Processors - Page 90

Things just got very, very interesting in the CPU space as Chinese manufacturer Hygon has just announced it is starting the production of Chinese-designed "Dhyana" x86-based processors that are reportedly based on AMD's Zen CPU microarchitecture.

How did this happen? According to TH, there is an x86 IP licensing deal struck between AMD and its Chinese partners. AMD has said it doesn't sell final chip designs to Chinese based companies, but the IP licensing deal allows them to sell their own processors made specifically for the server market in China.

These new Dhyana CPUs are as close as they can be to AMD's own EPYC server processors, with the only difference between the CPUs being their unique vendor IDs and family series numbers. Mic drop moment from AMD against Intel. But what about US laws against this type of thing, as we all know Chinese companies rip designs and IP all the time.

Intel is expected to launch their 8th-gen Whiskey Lake-U processors, with the new range of mobile CPUs to succeed the current-gen Kaby Lake-R processors. Intel will continue using its 14nm++ process, driving CPU clock speeds up and keeping the power consumption low.

Whiskey Lake-U will include the flagship Core i7-8565U that will succeed the Core i7-8550U, while the mid-range Core i5-8265U succeeds the Core i5-8250U. Both of these processors will be made on the 14nm++ note, with Intel pushing CPU clock speeds up through the optimized process design... but other than that, there's nothing else really new here.

What we will be getting from Intel with the 4C/8T chip in the new Core i7-8565U is an increase of 200MHz base CPU clock and 100MHz increase in the Turbo Boost frequency. As for the Core i5-8265U which is also a 4C/8T part, with a much larger increase in its Boost clock of 500MHz, while the base clock stays the same as the previous-gen Core i5-8250U.

AMD is set to have one of the most exciting quad-core CPUs of the last generation or so with its upcoming Ryzen 3 2300X processor, which has found its way into the hands of XFastest, who also grabbed the Ryzen 5 2500X that is around the corner.

The upcoming Ryzen 3 2300X is a 4C/4T part with no multi-threading support, 3.5GHz base CPU clock and 4.0GHz turbo CPU clocks. This is all made on the 12nm node with precision boost overdrive, and XFR enhanced. XFastest tested the Ryzen 3 2300X and its 4C/4T of budget CPU power under LN2 cooling, where the team reached a peak of 5663MHz (5.6GHz) but it wasn't stable for benchmark runs, so they dropped down to 5585MHz (5.5GHz) and it was fine.

Cinebench results for the overclocked Ryzen 3 2300X reached 895 points in the multi-core test, and 233 points in the single-core tests. CPU-z benchmarks reveal the 2300X scored 642.3 points in single-thread, and 2553.2 in multi-threaded benchmarks. If we compare this against the stock results of 690 points in Cinebench against 895 points when overclocked, we can see the Ryzen 3 2300X is knocking down some performance doors.

Intel is preparing its strange refresh/emergency release of their 9000 series based on the Coffee Lake-S architecture, with a new mid-range Core i5-9600K planned and coming soon.

The new Core i5-9600K is a 6C/6T processor with no Hyper-Threading and a slight CPU clock speed bump to usher in the new CPU. Intel will be offering up 100MHz in base clock and 200MHz in Turbo clocks over the Core i5-8600K, all staying within the same 95W TDP. The new SKUs were found in the Microcode Update and June 2018 8th Gen Core Family update, which means they're close, but not here just yet.

Intel's big selling point with the 9000 series CPUs will be the official introduction of the first 8-core CPU parts from Intel in the form of the upcoming Core i9-9900K which is an 8C/16T beast that will take on the AMD Ryzen 7 2700X processor. It's going to be a confusing release, compounded by the fact there'll be a "new" Z390 chipset launched at the same time.

AMD might not be securing all of the CPU market share right away with Ryzen, but it's causing enough of a stir to pull thousands of people away from the clutches of Intel, who is in a world of trouble as it is right now with its CEO resigning recently, and their 10nm process completely broken.

