Broadcom announced parts 1 mo ago "that are now sampling" (in Procera 43 Tier 1 accounts??). PHY SERDES chip is the "critical component" most likely gating supplier high volume 100 gbs ethernet card availability.
The rapid growth of bandwidth consumption across data center, enterprise and core networks is driving the transition of line cards and systems from 10G/40G to 100G. Poised for rapid expansion, the 100GE market will grow 5X by 2016, according to Infonetics.
* High-performance PHYs that support physical medium dependent (PMD) sub layers, such as 100GBASE-SR4 and CR4, are a critical component for this rapid growth*.
Optimized for high-bw nets, the dual-port Broadcom gearbox PHY can mux/demux data across eight 25 Gbps channels to (or from) twenty 10 Gbps channels. The device can also be configured to support eight bi-directional lanes at 10 Gbps for 10GbE or 40GbE pass-through applications. High performance 25G/10G input/output (IO) eases routing constraints by providing for longer physical links.
"Next generation 100G networks demand the high performance, low power technologies that Broadcom delivers," said Broadcom "These first to market gearbox devices.... enable high density 100G networks."
"We're seeing strong demand for high bw 100gbs solutions in data center, enterprise and core networks to handle the growth in network traffic," said Infonetics "The higher density and low power consumption of Broadcom's new PHY will allow network managers to transform their networks and achieve higher performance on the industry's most widely adopted copper and short range links."
Photon.."Personally, I was dumbfounded the concept of no lost business at the same time that revenue projections seemed to disappear into the future"..
I had the exact same observation...it appears that the Q3 billing slippage into Q4 was more than offset by the lack of Q4 2014 PL10000 platform availability (i.e., volume quantities of 100gbsE serdes phy for the 100gig line cards).
This should be resolved in 3 mo max...Broadcom is ace #1 supplier..
Analysts sandbagging next yrs sales to downside cause they arenow in "show me mode". There may have been some slippaqge in MSO CableCo land. Cox comes to mind..maybe Sandvine there...already discounted.
Trevor, you need to match your supposed spoon-fed technical DDR with BadAlone to some answers from the TeraDumm post. If you knew anything about round-trip latencies to Flash-based subsystems, you'd be half there
exactly what Diablo is doing..masking IP theft with "our 4 million gates vs Netlist 40 thousand" and "our slick guru sw algos...". If badalone has 4m gates, what's the perf/ watt hit for turning on/ off 4 million transistor junctions???...vs 40 thousand w/ dram access times???
I'm pretty sure that the Chuckie Korean Bank move is directly related to the 2 Board resignations over ..."finance and strategic dir issues"..could be an LBO which would screw shareholders or this Korean Bank would provide the Surety bond in case Netlist loses the 12/2/2014 stop ship injunction against Diablo/ Sandisk...
nice post CZ and I did watch that You Tube Baaaad to da Bone presentatiuon...If it's "DRAM like" response time, they use Netlist IP to emulate bios while read-writing from dram to flash. The boys in the audience gaaacked at flash subsystem in mem channel and wondered what's wrong with using the PCI express bus instead..Baad couldn't even finish talk..
MCS/ Diablo at block level now with 7us of latency. Gurus want whole rear blovck of memory available as well (e.g., for HANA, SAP's awesome in-memory RT DBMS...12-18 months we'll have 1 us delay..uh huh..is that before or after you re-pot MCS due to court-ordered STOP SHIP??
Netlist NVDimm latency 500x lower than MCS..bye bye 4m gate MCS...oh, btw, NVDimm is 28nm process step and MCS is...130nm...embarassing there, mate..
Netlist IP is in the kernal space as Netbios (UEFC) sw mod to kernal OS. Diablo also has an asic bridging DDR3 to Flash storage which uses the netlist bios hw register handshake for duping bios in the kernal os later on. There are a number of Netlist patents involved. In fact, Diablo references some of them in their literature. There has been fascinating discussion on how a mission critical RT DBMS requires persistance at the main mem- CPU level. To use ultimate MCS capabilities, you need need to rewrite not only OS but the apps as well. 2 NoNos there..only niche strategy for high-end, like Netlist.
sales 16.1, meet.. non- gaap - .08 eps vs loss forecast of -.01 to -.11 eps
100 gbs vendor fixed issue and now has delivered parts to Procera; U.S. Pie now 11 vs 33% last yr; cable customers on freeze; euro pushout in taking delivery of bookings..