AMD - a company barely a year younger than Intel, with very similar beginnings, had its ups and downs. The latest of the latter are great AMD Ryzen processors that have just hit our laptops. Let's look at the details of the architecture of AMD Zen, and the history of the company that today is triumphing in virtually every market segment.
AMD, more specifically Advanced Micro Devices Inc. was founded on May 1, 1969. The company's headquarters, which should not be particularly surprising to anyone familiar with the specifics of the technology industry, was then located in Sunnyvale, California - one of the cities that gave rise to the famous Silicon Valley. The founders of AMD were, as in the case of Intel, employees of the Fairchild Semiconductor company. For AMD, the founding group was Jerry Sanders, Edwin Turney, John Carey and Jack Gifford. The company showed its first project, a 4-bit shift register Am9300, the same year. Its sales began in 1970. At that time, nothing heralded the fierce battle between AMD and Intel.
The conflict was ... copying, and more precisely the process of reverse engineering. AMD released its first processor in 1975, called the Am9080. It was in practice an exact copy of the 8-bit Intel 8080 unit, shown a year earlier. Eventually, the companies agreed and AMD was given the opportunity to continue developing this processor.
The era of x86 systems
In the early 1980s, personal computers entered a completely new phase. Powered by 16-bit Intel 8086 processors began to flood the market, and computers timidly began to appear in spaces other than large corporations. At the time, the market-playing company was IBM, which forced Intel to sign a license agreement with another entity that could produce identical systems. Thus, from 1982, AMD could produce its chip versions using the x86 instruction set. It was also the first-ever situation in which the AMD chip showed higher performance than the one created by Intel. Am286, a copy of Intel's 80286, was faster than the original. The secret was the higher frequency. AMD "turned up" (of course this term was not used yet) the construction from 12 MHz to dizzying 20 MHz.
It is in this era that one should look for the causes of a kind of conflict that ignites technology fans to this day. Exchanges of sentences between lovers of Ryzen and Core, are reminiscent of what readers in the letters to technology editors showed the superiority of Am286 over 80286 (or vice versa). While the license agreement signed between the companies did not arouse great controversy in the case of the 80286 processor family, the next two generations of processors: 80386 and 80486 were, according to Intel, illegally copied by AMD. The case went to court, which admitted AMD.
End of copying
In 1996, AMD presented its response to Pentium systems. K5, because this system had such a seemingly bland name, was the first CPU of a California company with its own design. There is a joke in the name that comic fans will appreciate. Apparently, K came from the word Kryptonite. In this list, it is easy to guess who was the Superman.
Further years in the development of AMD abounded in interesting projects. K5 architecture, in its next incarnations, competed with Intel, but it was only K6 that managed to actually establish an effective fight with the Pentiums. In the end, AMD reached for a more friendly name - that's how Athlon was born. Although today Athlon processors are associated with low-end units for not too demanding, around 2000 the processors of this family were breaking performance records. Literally. Athlon 1000 was the first CPU in history to achieve a staggering 1 GHz clock speed. For less demanding users, AMD anticipated processors named Duron that competed on the market with Intel Celeron.
An innovative approach
In 2003, AMD followed a path previously unknown in personal computers. Introduced in September of that year, Athlon 64 introduced AMD64 instructions to the home desktop environment, opening the possibilities of 64-bit machines to ordinary PCs. The leading benefit, of course, was the ability to address more memory capacity. These processors immediately dominated the rankings of the fastest CPUs on the market. Contributed to this, among other significant structural changes - the RAM memory controller, previously located in the motherboard chipset, went directly to the CPU.
You didn't have to wait long for the next brave step of AMD. Already in 2005, the company showed the first truly dual-core desktop CPUs - AMD Athlon 64 X2. Processors using the K8 architecture literally destroyed the incredibly hot and power-eating dual core processors of the competition - Pentium D. AMD systems were cheaper, consumed less power, got less heat and above all - in most tasks they outperformed their rival. But the time of change was approaching inexorably …
While still on the wave of success, AMD purchased the Canadian company ATI, a manufacturer of graphics systems Radeon in 2006. This decision influenced the development of the company over the next years and is of great importance for its current position.
A decade of trouble
AMD had a great beginning in the 21st century, but everything began to change around 2006. Then, still performing well Athlon 64 X2 received a worthy rival in the form of Intel Core 2 systems - drawing on handfuls of mobile technology. In 2007 and 2008 both companies showed new processors. AMD Phenom (2007) represented a far-reaching improvement of the AMD architecture used in earlier systems. The changes included the use of shared third-level cache between cores. The processor's IPC has also increased, and the memory controller has also been improved. The company decided to stay with 65 nm lithography.
