GN Receives AMD Ryzen Leaks
GamersNexus received tables of upcoming AMD processors within the 17h 70h-7Fh family, following the Ryzen 3000 release. We’ve received multiple dense PDFs with thermal design and power requirements for new processors, along with some specification tables for the CPUs detailed in the document. These are all using product numbers rather than names in the table, but we can take a guess at a few.
Notably, the tables give visibility to a number of upcoming processors, including two 65W TDP parts within the 12C family. Although yet unnamed in an official capacity, we can assume that units 100-070 and 100-072 will likely be in the Ryzen 9 family. These are 65W parts built for single-socket configurations with AM4, with a listed maximum boost clock of 4350MHz for each. The all-core clock will obviously be lower. Power consumption is listed as 88W during the highest power state, keeping in mind that TDP is mostly an imaginary number and doesn’t relate to power since power is not referenced in AMD’s formula. The P1 state will draw 53W, according to the document, at 2800MHz. We're also not sure if these will be mobile or desktop parts.
|AMD Ryzen Leaks to GamersNexus.net|
|Socket Group / Revision||B, B0||B, B0|
|Stock DDR4 Frequency||3200||3200|
|L2 / L3||6144KB / 64MB||6144KB / 64MB|
|Max TCTL (*C)||100||100|
|Minimum Ambient (*C)||5||5|
|Boot Core Frequency||2200MHz||2200MHz|
|S0 C0 Pb||4350MHz||4350MHz|
|S0 C0 P0||3100MH
|S0 C0 P1||2800MHz
|S0 C0 P2||2200MHz
|S0 C0 FMIN||550MHz
Source: GN table
Intel Could Be Looking At Another 14nm Shortage
In a report by Digitimes, it appears Intel is staring down the barrel of yet another 14nm CPU shortage. Digitimes’ sources indicate that Intel’s production has fallen back once again, with OEMs and notebook vendors allegedly having to delay product launches until 2020. The culprit, it seems, is the new Comet Lake and Cascade Lake CPUs.
Ian Cutress with Anandtech reached out to Intel to see if the chipmaker would comment on the story from Digitimes.
“We continue working to improve the supply-demand balance for our PC customers. In the first half of 2019, we saw PC customer demand that exceeded our expectations and surpassed third-party forecasts. We have added 14nm output capacity and are ramping volume on 10nm with systems on shelf for holiday. While our output capacity is increasing, we remain in a challenging supply-demand environment in our PC-centric business. We are actively working to address this challenge, and we continue to prioritize available output toward the newest generation Intel Core i5, i7 and i9 products that support our customers’ high-growth segments,” says Intel.
The statement at least partially corroborates Digitimes’ story. Intel’s supply and demand situation with 14nm CPUs has been volatile since 2018, but has eased in recent months. However, Q4 looks like it could see Intel’s CPU supply constricted once again.
China Enters DRAM Market
There are currently three notable memory suppliers in the market: Samsung, Micron, and SK Hynix, with two of those headquartered in South Korea and Micron headquartered in the US. For a long time, the memory market hasn’t done anything except lose competitors, as the big players emerged and smothered smaller companies like Elpida. Nanya is still sort of around for SSD cache, but isn’t a notable contributor to DRAM supply in memory or GPU applications. GPU memory is primarily from Samsung or Micron, at this point.
Last year, the Chinese government was investigating these three firms for price-fixing when DRAM prices had skyrocketed, but was simultaneously trying to enter the DRAM space itself, raising questions. State-backed semiconductor manufacturer Changxin Memory Technology, based in Hefei to the west of Shanghai and Nanjing and run by CEO Zhu Yiming, has begun making memory on its own node that it has dubbed “10nm-class,” ending up with a chip larger than competitors within the same class. At 18nm versus 12-16nm competition, it’s not the smallest, but it is expected to be fabbed at 120,000 wafers monthly.
Corsair iCUE Updated
Last week, we reported on a recent version of iCUE causing gameplay issues for some users. After we published the video and article, it came to our attention that Corsair dumped a new version of iCUE, v3.20.80. Thanks to those who pointed it out.
Many users on Corsair’s forums are reporting that version 3.20 has resolved the previous issues. So, if you’ve been affected, be sure to get the latest version here.
