Intel has announced a significant advance for its neuromorphic research processor, named Codehi. The company has now extended its implementation of Loihi to 64 processors, allowing it to create a system with over 8 million neurons (8.3 million). This new configuration (codenamed Pohoiki Beach) delivers 1000 times better performance than conventional processors in applications such as sparse coding, graph search, and constraint satisfaction problems. Intel says the new Pohoiki range offers 10,000 times more energy efficiency than conventional processor architectures in this type of test.
Neuromorphic Computing is a subset of the computer that attempts to imitate the architecture of the brain with the help of modern technological analogues. Instead of implementing a typical CPU clock, for example. Loihi is based on a peaked neural network architecture. The basic Loihi processor contains 128 neuromorphic cores, three Lakefield processor cores (Intel Quark) and an off-chip communications network. Theoretically, Loihi can handle up to 4,096 on-chip cores and 16,384 chips, although Intel has announced that it does not intend to market a design of this size.
"Thanks to the Loihi chip, we were able to demonstrate a power consumption 109 times lower than that of a real-time graphics processor, and a power consumption 5 times lower than that of an inference material. IoT specialized, "said Chris Eliasmith. CEO of Applied Brain Research and Professor at the University of Waterloo. "Even better, while the network is multiplied by 50, Loihi retains performance results in real time and uses 30% more power, while the IoT hardware consumes 500% more power and is no longer in time. real."
The implementation of Pohoiki Beach is not the largest deployment planned for the neuromorphic chip. Intel announces its intention to roll out an even larger concept, dubbed Pohoiki Springs, which will offer "unprecedented levels of performance and efficiency for enhanced neuromorphic workloads."
We covered the advances and research in neuromorphic computing for several years at ET. The work done on these processors is closely related to the work done in AI and the machine intelligence in general, but it's not just about how to perform AI / ML workloads. on existing chips. The ultimate goal is to build processors closer to the human brain.
One of the quirks of computing is that the analogies between the functioning of the human brain and the operation of computers are widespread. Human brains and conventional computers do not overlap very little on their functioning. Transistors are not equivalent to neurons and the pulsed neural network that Loihi uses to transmit information about his own processor cores is intended to be closer to biological processes that humans and other animals use than traditional silicon.
Projects like this one have several long-term research goals, but one of the most fundamental is to better understand how the brain works to replicate some of their energy efficiency. The human brain operates at around 20W. The Exascale supercomputer, considered the minimum for advanced neural simulation of anything more complex than an earthworm, should consume megawatts of power per supercomputer. The difference between these numbers explains why we are primarily interested in the long-term energy efficiency and computing potential of the brain. Architectures such as Loihi are not just an effort to write programs that mimic what humans can do. the goal is to copy aspects of our neurology as well. This makes their progress a little more interesting.
Background image: Tim Herman / Intel Corporation
If you own Ryzen 3000 (or plan to become one soon), be aware that Bungie's popular MMO Destiny 2 does not currently work on the latest AMD processor. The reason this is the case currently is not clear, but according to the developer, a fix is in progress. However, an update of the motherboard will be necessary to solve the problem.
AMD has identified the problem affecting the ability to run Destiny 2 on Ryzen 3000 processors and has implemented a BIOS patch distributed to partners. In the coming days, players will be able to download the updated BIOS from their motherboard vendors.
– Help Bungie (@BungieHelp) July 12, 2019
The problem appeared a few days ago when Ryzen 3000, the first correctors, discovered that they could not run the game. A week ago, Redditor Trinsikk posted:
Destiny does not work on PC with the new processors of the Ryzen 3000 series. After clicking on the battlenet play button, the game is running and the Destiny 2 executable will show up in the task manager, but it only states that it uses 5 to 10% of the CPU, but never starts. Some discussions have appeared on the bungie forums and this seems to affect all variants of the mother / ryzen 3000 configurations. I have tried to reinstall Windows 3 times with full formats and reinstall the drivers each time. Bungie Halp.
