Amazon Web Services this week introduced Trainium2, its new accelerator for artificial intelligence (AI) workload that tangibly increases performance compared to its predecessor, enabling AWS to train foundation models (FMs) and large language models (LLMs) with up to trillions of parameters. In addition, AWS has set itself an ambitious goal to enable its clients to access massive 65 ‘AI’ ExaFLOPS performance for their workloads.

The AWS Trainium2 is Amazon’s 2^nd Generation accelerator designed specifically for FMs and LLMs training. When compared to its predecessor, the original Trainium, it features four times higher training performance, two times higher performance per watt, and three times as much memory – for a total of 96GB of HBM. The chip designed by Amazon’s Annapurna Labs is a multi-tile system-in-package featuring two compute tiles, four HBM memory stacks, and two chiplets whose purpose is undisclosed for now.

Amazon notably does not disclose specific performance numbers of its Trainium2, but it says that its Trn2 instances are scale-out with up to 100,000 Trainium2 chips to get up to 65 ExaFLOPS of low-precision compute performance for AI workloads. Which, working backwards, would put a single Trainium2 accelerator at roughly 650 TFLOPS. 65 EFLOPS is a level set to be achievable only on the highest-performing upcoming AI supercomputers, such as the Jupiter. Such scaling should dramatically reduce the training time for a 300-billion parameter large language model from months to weeks, according to AWS.

Amazon yet has to disclose the full specifications for Trainium2, but we’d be surprised if it didn’t add some features on top of what the original Trainium already supports. As a reminder, that co-processor supports FP32, TF32, BF16, FP16, UINT8, and configurable FP8 data formats as well as delivers up to 190 TFLOPS of FP16/BF16 compute performance.

What is perhaps more important than pure performance numbers of a single AWS Trainium2 accelerators is that Amazon has partners, such as Anthropic, that are ready to deploy it.

“We are working closely with AWS to develop our future foundation models using Trainium chips,” said Tom Brown, co-founder of Anthropic. “Trainium2 will help us build and train models at a very large scale, and we expect it to be at least 4x faster than first generation Trainium chips for some of our key workloads. Our collaboration with AWS will help organizations of all sizes unlock new possibilities, as they use Anthropic’s state-of-the-art AI systems together with AWS’s secure, reliable cloud technology.”

Source: AnandTech – Amazon’s Trainium2 AI Accelerator Features 96 GB of HBM, Quadruples Training Performance

Nowadays many cloud service providers design their own silicon, but Amazon Web Services (AWS) started to do this ahead of its rivals and by now its Annapurna Labs subsidiary develop processors that can well compete with those from AMD and Intel. This week AWS introduced its Graviton4 SoC, a 96-core ARM-based chip that promises to challenge renowned CPU designers and offer unprecedented performance to AWS clients.

“By focusing our chip designs on real workloads that matter to customers, we are able to deliver the most advanced cloud infrastructure to them,” said David Brown, vice president of Compute and Networking at AWS. “Graviton4 marks the fourth generation we have delivered in just five years, and is the most powerful and energy efficient chip we have ever built for a broad range of workloads.”

The AWS Graviton4 processor packs 96 cores that offer on average 30% higher compute performance compared to Graviton3 and is 40% faster in database applications as well as 45% faster in Java applications, according to Amazon. Given that Amazon did not reveal many details about its Graviton4, it is hard to attribute performance increases to any particular characteristics of the CPU.

Yet, NextPlatform believes that the processor uses Arm Neoverse V2 cores, which are more capable than V1 cores used in previous-generation AWS processors when it comes to instruction per clock (IPC). Furthermore, the new CPU is expected to be fabricated using one of TSMC’s N4 process technologies (4nm-class), which offers a higher clock-speed potential than TSMC’s N5 nodes.

“AWS Graviton4 instances are the fastest EC2 instances we have ever tested, and they are delivering outstanding performance across our most competitive and latency sensitive workloads,” said Roman Visintine, lead cloud engineer at Epic. “We look forward to using Graviton4 to improve player experience and expand what is possible within Fortnite.”

In addition, the new processor features a revamped memory subsystem with a 536.7 GB/s peak bandwidth, which is 75% higher compared to the previous-generation AWS CPU. Higher memory bandwidth improves performance of CPUs in memory intensive applications, such as databases.

Meanwhile, such a major memory bandwidth improvement indicates that the new processor employs a memory subsystem with a higher number of channels compared to Graviton3, though AWS has not formally confirmed this.

