Despite Computex’s imminence, there are still plenty of pre-show announcements and news items to discuss. This week’s anchor item is the “conversation” that Micron has been having with memory suppliers; specifically, China’s Anti-Monopoly Bureau has discussed DRAM pricing with Samsung and Micron, Hynix likely to follow. Connecting the dots isn’t too hard here, but keep in mind that there’s still nothing confirmed with regard to price fixing possibilities.
Separately, AMD’s B450 chipsets were detailed, passive AM4 coolers debuted, and JPR thinks cryptomining is waning, giving way to more affordable video cards for gamers.
Show notes are below the video.
We wrote a couple of scripts to scrape the data shown in this content, showing memory price trends for the year so far. We recently set forth on an information gathering mission to learn about how much it costs to actually buy different types of memory, allowing us to look at just how much the memory suppliers are making. They’re raking in record profits with record stock highs – just look at the below Hynix or Micron stock chart: Despite claimed cleanroom limitations, the companies are making record revenue. Today, we’re talking about why and how the memory industry is in the shape it’s in.
This is Part 2 of our RAM Report series. The first part aired previously, and dug deep into five years of memory price data and earnings results for memory suppliers. Be sure to read or watch that content if you haven’t already.
The headlining story for the past week covers the memory supplier class action that was recently filed (vs. SK Hynix, Samsung, and Micron), alleging conspiracy to fix prices. In contension for the headline story, Intel's 10nm process problems have grown more complicated, seemingly preempting the company's hiring of Jim Keller, former AMD Zen architect.
While researching GPU prices and learning that GDDR5 memory price has increased by $20-$30 on the bill of materials lately, we started looking into the rising system memory prices. RAM pricing has proven somewhat cyclic over the past few years. We’ve reported on memory price increases dating back to 2012, and have done so seemingly every 2 years since that time. This research piece pulls five years of trend data, working in collaboration with PCPartPicker, to investigate why memory prices might be increasing, when we can expect a decrease, and more.
DRAM prices are crazy right now. We’ve driven that point into the ground over the past few years, but pinpointing a “when” and a “why” is a difficult proposition. With the help of PCPartPicker, we’ve identified some general trends that seem almost cyclic, and provide some relief in pointing toward an eventual downturn.
China’s National Development and Reform Commission (NDRC) is looking into the possibility of DRAM price-fixing between the major memory and Flash suppliers, with specific interest from the Pricing Supervision Department of said commission. An official from the regulatory body, Xu Xinyu of NDRC, stated the following: “We have noticed the price surge and will pay more attention to future problems that may be caused by ‘price fixing’ in the sector.”
This comes following recent reports that Samsung initiated plans to increase supply by 20%, which still failed to meet rising demand. The NDRC told the China Daily, a state-run media outlet, that the NDRC has paid attention to DRAM pricing and demand over the past 18 months, and that memory suppliers are now under the eye of the NDRC. There are only four major suppliers in the industry, and those include SK Hynix, Micron, Toshiba, and Samsung.
ET News reports [English] that the price of silicon wafers, the raw material used in the production of 300mm semiconductors, has increased 20% year over year from major manufacturers SK Siltron and SUMCO.
SK Siltron is a recent acquisition of the SK Group, a massive South Korean conglomerate that also includes SK Materials (produces NF3 gas used in semiconductor production) and SK Hynix (a memory chipmaker that regularly appears in our articles on increasing NAND demand). SK Siltron was known as LG Siltron until January, when SK Group purchased 51% of shares from LG for $532 million, and then proceeded to purchase the rest as well (Chairman Choi Tae-Won personally secured 29.4%). LG Siltron sales had suffered since 2012 with an industry increase in silicon wafer production, as well as the aftermath of the 2008 financial crisis--but SK Group’s purchase immediately paid off.
SK Hynix has been busy as of late. We most recently covered their plans for expansion, which offered a cursory foretaste into what 2017 might hold for the semiconductor supplier. SK Hynix has also recently further delineated plans for 2017, trailing behind their still-fresh announcement of the industry’s first 8GB LPDDR4X-4266 DRAM packages aimed at next-generation mobile devices.
In revealing plans, SK Hynix intends to volumize production of new types of memory—not altogether unexpected. Their primary focus on NAND production and expansion over DRAM is most noteworthy, at least for impermanent future. As such, SK Hynix intends to start volume production of 72-layer 3D TLC NAND (3D-V4). For reference, SK Hynix’s 36-layer and 48-layer NAND were 3D-V2 and 3D-V3, respectively. Notable about SK Hynix’s fourth version of 3D NAND is that it will use block sizes of 13.5 MB over the 9 MB sizes of the second and third generation predecessors. Furthermore, SK Hynix intends to roll-out 256 Gb 3D TLC ICs by Q2 2017, with 512 Gb 3D TLC ICs coming in Q4 2017. SK Hynix’s new 72-layer 3D NAND should allow for higher capacity SSDs in smaller form factors and increase performance on a per IC basis.
