Throughout 2019, we’ve had fairly the reverse of efficiency in relation to the competitiveness of the fashionable performance-oriented desktop processor. This 12 months we’ve seen AMD introduce its Zen 2 processor designs, providing as much as 16 cores for mainstream use circumstances, and matching if not beating Intel in uncooked clock-for-clock efficiency. In an analogous vein, AMD’s Rome CPUs provide as much as 64 cores per socket on the server aspect, the place Intel can solely provide 28 (or 56 in some configurations), whereas on the similar time each shopper and server pushing PCIe 4.zero in preparation for upcoming GPUs and add-in accelerators. Intel has additionally launched plenty of merchandise, similar to a few 5.zero GHz all-core CPUs, an unlocked 28-core CPU, Xeon 9200, and the following era of server merchandise.

If you’ve been away from the CPU area this 12 months, right here’s the Year in Review.

January 2019

In early January we attended the annual CES commerce present in Las Vegas. It was the beginning of many keynotes for AMD’s CEO Dr. Lisa Su this 12 months, which additionally included Computex, E3, Hot Chips, and varied AMD associate occasions.

The AMD Keynote at CES 2019: Looking Ahead

Part of the keynote was that AMD was dedicated to delivering on its 7nm CPU platform later within the 12 months, particuraly with its Zen 2 structure, displaying some very spectacular benchmarks together with an 8-core AMD vs 8-core Intel operating a benchmark, scoring equally, with the AMD at a quantifiably decrease system energy. We additionally received the primary have a look at the 8-core Matisse bundle, together with predictions for a dual-die bundle providing as much as 16-cores. At this level AMD was maintaining quiet on core counts.

Intel’s keynote in contrast was the primary for the reason that new interim CEO, Bob Swan, had taken over. It wouldn’t be till later within the month that he can be confirmed because the full-time CEO, however not like the earlier CEO, he wasn’t on stage. Instead, SVP Greg Bryant was as much as speak about how Intel had launched a CPU providing a 2-core 5.zero GHz turbo (the i9-9900Ok, in October 2018), but in addition spoke about how Intel was dedicated to delivering 10nm in 2019 by its cell Ice Lake platform, new 3D stacked ‘Lakefield’ chip, and in 5G with its Snow Ridge design.

Intel’s Keynote at CES 2019: 10nm, Ice Lake, Lakefield, Snow Ridge, Cascade Lake

Most of this was based mostly on Intel’s Architecture Day, which was just a few weeks prior in December 2018. At CES Intel additionally mentioned its future for the server market with a quick protection of Cascade Lake and likewise Intel’s AI choices for that individual market.

Also at hand was Intel’s Project Athena, which Intel had launched final 12 months, with the concept to create a degree of necessities for cell gadgets to ensure efficiency, battery life, and different options that Intel believes its high-end clients want.

Also at CES, we discovered of a brand new CPU that Intel didn’t need many individuals to learn about. The Core i9-9990XE was meant to be a particular processor stored underneath wraps. With 14 cores all rated at 5.zero GHz, this tremendous binned chip (in response to companions) would solely be out there to Intel’s key companions by an public sale solely course of. Not solely that, however these CPUs can be bought to the companions with none guarantee. End-users who would need these components embody high-frequency merchants who need the perfect of the whole lot and might amortize the prices. We ended up reviewing the {hardware} later within the 12 months.

Intel Core i9-9990XE : Up to five.zero GHz, Auction Only

As for CPU launches in January, Intel was the primary participant right here. In order to spice up the launch of its 9th Gen Core Ok-series processors from October 2018, the corporate launched its 9th Gen F-series {hardware}. F-series components lacked built-in graphics, so we anticipated them to be cheaper, however they have been listed on Intel’s pricing on the similar value because the Ok components that did have built-in graphics (later within the 12 months the value did ultimately change). Along with the F components have been KF components that provided each overclocking performance however no built-in graphics.

