Further to our last piece which we detailed Intel's issue to motherboard vendors to follow with stock power settings for Intel's 14th and 13th Gen Core series processors, Intel has now issued a follow-up statement to this. Over the last week or so, motherboard vendors quickly released firmware updates with a new profile called 'Intel Baseline', which motherboard vendors assumed would address the instability issues.
As it turns out, Intel doesn't seem to accept this as technically, these Intel Baseline profiles are not to be confused with Intel's default specifications. This means that Intel's Baseline profiles seemingly give the impression that they are operating at default settings, hence the terminology 'baseline' used, but this still opens motherboard vendors to use their interpretations of MCE or Multi-Core Enhancement.
To clarify things for consumers, Intel has sent us the following statement:
Several motherboard manufacturers have released BIOS profiles labeled ‘Intel Baseline Profile’. However, these BIOS profiles are not the same as the 'Intel Default Settings' recommendations that Intel has recently shared with its partners regarding the instability issues reported on 13th and 14th gen K SKU processors.
These ‘Intel Baseline Profile’ BIOS settings appear to be based on power delivery guidance previously provided by Intel to manufacturers describing the various power delivery options for 13th and 14th Generation K SKU processors based on motherboard capabilities.
Intel is not recommending motherboard manufacturers to use ‘baseline’ power delivery settings on boards capable of higher values.
Intel’s recommended ‘Intel Default Settings’ are a combination of thermal and power delivery features along with a selection of possible power delivery profiles based on motherboard capabilities.
Intel recommends customers to implement the highest power delivery profile compatible with each individual motherboard design as noted in the table below:
Click to Enlarge Intel's Default Settings
What Intel's statement is effectively saying to consumers, is that users shouldn't be using the Baseline Power Delivery profiles which are offered by motherboard vendors through a plethora of firmware updates. Instead, Intel is recommending users opt for Intel Default Settings, which follows what the specific processor is rated for by Intel out of the box to achieve the clock speeds advertised, without users having to worry about firmware 'over' optimization which can cause instability as there have been many reports of happening.
Not only this, but the Intel Default settings offer a combination of thermal specifications and power capabilities, including voltage and frequency curve settings that apply to the capability of the motherboard used, and the power delivery equipped on the motherboard. At least for the most part, Intel is recommending users with 14th and 13th-Gen Core series K, KF, and KS SKUs that they do not recommend users opt in using the Baseline profiles offered by motherboard vendors.
Digesting the contrast between the two statements, the key differential is that Intel's priority is reducing the current going through the processor, which for both the 14th and 13th Gen Core series processors is a maximum of 400 A, even when using the Extreme profile. We know those motherboard vendors on their Z790 and Z690 motherboards opt for an unrestricted power profile, which is essentially 'unlimited' power and current to maximize performance at the cost of power consumption and heat, which does exacerbate problems and can lead to frequent bouts of instability, especially on high-intensity workloads.
