Processors - Page 98

Intel's new Core i9 range is getting flashed around, with the upcoming 10C/20T part hitting 5GHz with an AIO liquid cooler. Intel's upcoming Core i9-7900X has reached 5GHz with the AIO liquid cooler, and even 4.5GHz on air, all on 1.15V - and with all CPU cores enabled.

The new Core i9-7900X will arrive with 10C/20T of CPU power, 13.75MB L3 cache, and 3.3/4.3GHz for base/boost, respectively. But Intel's new Turbo Boost 3.0 technology will overclock the 10C/20T chip to 4.5GHz, which is a huge upgrade over the current high core count CPUs like the Core i7-6950X. Intel's upcoming Core i9-7900X will have a TDP of 140W, and its X299 chipset will feature 44 PCIe 3.0 lanes.

Intel will be pricing the new 10C/20T Core i9-7900X at $999, and considering the Core i7-6950X debuted at $1500 and is still $1500 on Amazon right now - all with the same 10C/20T, but at much lower clocks (3.0/3.5GHz for base/boost), it's a steal. AMD will be competing big time with their upcoming Ryzen ThreadRipper CPUs offering up to 16C/32T, but their spiffy new X399 chipset rocks 64 PCIe 3.0 lanes, beating Intel out by a fair chunk.

Intel is reportedly launching its new Core i9 range of processors next month, starting with the Core i9-7920X (12C/24T) but the 14C/28T, 16C/32T, and 18C/36T processors will be launching in October.

The first phase of Intel's new Core launch will see the 4/6/8/10-core models launch, while the second phase will see the 12C/24T chip launch, and then finally the mudslide of 14/16/18-core parts. Intel will reportedly begin accepting pre-orders of the new 4/6/8/10-core Kaby Lake-X and Skylake-X processors on June 19, shipping to consumers on June 26. All of these processors will work on the new X299 chipset.

Intel's upcoming 12C/24T processor in the Core i9-7920X will launch in August for a hefty $1199, while the super high-end Core i9-7980XE with 18C/36T of power will launch in October.

Intel has just made its upcoming Cannon Lake CPU architecture official, with its 9th-gen Core processor on track and made on the 10nm process.

Intel also took the time to announce that its second micro-architecture built on 10nm is taped out, with details on Ice Lake to follow. Intel is reacting in a scared fashion as I said months and months ago now, throwing out its 18C/36T processor at Computex 2017 without any details, a gimped X299 chipset compared to the ThreadRipper ready X399 chipset from AMD which has 64 PCIe 3.0 lanes compared to just 44 PCIe 3.0 lanes on X299.

The new Cannon Lake CPUs will be a die shrink of the Coffee Lake chips, so don't expect huge things from it, apart from lowered TDP and possibly higher clocks. Ice Lake will be the bigger one, something like the upgrade from Broadwell to Skylake, and is expected in 2H 2018.

While the consumer CPU world is tangled up in AMD's upcoming ThreadRipper that comes with up to 16C/32T, Intel is wrapping up their next-gen Xeon Platinum line with the flagship Xeon Platinum 8176 processor that rocks 28C/56T of processing power.

Better yet, Intel's upcoming Xeon Platinum 8176 can be used in dual CPU socket configurations, and with each CPU featuring 28C/56T, you'd be looking at 56C/112T of CPU performance. Intel's new Xeon Platinum 8176 comes with a stock 2.1GHz CPU clock, while a single core can ramp up to 3.8GHz, the all cores enabled Turbo clock hits 2.8GHz. It's a beast.

But what good are the specs of this 28C/56T processor without benchmarks? Intel's new Xeon Platinum 8176 scored 6525 points in Cinebench R15 (which supports more than 16 threads), blasting past the 44C/88T dual-CPU socket Intel Xeon E5-2696 v4 processor... amazing stuff.

It was rumored quite a while ago now, but it's back: the news of a future Intel processor with AMD graphics has been spotted.

The chip was on Sisoft as "Intel(R) HD Graphics Gen9; 694C:C0 (1720SP 47C 1GHz, 528kB L2, 10.4GB 800MHz)" but if you'll note the hardware ID "644C:C0" which is reportedly the hardware ID code for AMD. There is absolutely nothing known about this processor that is concrete, so treat this as a hot rumor - but the CPU is a 2T processor, presumably 2C/2T APU with AMD Radeon graphics. The rumors of the Intel/AMD GPU licensing deal could be true if this CPU is the real deal.