What we additionally received to see in January 2019 was the Xeon W-3175X: Intel’s fully-unlocked 28-core beast of a processor. Intel shipped plenty of press a full system to check this chip on, which included a somewhat beefy three-fan liquid cooler, an exhuberant $1500 motherboard, and a 1500W energy provide. All in all, the test-kit for this CPU was $7000+, together with the value of the processor, which was a cool $2999.

The Intel Xeon W-3175X Review: 28 Unlocked Cores, $2999

As maybe was to be anticipated, it knocked the socks off of any of the opposite processors we’ve ever examined, even placing some dual-socket configurations to disgrace with its excessive core rely, excessive frequency, and maybe not surprisingly, excessive energy draw. Only two motherboard distributors even dared to construct boards for it, ASUS and GIGABYTE, with solely ASUS prepared for launch day. EVGA ultimately constructed a board too, when the product stack crammed out a bit, however in the end we have been advised by these corporations that they anticipated the W-3175X to be bought largely by system integrators.

One of the ultimate evaluations to spherical off the month from us was our Cannon Lake overview. Intel had ‘formally’ shipped its first 10nm CPU merchandise in December 2017, however they have been no-where to be seen. Despite repeated requests to search out out the place this legendary CPU had gone, we discovered a couple of in China, with Lenovo producing a product for the training market. We pulled a couple of strings and received this Cannon Lake laptop computer in, and gave it a superb teardown in addition to a full structure evaluation.

Intel’s 10nm Cannon Lake and Core i3-8121U Deep Dive Review

Cannon Lake, Intel’s first era 10nm product, shipped in a twin core design with a disabled GPU (as a result of it didn’t work). It was extra energy hungry than the earlier era 15W laptop computer designs, and the components that ended up out there have been undoubtedly salvaged to make it appear to be it was able to go. Cannon Lake will perpetually stay a curio for CPU fans, and that’s the place it ought to keep.

Right on the finish of the month we had some excellent news. Intel had lastly put the Itanic to mattress. Intel’s Itanium product, initially launched in 2000, deployed the ‘Itanium’ structure for high-scale enterprise servers. The final era of merchandise, known as Kittson, have been launched in 2017 however even then have been solely a frequency bump over the 2012 Poulson merchandise.

The solely firm nonetheless deploying Itanium was HP, and January 2019 marked the beginning of the End-of-Life cycle for this {hardware}. HP (and others) have till January 30th 2020 to make the final orders of CPUs, which will likely be shipped by July 29th 2021. But because it stands, with no roadmap and no extra help, the Itanic has lastly bubbled its final puff of air.

Intel to Discontinue Itanium 9700 Kittson Processors, the Last of the Itaniums

We additionally reviewed the AMD Athlon 200GE towards the Intel Pentium G5400, in a battle of $60 {hardware}.

The $60 CPU Question: AMD Athlon 200GE or Intel Pentium Gold G5400? A Review

 

 

February 2019

No CPU launches to report from February, though discussions about Intel’s potential to supply sufficient shopper CPUs into the market have been slowly rising. Having solely launched 9th Gen Core {hardware} for Ok/F/KF processors and never a full stack had individuals confused as to the place Intel was going with the 9th Gen Core household.

What we did at AnandTech was overview a few particular OEM-only components from AMD. The Ryzen 5 2500X and Ryzen 3 2300X have been technically introduced in September, however attributable to their OEM-only standing (and maybe a little bit of region-specificity occurring), they may solely be obtained in particular pre-built programs from HP, Lenovo, or different huge distributors. We managed to pay money for the chips to match their efficiency.

The AMD Ryzen 5 2500X and Ryzen 3 2300X CPU Review

It did strike us as odd as to why AMD doesn’t broadly distribute them. Potential causes embody that these lower down components are restricted in quantity, or AMD wished to drive up the common promoting value of its processors in the marketplace – a tactic utilized by plenty of corporations. At this level it was unclear, and as we’ll learn additional, AMD did one thing a bit comparable with the Ryzen 3000 household.