Another variable Intel is recommending is that the AC Load Line must match the design target of the processor, with a maximum value of 1.1 mOhm, and that the DC Load Line must be ... CPUs
TSMC to Expand Specialty Capacity by 50%, Introduce 4nm N4e Low-Power Node With all the new fabs being built in Germany and Japan, as well as the expansion of production capacity in China, TSMC is planning to extend its production capacity for specialty technologies by 50% by 2027. As disclosed by the company during its European Technology Symposium this week, TSMC expects to need to not only convert existing capacity to meet demands for specialty processes, but even build new (greenfield) fab space just for this purpose. One of the big drivers for this demand, in turn, will be TSMC's next specialty node: N4e, a 4nm-class ultra-low-power production node. "In the past, we always did the review phase [for upcoming fabs], but for the first time in a long time at TSMC, we started building greenfield fab that will address the future specialty technology requirements," said Dr. Kevin Zhang, Senior Vice President, Business Development and Overseas Operations Office, at the event. "In the next four to five years, we actually going to grow our specialty capacity by up to 1.5x. In doing so we actually expanding the footprint of our manufacturing network to improve the resiliency of the overall fab supply chain." On top of its well-known major logic nodes like N5 and N3E, TSMC also offers a suite of specialty nodes for applications such as power semiconductors, mixed analog I/O, and ultra-low-power applications (e.g. IoT). These are typically based on the company's trailing manufacturing processes, but regardless of the underlying technology, the capacity demand for these nodes is growing right alongside the demand for TSMC's major logic nodes. All of which has required TSMC to reevaluate how they go about planning for capacity on their specialty nodes. TSMC's expansion strategy in the recent years has pursued several goals. One of them has been to build new fabs outside of Taiwan; another has been to generally expand production capacity to meet future demand for all types of process technologies – which is why the company is building up capacity for specialty nodes. At present, TSMC's most advanced specialty node is N6e, an N7/N6 variant that supports operating voltages between 0.4V and 0.9V. With N4e, TSMC is looking at voltages below 0.4V. Though for now, TSMC is not disclosing much in the way of technical details for the planned node; given the company's history here, we expect they'll have more to talk about next year once the new process is ready. Semiconductors
SK hynix Reports That 2025 HBM Memory Supply Has Nearly Sold Out Demand for high-performance processors for AI training is skyrocketing, and consequently so is the demand for the components that go into these processors. So much so that SK hynix this week is very publicly announcing that the company's high-bandwidth memory (HBM) production capacity has already sold out for the rest of 2024, and even most of 2025 has already sold out as well. SK hynix currently produces various types of HBM memory for customers like Amazon, AMD, Facebook, Google (Broadcom), Intel, Microsoft, and, of course, NVIDIA. The latter is an especially prolific consumer of HBM3 and HBM3E memory for its H100/H200/GH200 accelerators, as NVIDIA is also working to fill what remains an insatiable (and unmet) demand for its accelerators. As a result, HBM memory orders, which are already placed months in advance, are now backlogging well into 2025 as chip vendors look to secure supplies of the memory stacks critical to their success. This has made SK hynix the secnd HBM memory vendor in recent months to announce that they've sold out into 2025, following an earlier announcement from Micron regarding its HBM3E production. But of the two announcements, SK hynix's is arguably the most significant yet, as the South Korean firm's HBM production capacity is far greater than Micron's. So while things were merely "interesting" with the smallest of the Big Three memory manufacturers being sold out into 2025, things are taking a more concerning (and constrained) outlook now that SK hynix is as well. SK hynix currently controls roughly 46% - 49% of HBM market, and its share is not expected to drop significantly in 2025, according to market tracking firm TrendForce. By contrast, Micron's share on HBM memory market is between 4% and 6%. Since HBM supply of both companies is sold out through the most of 2025, we're likely looking at a scenario where over 50% of the industry's total HBM3/HBM3E supply for the coming quarters is already sold out. This leaves Samsung as the only member of the group not to comment on HBM demand so far. Though with memory being a highly fungible commodity product, it would be surprising if Samsung wasn't facing similar demand. And, ultimately, all of this is pointing towards the indusry entering an HBM3 memory shortage. Separately, SK hynix said that it is sampling 12-Hi 36GB HBM3E stacks with customers and will begin volume shipments in the third quarter. Memory
G.Skill on Tuesday introduced its ultra-low-latency DDR5-6400 memory modules that feature a CAS latency of 30 clocks, which appears to be the industry's most aggressive timings yet for DDR5-6400 sticks. The modules will be available for both AMD and Intel CPU-based systems.
With every new generation of DDR memory comes an increase in data transfer rates and an extension of relative latencies. While for the vast majority of applications, the increased bandwidth offsets the performance impact of higher timings, there are applications that favor low latencies. However, shrinking latencies is sometimes harder than increasing data transfer rates, which is why low-latency modules are rare.
Nonetheless, G.Skill has apparently managed to cherry-pick enough DDR5 memory chips and build appropriate printed circuit boards to produce DDR5-6400 modules with CL30 timings, which are substantially lower than the CL46 timings recommended by JEDEC for this speed bin. This means that while JEDEC-standard modules have an absolute latency of 14.375 ns, G.Skill's modules can boast a latency of just 9.375 ns – an approximately 35% decrease.