Graviton4 will be featured in memory-optimized Amazon EC2 R8g instances, which is particularly useful to boost performance in high-end databases and analytics. Furthermore, these R8g instances provide up to three times more vCPUs and memory than Graviton 3-based R7g instances, enabling higher throughput for data processing, better scalability, faster results, and reduced costs. To ensure security of AWS EC2 instances, Amazon equipped all high-speed physical hardware interfaces of Graviton4 CPUs.

It should be noted that Amazon offers up to 64 vCPU cores with its R7g instances (i.e., one full Graviton3 CPU). Therefore, to offer 192 vCPU cores, Amazon will need to either install two 96-core CPUs into one server (which marks one of the world’s first Armv9-based 2-way server design), enable three-way SMT on a 96-core CPU, or implement a very low-latency high-bandwidth interconnection between servers running 96-core cPUs.

Graviton4 R8g is currently in preview, these instances will be available widely in the coming months.

Sources: AWS, NextPlatform

Source: AnandTech – Amazon Unveils Graviton4: A 96-Core ARM CPU with 536.7 GBps Memory Bandwidth

Once the domain of external, bus-powered hard drives, these days the market for external storage has been almost completely consumed by portable SSDs. Rapid technological advancements in NAND flash technology (including the advent of 3D NAND) has allowed their capacity to eclipse 2.5-inch HDDs, all the while multiple improvements in host interface speeds (such as the move from USB 2.0 to 3.0, and onwards to 3.2 Gen 2 / Gen 2×2 / USB4) has made portable SSDs faster and more reliable than the hard drives they replace. All of which has helped to make portable SSDs the go-to solution for external storage, both big and small.

For our latest storage buyer’s guide, we’ve rounded up some of the best portable SSDs we’ve looked at in the past couple of years. What’s hot, what’s not, and what’s a good deal while still offering reasonable performance? We break that down and make our picks for the best portable SSDs on the market today.

Source: AnandTech – Best Portable SSDs: Holiday 2023

Asus this week formally introduced its ASUS Dual GeForce RTX 4060 Ti SSD OC Edition 8 GB graphics card, an unusual video card that also offers an M.2 SSD slot. First showcased earlier this year, the card is aimed at small form factor systems, giving system builders access to a full-bandwidth M.2 slot for additional storage without having to occupy a second PCIe slot.

Just like other GeForce RTX 4060 Ti graphics cards, the Dual GeForce RTX 4060 Ti SSD is based on NVIDIA’s AD106 GPU with 4352 CUDA cores (that can operate at up to 2595 MHz). But rather than wiring up all 16 PCIe lanes coming into the card to the GPU, Asus has wired up just 8, giving the GPU a PCIe 4.0 x8 interface while leaving the remaining 8 PCIe lanes to drive the M.2 slot.

An M.2 2280 slot is located on the backside of the card and can be easily accessed by the user. Internally, Asus is relying on PCIe bifurcation here, meaning there isn’t a PCIe switch or bridge chip present. Rather, the M.2 slot is fed directly via the PCIe lanes, allowing it to run at up to PCIe 5.0 speeds. The cooling system of the graphics card can contact the SSD using a thermal pad, so the cooler of the Dual GeForce RTX 4060 Ti SSD can cool down the drive to ensure its consistent performance even under high loads.

Compared to a prototype board that Asus demonstrated back in June, Asus has simplified the final product. Specifically, whereas the prototype featured two M.2 slots – one of which required removal of the cooling system – the commercial product only has one M.2 slot on the back that is easy to access.

Internally, the use of PCIe bifurcation does present some additional complications. The host needs to support bifurcation of a single PCIe x16 link down to x8 + x8, which virtually all modern AMD and Intel platforms do (and indeed, this is how most of these systems drive multiple x16 slots). But there are the rare exceptions, particularly with entry-level platforms. And even then, bifurcation at the device level isn’t something commonly used in consumer hardware, which is to say it’s not always well-tested. To that end, Asus offers a list of its own motherboards that are guaranteed to support its Dual GeForce RTX 4060 Ti SSD graphics card.

Otherwise, the Dual GeForce RTX 4060 Ti SSD OC Edition 8GB is a fairly typical RTX 4060 Ti card. Additional power is supplied via an 8-pin PCIe power connector, and Asus offers the standard 3x DisplayPort 1.4a + 1x HDMI 2.1b video outputs. Meanwhile, cooling for the 2.5-slot wide card is provided by a pair of axial fans.

