Anandtech News

**FEED** · 11-09-18, 03:28 PM

One of the interesting things to come out of the news in recent weeks is the march to double capacity memory. In today’s market, memory modules for consumer grade computers have a maximum of 16GB per module. This is unbuffered memory, and the standard for home computers and laptops. However recently there have been two major announcements causing that number to double from 16GB to 32GB: Samsung has developed double capacity ICs to drive up to 32GB per module with the same number of chips, but also a couple of DRAM vendors have found a way to put two times as many ICs on a 16GB module to make it up to 32GB. Both G.Skill and ZADAK fall into that latter category, and now we have both of these kits in the lab for review.

More...

**FEED** · 11-11-18, 06:28 PM

One of the announcements from last week involved Intel and its new Cascade Lake Advanced Performance category of processors to launch next year. These new processors will be based on combining two 24-core Cascade Lake-SP processors on a single package substrate to offer a single socket 48-core option with a total of twelve memory channels. The Cascade Lake-AP parts are going to be launched next year, and until then Intel is putting out some internal benchmark numbers.
Vendor Benchmark Results

When we are this far away from a product launch, all benchmark numbers should be taken with a grain of salt. This goes doubly so for vendor supplied benchmarks. However, Intel is on the warpath to promote what it sees as a new product family within its portfolio, even if it is only set to come out next year.
At the announcement last week, Intel offered Linpack and Stream Triad as two main high-performance metrics as comparison. Today Intel is also offering more ‘real world’ metrics. These metrics are, to quote Intel, ‘estimates based on pre-production hardware’. This means that the hardware is not ready yet, and these are values based on the engineering samples running but extrapolated to an expected benchmark value. Add another dump truck of salt on these numbers.
Intel’s official list of results are as follows:

Intel's Benchmark Numbers for 2S 48 Core Cascade Lake-AP
	Benchmark Type	Score vs 2S EPYC 7601
*These numbers were created by Intel
Linpack	Numerical Linear Algebra	3.4x
Stream Triad	Memory Bandwidth	1.3x
MILC	Quantum Chromodynamics	1.5x
WRF	Weather Forcasting	1.6x
OpenFOAM	Computational Fluid Dynamics	1.6x
NAMD (APOA1)	Molecular Dynamics	2.1x
YASK (ISO 3DFD)	HPC Kernel Tuning	3.1x

The slide with this data is in the gallery below.
In each case, Intel is comparing a dual-socket Cascade Lake-AP system with a dual socket EPYC 7601 system. Intel’s information slides go through how it set up all of its AMD systems in detail, however it does not disclose how the Cascade Lake-AP systems are set up by comparison, presumably as to not disclose any additional set-up numbers.

For the most part, we don’t put much stock into vendor supplied benchmark numbers. It’s easy for a vendor to claim a multiple when doubling particular compute resources, but when it comes to real world tests, companies like Intel have to try and promote its future products to potential customers. This is what this is. However, no matter how many numbers come out, these are impossible to verify independently. Wait until the AnandTech review, of course.
Intel also disclosed a number of ecosystem partners who are getting ready to deploy Cascade Lake-AP, as well as an offical declaration of the Cascade Lake-AP deployment at HLRN.

Cascade Lake-AP is set to be launched alongside the Cascade Lake-SP in the first part of 2019, although Intel states that Cascade Lake-SP will ship for revenue in 2018. This week we are at the Supercomputing trade show - hopefully there will be a demo somewhere that we’ll be able to see and talk about.
Gallery: Intel Offers More Cascade Lake-AP Performance Numbers

More...

**FEED** · 11-12-18, 10:26 AM

At AMD’s Next Horizon event this week, the company disclosed for the first time the layout of its next generation EPYC processor, the new Vega Radeon Instinct datacenter compute accelerators, as well as a strong confidence that its execution on 7nm will be a big win. If there’s anyone at AMD prepared to talk about execution, it’s the Chief Technology Officer, Mark Papermaster.

More...

**FEED** · 11-12-18, 10:26 AM

The Frontera supercomputer is the next generation high performance machine set to debut at the Texas Advanced Computing Center (TACC). As part of Intel’s HPC Forum, being held just before the annual Supercomputing conference, a number of disclosures about the design of Frontera (Spanish for ‘Frontier’) were made. One of which is certainly worth highlighting: this is not a supercomputer that is going to worry about performance per watt – this is all about the peak performance.

More...

**FEED** · 11-12-18, 11:32 AM

Earlier this year, Microsoft launched their lower cost, smaller Surface Go. With a display size of 10 inches, and weighing 1.15 lbs, or 515 grams, it’s a more portable version of the popular Surface Pro.
Today Microsoft is announcing a new LTE variant for those that want to use the convertible PC even more on the go. The Surface Go has been paired with the Qualcomm X16 modem, which is the same modem they used in the Surface Pro LTE.