March 2019

March was additionally a comparatively quiet month, with nothing occurring on the buyer processor entrance. There have been a number of predictions about AMD’s upcoming 7nm {hardware}, though nothing out of AMD itself.

What AMD did lastly put into retail was a high-speed model of its then current-generation Naples server {hardware} based mostly on its Zen microarchitecture. The AMD EPYC 7371 was a 16-core half that pulled an additional 30 W (200 W vs. 170 W) however enabled a a lot greater base frequency (3.1 GHz vs. 2.Three GHz) and turbo frequency (3.Eight GHz vs. 2.9 GHz) than some other EPYC in the marketplace. Beyond that, AMD said it had a 3.6 GHz all-core turbo.

AMD Launches High-Frequency EPYC 7371 Processor

While the product was introduced in November 2018, it truly was made out there in February 2019. While we didn’t get to overview the product straight, our buddy Patrick Kennedy at ServeTheHouse did do a overview and mentioned that it provided AMD’s best EPYC server product up to now, simply changing twin 8-core programs from Intel (a standard configuration) with a single chip.

 

April 2019

While the leaks and rumors are nonetheless flying for AMD’s Zen 2, April was all about Intel.

Intel’s huge announcement for the month was the launch of its new enterprise processors. The second era of the Xeon Scalable line is known as Cascade Lake, and the important thing updates right here are usually not solely the primary spherical of hardened safety updates for Spectre/Meltdown, however Intel additionally adjusted the core counts at particular value factors to supply 25% extra cores for a similar value. The new processors additionally help Intel’s Optane DCPMM, which permits for as much as 4.5 TB of reminiscence per socket with particular SKUs.

Intel’s Enterprise Extravaganza 2019: Launching Cascade Lake

The socketed Cascade Lake {hardware} matched the earlier Skylake by way of peak core counts (as much as 28), and IPC was nearly equal as nicely, with Intel leveraging will increase in core counts and small frequency lifts to present total frequency enhancements.

We additionally noticed the Xeon 9200 household being launched. This new line, sitting above the Cascade Lake Platinum 8200, basically places two 28-core silicon dies onto a single bundle. These chips aren’t socketed, and require to be soldered on board, however basically act as a dual-socket processor in a single socket (albeit, at 1/Three of the hyperlink pace). Intel made these chips as a way to provide high-density programs (224 cores in a 1U), but in addition to counter the upcoming the 64-core possibility from AMD.

Hands on with the 56-core Xeon Platinum 9200 CPU: Intel’s Biggest CPU Package Ever

Also on the books first was the Xeon D-1600 collection, a community targeted decrease core rely embedded chip for the sting market – a type of child brother to the D-2100 collection.

Intel Launches the Xeon D-1600 Family: Upgrades to Xeon D-1500

For customers, the discharge cycle meant that Intel began placing 45W TDP variations of the 9th Gen Core household into the market, for each desktop and cell. Finding them on desktop for the reason that launch has been somewhat difficult (and seemingly location particular).

Intel ninth Gen Core Processors: All the Desktop and Mobile 45W CPUs Announced

 

May 2019

At the start of May, we began to bid farewell to one in every of Intel’s most bold initiatives. Originally derived from Intel’s try and make an x86 graphics chip known as Larrabee, Xeon Phi had been on the run as a many-core design targeted in high-performance computing, specifically attributable to its 72 AVX-512 items. The final product announcement for Xeon Phi had the 10nm product going into the Aurora supercomputer, however Aurora is now getting Intel’s new XeHPC GPU as an alternative, known as Ponte Vecchio. For Xeon Phi, the final era that was given life was Knights Mill, which was nearly equivalent to Knights Landing aside from an optimization to go to AI.

The Larrabee Chapter Closes: Intel’s Final Xeon Phi Processors Now in EOL

Xeon Phi was an attention-grabbing product that slowly grew to become out of date as Intel launched AVX-512 on its major Xeon Scalable household of processors with devoted AI directions. With Knights Mill now being ushered into its End-of-Life program, the chapter on Larrabee has lastly closed, with the corporate now set to launch a spread of GPUs up and down the stack with its new Xe structure.