G.Skill's DDR5-6400 CL30 39-39-102 modules have a capacity of 16 GB and will be available in 32 GB dual-channel kits, though the company does not disclose voltages, which are likely considerably higher than those standardized by JEDEC.
The company plans to make its DDR5-6400 modules available both for AMD systems with EXPO profiles (Trident Z5 Neo RGB and Trident Z5 Royal Neo) and for Intel-powered PCs with XMP 3.0 profiles (Trident Z5 RGB and Trident Z5 Royal). For AMD AM5 systems that have a practical limitation of 6000 MT/s – 6400 MT/s for DDR5 memory (as this is roughly as fast as AMD's Infinity Fabric can operate at with a 1:1 ratio), the new modules will be particularly beneficial for AMD's Ryzen 7000 and Ryzen 9000-series processors.
G.Skill notes that since its modules are non-standard, they will not work with all systems but will operate on high-end motherboards with properly cooled CPUs.
The new ultra-low-latency memory kits will be available worldwide from G.Skill's partners starting in late August 2024. The company did not disclose the pricing of these modules, but since we are talking about premium products that boast unique specifications, they are likely to be priced accordingly.
MemoryThe CXL consortium has had a regular presence at FMS (which rechristened itself from 'Flash Memory Summit' to the 'Future of Memory and Storage' this year). Back at FMS 2022, the company had announced v3.0 of the CXL specifications. This was followed by CXL 3.1's introduction at Supercomputing 2023. Having started off as a host to device interconnect standard, it had slowly subsumed other competing standards such as OpenCAPI and Gen-Z. As a result, the specifications started to encompass a wide variety of use-cases by building a protocol on top of the the ubiquitous PCIe expansion bus. The CXL consortium comprises of heavyweights such as AMD and Intel, as well as a large number of startup companies attempting to play in different segments on the device side. At FMS 2024, CXL had a prime position in the booth demos of many vendors.
The migration of server platforms from DDR4 to DDR5, along with the rise of workloads demanding large RAM capacity (but not particularly sensitive to either memory bandwidth or latency), has opened up memory expansion modules as one of the first set of widely available CXL devices. Over the last couple of years, we have had product announcements from Samsung and Micron in this area.
At FMS 2024, SK hynix was showing off their DDR5-based CMM-DDR5 CXL memory module with a 128 GB capacity. The company was also detailing their associated Heterogeneous Memory Software Development Kit (HMSDK) - a set of libraries and tools at both the kernel and user levels aimed at increasing the ease of use of CXL memory. This is achieved in part by considering the memory pyramid / hierarchy and relocating the data between the server's main memory (DRAM) and the CXL device based on usage frequency.
The CMM-DDR5 CXL memory module comes in the SDFF form-factor (E3.S 2T) with a PCIe 3.0 x8 host interface. The internal memory is based on 1α technology DRAM, and the device promises DDR5-class bandwidth and latency within a single NUMA hop. As these memory modules are meant to be used in datacenters and enterprises, the firmware includes features for RAS (reliability, availability, and serviceability) along with secure boot and other management features.
SK hynix was also demonstrating Niagara 2.0 - a hardware solution (currently based on FPGAs) to enable memory pooling and sharing - i.e, connecting multiple CXL memories to allow different hosts (CPUs and GPUs) to optimally share their capacity. The previous version only allowed capacity sharing, but the latest version enables sharing of data also. SK hynix had presented these solutions at the CXL DevCon 2024 earlier this year, but some progress seems to have been made in finalizing the specifications of the CMM-DDR5 at FMS 2024.
Micron had unveiled the CZ120 CXL Memory Expansion Module last year based on the Microchip SMC 2000 series CXL memory controller. At FMS 2024, Micron and Microchip had a demonstration of the module on a Granite Rapids server.
Additional insights into the SMC 2000 controller were also provided.
The CXL memory controller also incorporates DRAM die failure handling, and Microchip also provides diagnostics and debug tools to analyze failed modules. The memory controller also supports ECC, which forms part of the enterprise... Storage
0 Comments