Microsoft Surface Go
	Surface Go Specifications
CPU	Intel Pentium Gold 4415Y (Kaby Lake-Y) 2 core, 4 thread, 1.6 GHz base frequency
GPU	Intel HD 615 24 EUs 850 MHz boost frequency
Display	10-inch PixelSense 1800x1200 3:2 aspect 216 Pixels Per Inch 10-point Multitouch Surface Pen support
Dimensions	245 x 175 x 8.3 mm 9.6 x 6.9 x 0.33 inches
Weight	515 grams (WiFi) 1.15 lbs (WiFi)
RAM	4 or 8 GB LPDDR3-1866
Storage	64 GB eMMC 128 / 256 GB NVMe SSD optional
Wireless	802.11ac with Bluetooth 4.1 Qualcomm Snapdragon X16 LTE Optional Cat 9 modem, 22 bands, 450 Mbps download speed
Battery	Up to 9 hours of video playback 24W Charger
Cameras	Windows Hello IR camera 5 MP Front Camera with 1080p video 8 MP Rear Camera with 1080p video
Ports	USB Type-C 3.1 Gen 1 with power delivery Surface Connect MicroSD Headset
Price	4GB/64GB $399 8GB/128GB $549 8GB/128 GB LTE $679 Windows 10 Pro $50 extra

The new Surface Go LTE is available for pre-order now, starting at $679, which is a $130 increase over the same 8 GB / 128 GB Wi-Fi model, or $729 for commercial customers which will probably mean a bump up to Windows 10 Pro.
The new model will be available in 23 markets by November 22, and more are coming after that.
Source: Microsoft

More...

**FEED** · 11-13-18, 10:12 AM

AMD has announced its new high-frequency EPYC 7371 processor designed for applications that benefit from high clocks. The CPU has 16 cores and is aimed at tasks like electronic design automation, high-frequency trading, and other. The EPYC 7371 can work in dual-socket configuration, thus offering up to 32 cores and 64 threads per box.
The AMD EPYC 7371 processor features 16 cores with SMT (spread across four eight-core Zen dies), 64 MB of L3 cache, an eight-channel DDR4 memory subsystem, and 128 PCIe lanes. The CPU features a 3.1 GHz default frequency, yet can run all cores at 3.6 GHz, or just eight cores at 3.8 GHz.
Being aimed at workloads that need a high single-thread performance, the EPYC 7371 takes advantage of its massive L3 cache along with its rather high frequencies. Meanwhile, its ability to work in dual-socket configuration is a major advantage that the EPYC 7371 has over other high-clock CPUs because the processor still enables 32 physical cores and 64 threads per system.

AMD EPYC Processors (2P)
	Cores Threads	Frequency (GHz)			L3	DRAM	PCIe	TDP	Price
Base	All	Max
EPYC 7601	32 / 64	2.20	2.70	3.2	64 MB	8-Ch DDR4 2666 MT/s	8 x16 128 PCIe	180W	$4200
EPYC 7551	32 / 64	2.00	2.55	3.0	180W	>$3400
EPYC 7501	32 / 64	2.00	2.60	3.0	155W/170W	$3400
EPYC 7451	24 / 48	2.30	2.90	3.2	180W	>$2400
EPYC 7401	24 / 48	2.00	2.80	3.0	155W/170W	$1850
EPYC7 7371	16 / 32	3.10	3.60	3.8	?	?
EPYC 7351	16 / 32	2.40	2.9		155W/170W	>$1100
EPYC 7301	16 / 32	2.20	2.7		155W/170W	>$800
EPYC 7281	16 / 32	2.10	2.7		32 MB	155W/170W	$650
EPYC 7251	8 / 16	2.10	2.9		120W	$475

What is particularly noteworthy about the EPYC 7371 processor is that it is not officially listed over at AMD’s website. Perhaps, this part was initially offered to select customers only as an “off-roadmap” SKU, but demand from HPC vendors and traders encouraged AMD to expand availability of the chip to a broader range of its customers.
AMD plans to start selling its EPYC 7371 chip in Q1 2019, so before its next-generation EPYC ‘Rome’ chips will become available. The manufacturer did not touch upon pricing of its chip as it will naturally depend on volumes and other factors.
Related Reading:

Source: AMD

More...