A aspect observe on different Intel information: technically the corporate launched its Xeon E-2200 household of CPUs in May as nicely. These are the Xeon analogues to the Core 9th Gen {hardware}, besides with enterprise options like ECC and vPro by default.

Intel Launches Xeon-E 2200 Series for Servers: 8 Cores, as much as 5.zero GHz

 

The finish of the May/starting of June is our annual journey to Computex, and Dr. Lisa Su took to the stage for AMD’s keynote. AMD for the primary time gave everybody particulars concerning the subsequent era of Ryzen: Ryzen 3000 constructed on 7nm chiplets utilizing Zen 2 microarchitecture. In the disclosures, AMD coated its new 8-core merchandise, in addition to its new flagship Ryzen 9 3900X with 12 cores utilizing two chiplets, as had been theorized by the earlier CES disclosure.

Dr. Lisa Su at Computex 2019: AMD Keynote Live Blog

Of course, everybody was asking two issues: (a) what’s new in Zen 2, and (b) the place’s the 16-core ? More on that later within the 12 months.

Intel additionally had its standard keynote for Computex, with Greg Bryant taking the stage. Last 12 months the corporate had an upset the place it confirmed a 28-core 5.zero GHz CPU with out mentioning it was overclocked (and appeared very cagy about it on the time). This 12 months there have been no such hangups, and the corporate introduced the Core i9-9900KS. This 8-core 5.zero GHz CPU truly does 5.zero GHz on all cores out of the field, with out the necessity for unique cooling.


Presented by Jon Carvill, now ex-Intel

Intel Announces 8 Core i9-9900KS: Every Core at 5.zero GHz, All The Time

Intel didn’t state a launch date on the time (it ended up being October), nor a value, with the value anticipated to be excessive. It appeared like apparent competitors for AMD’s 12-core and 16-core, so it did create various visitors.

At Computex, Intel additionally talked about that its subsequent era of X-series processors (Cascade Lake-X) can be coming to the high-end desktop market later that 12 months, as would its 10nm Ice Lake laptop computer processors. This would coincide with the official launch of the Project Athena 1.zero laptop computer specs.

Computex 2019: Intel Keynote with Gregory Bryant

The full SKU checklist for Ice Lake was launched, protecting 9W, 15W, and 28W {hardware}. To date (as of Dec 2019), we nonetheless have solely seen the 15W {hardware} come to market, so we expect to see 28W and 9W components in Q1 2020. The huge uplift for the high-powered components is within the graphics, providing as much as 64 EUs of the brand new Gen11 graphics. We received an opportunity to benchmark an early preview in August.

One different factor we discovered at Computex. We know from the earlier 12 months that AMD had executed a deal to create a three way partnership in China with a Chinese firm known as THATIC. This three way partnership had a license to request modifications to the Zen 1 structure, which AMD would approve, after which they may very well be manufactured by the three way partnership. These ‘Hygon’ processors have been constructed underneath license particularly for the Chinese market, and even right this moment on the finish of 2019 it’s uncommon to even see one, not to mention contact it or take a look at it. At Computex, we discovered one to look at.

Spotted at Computex: Let Bygones be Bygons, with a Sugon Hygon

Interestingly sufficient, we received the underside proper of the CPU translated. A literal translation is 

Using Cores to Calculate The Future

However, the second character used within the slogan is a homonym, which may very well be translated as:

Using Passion to Calculate The Future

For a Chinese CPU, the synergistic phrasing could be very poetic. Sugon, regardless of by no means having interacted with AnandTech earlier than, requested us to take down our brief article, and set us some odd paperwork that had nothing to do with AnandTech. The associate who had the CPU on the sales space not converse to us. That’s how ‘delicate’ a Hygon is. As a part of our Computex protection, we requested AMD about these components, about what’s totally different – we have been advised ‘when you take a look at one and discover one thing out, tell us and we’d affirm it’. Perhaps this can be a good time as any to say that now we have some Hygon for testing, and located some very distinct variations to Zen 1. Stay tuned in early 2020 for our overview.

