The x86 processor sector has experienced a huge evolution in the last three years. The arrival of AMD's Zen architecture marked the beginning of the end of an absolute domain by Intel, and with the launch of Zen + and Zen 2 we reached a situation of total competition between both companies.
As we have said on previous occasions the competition is always good, and in this case it was noted in a very clear way. Before the arrival of the first-generation Ryzen, based on Zen architecture, quad-core and eight-wire processors they formed the high end within the general consumer market, and processors with six or more cores had prohibitive prices for most users.
AMD completely changed that situation. Zen opened the doors on the arrival of processors with up to eight cores and sixteen threads at really affordable prices, marked a new trend in the sector and forced Intel to "evolve." The stagnation of the four cores and eight threads, which had been maintained since 2009, ended in 2017.
Since that year both Intel and AMD have been launching processors with an increasing number of cores and threads, shaping a "war" that has allowed prices to be reduced to levels that a few years ago would have seemed impossible. Do you think we exaggerate? Not at all, stop a second despite what a six-core and twelve-wire CPU cost in 2017 and what it costs today.
Are you lost? Quiet, we tell you with a clear example: the Core i7 6800K was around 400-450 euros, and today we can find a Ryzen 5 1600 by less than 100 euros. The difference is huge, and if we compare eight-core and sixteen-wire processors it would be even greater.
Intense competition is beneficial for the user because it encourages innovation and enhances price wars, but in the end it can also have an important counterpart, and is that the annual launch of new processors creates confusion among users, especially among the less advanced. They do not quite know what differences the new generations make compared to the chips they use, and it is difficult for them to determine when they should change the processor.
That is why we have decided to update our guide of equivalences of Intel and AMD processors, an article that has become one of the most valued by our readers and that we want to keep “fresh” so that it serves you as a reference point at all times. We will keep the classic format, and we will introduce the latest releases of both companies so that you have all the information you need.
Intel and AMD processors: architectures and manufacturing processes
Both Intel and AMD have used different architectures and manufacturing processes. The chip giant kept a tick-tock strategy until the arrival of the Kaby Lake series, which consisted of following an annual cycle of "process reduction-architecture change".
Under that Intel strategy I never used a manufacturing process more than twice. One year introduced a reduction of the manufacturing process ("tick") and the following year a new microarchitecture under that manufacturing process. The following year it reduced the manufacturing process again. This allowed him to introduce notable improvements in terms of CPI and efficiency, but with the jump at 14 nm he could not keep up.
From 2015 to today Intel has been using the manufacturing process of 14 nm, although it has been refining to improve gross performance significantly. His next big move in this direction will be the 10 nm process, a step that Ice Lake U and Y series have already taken, both of which are low consumption, but which have not yet been carried out in their line of desktop processors .
Unfortunately This process change will not occur in the short or medium term. Before the arrival of the Ice Lake-S series we will see the Comet Lake-S processors, a new generation of desktop processors that will arrive in the coming months and that will maintain the 14 nm ++ process.
After this simple introduction we are able to take a look at the different architectures and manufacturing processes that Intel has used so far. We have incorporated the Comet Lake series because its launch is just around the corner, and thanks to that we already know all its key specifications.
Intel architectures: general consumption processors
- Conroe and Kentsfield: architectures based on the 65 nm process and used in the Core 2 Duo 6000 and Core 2 Quad 6000.
- Wolfdale and Yorkfield: The process of 45 is based and used in the Core 2 Duo 8000 series and Core 2 Quad 8000-9000.
- Lynnfield and Nehalem: architecture based on the 45 nm process and has been used in the first generation Core i3, Core i5 and Core i7 processors (5xx series and above, except for the Core i7 980X, which comes in 32 nm).
- Sandy Bridge: It is based on the 32 nm process and was used in the Celeron, Pentium, Core i3, Core i5 and Core i7 second generation processors (2xxx series).
- Ivy Bridge: architecture based on the 22 nm process and was used in the Celeron, Pentium, Core i3, Core i5 and Core i7 third generation (3xxx series) processors.
- Haswell: It is based on the 22 nm process that has been used in the Celeron, Pentium, Core i3, Core i5 and Core i7 fourth generation (4xxx series) processors.
- Broadwell: architecture based on the 14 nm process that was used in the fifth-generation Celeron, Pentium, Core i3, Core i5 and Core i7 processors (5xxx series).
- Skylake: architecture based on the 14 nm process and used in the Celeron, Pentium, Core i3, Core i5 and Core i7 ranges of the sixth generation (6xxx series).