The company is reportedly preparing a mid-range Ryzen 3 2300X processor which is a 4C/4T part, with its CPU clocks bumped up to 3.5/4.0GHz thanks to no multi-threaded support. AMD's use of the 12nm process also helps, with tweaks to precision boost overdrive and XFR enhanced.

AMD's upcoming Ryzen 3 2300X should sell for between $120-$140 which will make it some truly insane value when you compare it to the Intel Core i5-7600K which costs over $220. But what about the performance of the Ryzen 3 2300X? This is what we need to see.

If there's a CPU that I'm excited for this year, it would have to be AMD's impending megatasking crusher, the Ryzen Threadripper 2990X and its monstrous 32C/64T of CPU power.

But what about the price? AMD seemed very coy during Computex 2018 in public and behind closed doors about the price of the new Ryzen Threadripper 2990X, but weren't shy from saying it would be more expensive than the current Ryzen Threadripper 1950X. Now there's a European retailer that has listed it early for 1509 EUR, which works out to around $1750 USD.

This is most likely a placeholder price, and I expect it to be more expensive than that - closer to, or over $2000. Hell, AMD could charge $3000 for the Ryzen Threadripper 2990X and it would still be an amazing price. Intel doesn't have anything close to it apart from the Xeon Platinum 8180 Processor which is a 28C/56T part that costs over $10,000.

Intel's unreleased and still not named 8C/16T processor is ri ghit around the corner, a new CPU that will be unleashed next to the Z390 chipset and now we might have the new name: the Core i9-9900K. Yeah, Core i9.

The latest rumor is directly from the salt shaker itself, with WCCFTech saying it will be the Core i9-9900K with 8C/16T. Intel will also reportedly launch the Core i7-9700K as a new 6C/12T processor, and the Core i5-9600K as a 6C/6T processor with HT obviously disabled.

I find lumping the new 8C/16T chip into the Core i9 family as extreme, and it will surely cause some marketing issues, but on the other hand this is a new sector of the market for Intel to enter. Intel doesn't have an 8C/16T processor in the desktop CPU market, a slice of the market that AMD has absolutely dominated in since the release of the Ryzen 7 1800X over a year ago now.

Intel CEO Brian Krzanich has made a shock exit from the company, resigning effective immediately after it was exposed he was involved in a "past consensual" relationship with an Intel employee.

Krzanich is being replaced with Intel CFO Robert Swan, after Krzanich took position as CEO back in May 2013. He was tasked with shifting Intel from a PC-centric company to a silicon/data centered company, something he achieved in short time. Krzanich was also previously the COO and executive vice president of Intel after he joined the company in 1982.

Intel Chairman Andy Bryant said in an official statement: "The Board believes strongly in Intel's strategy and we are confident in Bob Swan's ability to lead the company as we conduct a robust search for our next CEO. Bob has been instrumental to the development and execution of Intel's strategy, and we know the company will continue to smoothly execute. We appreciate Brian's many contributions to Intel".

Intel is acquiring the freaking Justice League of industry heroes for its new GPU division, lead by GPU architect and legend Raja Koduri.

Ex-RTG Marketing boss in Chris Hook, Athlon and Ryzen architect Jim Keller (who is also ex-Tesla) and now ex-Intel and Larrabee CPU architecture architect Tom Forsyth has announced he's re-joining Team Blue to work with Raja on Intel's new GPU architecture.

Intel's refreshed GPU division has some serious talent and brainpower behind it now, with Forsyth's recent announcement taking place on Twitter where he said: "I start at Intel shortly as a chip architect in Raja Koduri's group. Not entirely sure what I'll be working on just yet". Gives me shivers, the excitement for Intel's mysterious GPU division is growing, and I love it.

Earlier today news dropped about AMD's upcoming 2nd Gen Threadripper processors, and things are looking good for AMD. If these leaks are true then the top SKU will be the 32-core 2990X followed by a 24-core 2970X and a 16-core 2950X. According to the leak, the 32-core 64-thread CPU will have a base clock of 3GHz with a boost up to 4GHz and XFR allowing up to 4.2GHz with a 250W TDP. We see 64MB of L3 cache as well as 16MB of L2.