The first engineering sample of AMD's upcoming Ryzen ThreadRipper is reportedly here, with the Ryzen ThreadRipper 1920 processor rocking 12C/24T of CPU performance.

AMD's first Ryzen ThreadRipper 1920 chip will have its 12C/24T of power at 3.2GHz base clock, but we don't know what the boost clock will be under XKR. The higher-end 16C/32T processor without a name just yet, rumored as the ThreadRipper 1998X will have 3.4GHz base CPU clock, while under XFR it will hit 3.7GHz and hopefully, beyond.

There's also another 16C/32T chip at 3.1GHz base and 3.6GHz boost, another engineering sample. There's also the Ryzen 3 1200 processor, with its 4C/4T power at 3.1/3.6GHz for base/boost respectively.

Computex 2017 - GIGABYTE hosted their X299 OC gathering at Computex 2017 last week, bringing in a bunch of the world's leading overclocking enthusiasts and hit a new world record with the Intel Core i7-7740K: hitting 7.5GHz.

Legendary overclocker HiCookie reached 7.5GHz using liquid helium at -250C, pushing the Core i7-7740K to 7500MHz on a multiplier of 75, and a bus speed of 100MHz on the GIGABYTE X299-SOC Champion motherboard. They used a 16GB kit of Corsair Vengeance LPX DDR4-4333 RAM, and a Corsair AX1500i PSU.

Two new Exynos SoCs have leaked online suggesting that Samsung could be introducing new SoCs for mid-range devices.

The first processor is the Exynos 9610 which is a step below the company's current leading SoC, the Exynos 8895. The Exynos 9610 is expected to use the 14nm process similar to the one used by the Exynos 8890 inside of the Galaxy S7.

The 9610 doesn't use a combination of Samsung's Mongoose cores with A53 cores, but rather off-the-shelf A73 cores instead. The interesting development is the inclusion of ARM's new G72 GPU which was announced at Computex and should be a pretty sizable upgrade over the G71 in the previous generation. It also has Cat. 13 LTE which isn't quite as fast as what's in the Exynos 8895.

We're living in a world where our new graphics cards are made on the impressive 14nm and 16nm nodes, and we have companies like Qualcomm and Samsung making the latest and greatest SoCs on the small 10nm node - but IBM has trumped them all.

IBM has recently began testing their new 5nm process, with a huge 30 billion transistors - 30,000,000,000 transistors on something as small as your finger nail, just let that sink in. A couple of years ago the team hit 7nm and had 20 billion transistors which was impressive back then, but now we're at 5nm and 30 billion transistors. Amazing stuff.

The team at IBM hopes to see the impressive new 5nm chip helping their own cognitive computing efforts, the Internet of Things (IoT), and other "data intensive" tasks. The new chip could also help smartphones, with "two to three times" more battery life than we have now. We shouldn't expect that, but maybe a blend of improved battery life and speed - all balanced perfectly on 5nm. We shouldn't expect 7nm any time before 2018 at the earliest, so 5nm is still quite a while away.

Intel might have unveiled their new Core i9 range of processors at Computex 2017, led by the flagship Core i9-7980XE processor and its 18C/36T of power. Except, that monster will not see the light of day until 2018 according to an ASUS representative.

"Raja@ASUS" on the ASUS ROG forums replied to someone asking about the benefits of Intel's new 18-core CPU, to which he responded with: "The 18-core CPUs are not scheduled until later this year. Won't have them for a while. Either way, unless you're using the rig for rendering or encoding to make a living, no need".

I don't think we'll see the Core i9-7980XE until early 2018, which will let AMD have plenty of time to get ThreadRipper into the hands of consumers with the full 16C/32T of power and a full 64 PCIe 3.0 lanes compared to the 44 PCIe 3.0 lanes on X299 from Intel.

Computex 2017 - AMD had its big reveal of its upcoming Ryzen ThreadRipper CPUs this week at Computex 2017, but now we're hearing that the new 16C/32T variant will cost just $849... and if that's true, Intel is in for a world of hurt.

