Cerebras Systems just unveiled the largest chip ever at Hot Chips 31, which is a A Symposium on High Performance Chips hosted yearly with the biggest companies in the tech space involved.
The new Cerebras Wafer Scale Engine chip from Cerebras has a huge 1.2 trillion (1,200,000,000,000) transistors compared to the largest GPU in NVIDIA's GV100 and its 21.1 billion (21,100,000,000). NVIDIA requires just 815 square millimeters of die space for its GV100 and 21.1 billion transistors, while Cerebras Systems requires a huge 46,225 square millimeters of silicon.
Cerebras' new chip is so big that in comparison photos, the company compares its new 1.2T transistor chip side-by-side with a freaking keyboard. Yes, this new chip is as big as a desktop keyboard.
AMD released its game-changing EPYC Rome CPUs and in the hours afterwards, AMD shares spiked by a huge 16% in between Google and Twitter announcing they're using the new EPYC Rome chips in their servers.
AMD shares closed at $33.92 on Thursday, up by a huge 84% this year alone after riding the success of Zen 2 powering the new Ryzen 3000 series CPUs, and the new Navi-powered Radeon RX 5700 series graphics cards.
Patrick Moorhead, Founder, President and Principal Analyst of Moors Insights & Strategy had some comments on AMD's new EPYC Rome CPUs, where he said: "AMD took a big step forward today in the datacenter with its launch of the 2nd Gen EPYC processor and platform. It is a bigger leap forward than I had expected".
He continued, saying that AMD will make great headway with the second-gen EPYC offerings, adding: "AMD gained low, single-digit share with 1st Gen EPYC but I expect the company to gain more share with 2nd Gen EPYC with CSPs, enterprises, and HPC. Enterprises don't mass deploy any first gen product, they didn't deploy 1st Gen EPYC, but they will deploy the 2nd Gen EPYC".
AMD has had a busy year so far, launching its new Radeon VII on 7nm with 16GB of HBM2, and then came the Ryzen 3000 series launch powered by the exciting new Zen 2 architecture also on 7nm.
Just a couple of days ago AMD continued its winning streak with the release of its next-gen EPYC Rome CPUs, with up to 64C/128T of power that demolish Intel's current, and most likely next-gen offerings. The Zen 2 architecture powers the new EPYC processors, and now that it's here it's time for the Zen 3 hype train to begin.
AMD will be doing some exciting things with the Zen 3 architecture, something that will include the shift to 7nm EUV process which paves the way for even denser transistor counts, and could deliver even better power efficiency and higher CPU clock speeds. Better yet, Zen 3 could rock new ISA instruction sets.
On the eve of AMD's launch of its EPYC Rome CPUs, competitor Intel has announced that its future Intel Xeon Cooper Lake family of CPUs will have server versions with up to 56C/112T of power.
Intel detailed the future of its Xeon Scalable platform at its Data-Centric Innovation Summit that it's bringing its 56C/112T Xeon Platinum 9200 family to Cooper Lake. The reason? For the new bfloat16 instructions mixed with the high 28-core count CPU as well as the new Cooper Lake CPUs being socketed versus BGA only.
This means the new Xeon Platinum 9200 family on Cooper Lake is on the LGA4189 socket -- the same as the upcoming Ice Lake CPU coming after Cooper Lake. The new Xeon Platinum 9200 chips will have 8-channel DDR4 support and TDPs of up to
Intel embarrassingly bragged (maybe they were sniffing too much glue) that AMD's first Zen-based EPYC processor was "glued-together" in its own internal presentations back in 2017, but fast forward to its new CPUs and the 56C/112T Cooper Lake CPU has two dies on the same chip.
Intel has just announced a slew of new 10th-gen CPUs and will pack the most powerful integrated graphics in any thin-and-light notebook PC, which is a bold claim considering the power that Qualcomm throws around with Snapdragon 8cx.
We have 11 processors in total with most of the spread of CPUs packing 4C/8T, with TDPs of 9W, 15W, and 28W. We're looking at CPU clocks between 700MHz and 2.3GHz, while 4.1GHz is the peak single-core turbo frequency. Intel will be including an improved GPU with either 32, 48, and 64 EUs with up to 1GHz GPU clocks, as well as Wi-Fi 6 (which is great to see), up to 4 Thunderbolt 3 ports.