- Kaby lake: It is based on a 14 nm + process and is used in the Celeron, Pentium, Core i3, Core i5 and Core i7 seventh generation (7xxx series) ranges.
- Coffee lake: architecture based on the 14 nm ++ process that has been used in the Celeron, Pentium, Core i3, Core i5 and Core i7 ranges of the eighth generation (8xxx series).
- Coffee Lake Refresh: based on the 14 nm ++ process and used in the Celeron, Pentium, Core i3, Core i5, Core i7 and Core i9 ranges of the ninth generation (9xxx series).
- Comet Lake: architecture based on the 14 nm ++ process that will be used in the Celeron, Pentium, Core i3, Core i5, Core i7 and Core i9 tenth generation ranges (10xxx series).
Let's give it some context to better understand everything we have exposed. A processor Core i5 2400 use the architecture Sandy Bridge and is manufactured in 32 nm process. By cons, a processor Core i5 8400 is based on architecture Coffee lake and manufactured in 14 nm ++ process.
We continue with a couple more examples. The Core i7 860 is based on architecture Lynnfield and manufactured in 45 nm process, while the Core i7 7700 use the architecture Kaby lake and is manufactured in 14 nm + process.
I'm sure those examples they will have helped you to clarify the subject of architectures, manufacturing processes and nomenclatures. On the issue of performance increases and changes in core counting we will talk later, specifically when we focus directly on equivalences.
Before entering to see the key architectures of AMD I want to review the main architectures that Intel uses in the HEDT sector, a level that is aimed primarily at professionals and content creators. We have omitted the older models because they have ceased to make sense by being overcome by current models that are much cheaper and more powerful.
- Haswell-E: architecture based on the 22 nm process. It is used in the Core i7 Extreme 5000 series.
- Broadwell-E: architecture based on the 14 nm process. It is used in the Core i7 Extreme 6000 series.
- Skylake-X: architecture based on the 14 nm process. It is used in the Core i7 and Core i9 Extreme series 7000X and 7000XE, and also in the Core i7 and Core i9 series 9000X and XE.
- Kaby Lake-X: architecture based on the 14 nm + process. It is used in the Core i5 and Core i7 7000X series.
- Cascade Lake-X: architecture based on the 14 nm ++ process. It is used in the Core i7 and Core i9 10000X and XE series.
AMD architectures: general consumption processors
The giant Sunnyvale stayed in a complicated situation due to the fiasco that supposed the processors based on the architecture Bulldozer and its derivatives. AMD had achieved excellent results with the chips based on the K8 architecture, on which the mythical Athlon 64 and the Athlon 64 X2 were created.
Subsequently, he was able to withstand Intel's turn by resorting to the K10 architecture, but the Bulldozer series was a skid so large that its single-threaded performance was even below the previous generation models. AMD managed to improve the situation a bit by polishing this architecture and launching new revisions (Piledriver, for example), but did not achieve the expected result.
AMD remained in a secondary position for many years, until finally, in 2017, went back through the big door with Zen, an architecture that allowed it to shape the excellent Ryzen 1000 series, and that it has maintained as the basis of its latest renovations.
During the last years AMD has also combined process reductions and architecture changes. With the arrival of Zen 2 the company has managed to make the leap at 7 nm and beat Intel, but we must bear in mind that he has achieved it with a MCM architecture (multi-chip module), and not with a monolithic core architecture. The first is simpler and allows to face process reductions with greater guarantees, while the second is much more complicated in that regard.
If you have doubts on this subject I invite you to review the article that we published at the time dedicated to deepen on the key aspects of a processor, including from frequencies and cache to cores, architecture and manufacturing process.
That said, we enter, without further delay, to review the AMD main architectures in the general consumption processors sector:
- K8: a mythical architecture on which processes of 90 nm and 65 nm have been used. He gave life to Athlon 64 X2 and Sempron processors.
- K10: based on the 65 nm, 45 nm and 32 nm process. It was used in the Phenom, Phenom II, Athlon X2, Athlon II and Sempron processors.
- Bulldozer: based on the 32 nm process. It has had several revisions and is used in the AMD FX, Athlon II X4 (and lower) processors and in the 4000 series and higher APUs (up to the 9000 series).
- Zen: It is based on the 14 nm process and is used in the new Ryzen 3, Ryzen 5 and Ryzen 7 1000 series processors, as well as the Ryzen Pro 1000 series, ThreadRipper 1000 series and the Ryzen 2000 series APUs.