At this point, Asus has not announced a release date or a suggest price for the card. So it remains to be seen when it will become available.

Source: AnandTech – Asus Intros GeForce RTX 4060 Ti Video Card With Integrated M.2 SSD Slot

With the launch of AMD new Ryzen Threadripper 7000-series processors and associated TRX50 platform comes the need for new memory kits. Thanks to UDIMMs and RDIMMs requiring different slot designs under DDR5, AMD has opted to go exclusively with RDIMMs for their latest generation of Threadripper processors. Which for memory makers has included putting together overclockable registered memory modules with ECC that support EXPO timings, such as the kits G.Skill and V-Color have announced this week with their Zeta R5 Neo and v-color DDR5 OC R-DIMM memory kits.

G.Skill’s Zeta R5 Neo product family and v-color’s OC R-DIMM are exclusively designed for AMD’s Ryzen Threadripper 7000-series processors. Both vendors are currently offering two types of kits: a 64 GB quad-channel kit consisting of four 16 GB RDIMMs, and a 128 GB quad-channel kit featuring four 32 GB RDIMMs. For systems using the TRX50 platform, a single kit suffices, while the high-end workstation-class WRX90 platform requires two of these quad-channel kits to fully populate all eight channels.

G.Skill’s modules with ECC are rated for of DDR5-6400 CL32 39-39-102 at massive 1.40 Volts, whereas v-color promises data transfer rates of up to 7200 MT/s, never mentions timings or voltages, but stresses that its kits are aimed at TRX50-based machines.

G.Skill and v-color note that their overclockable RDIMMs, similar to other premium memory modules, are built using hand-selected memory ICs, making them better suited for overclocking. G.Skill’s modules come with very basic heat spreaders, which in case of DDR5 may become a limiting factor for overclocking potential. By contrast, v-color’s modules come with rather serious heat spreaders akin to those used for some server-grade modules.

All the overclockable RDIMM modules from G.Skill and v-color come with AMD’s EXPO profiles for simplified setup, with G.Skill using DDR5-6400 CL32 timings, while v-color runs at DDR5-7200 at unspecified timings.

The exact overclocking potential of G.Skill’s and v-color’s Threadripper memory kits remain to be seen. The actual I/O die powering Threadripper is the same I/O die as AMD’s other server processors, which is to say it’s been optimized for stability over performance – not to mention supporting a much larger amount of memory. Still, as we’ve seen on AMD’s consumer CPU platforms, the company’s memory controllers are no slouches. All the while memory vendors are already offering high-clocked RDIMM kits for Intel’s rival Sapphire Rapids Xeon platform.

G.Skill’s Zeta R5 Neo memory kits designed for AMD’s Ryzen Threadripper 7000-series processors are already available at Newegg: the 64 GB kit is priced at $530, whereas the 128GB kit costs $1,070. This is of course a huge price premium, but this can be explained by the fact that we are talking about unique memory kits designed for very specific high-end CPUs.

As for v-color’s DDR5 OC R-DIMM memory kits for AMD’s Ryzen Threadripper TRX50 platform, the company says that they will be available in November on its website.

Sources: G.Skill, v-color

Source: AnandTech – G.Skill and V-Color Unveil Factory Overclocked ECC RDIMMs for Ryzen Threadripper 7000

AMD’s top-of-the-line Ryzen Threadripper Pro 7995WX is a unique workstation-grade processor that not only offers unbeatable performance in programs that can take advantage of all of its 96-core might, but it also supports overclockability. As it turns out overclocking potential of the CPU is quite huge as it can hit 4.80 GHz on all cores with air cooling and 6.0 GHz on all cores with liquid nitrogen.

AMD’s Ryzen Threadripper Pro 7995WX is a quite formidable processor that packs 96 Zen 4 cores clocked at 2.50 GHz by default (and which can push one core all the way to 5.10 GHz with decent cooling), an eight-channel DDR5 memory subsystem, and 128 usable PCIe 5.0 lanes. But AMD decided to show that the processor can do much better with advanced and extreme cooling. To do so, the company teamed up with Bill ‘Sampson’ Alverson, an accomplished overclocker from the U.S., who took an Asus Pro WS TRX50-Sage WiFi motherboard, four G.Skill DDR5-6400 CL32 memory (which he set to work at 3200 MT/s CL32 mode), and an Asus Thor 1.5 kW power supply.

