Aquantia Launches New 2.5G/5G Multi-Gigabit Network Controllers for PCsby Ian Cutress on December 14, 2016 2:00 PM EST
- Posted in
Frequent readers of AnandTech will have recognized that our coverage of higher speed network interfaces over copper, particularly 10G, for the home and business has increased recently. This is perhaps no surprise: 1 Gbit speeds for wired networking have been the norm for fifteen years on consumer level products, but there are certainly some users that require faster network availability in their home/business due to the increased number of devices and size of content, particularly video content, that has exploded this decade.
Arguments aside about external network availability and infrastructure, streaming inside the home or in the office typically comes down to a couple of key factors – the size of the networking backbone and the capability of each device connected to it. For consumer infrastructure that stays on copper, 1Gbit or teamed connections has been the limit for long distance Cat 5a/Cat 6 cabling, which is fairly cheap, and some solutions have moved to the more expensive 10GBase-T solutions which are power hungry, more enterprise focused and simultaneously a lot more expensive by at least an order of magnitude. 10GBase-T also requires Cat 7 cabling for any serious distance requirements.
Back at the Supercomputing 2015 show, there were several talks about the utility of 2.5G and 5G copper networks. These discussions were primarily focused in the data-center and high-performance categories rather than the consumer, and at the time I commented that it was likely where these standards were to stay as it would require a significant step up to drive a 2.5x/5x adoption for the consumer over a 10x adoption. It was expected that the cost of 10G would start to come down, but over 2016 it has remained consistent, much to the chagrin of enthusiasts.
At the time in 2015, the 2.5G/5G standards were not yet ratified by IEEE. There were chips in the market, solely from Aquantia, for enterprise configurations that were happy to go with an evolving standard for their solution. From September 2016 this changed, and the standards have been ratified with Aquantia, Intel, Cisco and others all involved in the specification. Aquantia’s earlier generation silicon adhered to the standard, and has been deployed in a number of enterprise backbone deployments to the tune of 5M ports a year. Today’s announcement surrounds the launch of two new controllers based on the multi-gigabit standards aimed at more consumer level solutions – specifically ‘client connectivity in enterprise, gaming and SMB applications’.
The chips are small, 11x7mm, and use Aquantia’s expertise in both MAC/PHY management to produce the AQtion brand (as in, ‘Action’), to complement the AQrate (‘accurate’) currently in the ecosystem. The two AQtion controllers are single channel, single port solutions and differ in their speed support:
The AQC107 will support 10GBase-T on Cat 6 copper cables, adhering to the 8 02.3an standard, although at a different set of distance limitations. Both chips are targeting a large market of motherboards, routers, office PCs, switches, routers, dongles and anything that a consumer or small/medium business might purchase. This includes add-in PCIe cards. Both chips will support up to PCIe 3.0 x4 connectivity, going down to PCIe 2.0 x1 as required (peak rates will be limited in this mode, obviously).
On price comparisons to current gigabit solutions, Aquantia was unwilling to release the information given that individual contracts will govern the price but they were willing to say that a number of partners are already in production of their hardware. Aquantia wanted to make it clear that they will leave it up to their partners to market and promote the feature set of the controller. This will become important, especially as the motherboard PC networking controller market has a number of options.
Currently the big players are Intel and Realtek, using the I219-V/LM and RTL8111E/F families respectively. Certain versions of the controllers have more management functions, or prioritize some capabilities over the others, and key defining elements are both cost and consumer requests (distributors or clients). Rivet Networks is also a significant player with their Killer brand of networking ICs, focusing primarily on the gaming crowd with controller specific QoS features that are promoted to gamers and prosumers, and used extensively with MSI laptops and MSI/Gigabyte motherboards. Technically Realtek has the Dragon brand here, although we’ve only seen them on one generation of ECS motherboards so far. It’s important to note that only the Killer ICs come with any form of default controller software aimed at consumers for QoS and traffic shaping; Intel has management software at a more admin level. We asked Aquantia about additional consumer software, and was told that currently none is bundled but they are aware of the current market.
For now, the AQtion 2.5G/5G controllers coming to market look to be a premium component, destined for high-end notebooks/PCs, and if the pricing is right, more expansive than the current array of 10G integrated options. One of the issues Aquantia will have, which they also acknowledge, is the switch problem that currently stops 10G being more widespread – the lack of consumer grade and consumer budget level switches. We were told that there are some enterprise models of 2.5G/5G switches currently for more backbone type of work, and it will be up to Aquantia’s partners to spot opportunities in the consumer market. From a personal perspective, the switch side of the equation will be the slowest to change and be a defining aspect for the widespread adoption of this technology. We’ve seen this with 10G, or the fact that the Killer gaming NICs do not have corresponding switches/routers to assist in a number of features that might become irrelevant in a general network. Publicly Aquantia isn’t stating which switch developers they are working with, and as before, leaving those companies to decide/announce their product lines, but I think the switch aspect will be more important to watch over 2017.