- Zen +: It is based on the 12 nm process and is used in the Ryzen 3, Ryzen 5 and Ryzen 7 2000 series processors, as well as in the Ryzen Pro 2000 series and ThreadRipper 2000 series and in the Ryzen 3000 series APUs.
- Zen 2: an architecture based on the 7 nm process that is used in the Ryzen 5, Ryzen 7 and Ryzen 9 3000 series processors, as well as in the Ryzen Pro 3000 series and ThreadRipper 3000 series.
Unlike Intel, which separates the architectures of its general consumption and HEDT processors even though they have an obvious common foundation, AMD groups them. Thus, the 1000 series Threadripper use the Zen architecture and are manufactured in 14 nm process, the 2000 series Threadripper use the Zen + architecture and are manufactured in 12 nm process and the 3000 series Threadripper use the Zen 2 architecture and come in the process of 7 nm
Intel processors: series and features
We have already seen all the keys to the architectures that Intel has used in recent years, information that it gives us the base we need to enter to review the series and characteristics that have the different processors of the giant of the chip.
We adopt a simplified approach for you to find Easier to assimilate. In the final section of this guide we will complete it with a precise and simple relationship of equivalences to understand and consult.
- Core 2 Duo: these are old processors with two cores and two threads. They are very old and have been widely surpassed, but they still perform well with games of the Xbox 360 and PS3 generation, and also with less demanding applications.
- Core 2 Quad: It is an evolution of the previous ones that raises the maximum of cores and threads to four. They can move current games thanks to their four cores, but not completely optimally.
- Intel Celeron: They are economical processors with two cores and two threads that cover the basic level. They offer good performance in general office automation, multimedia and web browsing, and also with less demanding games.
- Intel Pentium: Skylake architecture-based models have two cores and two wires and, in general, do not offer a significant performance improvement over the Celeron. With the arrival of Kaby Lake, Pentium G4560 and higher mount two cores and four threads, which makes them a solid choice for cheap multimedia PCs. They perform well in most games of the current generation, except for the most recent ones that require four cores.
- Intel Core i3: until the 7000 series (Kaby Lake) they have two cores and four threads until the generation. With the arrival of Coffee Lake they have made the leap to the four cores, and it is rumored that with Comet Lake they will have four cores and eight threads. They have a high CPI and offer good performance in general, which makes them an interesting option to assemble low-cost gaming equipment. They serve to work and to play.
- Intel Core i5: It remains one of the best value-for-money ranges Intel offers today. Models based on Kaby Lake and earlier come with four cores and four threads, but with the arrival of the Coffee Lake architecture they have made the leap to the six cores and six threads. They also serve to work and play, and can move any current game with full guarantees. It is said that Comet Lake (Core 10000) will raise the count to six cores and twelve threads.
- Intel Core i7: as in the previous case there was an important leap. Until the 7000 series (Kaby Lake) it had a four-core and eight-wire configuration. With the arrival of the Coffee Lake Intel architecture, the count was increased to six cores and twelve threads, and the 9000 series has configured them with eight cores and eight threads. They offer exceptional performance and can with anything. They are prepared to completely overcome the transition marked by PS5 and Xbox Series X. When the arrival of Core 10000 processors is expected, an increase to 8 cores and 16 threads is expected.
- Intel Core i9: It has become Intel's new top of the range in the general consumer market. They debuted with the 9000 series (Coffee Lake Refresh), offer high performance and have 8 cores and 16 threads. They can with anything and have a long lifespan ahead. With the launch of the Core i9 10000 series we will see a renewal that will improve the specifications to 10 cores and 20 threads.
- Intel Core HEDT series: They are high-performance processors that have between six and eighteen cores, and that thanks to HyperThreading technology can work with a thread with each core, which leaves us configurations of up to 18 cores and 36 threads. They are aimed at the professional sector and use a specific platform that allows them to mount RAM configurations in quad channels and have a greater number of PCIE lines.
AMD processors: series and features
As in the previous case we have cleared any possible doubt about the most recent AMD architectures, so we have everything we need to start seeing the different series of AMD processors and to review its most important characteristics.
We will keep the previous format so that it is easy for you to consult and compare, although you must bear in mind that there is no total correlation and therefore It is not a comparison, but a descriptive list. In the end you will find the exact equivalences, that is, the comparison.