At first, Sampson tried to overclock AMD’s Ryzen Threadripper Pro 7995WX using a giant IceGiant ProSiphon Elite air cooler with four huge fans. He managed to hit 4,791 MHz in Cinebench R23 and 4,875 MHz in Cinebench R15. Then, he took a custom liquid cooling system and pushed the 96-core CPU to 4,966 MHz in Cinebench R23; 5,000 MHz in Cinebench R15; and a whopping 5,265 MHz in GPUPI 1B for CPU.

But the most interesting part is of course usage of liquid nitrogen cooling as −196°C (− 320 °F) certainly removes most of thermal constraints. The results are quite impressive: Sampson has managed to push AMD’s Ryzen Threadripper Pro 7995WX all the way to 5,550 MHz with LN2 cooling, which is over 584 MHz higher compared to his result with custom liquid cooling. The CPU scored 186,756 points in Cinebench R23, up from 100,000 points for the CPU at default clocks. The processor consumed 980W at such an extreme frequency, which is up dramatically from its 350W TDP. Meanwhile Elmor and Der8auer have managed to overclock his Ryzen Threadripper Pro 7995WX to 6.0 GHz wih LN2 and scored 201,501 points in Cinebench R23.

Meanwhile, to somewhat simplify overclocking botg overlockers had to reduce the number of memory modules to two and lowered its data transfer rate to 3,000 MT/s, which somewhat reduced the value of such overclocking (after all, nobody uses this CPU with only two memory modules). Meanwhile, considering the fact that AMD wanted to achieve the maximum clock possible for its 96-core processor, playing with memory configuration is a fair thing to do.

Source: AnandTech – AMD’s 96-Core Ryzen Threadripper Pro 7995WX Hits 6.0 GHz on All Cores with LN2

Western Digital is announcing today that it has begun volume shipments of its 24 TB hard drives based on conventional magnetic recording (CMR) to its full customer base. In addition, the company started to ramp up production of its 28 TB HDDs featuring shingled magnetic recording (SMR) that will be used by select customers.

The new lineup of 3.5-inch 7200 RPM hard drives includes Western Digital’s Ultrastar DC HC580 24 TB and WD Gold 24 TB HDDs, which are based on the company’s energy-assisted perpendicular magnetic recording (ePMR) technology. Both of these drives are further enhanced with OptiNAND to improve performance by storing repeatable runout (RRO) metadata on NAND memory (instead of on disks) and improve reliability.

The company is also preparing their Ultrastar DC HC680 SMR 28 TB hard drive, which relies on shingled ePMR and OptiNAND.

When it comes to performance, the new hard drives are a tad faster than predecessors due to higher areal density. Meanwhile, per-TB power efficiency of Western Digital’s 24 TB and 28 TB HDDs is around 10% – 12% higher than that of 22 TB and 26 TB drives, respectively, due to higher capacity and more or less the same power consumption. For CSPs and hyperscalers, the main value advantage of the new 24 TB drives is their higher capacity, which enables up to 612 TB of raw storage per rack using 102-bay 4U chassis.

All three models feature 10 platters and largely re-use the company’s current enterprise-grade platform for multi-drive environments, which was introduced over a year ago. The platform traditionally features a top and bottom attached motor as well as RV sensors to ensure consistent performance in vibrating multi-HDD chassis. The latest Ultrastar HDDs are generally being aimed at cloud service providers, hyperscalers, and large enterprises, whereas WD Gold are going primarily to VARs, enterprises, enterprise-grade NAS, and small businesses that need to store loads of data.

Western Digital is shipping WD Gold 24 TB hard drives through their distribution channel, so expect them to hit the market shortly. The Ultrastar DC HC680 and HC580 HDDs are now in the qualification phase with select CSPs, hyperscalers, and OEMs. They are also available for enterprises that deploy these devices internally. It looks like WD is just shipping the SATA versions of these drives initially, as the company is separating noting that that the SAS versions of the Ultrastar DC HC680 and HC580 HDD will be available in the first quarter of 2024.

“With these new offerings, Western Digital is once again proving that hard drives are not just keeping pace, they are forging a path forward, ensuring that data-intensive applications of today and tomorrow have a strong foundation to build on while the industry prepares for HAMR,” said Ed Burns, research director of HDD and storage technologies at IDC. “We are seeing strong momentum for Western Digital’s SMR HDDs and believe that SMR adoption will continue to grow as their new 28TB SMR HDD offers the next compelling TCO value proposition that cloud customers cannot ignore.”