The race to invest in semiconductor technology is unabating, it would seem. SK Hynix, the world’s second largest memory chipmaker (after Samsung), has announced plans to construct a new memory semiconductor fab in Cheongju, South Korea. The company will also upgrade DRAM facilities in China, with the total outlay summing $2.6 billion. This comes after global chipmakers like Samsung, Toshiba, and TSMC have spurred investments of their own to expand production.
In additional hardware news to what we published yesterday -- a look at Intel's Kaby Lake (7600K, 7700K, etc.), the X2 Empire unique enclosure, and Logitech's G Pro mouse -- we are today visiting topics of Samsung's GDDR6, SK Hynix's HBM3 R&D, PCIe Gen4 power budget, and Zen's CCX architecture.
The biggest news here is Samsung's GDDR6, due for 2018, but it's all important stuff. PCI-e Gen4 is looking at being fully ratified EOY 2016, HBM3 is in R&D, and Zen is imminent and finalized architecturally. We'll talk about it more specifically in our reviews.
Update: Tom's misreported on PCI-e power draw. The Gen4 PCIe interface will still be 75W.
Anyway, here's the news recap:
Memory manufacturer Samsung is developing GDDR6 as a successor to Micron's brand new GDDR5X, presently only found in the GTX 1080 and Titan XP cards. GDDR6 may feel like a more meaningful successor to GDDR5, though, which has been in production use since 2008.
In its present, fully matured form, GDDR5 operates at 8Gbps maximally, including on the RX 480 and GTX 10 series GPUs. Micron demonstrated GDDR5X as capable of approaching 12-13Gbps with proper time to mature the architecture, but is presently shipping the memory in 10Gbps speeds for the nVidia devices.
Samsung indicates an operating range of approximately 14Gbps to 16Gbps on GDDR6 at 1.35V, coupled with lower voltages than even GDDR5X by using LP4X. Samsung indicates a power reduction upwards of 20% with post-LP4 memory technology.
Samsung is looking toward 2018 for production of GDDR6, giving GDDR5X some breathing room yet. As for HBM, SK Hynix is already looking toward HBM3, with HBM2 only presently available in the GP100 Accelerator cards. HBM3 will theoretically run a 4096-bit interface with upwards of 2TB/s throughput, at 512GB/s per stack. We'll talk about this tech more in the semi-distant future.
Tom's Hardware this week reported on the new PCI Express 4.0 specification, primarily detailing a push toward a minimum spec of 300W power transfer through the slot, but could be upwards of 500W. Without even talking about the bandwidth promises – moving to nearly 2GB/s for a single lane – the increase of power budget will mean that the industry could begin a shift away from PCI-e cables. The power would obviously still come form the power supply, but would be delivered through pins in the PCI-e slots rather than through an extra cable.
This same setup is what allows cards like a 750 Ti to function only off the PCI-e slot, because the existing spec allows for 75W to push through the PCIe bus. PCI-e 4.0 should be ratified by the end of 2016 by the PCI-SIG team, but we don't yet know the roll-out plans for consumer platforms.
AMD also detailed more of its Zen CPU architecture, something we talked about last week when the company camped out near IDF for an unveil event. The Summit Ridge chips have primarily been on display thus far, showing an 8C/16T demo with AMD's implementation of SMT, but we haven't heard much about other processors.
AMD is ditching modules in favor of CPU Complexes, or a CCX, each of which will host four CPU cores. Each CCX runs 512KB of L2 Cache per core, as seen in this block diagram, with L3 sliced into four pieces for 8MB total low-order address interleave cache. AMD says that each core can communicate with all cache on the CCX, and promises the same latency for all accesses.
It looks like the lowest SKU chips will still be quad-cores at a minimum.
Host: Steve "Lelldorianx" Burke
Video: Andrew "ColossalCake" Coleman
One of newest memory technologies on the market is HBM (High Bandwidth Memory), introduced on the R9 Fury X. HBM stacks 4 memory dies atop an interposer (packaged on the substrate) to get higher density modules, while also bringing down power consumption and reducing physical transaction distance. HBM is not located on the GPU die itself, but is on the GPU package – much closer than PCB-bound GDDR5/5X memory modules.
We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.