Reply to thread

Message: <blockquote data-quote="ibnanv" data-source="post: 21319060" data-attributes="member: 218596">[YOUTUBE]PN93G6Rg2ek[/YOUTUBE]AMD’s new 32-core Naples CPU will attack Intel’s data center monopoly.AMD formally announced its  new Naples server CPU today, with up to 32 cores, eight memory channels,  and its new, 128-lane Infinity Fabric. Ryzen’s consumer performance may  be impressive, but this is a straight server play, with AMD hoping to  capitalize on strong performance in the 1-2 socket market.The  company’s strategy is similar to the one it adopted eleven years ago.  Back in 2006, AMD’s server business was one of the company’s biggest  success stories. Beginning in 2001, with the launch of Athlon MP and the  760MP chipset, AMD had slowly taken market share aware from Intel,  particularly once Opteron launched. Its share of the worldwide server  business peaked in Q2 2006, at just under 25% of the market. Today, <a href="https://www.extremetech.com/tag/amd" target="_blank">AMD</a>  has less than 1% of the market, which is one of the reasons it’s so  eager to rejoin that particular fight. Intel’s data center business was  worth $17.2 billion in 2016, up 8% from 2015. If AMD can capture 5% of  that business, that’s $860 million of additional revenue per year, and  it’s revenue that comes with a nice margin.<img src="https://www.extremetech.com/wp-content/uploads/2017/03/Naples-Design-640x266.jpg" alt="" class="fr-fic fr-dii fr-draggable " style="" />AMD’s Naples (that’s the  design name, not the marketing brand) server CPU is a multi-chip-module  design rather than a unified 32-core chip. This makes smart sense for a  number of reasons. First, GlobalFoundries has no experience building  a 32-core processor on 14nm, and the sheer size of the die would be a  bunker-buster. Instead, AMD has glued together multiple eight-core CCXs.  Each CCX has two memory controllers and there are four in a 32-core  chip, which is how AMD hits its 8-channel memory target. Total memory  bandwidth per CPU is 170 GB/s.<a href="https://www.extremetech.com/wp-content/uploads/2017/03/AMD-Naples-Head.jpg" target="_blank"><img src="https://www.extremetech.com/wp-content/uploads/2017/03/AMD-Naples-Head-640x336.jpg" alt="" class="fr-fic fr-dii fr-draggable " style="" /></a>At  its Ryzen tech day, AMD showed demos of its Naples CPU platform  performing a sparse matrix calculation on a 3D data set. When both AMD  and Intel chips were held to the same configuration — 44 cores,  DDR4-1866 — the Intel CPU took 35 seconds to complete a workload where  Naples took 18 seconds. When given its head and allowed to run all 64  cores and DDR4-2400, AMD cut another four seconds off its time. That’s  fairly significant performance, though we have to note that AMD released  no details on how its Naples CPU was clocked, what clock speeds the  chip will target, or even how the two servers were specifically  configured. All such benchmark results should therefore be taken with a  grain of salt.<a href="https://www.extremetech.com/wp-content/uploads/2017/03/Workload2.jpg" target="_blank"><img src="https://www.extremetech.com/wp-content/uploads/2017/03/Workload2-640x354.jpg" alt="" class="fr-fic fr-dii fr-draggable " style="" /></a>Then again, after seeing Ryzen’s performance in desktop applications,  the idea that it could serve up substantially better performance in  memory bandwidth-bound applications isn’t surprising. Any test that can  leverage memory bandwidth is going to love Naples’ 170GB/s of  connectivity.Why AMD is focusing on the server marketIt  may seem odd for AMD to prioritize breaking into servers over mobile,  given that the mobile market’s revenue is much larger than the data  center space, but there are some specific reasons for AMD to target this  market.First, Ryzen / Zen is unabashedly a server and  workstation CPU. All of our benchmarks show that Ryzen is most  competitive in computational workloads, particularly rendering tests.  AMD is working on a mobile Ryzen, to be fielded with Raven Ridge later  this year, but we still don’t know much about the GPU solution that chip  will use or what configurations AMD will offer. Bringing up the CPU and  GPU together on 14nm is more complex than launching a CPU on 14nm, so  it makes sense for AMD to target the easier lift.Second, the data  center market is the only PC space that’s really expected to grow in  the next few years. Consumer PC sales have fallen every year since 2011.  The best forecasts for this year are either flat or predict a smaller  decrease than previously. AMD is going to enter the mobile market with Ryzen, but it doesn’t make sense to prioritize launches in a contracting business.Third,  AMD has an opportunity to shine with Naples that it may lack in mobile.  It is unlikely, for example, that any mobile APU will offer the same  advantages in core count and memory bandwidth the server part can  achieve.                                  Since  Ryzen launched, AMD has occupied far more spots on Amazon’s Top 20 CPU  sales, and while this doesn’t tell us anything about sales volume or  revenue, seeing AMD nearly double its representation on that list is  still a sign that Ryzen is in strong demand. (As of this writing, the  Ryzen 7 1700 is in sixth place, the 1800X is in 8th, and Ryzen 7 1700 is  in 12th place. The other spots are occupied by Piledriver and a lone  APU SKU).Naples is unlikely to move in these kinds of volumes.  But if server manufacturers pick up the chip the way consumers have, AMD  could enjoy significantly improved financials by the end of the year.</blockquote>

Verification: Winadiyakata thappara keeyak tibeda?

Top Bottom