It's not unusual to have some bugs to eradicate, but it's a little unusual: we do not remember the last time the UEFI update was required to eject the bugs of a specific game. In this case, there may be a problem with the way the instructions are issued on the AMD processors. according to Michael Larabel, there are signs of Linux hiccups on new AMD processors also, and the new processors have problems starting some Linux distributions. It seems possible that the problems are related. If this is the case, the updates to the motherboard should solve the problem.
AGESA updates have been shown to improve both the performance and compatibility of Ryzen processors. At the first launch of Ryzen, AMD stated that gaming performance and overall compatibility would improve over time. That was indeed the case – our initial Gigabyte board that we tested in 2017 on Linux until later updates solved the problem, and the overall performance of the first-generation Ryzen improved. with time. When we looked at CPU performance later in 2017, we found that he had picked up speed in several tests.
The situation with the X570 has been much better. AMD's single-threaded overall performance is stronger with the Ryzen 3000 family than it was when the first generation debuted, but there are obviously some compatibility issues to be resolved. If you're a Destiny 2 player, watch for updates to the vendor support page on your motherboard.
The AMD Ryzen 7 3000 family has fallen earlier this week, carrying a heavy blow to the entire stack of high-end products from Intel. The Ryzen 9 3900X destroys the performance of the Core i9-9900K at 500 USD, while the Ryzen 7 3200 USD at 329 USD is slightly faster than the 9900K in most tasks, at 66% of the price. This is not an excellent position for Intel, and the company is definitely considering its own competitive response.
A recently disclosed roadmap claims to show what this response will be. As always, these leaks must be taken with a grain of salt, especially given the unusual nature of some of the predictions made by the slide. There is a new socket that we have never heard of before (LGA 1159), Hyper-Threading is back in the high end (after being removed from all the chips of the 9th generation, except the most high-end ), and supposedly, the new processors are built on "14nm +++", even though Intel has already indicated that it was abandoning this nomenclature. High-end cores also lack IGP, which is not a step that Intel was previously willing to take, except in case of severe need, to improve its performance.
It's not clear if the supposed 10-core chips would have a disabled but present GPU core, or if Intel actually built a totally different die without any processor. The leak also implies that Intel will add Hyper-Threading to the entire stack of products, which the company did not want to do. Finally, this leak discloses the maximum all-in-one increase frequency, the information generally stopped by Intel.
Season this dish well, in other words.
The use of five-digit model numbers is one of the cases of rumors in previous reports, but as the previous one was concerned with moving parts, it was not possible to compare the lists disclosed to check whether they were 'agreement. The TDPs are believable, mainly because the Intel TDPs are derived from the base clock, and the base base clocks here could allow a 10-core chip. The Core i7-9700, for example, is an 8C / 8T CPU with a base clock at 3 GHz in a TDP of 65W. A 2.7GHz 10C / 20T processor in a 65W TDP does not seem impossible based on Intel's already existing product lines.
Intel has never publicly explained why it removed the hyper-threading of its 9th generation processors, Although some speculation was made, this decision was linked to the security flaws reported by Specter and Meltdown that were reported in January 2018. Whatever its reasoning, adding functionality to its 10th generation processors would allow Chipzilla to recover additional performance for a relatively long time. trouble of small power. Hyper-threading is generally supposed to offer additional performance between 1.1x and 1.2x, although some games and other applications actually see a very small penalty when the feature is enabled.
If these leaks are accurate, it would mean a realignment of the entire stack of Intel's competitive products. The addition of Hyper-Threading would give the Core i3 and i5 chips a more powerful jog for attacking the Ryzen 3 and Ryzen 5 families of AMD. Chips such as the Core i3-10350K offer 4C / 8T and a boost of 4.6 GHz for $ 179. The Core i5-10500 (6C / 12T, 3.1GHz base, 4.4GHz all-core boosted) at $ 199 would address the Ryzen 5 3600 (6C / 12T, 3.6GHz base, 4.2GHz boost), also at $ 199.