There will still be the same Core i3, Core i5, and Core i7 families of processors, so we'll be seeing the Coppre i3-1000G1 through to the Core i7-1068G7. Each part of the name in the new 10th-gen CPUs has something behind it, with Intel breaking it down:
We all know that since AMD had launched the Ryzen 3000 series, there would be non-X model CPUs coming eventually. If the recent posting by EEC is to be believed, we may be seeing many new models showing up very soon.
EEC lists several models which are ranging from the top end to the more midrange, along with several Ryzen PRO models which will be added to the stack. Now, this is a translation of the EEC listing, but being the way the data is listed and the information available, it lends a bit of credence to the listing being an excellent lead to what we may see launching soon.
AMD's new Ryzen 3000 series has released with significant fanfare and some excellent performance. However, like any considerable platform launch, it also was shaken with several issues from improper throttling to incorrect reading and even select games not wanting to start at all. The latter is the case with Destiny 2 which players were understandably upset and lambasted AMD for a fix as soon as the issue was discovered.
AMD's Robert Hallock has been a beacon of knowledge for AMD users on social, especially Reddit, where this very story broke.
Intel has released the latest drivers for its integrated graphics, and inside there is a nice surprise: a tease of its upcoming Xe-branded GPUs in the form of DG1 and DG2 codenames.
The 'DG' in both DG1 and DG2 should mean discrete graphics, while there is one variant of the DG1 and 3 variants of the DG2. The variants are broken into LP and HP could mean low-powered and high-powered, but it could also mean low-profile and high-profile... maybe.
Inside of the DG2 family there are three different numbers: 128, 256, and 512 -- this could be the EUs, or execution units. The DG1 variant seems to be the higher-end of the four SKUs.
Here's what we're looking at:
- iDG1LPDEV = "Intel(R) UHD Graphics, Gen12 LP DG1" "gfx-driver-ci-master-2624"
- iDG2HP512 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
- iDG2HP256 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
- iDG2HP128 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
AMD has been pumping its new Ryzen 3000 series CPUs into the market, based on the new Zen 2 architecture and 7nm node, but where are the third-gen Ryzen Threadripper CPUs? First we thought they were axed, but then the glimpses of a monster 64C/128T next-gen Ryzen Threadripper started... and now they're back again.
The Ryzen Threadripper 3000-series CPUs has turned up in a UserBenchmark result as a 16C/32T part as codename Castle Peak, with 3.6/4.0GHz base and boost CPU clocks, respectively. This is an engineering sample so CPU clocks are generally lower and will improve before the CPU is in its final, and retail stages.
Even in engineering form the new Castle Peak-dubbed 16C/32T chip offers an overall performance increase of 11% in single-core tests compared to a Ryzen Threadripper 2950X and up to 18% more in multi-threaded tests. It is about identical with the Ryzen 9 3900X in single-core tests and 35% faster in multi-core.
The two new third- and fourth-gen Threadripper codenames:
- AMD Family 19h Models 00h-0h, "Genesis"
- AMD Family 19h Models 20h-2h, "Vermeer"
AMD is kicking all sorts of ass right now, so much so that I'm sure Lisa Su is changing her shoes a few times a day -- and especially so in major Asian markets where AMD has crossed the 50% market share line.
Danawa is one of the largest retailers in South Korea with its internal firm Danawa Research seeing AMD Ryzen CPUs overtake Intel Core CPU market share in the last few days, with the surge across 50% thanks to Zen 2 chips launching on 7/7. The most popular CPU was the Core i5-9400F followed closely by the new Ryzen 7 3700X, while the Ryzen 5 3600X had slightly less, and the high-end Ryzen 9 3900X being surprisingly popular with more sold than the Core i5-9600K for July 6-11.
Then we have some data from BCN Ranking which uses data from major outlets in Japan that AMD has crossed the milestone 50.5% market share, Intel has fallen from grace in Japan where it did dominate in October 2018 with 72.1% and is now sitting at just 49.5% in July 2019.