As in the case of Intel, the history of AMD processors is also very interesting, and very rich. A few months ago we published an article dedicated to reviewing ten processors from the chip giant that, by their own merits, have gone down in the history of the PC world, and today we are going to do the same con ten AMD processors that also have an important space reserved in said history.
AMD (Advanced Micro Device) was founded on May 1, 1969 by Jerry Sanders. He was not alone, this electrical engineer was the marketing director of Fairchild Semiconductor, and to carry out his project he had the support of seven colleagues from that company. This adventure went through different stages that gradually molded the giant of Sunnyvale.
In its first stage, AMD served as a second supplier of microchips designed by Fairchild and National Semiconductor. This means that AMD processors did not exist at that time, since the company focused on the production of logic chips designed by another company that revolved around a clear objective: to comply with the military quality standard required by the United States. United. I know it might sound strange today, but we’re talking about the sixties, a time when it was very difficult to get really reliable microchips, and therefore this business was viable.
It may surprise you, but at the time, AMD processors bet on the RISC architecture. The Sunnyvale giant wanted to push such architecture into the mainstream consumer market, and did it with AMD 29000 series, a family that competed, at the time, with Intel’s 80960 (RISC) line, although in the end we already know that both companies focused on the x86 architecture.
In order to use this architecture in its processors, AMD had to sign a licensing agreement with Intel. This agreement had a relatively simple first phase, but everything got complicated over time. That agreement had critical moments that put AMD on the ropes, since when Intel realized the growing weight of the x86 CPU sector it tried to do everything possible to remain as the only player within it, in fact not He even shared the details of the 386 processor with the Sunnyvale company.
As you may have imagined, AMD did not stay to see them coming. The company took Intel to court, as it viewed it as a violation of the contract that both signed. In the end, a court agreed with AMD, and Intel had no choice but to pay a large sum of money to AMD. From that moment, a fierce struggle began between the two companies to lead the x86 processor market, both at the general consumer level and at the professional level.
AMD processors have had truly brilliant moments, although it is true that, if we make an average assessment and take into account all the key moments that the industry has experienced, Intel has been in the lead more times, and longer, a reality that has allowed it to reach this day as a true silicon giant, although it is also true that, right now, AMD processors offer a more solid value in price-performance ratio, and that the company that runs Lisa Su is not making it easy for the chip giant.
After this brief, but necessary, introduction we are ready to discover those ten AMD processors that marked a turning point, and that have managed to go down in the history of the PC world on their own merits.
We are before a clone of the Intel 8086 that AMD managed to develop by reverse engineering, and that shared its most important characteristics with it.
It was launched in 1974 with a production cost of 50 cents for each chip. A silicon wafer produced around 100 functional chips, which were then sold to the military market at a price of $ 700 each. Yes, it was very profitable.
It deserves to be on the list of the most important AMD processors in the PC world because it was the first from the company to use the x86 architecture.
We are facing another clone of an Intel processor, specifically the 386 of the chip giant. This AMD processor arrived in 1991 with a 32-bit configuration and shares, broadly speaking, the foundations of the Intel model, although with an important nuance, and that is that the Sunnyvale company was able to market a model 100% compatible with the Intel chip, but more powerful.
The Intel 386 ran at 33 MHz, while the AMD chip was capable of 40 MHz. At the time, that 7 MHz difference was significant enough to have a significant impact on raw performance and processor life.
We can say that this processor was a great victory for AMD, not only because the company was able to develop a 386 more powerful than Intel’s and totally compatible with your platform (it worked on the same motherboards), but also because won a major court battle against Intel, which had attempted to block the release of AMD’s 386.
Intel wanted limit x86 license to 286. If he had gotten away with it, the future of AMD processors would have been very different from what we all know.
We are facing the first generation of x86 processors fully designed by AMD, and also against the direct rival of the Intel Pentium, one of the best generations of processors that the Santa Clara company has launched in its entire history.
If we focus on the architecture, we can say that the AMD K5 series was closer to the Pentium Pro than the Pentium design. This generation had a RISC processor with an x86 decoder unit which transformed all x86 commands into RISC commands. It added five integer units and an integrated FPU, and had speculative execution out of order to improve performance.
In general, this architecture was superior to that used in the Pentium series, but its design was very complex and AMD could not overcome all the challenges that lay ahead in time. This, together with some errors at the L1 cache level in the first units, affected the number of chips it could produce, forced its launch to be delayed to 1996 and did not convince the PC manufacturers to give you a reference. estimates that a K5 at 100 MHz performed like a Pentium at 133 MHz, but from what we have commented it ended up being a failure for AMD.
Without a doubt one of the most interesting and competitive AMD processor families. This generation started from the base of the previous one, that is, of the AMD K5s, but AMD did not make the same mistakes and it ended up being a success.
This was made possible by a much simpler base architecture. The architecture that these AMD processors used is known as Nx686, maintained compatibility with motherboards Socket 7, raised the frequencies considerably and integrated the MMX multimedia instructions, which marked a real revolution in the second half of the nineties.
His work frequency was from 166 MHz to 300 MHz. In terms of performance they were above the Pentium MMX, but they did not exceed the Pentium II, which had a more powerful floating point unit and reached much higher working frequencies.
We are facing a generation of AMD processors that brings uses a major revision of AMD’s K6-II architecture, used in processors that rivaled the Pentium II.