If this graphic is genuine, it would mean that Intel is taking the fight against AMD seriously for its entire product line. Putting HT back on the Core i7 and adding it for the first time to the Core i3 and Core i5 of the desktop would bring them extra performance at low and midrange speeds. Adding two more cores to the top of the desktop market would give the Core i9-9900K substitute enough power to beat the Ryzen 7 3700X and 3800X. We do not know if such a chip could beat the 12-core Ryzen 9 3900X, not to mention the Ryzen 9 3950X that will arrive in September, but this stack of products – so true – would improve Intel's global competitiveness across the entire product line, not just the high end.
At Semicon West 2019, a panel of industry experts debated whether Moore's law – the big prediction provided by Gordon Moore, said the number of components per integrated circuit would double over a period predictable time (originally 12 months, later extended to 24 months) – was still alive. Over the past decade, whether Moore's Law was viable in the long run or whether it was already dead and replaced by other methods of scaling up has become more commonplace.
Not surprisingly, there are many points of view on the subject. I have discussed in the past that if the text of Gordon Moore's prediction of 1965 did not change, how did people including it has a lot. Almost from the beginning, Moore's law – which initially provided for the density of transistors – was expanded to include performance predictions. The concept of scaling nodes has also changed over the years; It's been 20 years since node names refer to a specific entity size. When TSMC, Samsung, or Intel talk about a new node, they mean they have combined a combination of new technical approaches, smaller feature sizes (in some cases), significant changes, and manufacturing enhancements that Collectively justify have developed a new way of manufacturing transistors.
This kind of definition leaves room for a lot of play. According to Aart de Geus, co-CEO of Synopsys:
Moore's Law is based on the behavior of an exponential who has a technical back to the exponential that has revolutionized what humanity can do. The reason I say it 's totally alive, is that we are facing an extra one or two decades of incredible opportunities that will economically boost the thrust of technology without stopping. Maybe it is not exactly the same curve as that drawn by Moore, it does not matter.
Victor Peng, CEO of Xilinx, singled out the traditional scaling up of Moore's Law that offered advantages in terms of performance, power, and surface (that's Moore's Law + Dennard Scaling , and the fact that the CEO has handled the issue in this way testifies to the concepts have been mixed over the years.According to him, when it was possible to see benefits in these three areas over time, now companies had to focus on one or two of them and in that sense Moore's law was no longer working.
Lisa Su, CEO of AMD, who recently launched the most revolutionary processors his company has opened the door in 15 years, does not see that Moore's law is disappearing, but think that it is slowing down and that companies need to integrate more and more technologies outside the traditional field of the scaling the nodes to succeed. Other industry stakeholders shared their own points of view (Semi Engineering see you). In your opinion, the evolution of Moore's Law depends on the other technologies used by your industry to continue to advance silicon, be it NAND 3D stacking, advanced packaging techniques, or new hardware technologies. offering better performance.
Yet, despite the wide debate over whether Moore's law can be properly described as "dead," I think the very fact that we are discussing this point is itself instructive. If you have been a computer enthusiast for quite a long time, you remember a time when even asking the question "Is Moore's Law Dead?" You would have brought a funny look. Whether Moore's Law mentions transistor densities or that it also includes aspects of performance and power, these are factors that people have been debating and confusing for decades. The fact that it existed and worked was, however, inviolable. Today, we see people who are openly looking for other methods of definition that allow Moore's Law to Carry on to be true, mainly because of the classic definition no longer works in fact.
But once you have reached the point where the old definition actually works, you have arrived at a transition point. It is possible to retain the concept of Moore's Law to describe the equivalent improvement, but we have thus changed the fundamental nature of what Moore's Law was supposed to be. In the end, Gordon Moore's foundational article becomes a sort of ship of Theseus's problem – when is Moore's law itself no longer Moore's law? When did we dissociate the technical reality from what was originally described so perfectly from what the law supposedly refers to who has more to do with it? ;other?