The leak also mentioned that an unknown motherboard vendor said that AMD originally planned to launch the 16-core and 24-core variants, but is now considering nixing the 2950X.

The most tantalizing thing we see is the 6399 CINEBENCH score with the CPU overclocked. Supposedly overclocking the CPU was easy enough, but the fact that the 4.2GHz score is 300 points lower than the 4.1GHz score makes us think that the CPU was perhaps throttling under the H150i Pro they used. The overclock was achieved at 1.38v and resulted in a 4.12GHz OC on all cores.

AMD is a master troll, and this time the company is targeting Intel's super-exciting and 40th anniversary CPU, the Core i7-8086K.

Intel is giving away 8086 limited edition processors as part of the celebration of the 8086 processor from 1978, and AMD has now swooped in offering to swap Core i7-8086Ks that people win in the sweepstakes (there are 40 being given away) with its flagship Ryzen Threadripper 1950X processor.

There is a big win here for anyone who swaps the 8086K, as it is a 6C/12T processor that retails for around $469, while the Threadripper 1950X retails for $946, nearly $500 more.

Intel unveiled its Core i7-8086K processor during Computex 2018 a couple of weeks back now, but now Silicon Lottery is offering binned 8086K processors for serious overclockers.

The company is offering the $430 processor which runs at 5GHz out of the box in a special binned variant that rocks up to 5.3GHz, but costs double the price. The 5.3GHz guarantee with 6C/12T of CPU power will cost you $859.99.

There are four different variants on offer: stock 8086K at 5GHz but delidded, then the final three are all delidded but offering 5.1GHz for $499.99, 5.2GHz for $589.99, and finally, 5.3GHz guaranteed for $859.99.

AMD has nailed its Zen CPU architecture that has not only been keeping gamers and megataskers happy, but the company has extended it into servers where Intel dominates with Xeon.

It was only recently that AMD reached the impressive 7nm node, all before Intel is stable on 10nm (which it has been trying for years now) and now EPYC is about to make the big leap to 7nm with EPYC on 7nm silicon being tested in AMD labs right now, samples headed out in 2H 2018, and its big launch in 2019 right around the corner.

AMD's first-gen EPYC processor family was named Naples, while the second-gen EPYC is called Rome, and the third-gen EPYC will be called Milan. EPYC's names come from famous locations in Italy, with Milan arriving closer to 2020 on the improved 7nm+ node with the revamped Zen 3 architecture.

AMD has been slaying it in all areas of the CPU market with its Ryzen family of processors, but the company is working on some new budget/mid-range Ryzen CPUs that will impress.

The new second-generation Ryzen 3 2300X and Ryzen 5 2500X are reportedly on their way, both being quad-core CPUs with boost clocks of up to 4GHz, but it's their price that will impress the most.

We should expect the Ryzen 3 2300X to arrive as a 4C/4T processor with 3.5/4.0GHz base and boost clocks, and 10MB of cache, 65W TDP and $129-$139 price. The mid-range Ryzen 5 2500X would be a 4C/8T chip with 3.6/4.0GHz base and boost clocks, 18MB of cache, 65W TDP and a mouth watering price of just $149-$159.

Intel PR needs some work post-Computex after its ZOMG 28C/56T at 5GHz fiasco, but we can expect a more realistic 22C/44T processor later this year that will launch next to the Coffee Lake-S 8-core CPUs on the Z390 chipset.

The new 22C/44T processor will fall into Intel's Skylake-X family, where we can expect CPU clocks of somewhere between 2.6GHz base and 4GHz boost, with 24.75MB of L3 cache and 44 PCIe 3.0 lanes. It'll arrive on the same LGA 2066 socket and X399 chipset, which means you won't need to upgrade your motherboard to the beasty LGA 3647 that the 28C/56T processor will need on the Cascade Lake-X platform.