**FEED** · 11-13-18, 10:12 AM

One of the interesting elements in NBase-T networking is how and when companies are releasing new multi-gig 2.5G+5G+10G controllers to the market. This week Buffalo lifted the lid their newest card, which is using a unique controller implementation we’ve not seen in the wild before.
The LGY-PCIE-MG expansion card is a PCIe 2.0 x4 card designed for commercial systems with a standard RJ-45 port but capable of 1G, 2.5G, 5G and 10G networking connectivity. The key part of the card, the MAC and PHY, comes from a combination of a Tahuti Networks TN4010 multi-speed MAC and Marvell’s Alaska M 88E2180 PHY. Up until this point, any multi-gig Base-T working implementation has been using Aquantia solutions, or we’ve recently seen Realtek controllers for 2.5G, but now we have a true second player in the multi-gig space. The Marvell controller was actually announced back in May 2017, so it seems there has been a long lead time before coming to market. The obvious players missing from this set are Broadcom, Qualcomm, and Intel.
The Buffalo LGY-PCIE-MG card is designed to work with operating systems from Windows 7 and up, as well as Windows Server 2012 and up. Power consumption is listed in the specification sheet at 6.1W, and has an operating temperature window of 0-55C (which seems low?). The RRP of the card is £155 (UK), which is substantially higher than current Aquantia solutions on the market, which are around £100.
Related Reading

More...

**FEED** · 11-13-18, 02:29 PM

Much to our surprise, we saw a Qualcomm Centriq server from an OEM at Supercomputing this week.

More...

**FEED** · 11-13-18, 02:29 PM

Part of AMD’s recent pre-announcement of its next generation Rome server processors, using 7nm chiplets and running up to 64 cores, is that the new processors will be compatible with current motherboards on the market. However, one of the new features of Rome is the use of PCIe 4.0. PCIe 4.0 has different standards for on-board signalling in order to get the required speed, so even though the processors are backwards compatible with PCIe 3.0, we expect new motherboards and new systems to be developed with PCIe 4.0 specifications in mind. Today, we saw the first early revision of such a motherboard.

More...

**FEED** · 11-13-18, 03:41 PM

Western Digital on Monday introduced its new special-purpose SSD designed to speed up in-memory processing applications. The new Ultrastar DC ME200 Memory Extension Drive uses proven hardware and comes with special software that creates virtualized memory pools.
From hardware standpoint, the Ultrastar DC ME200 belongs to the Ultrastar SN200 family launched about two years ago. The drive is based on a proprietary controller and paired with Western Digital’s 128 Gb planar MLC NAND memory. The drive comes in U.2 and PCIe 3.0 x8 HHHL card form-factors, making it drop-in compatible with most existing x86 servers. WD is offering 1 TB, 2 TB, and 4 TB versions of its Ultrastar DC ME200 drives, all rated for up to 17 drive writes per day over a three-year period.
Meanwhile when it comes to speed, the special-purpose SSD should perform in line with the SN200-series. But the key thing about performance of the Ultrastar DC ME200 are not sequential reads or writes, but ability to speed up select applications that use in-memory processing.

As noted above, the Ultrastar DC ME200 comes with software that allows the NAND flash to be “seen” as system memory by the operating system. The software supports advanced prefetch algorithms that the company claims enables near-DRAM performance in applications that use in-memory processing (e.g., Redis, Memcached, SGEMM, MySQL, etc.). Western Digital recommends to install one Ultrastar DC ME200 per socket, essentially extending system memory capacity upwards of eight-fold.

By using the Ultrastar DC ME200 owners of datacenters can potentially reduce their spending on DRAM while still ensuring high performance for their large databases that require a lot of system memory. For example, a server featuring 256 GB of DDR4 and a 2 TB memory extension drive costs 25% less than a system with 1.5TB of DDR4.

Western Digital’s Ultrastar DC ME200 memory extension drive are sampling with select customers. Pricing has not been announced because it depends on volumes and other factors.

Western Digital Ultrastar DC ME200 Specifications
		Ultrastar DC ME200
Capacities		1 TB 2 TB 4 TB	1 TB 2 TB 4 TB
Form Factors		HHHL add-in card	2.5-inch U.2
Interface		PCIe 3.0 x8 (NVMe 1.2)	PCIe 3.0 x4 (NVMe 1.2)
Controller		Proprietary
NAND		128 Gb MLC made using 15 nm process technology (?)
Write Latency 512 B		20 ms
Power	Idle	9 W
Operating	25 W
Endurance		17 DWPD
Supported Processors		Intel Xeon E5-x6xx v3 or later, E7-x8xx v3 or later Intel Xeon Bronze, Silver, Gold, or Platinum
Supported Operating Systems		Linux 64-bit OS RHEL 6.5, 6.6, 6.7, 6.8, 6.9, 7.0, 7.1, 7.2, 7.3, 7.4 SLES 11-SP4, 12, 12-SP1, 12-SP2 Fedora Core ver. 4 to 27; Open SuSe ver. 10 to 11 Ubuntu Server ver. 16.04 to 17.10; Debian 9.5.0
MTBF		?
Warranty		Three years