It would be quite amusing if, in 500 years, AI-based hard light structures, which serve as frames for large conglomerates made up of human and business firms, are still talking gleefully about Moore's Law advances, describing improvements to the latest quantum peanut butter. Computer singularity – based on Gordon Moore technology that I have never heard of and circuit designs that I could not imagine.
Given how marketing works, I would not bet against that.
As Moore's law slowed down, manufacturers focused on optimizing other aspects of their devices. The reduced transistors design and smaller chip construction no longer bring the performance and efficiency improvements that it previously allowed. The improvement in packaging technology, on the other hand, offers a real potential for improving performance and reducing power.
At Semicon West, Intel unveiled several new tools in its proverbial toolbox to answer advanced packaging issues. In the past, we discussed Intel's two-dimensional technology for connecting technology packages (EMIB, also called integrated multi-chip interconnect bridge) and the three-dimensional technology used for future parts, like Lakefield (Foveros):
Intel has unveiled a new system allowing it to deploy together EMIB and Foveros, in the same package. Nicknamed ODI (Omni-Directional Interconnect), Intel says it will be an element of differentiation in the years to come.
"Our vision is to develop technological leadership to connect chips and chips in a package that matches the functionality of a monolithic on-chip system," said Intel vice president Babak Sabi. "A heterogeneous approach gives our chip architects unprecedented flexibility to combine and match IP blocks and processing technologies with various memory and I / O elements in new device shape factors."
Intel has released a video showing how EMIB, Co-EMIB and Foveros can be combined to create a single product. As a reminder, EMIB is a very small interlayer embedded in a substrate. This layer connects to two PHYs and provides the same type of physical connection as an HBM interlayer for a fraction of the cost. It is said that EMIB costs 0.3 picojoules (pJ) / bit of data transferred. Foveros, which allows Intel to stack chips into 3D stacks face-to-face, allowing for higher scaling and even smaller relief steps. The energy cost for data transfer via Foveros would be even lower than that of EMIB, at 0.15 pJ / bit. Co-EMIB combines Foveros and EMIB in the same technology and deployed in the same design.
Omnidirectional interconnection allows the most packaged chips to communicate with other chips horizontally, as in EMIB, or vertically, via TSVs, such as Foveros. A unique feature of ODI, however, is the use of large TSVs to feed the top of the matrix. Intel writes: "Much larger than traditional TSVs, large vias have lower resistance, providing a more robust power supply simultaneously with higher bandwidth and lower latency enabled by stacking. approach reduces the number of required TSVs in the base, freeing up more area for active transistors and optimizing chip size. "
Intel has also introduced a new approach, MDIO, based on its advanced interface bus and offers twice the spindle speed and bandwidth density.
The cynical way to look at this situation is that Intel is pushing to talk about its packaging enhancements because there is no processor on which to brag. There is probably a little truth in this situation – but less than you think. Experts had predicted that companies would opt for packaging optimization because of Moore's law downturn for years. AMD's adoption of chiplets is an example of how businesses of all kinds, including those who are undeniably at the top of their game, are exploring new technologies. And advancing to exascale requires that we find ways to move from point A to point B with ever smaller amounts of electricity.
It's unclear when products using technologies such as Co-EMIB or ODI will hit the market, but Intel is already talking about it as part of its own projects. It would not be crazy to suspect a connection between the two.
After months of rumors, revelations and previews, AMD's moment of truth in 7 nm has finally arrived. The third-generation Ryzen 3000 family, 7 nm, is the first opportunity for AMD to seize the top of the processor market in nearly 15 years.