As a part of our Computex protection, we have been additionally given the chance to take part in a roundtable with AMD’s CEO, Dr. Lisa Su, and quiz her on the exhibits bulletins. The questions on this interview are assorted, because of the nature of a roundtable dialogue not often staying on the identical subject, however Lisa provided some good gems of information.

All Ryzen: Q&A with AMD CEO Dr. Lisa Su

 

June 2019

Technically Computex rolled into June, but in addition at the start of June is E3. Normally AnandTech doesn’t cowl E3, nevertheless AMD determined to have a giant 2-day NDA tech occasion simply earlier than the present began. This is common earlier than a giant launch, and often it’s a battle for the corporate that does the tech occasion to maintain everybody quiet prematurely. Nonetheless, AMD took the time to undergo the brand new Zen 2 microarchitecture in-depth, with key architects readily available to reply questions. So after the primary full day of shows, we solely had 10 hours between the top of the shows and embargo raise. I swore to some $deity that I used to be destined to have it written up in time. Here’s my 7000-word evaluation of AMD’s Zen 2 microarchitecture.

AMD Zen 2 Microarchitecture Analysis: Ryzen 3000 and EPYC Rome

Aside from making the core wider, AMD had elevated the micro-op cache dimension, launched a brand new fashion of department predictor, totally enabled twin AVX2 pipes, improved the load/retailer balancing, and a greater effectivity by utilizing 7nm for the cores however 14nm for the IO die. AMD claimed a direct +15% efficiency uplift from Zen1 to Zen2, which might put it straight within the firing line with Intel’s Skylake (now Coffee Lake) 14nm++ processors for uncooked clock-for-clock efficiency. Promising a launch date of seven/7 (as a result of 7nm), the whole lot was heating up, to the purpose the place individuals have been now beginning to ask about Threadripper.

During the E3 occasion, AMD additionally introduced that alongside the 8-core and 12-core {hardware}, it will be launching a 16-core counterpart, the Ryzen 9 3950X. This didn’t actually shock anybody, aside from its low TDP (solely 105 W, 10 greater than the 12-core), and the very fact it was listed as having the best single-thread frequency of any Ryzen 3000 chip. AMD listed it as ‘coming September’, however it in the end got here out on the finish of November attributable to unprecedented demand of 7nm chiplets.

AMD 16-Core Ryzen 9 3950X: Up to 4.7 GHz, 105W, Coming September

To cap off the Computex/E3 bulletins, Wendell from Level1Techs and I sat in our resort in Los Angeles to speak concerning the bulletins and our opinions of them, for his channel. Video is embedded under.

Fireside Chat with Ian and Wendell: Ryzen 3000, Zen 2, Navi, Xeon W

 

The remainder of the month of June was all for Intel. The firm lastly launched the remainder of its 9th Gen Core processor stack, crammed with non-Ok and non-F processors, in addition to Pentiums and Celerons. There have been no actual surprises in frequencies or properties of those CPUs, nearly mirroring what we noticed on the 8th Generation {hardware} with small frequency changes and equivalent pricing.

We additionally noticed Intel launch the Xeon W-3200 collection on the again of the brand new Mac Pro launch. The W-3200 household makes use of Intel’s LGA-3647 socket, and turns into a less expensive strategy to allow a 28-core workstation with out happening the Xeon Scalable route. Intel additionally began to supply the Xeon W household with M variations for middle-tier reminiscence help, permitting the utmost quantity of reminiscence to be supported go up from 1 TB to 2 TB (for an additional $3000).

Intel Cascade Lake Xeon W-3200 Launched: Server Socket, 64 PCIe 3.zero lanes


Related Tech News:

Shares