Intel, which already has Core 2 Duo and Core 2 Quad managed to return to the game for good, competed with AMD for 2006 and 2007 with mixed efficiency - then the dispute Athlon 64 X2 / Phenom or Core 2 Duo / Core 2 Quad was one of the topics who kindled technology forums. By the time. At the end of 2008, when AMD was preparing for the release of Phenom II processors, i.e. the improved Phenom CPUs moved in 45 nm, there was a real earthquake: Intel showed Nahalem systems, i.e. the first generation of Core i7 / i5 / i3 processors. The number of changes that affected Intel systems turned out to be so large that these processors left competitors (and predecessors) behind.
In the following years, Intel refined its Core processors using a development model called tick-tock. Meanwhile, AMD managed to release the powerful Phenom II X6 - the first true six-core system for desktops, and then again looked for a "different path" as before with 64-bit or dual-core processors. The concept was not a bad theory - a combination of CPU and efficient GPU (to create which AMD had the appropriate technology thanks to the purchase of ATI) in one product. This type of unit was called APU (Accelerated Processing Unit). The designation was changed again by introducing chips marked simply A (A10, A8, A8, A4).
The Bulldozer architecture was a milestone for AMD units. The architecture itself had many interesting features, especially when it comes to the approach to work on many threads. While SMT (or HT at Intel) increases the surface of one core, but allows you to work on two threads at once, only in Bulldozer systems were the basis of two cores, which allows you to keep two threads, but such a set (called a module) was deprived some "unnecessary" elements. And so for every two "cores" (or rather for one module), there was one calculation unit for floating-point numbers. The same was done with decoding and downloading systems - one for each module. The result was processors that not only gave way to Intel processors in terms of performance but also drew a lot of power. Railroad modifications to this architecture have failed to take an equal fight with Core. AMD has been extensively developing this architecture from 2011 to 2017. In the meantime, already under the new CEO, Dr. Lisa Su (in the position since 2014), work was underway on something completely new.
How did it happen that AMD managed to survive the decade of Intel's dominance at all? Well, the rescue turned out to be, among other consoles. Both Sony and Microsoft, for their consoles, respectively PlayStation 4 and XboX One decided to use AMD APU precisely. These systems had everything needed to create a good gaming platform at low cost - a high degree of component integration, good iGPU performance, and tolerable CPU capabilities. Without going into details - the Jaguar systems, landing in consoles, provided the company with a financial drive to conduct research.
In February 2017, AMD CEO Dr. Lisa Su came on the scene and announced an earthquake in the industry. During the live presentation, further details of the architecture of the new CPUs made in the completely new Zen architecture were revealed. AMD Ryzen systems hit the market in March of the same year and AMD has competed with Intel for the first time in many years as an equal.
The secret to building AMD Ryzen processors was obvious - taken from existing solutions. Instead of clustered multi-threading, SMT has appeared in Zen architecture, which, like Intel's HT, allows two threads to work on a single core at the same time. This gives a better effect than the previous solution, because here when you need single-threaded performance, the whole core is working on obtaining it, while only a part of the module dealt with it before. The second secret was visible in the specification. AMD once ridiculed jokingly for the concept of "vinegar roots" none of these jokes made themselves. Well, it went even further. Among the new system palette, there are 8-core and 16-thread units - something so far only found on HEDT (high-end desktop) platforms. It turned out that in single-thread tests, these systems are slightly inferior to competitors' processors, but where the need to work many cores at once, they often proved to be faster. The construction of the cache, CPU predictive mechanisms and, crucial for performance, the executive units were completely rebuilt. As a result, we received a very efficient CPU, which over time covered the entire market space.
Since 2017, Zen architecture has been systematically developed, which resulted in its improvement called Zen + in 2018 and a much deeper transformation in the form of Zen 2 in 2019. And this last incarnation is the one that really conquered the market. It is known as the third generation or simply AMD Ryzen 3000. Below we explain all the complexities.
AMD Ryzen - how to distinguish them?
As in the case of Intel systems, a lot can be read in the AMD stable by the name of the processor. It contains a number of numbers and letters that allow you to understand the systematics of specific systems, their purpose, and probable price.
The name of each Ryzen processor obviously starts with ... the word Ryzen (not very revealing, right?), But it is followed by a single digit: 3, 5, 7 or 9. It determines what segment covers the given CPU among its platform. Attempts to relate this to Intel units are only seemingly accurate and often depend on specific models. While cheaper representatives of the Ryzen 7 line can actually compete with Core i7, the strongest often break down into the Core i9 silicon blades. Ryzen 9 rivals must look rather among the HEDT systems. Therefore, this numbering is mainly useful for internal comparisons as part of the AMD offering.