This review assumes that you usually know Ryzen and Zen Architecture 2; Our deep dive on the subject in E3 will help you upgrade if you need a quick reminder. To recap: AMD today launches its family Ryzen 7 3000, a new and complete update of its previous generation Ryzen products. These new chips are built on TSMC's 7-nm process node and utilize a new "chiplet" concept that separates I / O circuits and DRAM controllers that do not particularly benefit from losses and keeps them in a state of the art. common chip and connected by all chiplets. at
The Ryzen 7 3000 family does not drastically alter the number of AMD cores nor the overall positioning of the product in the midrange and low frequency markets, but introduces a new 12-core desktop processor with a 16-core chip still waiting. AMD has developed an extremely aggressive product line against Intel and clearly intends to be beaten by its main competitor.
Pricing says it all. AMD has positioned its 12-core Ryzen 9 3900X to face Intel's Core i9-9900K. The eight-core Ryzen 7 3800X will be attacking Intel's Core i7-9700K eight-core, while the $ 329 Ryzen 7 3700X is in a category at this price. The cheaper parts of the stack will face their counterparts Intel positioned equivalently.
AMD has sampled two of its third-generation Ryzen processors for this launch – the Ryzen 9 3900X and the Ryzen 7 3700X. The company is particularly proud of this product and believes it has a superior position in the overall stack.
For 18 years that I have covered the computer industry, AMD has never beaten Intel at a node, until now. Nearly 20 years ago, AMD effectively attached Intel when the two companies started shipping 180-nm parts at about the same time. This has never happened again. Subsequently, Intel migrated to new nodes, or even years before its competitor, until today.
Even taking into account that Intel's 10nm and TSMC's 7nm are considered roughly equivalent, AMD is on the verge of having 7nm processors and GPUs on the market in volume . Intel currently has only one Core i3-8121 10 nm, delivered without GPU and used only in a handful of low-end systems.
It's a clear and solid victory for AMD.
The Ryzen 3000 family will also be the first family of desktop processors to integrate up to 12 platforms on a mainstream platform. AMD had originally planned to change the number of processors beyond 8, with tracking of its Bulldozer family, but eventually decided to focus on improving the performance of this architecture in terms of reduced power consumption, while zen architecture was designed to replace it. While Intel and AMD have been offering processors with more than eight cores for many years, neither company has ever brought as many chips to their traditional desktop platform. Intel reserves them for its HEDT family (High-End DeskTop) and AMD has deployed them as part of its Threadripper product line.
AMD has not hidden its ambitions for Zen Architecture 2. At this year's E3, Travis Kirsch, director of AMD's customer management, said that this Was not anyone to buy an Intel processor to eleven Ryzen who had debuted. This puts a high bar for AMD to erase, given the mediocrity of its processors compared to Intel a few years ago.
When AMD launched Ryzen in 2017, it forced Intel to modify its entire product line, introducing Hyper-Threading on the low-end processors of the Pentium family and adding extra cores to its Core i3, i5 and i7 processors. After almost 6 years of static kernel counting, the market has changed dramatically in the last two years.
The is not AMD is not a target for Intel, but we particularly want to compare some places, including:
2700X against 3700X: The 2700X shows a slight improvement over the original Ryzen 7 1800X, but AMD has announced at least an overall gain of at least 1.15 times greater than the 2700X. It is difficult for a company to succeed these days, but we have speculated that there may still be some fruit at hand after the frantic rush to market the Ryzen. # 39; origin. AMD has promised an increase of 1.15 times that of the 2700X. We will therefore check what the company actually delivers.
Improved power consumption: 7 nm is expected to bring major improvements in overall power and performance per watt. We will check to see how AMD compares to Intel and its own previous processors on this point.
Global Leadership in Performance: AMD has queued up to 12 Ryzen cores to match the Core i9-9900K and you do not have to be a CPU engineer to guess how. it the comparison will go. What will be interesting, is to see how AMD's single-threaded overall performance is or is not stronger.
1080p game performance: If Ryzen had 1080p, his "weakness" depended largely on the games you played and the GPU you used them, but some games did not turn as fast on the new Zen architecture of AMD compared to Intel. This situation has apparently improved – we will monitor to see how much.
Our test configuration is complex enough to deserve a graph.