Source: AnandTech – Western Digital Releases 24TB Ultrastar & Gold Hard Drives, 28TB SMR Drives Ramping

Corsair has introduced its MP700 Pro-series solid-state drives with a PCIe 5.0 x4 interface that offer sequential read speed of up to 12.4 GB/s. The new high-end drive family offers a trio of cooling options, including active air cooling and liquid cooling. The SSDs will complement the company’s MP700-series PCIe Gen5 drives and will offer a new level of performance for PC enthusiasts.

Corsair’s MP700 Pro drives are based on Phison’s PS5026-E26 controller as well what initial teardowns have discovered to be 232 layer Micron 3D TLC NAND memory. Compared to their earlier E26-based MP700 drive, Corsair has been able to increase the drive’s maximum sequential read speed to 12.4 GB/s and maximum sequential write speed to 11.8 GB/s thanks to the use of faster NAND, which offers a higher 2000 MT/sec transfer rate (versus the original’s 1600 MT/sec rate). In addition, the company increased random read speed to 1.5 million IOPS and random write speed to 1.6 million IOPS, which is comparable to speeds offered by enterprise-grade SSDs.

As far as capacity points are concerned, Corsair’s MP700 Pro are currently available in 1 TB and 2 TB configurations, whereas 4 TB versions will be available sometimes in the future.

In addition to very high out-of-box performance, there is another selling point that Corsair’s MP700 Pro drives have. They are available with three different cooling system to cater different PC form-factors and audiences. For high-performance laptops and compact desktops, Corsair offers MP700 Pro with a simplistic graphene heatspreader; for high-performance desktops, the company will offer these drives with an aluminum heatsink and a fan to ensure sufficient cooling and consistent performance even under high loads; whereas for owners of PCs with custom-built liquid cooling the company will offer a SSDs with a waterblock.

When the first M.2-2280 solid-state drives with a PCIe 5.0 x4 interface hit the market earlier this year, the majority of PCIe Gen5 SSDs topped out at a 10 GB/s sequential reads and writes since fast 3D NAND chips were in short supply. Now that 3D NAND with a 2000 MT/s interface is getting more widespread, Corsair and other manufacturers of SSDs are rolling out drives that can hit 12.4 GB/s read speeds.

In fact, Sabrent and some other makers are working on even faster SSDs with around 14 GB/s read speeds, so it is well possible that we are going to see a yet another ‘sub-generation’ PCIe Gen5 drives that will saturate a PCIe 4.0 x4 interface.

Source: AnandTech – Corsair Releases MP700 Pro SSDs: Up 12.4 GB/s With Three Cooling Options

SK hynix had started volume shipments of its LPDDR5T-9600 memory for high-end smartphones, the company announced this week. So far, the company’s LPDDR6 ‘Turbo’ memory with a 9600 MT/s data transfer speed has been certified to work with two range-topping mobile application processors from Qualcomm and MediaTek.

SK hynix’s LPDDR5T-9600 memory is available in 16 GB packages with a VDD voltage range of 1.01V to 1.12V and a VDDQ of 0.5v. Notably, this VDD range is just slightly over the LPDDR5X specfication (1.00V to 1.1V), which shouldn’t be a serious problem, but likely warrants some extra compatibility testing with existing chips.

Smartphone and SoC manufacturers have a great incentive to validate SK hynix’s LPDDR5T-9600 and Micron’s LPDDR5X-9600 memory as these modules offer a 76.8 GB/s peak bandwidth, up 12.5% from 68.2 GB/s offered by LPDDR5X-8533.

So far, SK Hynix’s LPDDR6T-9600 modules have been certified by Qualcomm for its Snapdragon 8 Gen 3 mobile SoC as well as MediaTek for its Dimensity 9300 and some subsequent processors. Meanwhile, SK Hynix confirmed that it had begun shipments of its LPDDR6T-9600 devices to Vivo, which will use it for X100 and X100 Pro mobile application processors.

Following the latest trends, SK hynix mentions that its very fast LPDDR5T-9600 memory will be particularly useful for on-device AI applications. And, of course, faster DRAM is always welcomed for graphics intensive mobile applications, such as games.

“Smartphones are becoming essential devices for implementing on-device AI technology as the AI era kicks into full swing,” said Myoungsoo Park, Vice President and Head of DRAM Marketing at SK hynix. “There is a growing demand for high-performing, high-capacity mobile DRAMs in the market. We will continue to lead the premium DRAM market based on our technological leadership in AI memories, while staying in tune with market demands.”