Intel's upcoming flagship Core i9-7980XE will cost $1999 and offer 18C/36T, slightly more than AMD's flagship Ryzen ThreadRipper and 16C/32T of CPU grunt. But if the $849 price is right, AMD will be charging less than half the cost, at $1150 cheaper. For that cost, you could get the motherboard and a good kit of DDR4 RAM and still have change left over.

Intel's upcoming 16C/32T chip is $1699, which is still massively more expensive than the rumored $849 price on the Ryzen ThreadRipper 16C/32T processor. If we are going to be in a world where AMD is charging $849 for a processor with similar performance to Intel's CPU that costs $2000... well... we're going to be in for a wild ride.

Computex 2017 - AMD unleashed their new ThreadRipper CPUs at Computex, offering a full stack of multi-threaded processors with up to 16C/32T, and 64 PCIe 3.0 lanes through the new X399 chipset - but, Ryzen 7 pricing is now dropping big time.

AMD's current Ryzen 7 1800X, Ryzen 7 1700X, and Ryzen 7 1700 have all had price drops - and are in effect on Amazon and Newegg. The AMD Ryzen 7 1800X was $49 and can now be found for $459, while the Ryzen 7 1700X was $399 and is now just $349, while the Ryzen 7 1700 drops from $329 to just $299.

I thought it would be Intel that would drop the prices on their Core range of processors, but this puts AMD in a great position ahead of the big launch for ThreadRipper.

Computex 2017 - AMD unleashed their new HEDT platform at Computex 2017, detailing ThreadRipper and its new X399 chipset. Intel fought back with its new Core X-Series announcement and the new $1999 flagship CPU with Core i9-7980XE.

AMD might not have a consumer CPU with 18C/36T, but it will be offering 16C/32T with their flagship ThreadRipper 1998X processor. Interestingly, AMD has nailed the TDP on ThreadRipper where they're offering far superior TDP numbers on their processors. For example, Intel's new 6C/12T right up to the 10C/20T have 140W TDPs, while AMD sips 125W. It's only 15W, but traditionally it has been Intel that has superior TDP numbers.

There will be 9 different Ryzen ThreadRipper CPUs in total, with the 'X' models featuring XFR (Extended Frequency Range), which allows the ThreadRipper CPUs to scale their clock speeds past the traditional boost clocks, and it scales the CPU clocks with cooling. So if you have air cooling, you'll hit a certain frequency, but AIO water cooling and LN2 setups will benefit more with XFR.

Computex 2017 - AMD might not have Radeon RX Vega anywhere near ready for consumer and gamers' consumption, but that didn't stop them from unleashing their new ThreadRipper CPU at Computex.

AMD detailed their new ThreadRipper in Taipei, announcing it has a whopping 16C/32T of CPU performance, 64 PCIe 3.0 lanes (more on why this is the most important part in a minute), quad-channel DDR4 support, and their new X399 chipset for HEDT.

The new X399 chipset supports 64 PCIe 3.0 lanes which is perfect for 2/3/4-way GPU setups (4-way Radeon RX Vega anyone?!) and the world of M.2 SSDs that use those precious PCIe lanes. Not only that, but we have quad-channel DDR4 support with up to 8 x DIMMs on high-end motherboards.

Computex 2017 - Intel has just launched its new HEDT offensive with the introduction of the new Core i9-7980XE processor offering 18C/36T of CPU performance for a whopping $2000.

Intel's upcoming Core i9-7980XE will have 18C/36T of CPU performance at a yet unknown frequency, but their new 10C/20T chip in the Core i9-7900X sees a base clock of 3.3GHz, Turbo Boost 2.0 clock of 4.3GHz, and scales right up to 4.5GHz under Turbo Boost Max 3.0 tech.

The new Core i9 Extreme processor will feature support for Intel's AVX-512 instruction set, Turbo Boost Max 3.0, 44 PCIe 3.0 lanes, quad-channel DDR4 support at up to DDR4-2666, rebalanced Intel Smart Cache hierarchy, and support for Intel's Optane memory.