We should see a huge boost in TDP where the 22C/44T chip will require between 165-200W in stock form, and I'm sure much more power when it's pushed to its limits through overclocking. As for its release, we should expect sometime in Q3 2018 with a launch price of somewhere in the $2000 range.

Computex 2018 - Intel unveiled their knee jerk reaction to the news of AMD's impending Threadripper 2 announcement, which until AMD's own Computex 2018 press conference the industry thought it was going to be 24C/48T only.

But it seems Intel's announcement was beyond knee jerk and into the ridiculous, as it is now being reported that the company "forgot" to mention that their "new" 28C/56T processor was overclocked to 5GHz and most tech press thought it was a monstrous 5GHz chip... except it's not.

Our good friend Paul Alcorn from Tom's Harware is reporting that he met with Intel reps yesterday that said that "in the excitement of the moment" they "forgot" to say that the 28C/56T processor was overclocked. Intel PR seems to have all failed here, as the impression most people got was that this was a 5GHz behemoth that would crush Threadripper 2.

Computex 2018 - Qualcomm has announced its new Snapdragon 850 processor, a new chip that was specifically designed for Windows 10 Always Connected PCs, and has some nifty upgrades and improvements under the hood.

The new Snapdragon 850 will be exclusive to Windows 10 machine and will not be used in smartphones or tablets, even though it is a customized Snapdragon 845. We have a 30% performance improvement, 20% improved battery life, and 20% faster peak performance over the previous-gen Snapdragon 835, which is great.

It's built on the second-gen 10nm process, the same node that the Snapdragon 845 was made on, with the same X20 modem for LTE connectivity. Qualcomm has performed both hardware and software tuning to the Snapdragon 850 so that it will perform PC tasks better than the current Snapdragon 845.

Computex 2018 - AMD has just announced its new second-gen Threadripper CPU, and it is truly a monster that just so happens to be right around the corner.

The new Threadripper arrives on the same socket as the original Threadripper, but doubles the core count to 32C/64T for some extreme megatasking. AMD is using the new 12nm node with the Zen+ architecture on the second-gen Threadripper, which beats out Intel's just-announced 28C/56T processor.

AMD's new Threadripper will be coming in 24C/48T and 32C/64T with the company showing off the new CPUs on air cooling here at Computex, which is very impressive. The new Threadripper CPUs will be launching in Q3 2018, so we don't have long to wait at all.

Computex 2018 - Intel officially teased their new beast today at Computex 2018, with their new Cascade Lake-X HEDT processor coming 'at the end of this year'.

Intel's new beast is a huge 28C/56T processor that will compete with Threadripper 2 later this year from AMD, and ASUS was showing it off smashing through some amazing Cinebench results. During the Intel keynote, a GIGABYTE motherboard was used to run it at 5GHz. Intel said that this particular 28C/56T processor will be running at 5GHz, which is absolutely incredible.

Threadripper can't get close to 5GHz right now, so if Intel can get to the market with 28C/56T at 5GHz then AMD is going to be in for a bad time before Threadripper 2 is even unveiled. Wow, Intel... wow.

Computex 2018 - AMD is pushing for a strong Computex this year, with an exciting unveiling of Vega 20 on 7nm with 32GB of HBM2, but also their new project: Fenghuang 15FF, a new graphics chip for APUs,

Fenghuang 15FF is the internal codename for the graphics chip, with the current codename of DG02SRTBP4MFA applied. This chip should be part of the Zen+ APU family with 4C/8T of CPU power, 3GHz clock speed, and 2GB of HBM2 clocked at 1200MHz (2.4GHz effective).

AMD's new chip sports 28 CUs and 1792 SPs (up from 24 CUs and 1536 SPs on Vega 24) while there's 2GB of HBM2 at 2.4Gbps offering 307.2GB/sec memory bandwidth, which is up from the 1.6Gbps and 204.8GB/sec offered on Vega 24.