AMD processors based on the K6-III architecture have been chosen to fill this position because they were the first of the company to integrate 256 KB of L2 cache in the processor package. This was a major turning point, as moving the L2 cache from the motherboard to the processor greatly reduced latency and significantly improved performance.
They arrived in 1999, they used socket 7, they had the MMX and 3D Now! and they were capable of operating at a maximum of 550 MHz. They couldn’t really beat the Pentium III at the same frequency, as they relied on the motherboard’s built-in L3 cache setting to catch up with those, but they offered good performance and were very affordable.
6.-AMD Athlon K7
We are at the beginning of a true revolution by AMD that ended with the AMD processors that we will see in the next point.
With this generation of processors from AMD he had a clear objective: improve the floating point unit, one of its big pending accounts to surpass Intel in gross performance, and we can say that it was able to fulfill it without major problem. The first AMD processors based on this architecture arrived in the second half of 1999 with a frequency of 500 MHz, but they managed to be the first to break, later, the GHz barrier.
The AMD Athlon K7 series of processors used socket slot A, which was not electrically compatible with Intel’s Pentium III, and many models had a card design which not only facilitated the dissipation of the heat generated, but also gave it the necessary space to integrate a whopping of 128 KB L1 cache and 512 KB L2 cache. The model that exceeded the GHz frequency came to be considered the most powerful x86 CPU of the moment.
7.-AMD K8: Athlon 64
A myth, and without a doubt one of the best architectures AMD has created so far. With this architecture, AMD processors made the jump to 64 bits, thanks to the incorporation of the AMD64 instructions, and achieved what seemed impossible, to put Intel on the ropes.
The AMD Athlon 64, processors based on the K8 architecture, were compatible with 64 bits, had a cooler operation than the Pentium 4 and offered, in addition, much higher performance. To give you an idea of the enormous difference between the two, a simple comparison is enough, and that is that the Athlon 64 2000+ at a real frequency of 1 GHz was capable of performing at the level of a Pentium 4 at 2 GHz, while the Athlon 64 3000+, running at 2 GHz, outperformed a Pentium 4 at 3 GHz.
As Intel tried to stretch their hotter and slower Pentium 4s, pulling extreme frequency increases in a senseless bid to push the MHz race at AMD to the limit, they showed that its K8 architecture literally played in another league.
The Athlon 64 also incorporated the Cool’n’Quiet technology, that reduced the voltage and the frequency of work of the processor when executing less demanding tasks, which achieved a significant reduction in consumption. They started to arrive in 2003, I think it is fair to consider it as the best family of AMD processors in all history.
8.-AMD Athlon 64 X2
If the Athlon 64s were a coup of authority, the Athlon 64 X2s were the definitive wake-up call Intel needed to understand that the thing was very serious, and that she had to recharge if she didn’t want to be completely overwhelmed. His arrival took place in 2006.
The Athlon 64 X2 maintain the base of the K8 architecture, but use a monolithic dual-core design which was light years from the MCM design (two cores glued) of the Intel Pentium D, which were nothing more than two interconnected 64-bit Pentium 4s.
These new processors integrated two cores in a single package. Each core had its own resources available and could handle one thread. It also had nothing to do with the HyperThreading technology that Intel had used in the Pentium 4, which allowed the kernel to handle, in addition to a process, a thread. The most powerful versions reached the 3,200 MHz per core.
I know what you’re thinking, why do we put a low-power, low-performance architecture that rivals the Intel Atom series on a list like this? Well, very simple, because it has been the basis of PS4 and Xbox One, and also PS4 Pro and Xbox One X.
The Jaguar architecture is the successor to Bobcat. Like that one, it maintains a very low consumption and presents an excellent cost-core ratio. This made it the ideal choice for next-generation consoles, as it met all the requirements established by Sony and Microsoft designs: low power consumption, high number of cores and low cost. To show a button, the PS4 APU, which mounts an eight-core Jaguar CPU and a Radeon GPU superior to the HD 7850, cost approximately only 100 dollars.
Both PS4 and Xbox One, and their successors, PS4 Pro and Xbox One X, mount a Jaguar CPU of eight cores at frequencies ranging from 1.6 GHz by PS4 up to 2.3 GHz Xbox One X. It has a very low CPI, but it has been able to bring to life the entire current generation of consoles, so it deserves to be, without a doubt, on this list.
It was clear that the last place should be reserved for this architecture. We could have put the Zen 2 architecture here, which has been the one that has allowed AMD to catch up with Intel in terms of IPC and give a huge blow to the chip giant, but we have opted for the first generation for a very simple reason: represented a huge revolution, a true turning point, allowing AMD to initiate a seemingly impossible recovery.
Zen architecture abandons the monolithic core design in favor of an MCM (multi-chip module) design in which we find the CCX units. Each CCX drive has a total of four cores with its own L1, L2, and L3 caches. We can combine a maximum of four CCX units interconnected through an Infinity Fabric system, which would leave us a chip with 16 cores, but keep in mind that each block of four cores can only use its 8 MB portion of L3 cache.
Thanks to the changes that AMD introduced at the design and architecture level, Zen exceeded the Bulldozer architecture IPC by up to 52%, made the leap to 14 nm and gave life to processors of up to 16 cores and 32 threads with very competitive prices. It did not exceed Skylake’s CPI, but in price-performance value it was unrivaled. One of the best families of AMD processors out there.