- AMD Athlon 64 X2: were, at the time, the rivals of the Core 2 Duo, although they offered a lower performance. They add two cores and two threads, they can also move applications and demanding games from previous generations.
- AMD Phenom II: its launch came at a time of transition, so they rivaled the Core 2 Quad and the first generation Core (Lynnfield). They add between two and six cores and offer a higher gross performance than the Athlon 64 X2. They are also outdated, but models with four and six cores can still get an acceptable experience in many games and applications.
- AMD Athlon: we can find versions that have between two and four cores. The performance of Bulldozer-based versions and their derivatives is good in any basic task, and quad-core models offer acceptable performance in games that are not very demanding.
- APUs: they are solutions that integrate processor and graphics unit in the same package. There are very varied configurations for both CPU and GPU architecture and specifications. For example, the less powerful and older models are based on Bulldozer at the CPU level and Terascale 3 at the GPU level, while the most current ones use the Zen + architecture and Radeon Vega at the processor and graphics unit level. These versions mount processors of up to 4 cores and 8 threads and graphic units with 704 shaders, which makes them quite powerful solutions and capable of moving current games with guarantees.
- AMD FX 4000: they add two complete modules and have four integer cores at very high working frequencies, in addition to an unlocked multiplier. They offer acceptable performance in less demanding games.
- AMD FX 6000: They have three complete modules and have six integer cores at very high working frequencies, in addition to an unlocked multiplier like the previous ones. Their performance is good, but they do not offer a totally optimal experience in current games.
- AMD FX 8000-9000: They have four complete modules and eight integer cores. They also work at a very high frequency and support overclocking. They still offer good performance and can with current games.
- Ryzen 3: Zen architecture marked a huge leap at the IPC level compared to Bulldozer (52% more than the first generation models). These models have four cores and four threads. They are very economical and can move any current game with guarantees.
- Ryzen 5: There are three variants, the 1500 and lower models, which add up to four cores and eight wires, and the 1600,2600 and 3600 models, which have six cores and twelve wires. AMD launched a Ryzen 5 3500 with six cores and six threads, but its availability has been very limited. Their performance is very good, they can with current games optimally and are prepared to work with demanding multi-threaded applications.
- Ryzen 7: Add 8 cores and 16 threads in its three generations (1000, 2000 and 3000 series). They offer excellent performance in any scenario and are prepared to smoothly overcome the transition that will mark the new generation of consoles.
- Ryzen 9: We have two versions, the Ryzen 9 3900X, which has 12 cores and 24 threads, and the Ryzen 9 3950X, which adds 16 cores and 32 threads. They are the most powerful that exist in the general consumer market, and can with anything.
- Ryzen Threadripper 1000: They are high-performance processors that use Zen architecture and have up to 16 cores and 32 threads. They are integrated in a higher platform, and thanks to this they can use quad-channel memory configurations and offer a greater number of PCIE lines.
- Ryzen Threadripper 2000: an evolution of the previous ones based on the Zen + architecture. They add up to 32 cores and 64 threads and use the same platform. They are designed for professionals who use very demanding multithreaded applications (rendering and content creation, for example).
- Ryzen Threadripper 3000: It has been the latest evolution of AMD high performance processors. They have up to 64 cores and 128 threads and use a platform that supports quad-channel memory and offers a large number of PCIE lines.
Equivalences between Intel and AMD processors
It has been a long journey but we already have clear the different architectures used by Intel and AMD processors of different generations, the evolution that they have lived during the last years and we also know its most important characteristics.
We are ready to see a more direct and concrete relationship of equivalences. We have covered many generations to help even users who have very old processors, but this has an important disadvantage, and we cannot make an individualized list with each processor available in each generation and range and its direct equivalence, since it would result in a huge text.
It would also be counterproductive and confusing, as well as difficult to consult. That is why we are going to offer a more general relation of equivalences, starting from each one of the series that we have seen and offering concrete examples that will serve as reference points. If you have any questions you can leave it in the comments and we will help you solve it.
- Core 2 Duo: They are quite old processors that are limited by their IPC and its two cores. They surpass the Athlon 64 X2, but they are obsolete. The most frequently used models are close to the Core i3 500 series, although their overall gross power is lower.
- Core 2 Quad: its four nuclei have allowed them to endure the passage of time better. The most powerful models, such as the Core 2 Quad Q9450 and higher, offer acceptable performance and remain close to a Core i5 750. Its direct rival is the AMD Phenom II X4, although thanks to their higher working speeds the latter offer a superior performance For example, the Phenom II X4 965 performs more than the Core 2 Quad Q9650.