At Computex 2017, ARM announced two new CPUs and a new GPU to update their product line for devices coming in 2018. ARM always announces their newest processors ahead of time because their customers, the chip makers, need to build the chips with those new processors.

The two new CPUs that ARM has announced are the Cortex-A75 and Cortex-A55 which are supposed to replace the Cortex-A73 and A53 in smartphones today. ARM says that their 'big' processors like the A75 will get refreshes once every year while the smaller or 'LITTLE' CPUs will get refreshes every 2-3 years, like we are seeing now with the A55.

The interesting part about these new processors other than the obvious performance and power improvements is that ARM's CPU cores can now be configured in a single cluster in odd numbers like 1 + 7 or 2 + 5 which wasn't possible before. ARM's new DynamIQ technology helps to make these new cluster combinations possible. These combinations can also be made within a single cluster of cores rather than multiple clusters as long as they are under 8 cores total. More clusters are possible, but most people will rarely see over 8 in a smartphone.

When we first reported that AMD would be calling its now current-gen processors Ryzen, the name just didn't stick at first - but here we are, with a slew of Ryzen processors that are starting to make Intel squirm.

But according to a new trademark filing, it looks like AMD is working on new products called Kyzen, Aragon, Pharos, Promethean, Zenso, and CoreAmp. All of these product names were trademarked in March, with a few of them already released. What can we expect from these new, interestingly named AMD products?

Well, Kyzen sounds just like Ryzen - so maybe this is a new CPU from AMD, or maybe even a new APU. I have a 'cryo' feel from Kyzen, so maybe we're looking at a heavily overclockable Ryzen processor that will be released as Kyzen? Who knows. Next up we have Aragon - very Lord of the Rings. Promethean will most likely end up as a chipset, as AMD named their Ryzen chipset Promontory - with Promethean being very close to that.

Intel is preparing its 10C/20T processor to directly compete against AMD's Ryzen 7 1800X and their upcoming ThreadRipper CPUs that will feature 16C/32T of CPU power, but Intel could have it in the bag with its Core i9-7900X.

According to new benchmarks, the Core i9-7900X will be clocked at 3.3/4.3GHz for base/boost, respectively. This is a huge deal for Skylake-X, as the IPC performance at 4.3GHz and 10C/20T of power should pummel every other CPU into dust. Better yet, SiSoft's benchmark picked up the CPU as 4.0GHz base clock and 4.5GHz boost, which would be absolutely insane - especially for a 10-core processor.

We should expect 13.75MB of L3 cache, 1MB L2 cache per core (10MB total), and a 175W TDP for the upcoming Core i9-7900X. We should expect Intel to unleash its new Skylake-X parts at Computex next week, alongside the new X299 chipset and a slew of new motherboards.

Intel has been on the CPU offensive since AMD launched their new Ryzen platform, and now we're hearing about their next-gen Xeon Scalable Platform, otherwise known as Cascade Lake.

Cascade Lake was shown off during the SAP Sapphire conference, with Intel teasing that its next-gen platform will arrive in 2018. Cascade Lake-SP is the Skylake-SP refresh, with Intel tapping the 14nm+ node. Cascade Lake will support up to a mammoth 6TB of Optane DIMMs, which is a huge plus for Intel.

Intel has teased: "Intel persistent memory will allow users to improve system performance dramatically by putting more data closer to the processor on nonvolatile media, and do it in an affordable manner. This will truly be a game-changer when it comes to the way applications and systems are designed".

Google has just unveiled its second-generation tensor processor, something that packs 45 TFLOPs of performance per chip, with four of them placed onto a tensor processor unit (TPU) module for a total of 180 TFLOPs.

The massively powerful systems are built for machine learning and artificial intelligence, and Google is pushing it into the cloud with their TPU-based computational powerhouse systems to be made available to Google Cloud Compute later this year. Google's first-gen Tensor processors were already 15-30x more powerful, and a huge 30-80x more power efficient than CPUs and GPUs for these types of workloads.

These new TPUs are "optimized for both workloads, allowing the same chips to be used for both training and making inferences. Each card has its own high-speed interconnects, and 64 of the cards can be linked into what Google calls a pod, with 11.5 petaflops total; one petaflops is 1015 floating point operations per second", reports Ars Technica.