Originally, our plan was to test all AMD processors on the same platform, but this posed a problem with some of the first available motherboard UEFIs and the tight schedule for this review. AMD provided our kits with a fairly reasonable test window, but the volume of hardware we had to test and retry was considerable. The 430.86 WHQL driver from Nvidia and a Gigabyte Aorus RTX 2080 were used for all game tests.
We chose to use 16GB of DDR4-3600 for Ryzen 3xxx processors, but we have limited older processors to 16GB of DDR4-3200. It was partly a practical decision – we simply did not have time to validate the fastest clocks on every processor – but we also wanted to compare what was a high end configuration on the different chips that we were using against the advantage of equipping Ryzen processors with faster RAM.
All systems have been tested with Windows 10 May 2019 Update with all installed security patches. All security patches were installed and UEFIs were used for each motherboard. All the results presented in this review are new, nothing was drawn from the previous cover. Some of our reference versions have been updated accordingly and we have modified our hand brake coding test – the results are not comparable to previous iterations of the reference.
Note: our results on the 9700K are slightly lower than expected for non-game performance tests. This is under investigation. The game benchmarks with the chip were pretty good. The UEFI settings appear to be correct.
Our power tests were treated a little differently than our standard tests. All test benches were equipped with 32 GB of DDR4-3200 for this test. We have always used Prime95 for power testing, but Prime95 has recently released an update that changes the power of Intel and AMD processors. Previous iterations of the application provided the ability to test medium-length FFTs and positioned it as the most energy-hungry application of the benchmark. The new version of the test has a new option for small FFTs. Power amplifications on AMD and Intel processors are tested with this new version of the program, but they increase much more on Intel processors.
For this reason, we tested the power supply in both versions of the program, as well as in Cinebench R20 when rendering the multicore scene.
Our results are included in the slideshow below.
In fact, one of our graphics is not properly sized for the slideshow, so it's placed just below it. All d & # 39; other is integrated in the slideshows below.
The Blender 1.02Beta reference offers a range of scenarios and scenes for comparison – and Ryzen treats them all. Even before we reached the 12-core Ryzen 3900X, the Ryzen 7 3700X is 1.13x – 1.16x faster than the 2700X and 1.05x – 1.1x faster than the Core i9-9900K. The Ryzen 9 3900X is 1.34x – 1.4x faster than the Core i9-9900K.
The offside performance of the Ryzen 7 3000 family is, in a nutshell, excellent. The Core i9-9900K was dethroned in almost all cases by a processor costing 65% more. If you're ready to spend $ 500, the Core i9-3900X does not beat the Core i9-9900K as much as the neutralizer. The price / performance benefit offered by AMD at $ 500 is the same as Threadripper's offer to Xeon, but for a fraction of the price.
Those considering the Ryzen 7 3700X as an upgrade to a 2700X or a 1800X can still expect significant performance gains. The improvement of 1.15 times predicted by AMD compared to the 2700X Before the launch was, in hindsight, at lowball.
Now that we've talked about off-game performance, let's take a look at the game. We've tested eight titles in 1080p, 1440p, and 4K, using high-end video settings that match the options of a high-end GPU. The results are in the slideshow below. Each slide can be clicked to open it in a new window if you wish.
Intel still has an advantage in 1080p and the 9700K is a surprisingly powerful player in the video game, but it is clear that the third generation Ryzen processors reduce the gap between them and Chipzilla. The geometric mean of our 1080p test results for the 9700K is 115 frames per second, compared to 109 frames per second for the Ryzen 7 3700X and 101 frames per second for the Ryzen 7 2700X. At this point, Intel's advantage in 1080p is reduced to 6%. A margin of error of 3 to 5% is not considered unusual in the benchmark tests, and although we did not claim that Intel's performance advantage is simply due to the margin of error – it's all too consistent for that – there's just not a lot of "there". "The.