Source: AnandTech – SK hynix Ships LPDDR5T: 9600 MT/s Memory for Smartphones

AMD has announced the Ryzen Embedded 7000 Series processor family at Smart Production Solutions 2023, primarily targeting industrial markets such as automotive, robotics, and edge solutions. Built on the Zen 4 architecture and incorporating integrated Radeon graphics, it offers a new level of performance and functionality for the embedded market. Announced today are five new SKUs, ranging from 12C/24 down to 6C/12T, with three models with a 65 W TDP and two overclockable X-series models with a higher 105 W TDP.

Designed for industrial applications such as automotive, edge robotics, and IoT, AMD’s Ryzen 7000 Embedded series finally includes integrated graphics via their Radeon RDNA™ 2 graphics, reducing the need for discrete GPUs. The Ryzen Embedded 7000 series processors also feature support for multiple operating systems, including Windows Server and Linux/Ubuntu, as well as  Windows 10 and 11.

Touching on the specifications of the five new Zen 4-based Ryzen 7000 Embedded SKUs, AMD has opted to launch two overclockable X series variants, the Ryzen Embedded 7700X (8C/16T) and the Ryzen Embedded 7600X (6C/12T). These processors shouldn’t be confused with their Ryzen 7000 desktop counterparts. However, the Ryzen Embedded 7700X does have 8 x Zen 4 cores, 32 MB of L3 cache, with a base frequency of 4.5 GHz and a boost frequency of up to 5.4 GHz; this puts it in a similar position to the Ryzen 7 7700X, which is virtually identical in specifications.

The same can be said about the Ryzen Embedded 7600X, which is a 6C/12T part with a base frequency of 4.7 GHz, a turbo frequency of 5.3 GHz, 32 MB of L3 cache, with both X series Embedded chips coming with a 105 W TDP. Moving down the stack to the non-X series CPUs, the Ryzen Embedded 7945 has 16 x Zen 4 cores, with a larger 64 MB L3 cache, but has lower base clock speeds of 3.7 GHz, but has a 5.4 GHz turbo at just 65 W.

Coming in below the 7945 is the Ryzen Embedded 7745, an 8C/16T part with a 3.8 GHz base and 5.3 turbo frequencies, with 32 MB of L3 cache, all within a 65 W TDP. Offered as the entry-level chip to the Ryzen 7000 Embedded series is the Ryzen Embedded 7645, which has 6 x Zen 4 cores with a 3.8 GHz base frequency, with a turbo of up to 5.1 GHz.

The AMD Ryzen Embedded 7000 series processors have multiple platform options, including conventional AM5 socketed X670 and B650 motherboards. Interestingly, AMD includes support for an X600 “activator” socketed board, which provides the full 28 PCIe 5.0 lanes from the processor. The only caveat with not using PCIe lanes to connect to the chipset is that the X600 activator boards don’t have a chipset and thus don’t include any benefits such as I/O. Combining one of the Ryzen Embedded 7000 series processors with an X670 motherboard provides up to 24 x PCIe 5.0 lanes and 12 x PCIe 4.0 lanes for devices such as PCIe slots, M.2 storage, and other connectivity.

Unveiling the Ryzen Embedded 7000 series processors at the Smart Production Solutions 2023 event in Nuremberg, Germany, AMD’s Corporate VP and GM of the Embedded Processors Group, Rajneesh Gaur, went on to say, “With the integration of key performance features and a scalable design, Ryzen Embedded 7000 Series processors are ideal for a broad range of applications from advanced robotics and instrumentation design to power control, video surveillance and more.”

AMD hasn’t provided retail pricing at the time of writing, which is typical for their Embedded solutions. ODM and OEMs are expected to start shipping Ryzen Embedded 7000 products imminently.

Gallery: AMD Unveils Ryzen Embedded 7000 Series: Embedded Zen 4 for Industrial And Edge Solutions

Source: AnandTech – AMD Unveils Ryzen Embedded 7000 Series: Embedded Zen 4 for Industrial And Edge Solutions

Although AMD doesn’t have any major HPC announcements for this year’s SC23 show, the company isn’t starting this week empty-handed. For the workstation market the company is introducing its latest (and likely final) Radeon Pro W7000 series card, the Radeon Pro W7700.