Then we find the ordinal number consisting of four digits. The first one defines the processor's generation, while the next ones organize it within the series. Generations of Ryzen systems, although we have barely seen three at the moment, are clearly different. We will return to the issue of generation in a moment.
For example, Ryzen 7 3700 is a higher positioned processor than, for example, Ryzen 5 3600X. Where did that X come from? Well, you can also find the letter at the end of the name. There are not as many of them as Intel, and their explanation is relatively simple.
U - are mobile processors for the so-called mainstream. We find them rather in thin and light constructions, possible models for everyday tasks, not covering such categories as processing multimedia materials or advanced games. These processors are quite efficient, but it is difficult for players to recommend them. They are relatively cool and their TDP is only 15W. They do not reach high frequencies, hence their limited performance. However, they have a feature that makes them suitable for simple or older games relatively well - their integrated graphics system is based on Vega architecture, quite efficient and allowing fun at reduced details with some great classics. Of course, as always in the case of "integration", a lot depends on the quality of the RAM used. Some manufacturers have limited the capabilities of this iGPU using only one-channel memory on their laptops. No matter what - Ryzen U series processors of any generation are not systems for players.
H - Ryzens marked in this way are mobile systems for more demanding users. We will find them, among others on laptops for players, mobile workstations etc. wherever you need more performance. How do they differ from U-class systems? Structurally, generally ... nothing. However, they have much higher TDP, which allows them to spread their wings. In theory, such Ryzen 7 3750H is not particularly different from Ryzen 7 3700U - however, the former more often achieves higher clock speed, more aggressively managing the Turbo mode.
HS - currently only in the Ryzen 4000 series. This means the version of the H mobile processor but designed to be installed in a thin and slim (Slim, hence S) laptop design. Therefore, these systems have a slightly lower TDP than their H series counterparts of the same generation.
G - are Ryzen desktop processors with integrated graphics. They are great at the heart of an inexpensive multimedia computer, but due to the smaller number of cores than in other desktop models, they have some performance deficiencies that remove them from truly gaming applications. Especially since they also have a limited number of PCIe lanes. Interestingly, they have an unlocked multiplier, so you can overclock them.
no letters - typical desktop Ryzens. Contrary to what Intel markers may have become accustomed to, these processors have an unlocked multiplier, so they are prone to overclocking.
X - models that can be said to be slightly overclocked at the factory. By design, they are no different from processors without any letter at the end of the name. However, they are characterized by a higher frequency of work, and thus, higher efficiency. Here we also have the multiplier unlocked.
Let's also discuss the existing generations of AMD Ryzen processors. The first of these is the one that debuted at the beginning of 2017 changing the face of the processor market. These are units using the first Zen architecture. They were made in a 14 nm technological process. Their successors, i.e. the second generation, are units from 2018, are characterized by a slightly streamlined Zen + architecture and a more modern 12 nm FinFET production process. They also officially support faster DDR4 2933 MHz memory (previously only 2667 MHz). The third generation is a much bigger improvement. Ryzen 3000 processors, which debuted in the summer of 2019 use the Zen 2 architecture - which brings with it a whole lot of improvements, in particular increased IPC (instructions in the clock cycle), which allowed the CPU to catch up with Intel systems in single-threaded tasks. Ryzens 3000 have also brought much more L3 cache memory and a new RAM memory controller, working officially even with 3200 MHz DDR4 modules. However, the biggest change is the technological process - barely 7 nm. Thanks to this, the processor cores are smaller and you can simply fit more of them on the same surface, the processor heats up less while drawing less electricity. The fourth-generation has already been announced ... and here is a catch.
There is a kind of shift in the case of mobile systems. And so, Ryzen 2000 laptop processors are not, as the name would suggest, cousins of e.g. Ryzen 7 2700X, i.e. a representative of the second generation, and units closer to e.g. Ryzen 5 1600X. They use the first generation of Zen architecture and are created in the 14 nm technological process. Ryzen 3000 mobile models, on the other hand, are units using the Zen + architecture, thus closely related to the second generation desktop systems. The new mobile Ryzens 4000 that are now appearing are laptop equivalents of the third generation of our desktops.
Another thing is that not every laptop has to have a mobile CPU, which we have proven many times with our designs. Desktop systems are obviously more efficient, but they also require efficient cooling that most laptops can't provide. Most, but not all.