In the opinion of this reviewer, Ryzen's "weakness" in 1080p games has always been a little exaggerated. With Zen 2 on the market, it has gone from a minor point to a non-problem.
Finally, we turn our attention to power. As many of you know, the behavior of Intel processors can vary greatly depending on how the motherboard manufacturer has programmed its boards and settings implemented by the motherboard. In our case, the two Intel motherboards tested seem to implement the thermal and current limits expected by the manufacturer of the chip, but this has a certain impact on the behavior of our Intel processors under load.
After a relatively short period of time (usually 8 to 20 seconds), the Core i9-9900K, 9700K and 8086K will all be back from the hard metaphoric controller. The Ryzen 7 2700X, 3700X and Ryzen 9 3900X do not behave this way. Where the 9900K slows down even before reaching Cinebench R20 multicore rendering, Ryzen processors keep a full clock and power consumed throughout.
Prime95 29.4b8 energy consumption. The Ryzen 7 3000 processors run for a long time in slow motion – well above their Intel counterparts and above the Ryzen 7 2700X. This could improve with subsequent revisions to the UEFI or could result from the relatively powerful PCIe 4.0 chipset.
In Prime95 29.4b8, Intel processors reach powers between 190W and 205W for relatively short periods before returning to lower power. This is somewhat different from their behavior in Prime95 29.8b5.
In Prime25.8b5, Intel processors reach higher power and then slow down faster. This behavior on our Core i9-9900K was consistent during reboots and multiple tests. Reduced to 140W, the Core i9-9900K consumes less power than any other processor we have tested. So we decided to add a more realistic workload.
In Cinebench R20, we still see processors bursting and retreating, but they do it less aggressively and the turbo clocks are kept longer. However, none of the Intel processors tested would take its turbo clock to finish rendering the CB20 in multithreaded mode. This behavior is settable in UEFI – we just use the default settings set by Asus for its motherboards.
AMD processors do not behave this way, but the fact that they reach higher powers does not mean that they use more total energy. The Ryzen 7 3700X and Ryzen 9 3900X complete the CB20 rendering faster than the Core i9-9900K, reducing total power consumption. The Ryzen 7 3700X also represents a significant improvement over the 2700X: it consumes approximately 80% of the wall power in all cases.
In the spring of 2005, AMD launched its dual-core Opteron and Athlon 64 X2 processors and launched what would later be known as its golden age – a time when it was heavily challenged Intel on desktops, servers and workstations. Over the past 14 years, Advanced Micro Devices has never been closer to an equivalent moment.
Not until now.
When Intel launched the Core i9-9900K, we said that it was the 2700X of great products, but that AMD had a lethal grip on the performance category per dollar. This is no longer the case. Dollar for dollar, the Core i9-9900K is annihilated in multithreaded applications by the Ryzen 9 3900X and slightly exceeded by the overall performance not related to the game by the $ 329 Ryzen 7 3700X.
Intel does not have an easy short-term answer here. There are rumors about a 10-core desktop computer at the horizon, but the high-frequency 14 nm TDPs and number of cores are not favorable to anyone. AMD already has a 16-core desktop chip coming in September, and while not as good as its Threadripper counterpart due to limited memory bandwidth, it will be scalable enough. Like the fleet concept, the existence of this 16-core processor is tangible proof that AMD has gasoline in the tank and products that it can introduce to improve performance.
Zen 2 is a tremendous victory for AMD. This may not be a absolute victory – Intel retains a small lead in 1080p games, and the 9700K defends itself for these reasons, if you're obsessed with optimizing performance – but if you look at the test results in 2005, you'd find that AMD did not literally win every one of them at the time.
We would like to see if we can reduce idle consumption of the Ryzen 7 family, and the relatively high power of the single processor is a bit confusing, but the performance efficiency of the Ryzen 7 3700X and 3900X is, in an excellent word AMD has more than provided the promised improvements.
We will have more to say about Ryzen 7, the X570 platform, Intel and the overall performance of these new processors in the days to come. Stay tuned.