AMD Could Claim 10% Server Market Share By the End Of 2020
Digitimes, citing “market observers,” reports that AMD could claw away as much as 10% of the server market by the end of 2020. According to Digitimes, AMD has won contracts from IBM, Dell, and Nokia for its second-gen Epyc processors. As it stands currently, AMD represents roughly 5% of the server market.
Priced from $500 - $6,950, AMD’s Epyc line up offers significant value while also delivering competitive performance compared to Intel’s Xeon counterparts. This hasn’t gone unnoticed among server customers -- Dell recently announced five new server designs, built from the ground up, using AMD’s Epyc chips. Epyc has also secured a number of world records, including one for real-time 8K HEVC encoding.
Less than two years ago, Intel held roughly 99% of the server market; the inroads AMD has made with Epyc in this regard really can’t be overstated.
Intel Memory & Storage Day: Roadmap for Optane DIMMs, SSDs, and 144L NAND
Intel held its Memory and Storage Day in Seoul, South Korea, where Intel doubled down on its intent to fill the performance gap between storage speed and memory capacity, aiming to bring data closer to the processor.
The focal point of Intel’s press event was Optane, which will see a slew of new products using the technology, as well as a new Optane technology development line at Intel’s facilities in Rio Rancho, New Mexico. Intel also announced a second-generation of Optane DC Persistent Memory, code-named “Barlow Pass,” scheduled for release in 2020 to coincide with the release of Intel’s next generation Xeon Scalable processors.
Intel also announced a successor to its 660p SSD, the 665p SSD that uses Intel’s new 96-layer QLC NAND. Intel didn’t disclose and pricing or availability, but we could probably expect to see it in laptop designs early next year. Intel also demonstrated its 144-layer QLC NAND, using 5 bits per cell, that is also slated for commercial availability in 2020.
Rumor: Intel’s i9-9900KS Surfaces With $600 Price Tag
While Intel hasn’t officially disclosed the price of its forthcoming super-binned i9-9900KS, we expect it to be fairly steep. Listings at both Australlian and U.S. retailers (which are likely placeholders for now) somewhat confirm that suspicion.
A listing at mwave.com currently has the i9-9900KS at 899 AUD, which converts to roughly $670. Twitter user momomo_us found a U.S. listing with a $604 price tag. While we won’t have an official price until next month when the i9-9900KS ships, both listings seem to suggest that the CPU won’t be had for less than $600 -- a ~$100 premium over the i9-9900K.
More Rumors: AMD 550 Chipset, RX 5300 XT, and GTX 1660 Super Debut
This past week was one rife with rumors, so we’ll just go over them here quickly. So take them with the usual amount of salt, although they are interesting.
AMD’s B550 chipset has surfaced thanks to HP, with listings showing up via German retailer Alternate. The listings show two systems: the Pavilion Desktop TP01-0006ng and Desktop M01-F0017ng. Both systems list the “AMD Promontory B550A” chipset, which is set to be produced by ASMedia. The most prominent feature for AMD’s B550 -- or lack thereof, rather -- is the lack of PCIe 4.0 support. Additionally, the listings show the Radeon RX 5300 XT, with the only listed specs being 4GB of GDDR5 memory. The aforementioned systems could debut as soon as October, according to the listings.
Separately, HKEPC, a Hong Kong-based media outlet, seemingly outed more specifics for the 550 chipset in a review of ASRock's X570 Creator motherboard. The 550 chipset supposedly supports 2x USB 3.2 Gen 2 (10 Gbps), 8x SATA III ports, and an unspecified amount of USB 3.1 Gen 1 ports. Relatively unsurprising. The 550 looks to provide 4x PCIe 3.0 lanes to the processor, which is consistent with past chipsets. Interestingly, the 550 chipset supposedly supplies 4x PCIe 3.0 lanes and 8x PCIe 2.0 lanes for general purpose.
Finally, we recently discussed the possibility of a GTX 1660 Super, and the alleged card is back in the news, this time with a rumored October launch date. VideoCardz, citing sources at ASUS, claim that the company is preparing three models for a launch next month.
Editorial: Eric Hamilton
Host, Additional Reporting: Steve Burke
Video: Josh Svoboda, Andrew Coleman