher clock speeds, slower cache, etc.

After that, we all know what happened. AMD dropped off the radar, struggled to make headway in even the low and mid-range of the CPU market and Intel took over the world. Intel worked on improving the efficiency of its architecture, kept costs up and used the money to invest in its IoT and mobile efforts, among other things.

And now the new 8-core chip

Ryzen is exactly what AMD should have done many years ago, but for various reasons, couldn’t. Ryzen is effectively the same as an 8-core Intel chip, where each core is actually independent. AMD also finally shifted from its power-hungry, expensive and inefficient 32nm manufacturing process and is now on a 14nm manufacturing process.

As AnandTech put it, Ryzen is just a high-performance x86 core with some AMD tweaks, just as an Intel Core iX series processor is a high-performance x86 core with some Intel tweaks.

AMD’s best Ryzen 7 CPU, the 1800X is 9 percent faster in multi-threaded tasks than Intel’s only consumer grade 8-core chip, the Core i7 6900K. That performance difference can largely be attributed to the AMD chip’s higher clock speed, rather than simply to IPC or architectural improvements.

Intel isn’t a rookie, it could have built and sold a better chip than AMD’s Ryzen for the price of the Ryzen a long time ago. It’s CPUs have been on a 14nm manufacturing process for years now and Intel’s architecture has been more efficient than AMD. 

However, it speaks volumes about Intel’s pricing policies, that AMD is able to pull off a true 8-core chip at half the price, and even a third, as in the case of the Ryzen 7 1700.

So, it’s Ryzen for the win now?

We’ll go out on a limb and say: Yes. Even before the reviews are in, yes. CPU performance has stagnated for over 5 years and the only real jump is on the pricing front, and that’s exactly what AMD has delivered.

If you’re a video editor, a game designer, a 3D artist or a programmer or of any other profession that’s dependent on multiple cores, Ryzen is a no-brainer. The performance difference the extra cores give you are staggering when compared to Intel’s offerings at the same price.

The amount of money you save can buy you the rest of your PC or even a second CPU.

If you’re a gamer, it’s a toss up. Intel is a better option because its CPUs are clocked higher, which matters to games. But, as AMD showed in its streaming demo, more cores means more CPU resources for other tasks, such as game-streaming or video recording. Also, the advantages of a higher clock speed are slight. Once 8-core processors go mainstream, it’ll also be easier for game developers to send more tasks to the CPU, like AI.

When the benchmarks are in, this could likely be the scenario that we’ll see: AMD thrashing Intel in multi-threaded workloads like rendering and Intel sneaking ahead slightly in single-threaded workloads and gaming.

The real battle

The real battle for AMD will be in the mobile devices space. Every year we receive reports that PCs are dying and, in particular, the desktop market. Laptops and 2-in-1s are where the PC market is right now and AMD has, so far, offered nothing on that front.

Intel’s CPUs are extremely thrifty when it comes to power consumption today and they incorporate a lot of features that enable a better mobile experience. Now Intel may have got out of the mobile chipsets race, leaving that space open for Qualcomm and MediaTek. But, Intel is laser focused on fanless designs and is aggressively pushing its Core m series of processors, which are getting design wins in the 100s. This is definitely an additional advantage Intel has over AMD. 

AMD is planning to release laptop SKUs over the coming year; these will have AMD GPUs integrated. Given AMD’s shift to a more efficient 14nm process, power consumption should be much lower than AMD’s previous offerings. However, it remains to be seen if Ryzen can match Intel in the battery life department.

The lower prices will tempt many OEMs to be sure. But is it enough?

I don’t know, but I’ve certainly got my hopes up. Intel’s had a long, lazy and luxurious cruise till now. It’s time someone rocked their boat.