- Intel Core x00 series: They are the first generation Core. Up to the Core i5 (inclusive) we can make an approximate equivalence with the Core 2 Quad Q9450 and higher, and also with the AMD Phenom II X4 and FX 4100. Higher models, such as the Core i7 860, can handle eight wires thanks to HyperThreading, so they are at a level similar to that of the FX 8100 and 6100. The AMD Phenom II X6 also fits here, which adds six cores.
- Intel Core 2000: marked an important leap in terms of performance compared to the previous generation. The Core i3, which has two cores and four threads, roughly equals the FX 4300, the Core i5, with four cores and four threads, has the closest equivalence to the FX 6300, and the Core i7, which has four cores and eight threads are assimilated to the FX 8350, although these are inferior in gross performance. As a reference of interest, I remind you that the Pentium G4560, which adds two cores and four threads, offers a performance similar to the Core i5 2500 in applications that take advantage of four cores thanks to its higher IPC.
- Intel Core 3000: they maintain the same core count and overall performance as the previous generation, so their closest equivalents are exactly the same.
- Intel Core 4000: They do not raise the number of cores, but they did bring a jump at the level of CPI and working frequencies, so they offer higher gross performance. They outperform the FX 8300, FX 6300 and FX 4300, but fall short of the first-generation Ryzen processors (1000 series).
- Intel Core 5000: we are facing a "tick" (reduction of manufacturing process). There was no increase in the number of cores and no gross yield, so we keep what was seen in the previous point.
- Intel Core 6000: another generation that did not bring an increase in the number of cores, although it compensated with a higher CPI and higher working frequencies. Its closest equivalent is the Ryzen 2000 series in terms of IPC, but it must be remembered that this generation has more cores and threads. For example, the Ryzen 5 2600 has a monohilo performance similar to the Core i5 6600, but the first one has six cores and twelve threads, and the second only has four cores and four threads.
- Intel Core 7000: Both IPC and core count are maintained, although Intel achieved a small increase in performance compared to the previous generation by raising working frequencies. Their gross performance is slightly higher than that offered by the Ryzen 2000 series processors, but they have less multi-threaded potential. For example, the Ryzen 7 2700X has a lower performance in mono wire compared to the Core i7 7700K, but the first one adds 8 cores and 16 threads and the second is limited to four cores and eight threads.
- Intel Core 8000: represents another small advance in gross performance pulling frequencies, without changes in the CPI. The most important novelty is in an increase in the maximum number of cores that affected the entire series. The Core i3 has four cores and four threads, the Core i5 six cores and six threads and the Core i7 add six cores and twelve threads. In gross monohilo performance they are practically at the same level as the Ryzen 3000, but the latter have a higher multithread potential. For example, the Ryzen 5 3600 is equivalent to a Core i7 8700.
- Intel Core 9000: We remain unchanged at the CPI level. Intel pulled frequencies again and increased cores to offer greater performance. The Core i3 and Core i5 had no changes, but the Core i7 went from six cores and twelve threads to eight cores and eight threads. The Core i9 add 8 cores and 16 threads. Its monohilo performance is a little above the Ryzen 3000 for its higher clock frequencies, but the latter have higher multi-thread configurations, since they reach 16 cores and 32 threads. We go with examples of direct equivalences, a Core i9 9900K is equivalent to a Ryzen 7 3800X, and a Ryzen 5 3600X is above a Core i5 9600 thanks to its six cores and twelve threads (the second only has six cores and six threads) .
- AMD Ryzen 9: It has no direct rival from Intel, since we talk about configurations of up to 16 cores and 24 threads. With the arrival of the Comet Lake-S series, Intel will launch the Core i9 10900K, a chip with 10 cores and 20 threads that will still not be at the level of the Ryzen 9 3900X, which adds 12 cores and 24 threads.
- Intel Core HEDT series and Threadripper: First-generation Threadripper processors have an IPC comparable to that of the Core Extreme based on Broadwell-E, but they are a little behind the current Skylake-X. On the other hand, the second generation Threadripper have shortened distances in terms of IPC, but thanks to their greater number of cores and threads (18 and 36 the most powerful model of Intel and 32 and 64 the most powerful of AMD) they are superior in terms general The 3000 series Threadripper have raised the CPI again and thanks to the increase in the maximum number of cores and threads (64 and 128 respectively) they have become the most powerful in their category, so much so that they have no direct rival from Intel, at least for now