Based on AMD’s latest-generation RDNA 3 architecture, the W7700 slots in between the Navi 31-based Radeon Pro W7800 and the Navi 33-based Radeon Pro W7600. As the final member of the lineup, the Navi 32-based card is designed to split the difference between AMD’s high-end and low-end Radeon Pro cards, offering a more modest level of performance for a far more modest price tag. Priced at $999, AMD is looking to hit a sweet spot in terms of hitting the high-end workstation market, edging out more expensive NVIDIA cards with a combination of higher performance and lower pricing.

Under the hood, the Radeon Pro W7700 looks a lot like a Radeon RX 7800XT that’s been dialed down to make a lighter TDP. We’re looking at a Navi 32 GCD with just 48 CUs enabled, and a more modest peak clockspeed of around 2.3GHz. The total rated throughput of the card is 28.3 TFLOPS FP32, trailing the more powerful (and more power-hungry) consumer cards a bit.

The Radeon Pro W7700 is paired with Navi 32’s full suite of Memory Cache Dies (MCD) however, giving it a better-than-usual ratio of FLOPS to memory bandwidth. With 16GB of GDDR6 attached to a 256-bit memory bus, the card offers 576GB/second of memory bandwidth – and ECC-protected memory, at that, a feature carried over from the high-end Radeon Pro models.

As noted earlier, AMD is aiming a bit more conservatively in terms of power consumption for this card, looking to offer something between the toasty 250W+ flagship cards, and the more petite 130W W7600. The end result is that AMD has set the TBP for the W7700 at 190W, allowing for the construction of an easily-cooled dual slot card.

In terms of features, as the fifth RDNA 3 video card in AMD’s Radeon Pro lineup, the W700 is on par with its peers. Like the flagship parts, we’re looking at a fairly performant version of AMD’s RDNA 3 architecture, with the company looking to leverage various optimizations and new features throughout the card to offer better performance, and for specific tasks, new features. The core architecture includes a significant update to the CUs to improve the total number of compute FLOPS attainable, as well as optimizations to improve the kind of dense matrix math used in AI inference calculations. As well, the ray tracing units have been updated to provide better performance and a larger overall ray rate.

RDNA 3’s marquee video feature of AV1 encoding support also makes it over to the card as well. The W7700 can encode a single AV1 stream at up to 8K60, a very useful feature as AV1 adoption increases – though it doesn’t get the more powerful dual media engine found on the flagship W7800/W7900’s cards.

Meanwhile, AMD’s ace in the hole on the display output side of matters is their support for DisplayPort 2.1 connectivity, something all of NVIDIA’s cards in this generation lack. Still, Pro users intending to take advantage of DP 2.1 will want to pay close attention to the specifications for AMD’s Radeon Pro lineup, as support is not consistent throughout the cards. Whereas the cheaper W7500/W7600 only have UHBR 10 data rate support, the W7700 gets faster UHBR 13.5 support for higher resolution/color depth displays. Still, this is a step down from the W7800/W7900, which can uniquely drive a max-bandwidth UHBR 20 display.

As for the physical W7700 card itself, it is a fairly typical-for-workstations dual slot blower design. AMD is using their standard DisplayPort configuration here, equipping the card with a quartet of full-sized DisplayPorts. In terms of length, the W7700 is a bit shorter than its W7800/W7900 counterparts, though AMD hasn’t sent over the exact dimensions for the card. And while not shown in any of AMD’s pictures, the card is fed additional power a single 8-pin PCIe power connector.

Wrapping things up, the Radeon Pro W7700 is getting a hard launch, and will be available today in stand-alone card form from the usual PC parts retailers at an MSRP of $999. Meanwhile, OEMs are expected to start shipping the card in their workstations beginning later this month.

Gallery: Radeon Pro W7700 Press Deck

Source: AnandTech – AMD Intros Radeon Pro W7700: Rounding Out RDNA 3 For Workstations

In a quite unexpected turn of events, Valve has introduced a revamped version of its Steam Deck console that got a larger OLED display, a more efficient system-on-chip, a higher-performance memory subsystem, a 1 TB storage option, faster Wi-Fi 6E, and higher-capacity battery. While performance target of the portable gaming device remained the same, the new unit might just be faster than the original one due to faster memory. Meanwhile, the cheapest 256GB version of the console, which retained an LCD screen, now costs $399.

The main difference between the new Steam Deck OLED and the original Steam Deck is the larger 7.4-inch OLED display. The new screen retains a 1280×720 resolution, but it has a 600 nits typical brightness (a major improvement) and also supports HDR with a maximum zonal luminosity of 1000 nits, according to Tom’s Hardware. The new monitor can also cover 101.8% of the DCI-P3 color gamut, up from 48.5% in case of the original Steam Deck, our colleagues found out. Despite the larger display, the console retained dimensions of the original one, but got 29 grams lighter since OLEDs do not need a backlighting module and a display driver with all the appropriate circuitry.

Another major change from the original one is that Steam Deck OLED uses AMD’s new semi-custom SoC with four Zen 2 x86 cores and an RDNA 2-based GPU with 512 stream processors that is now made on TSMC’s N6 process technology. Usage of N6 made the SoC a little cheaper to produce and allowed to retain the original 4W – 15W power envelope without the need to vary GPU clock from 1.0 GHz to 1.60 GHz, which may have a mild positive effect on performance in certain games.

An interesting detail about the new unit is that it uses a 16 GB LPDDR5-6400 memory subsystem with a 102.4 GB/s of peak bandwidth, up from 88 GB/s bandwidth in case of the original one. We can only wonder whether a 16% memory bandwidth boost will significantly affect gaming performance, but we feel it could bring a boost in certain bandwidth-hungry scenarios. Alternatively, faster memory may be just a bit more power efficient.

Speaking of power, it should be noted that the Steam Deck OLED also comes with a 50 Wh battery, up from a 40 Wh battery in the original one, so expect the new one to have a longer battery life. Meanwhile, the new console comes with a longer 2.5-meter power cord.

Yet another notable improvement of the Steam Deck OLED is 1 TB storage subsystem (and the M.2-2230 drive can still be replaced with something different) on the range-topping model. Meanwhile, to download games faster, Steam Deck OLED comes with a Wi-Fi 6E adapter, up from Wi-Fi 5 on the original model.

Valve launches its Steam Deck OLED in time for the holiday shopping system and in addition to make the product more attractive overall, it also makes the platform more accessible. The older 64 GB eMMC model with an LCD screen will be available for $349 while supplies last, whereas the 256 GB LCD version will cost $399 from now on (down from $529). The new Steam Deck OLED 512 GB is priced at $449, while the Steam Deck OLED 1 TB carries a $649 MSRP.

Source: AnandTech – Valve’s Steam Deck OLED: Bigger Display, New SoC, Faster Memory, More Storage, Same Performance

The path to high-capacity RDIMMs for servers has primarily been through 3D stacking (3DS) of DRAM dice using Through-Silicon Vias (TSVs). However, this has presented significant challenges in packaging (driving up the cost), and has also not been efficient in terms of energy consumption. The demand for large memory capacity RDIMMs is being primarily driven by the sudden emergence of large-language models (LLMs) for generative AI and increasing CPU core counts. Both of these require significant amount of DRAM to keep pace with performance requirements. Keeping these in mind, Micron is introducing 128 GB DDR5 RDIMMs capable of operating at up to 8000 MT/s today, with mass-production slated for 2024.

Micron has recently started fabricating 32 Gb monolithic DDR5 dice using its proven and mature 1β technology. The new dice have a 45%+ increase in bit density, and are capable of reaching up to 8000 MT/s while also operating with much more aggressive timing latencies compared to the standard JEDEC specs. The company is claiming that it improves energy efficiency by as much as 24% compared to the competition’s 3DS TSV offerings, and the faster operation can also help in faster AI training times. Avoiding 3DS TSV allows Micron to optimize the data input buffers and critical I/O circuits better, while also reducing the pin capacitance on the data lines. These contribute to the reduced power and improved speeds.

Micron has been doubling its monolithic die density every 3 years or so, thanks to advancements in CMOS process as well as improvements in array efficiency. The company sees a clear path to 48 Gb and 64 Gb monolithic dice in the future with continued technological progress. Micron is also claiming that its 1β node has reached mass production ahead of the competition, and that it has had the fastest yield maturity in the company’s history. Dual-die packages and tall form-factor (TFF) modules using 1β DRAM are expected to enable 1TB modules in the near future.

Along with the announcement of the 128 GB RDIMMs using 1β technology, the company also laid out its roadmap for upcoming products. HDM and GDDR7 are expected to dominate bandwidth-hungry applications, while RDIMMs, MCRDIMMs, and CXL solutions are in the pipeline for systems requiring massive capacity. LPDDR5X, and LPCAMM2 solutions going up to 192 GB are expected to make an appearance in power-sensitive systems as early as 2026.

Source: AnandTech – Micron Introduces 128 GB DDR5-8000 RDIMMs with Monolithic 32 Gb Dice