On performance metrics, Aquantia have told us that the AQ107 can achieve 9.5 Gbps in each direction in the 10G mode with a CPU utilization of 12-20%, and in 5G mode it can do 4.6 Gbps in each direction with 6-14% CPU use. Due to the higher clock rate of the controller, in 1G mode the controller is quoted as having has lower latency than standard 1G controllers. The AQC107, in 5G mode, will have a typical power consumption around 3W when in use.
We were told that the chips support Wake-on-LAN and Energy Efficient Ethernet standards for power saving (wake-up in the microsecond range), as well as various offloads for reducing CPU use (LSO, LRO, header checksum, receive side scaling). QoS features allow for 32 queues/traffic in each direction, as well as rate shaping capabilities. Drivers will be available for all major flavors of Windows (7 to 10, 32/64-bit), Linux and macOS.
Aquantia stated that they are using their deep contacts at TSMC and GloFo for their product range, but stopped short of stating which process or which node they are using. Behind the scenes we have already have a sniff of upcoming products using Aquantia’s solution that are due to be announced over the next few months.
Post Your CommentPlease log in or sign up to comment.
View All Comments
willis936 - Thursday, December 15, 2016 - link10G is expensive and homes have been sitting on 1G for nearly 2 decades. It's believable that there will soon be a day where people want more than 1G networks. Wireless is convenient and the performance is impressive but it won't match the bandwidth of 8 copper lines.
azazel1024 - Thursday, December 15, 2016 - linkTo add another counter point, what do you think those wireless routers are connected with? 10GbE does exist, but it uses a lot more power and is more expensive. You are talking typically 5-7w per port. As mentioned in this article, the higher end chipset uses 3w for 5GbE and the lower end chipset is likely to use a little less.
I assume 2.5GbE is even less than that.
5-7w is quite a bit of power for a laptop, especially something like an ultrabook. So I doubt you'll be seeing 10GbE chipsets in laptops. You might see 2.5/5GbE chipsets in laptops soonish, especially higher end ones.
You also get in to the switch layer, 5-7w per port for something like a 16 port switch needs active cooling and a big power supply. 3w per port MIGHT be able to get away with good passive cooling. A 2.5GbE switch is almost certainly going to be able to get away with passive cooling.
I have little doubt that wireless routers are moving towards at least 2.5GbE ports. A current 4:4 MU:MIMO router in an ideal use case as a bridge can absolutely hit 1000Mbps over wireless, saturating the port. Even in operation as a multi-user device, it can likely hit 700-800Mbps in a good use case. As 802.11ac moves steadily closer to wide spread MU:MIMO clients as well as eventually 160MHz, you'll see a lot more instances where the Ethernet port on the wireless router is holding back wireless performance. Once 802.11ax comes along in a few years with it's promised increases in spatial efficiency and wide channels, a 1Gbps port is going to be a serious hindrance.
On a consumer router, a 10GbE NIC, let alone FIVE ports supported at that speed is a serious cost and power consideration. Your typical current Wifi router is looking at about 3-5w of power consumption for it's 5 GbE ports and another 3-15w for the SOC and radios. Moving to 10GbE means bumping the switching fabric up to consuming 25-35 watts, even with active cooling, not something you can fit in a typical router box, not with the 5-15w you have for the SOC and radios, let along the extra routing performance on a SOC that you'd need. Even 3w a port would be stretching it, but I bet you could easily shoe horn in NICs running in 2.5GbE mode and an upgraded routing SOC that can handle 2.5Gbps of throughput in a relatively ideal routing setup.
Sure, few people have connections like that to the internet, but it is like winter...it is coming. A few "elites" have actual gigabit internet connections and a handful of small providers are trialing and looking at 2Gbps internet connections in the US and around the world (my 100/100Mbps connection is jealous). A fair number of users need that performance at home.
Also, 10GbE works at a maximum distance of 55 meters over Cat6 and 45 meters over Cat5e. Max and under ideal use cases. You've got a bundle of Cat6 together running next to some power lines and telephone lines and I'd bet anything you probably couldn't get 30 meters on Cat6 for 10GbE.
5GbE can run on Cat6 in a typical installation (that includes bundles) at 100 meter distances, 2.5GbE on Cat5e to 100 meter distances. The home user can likely run 10GbE just fine, but the typical business user can't and frankly a few of my less than ideal runs in my house on Cat5e might not manage 10GbE. Like the line from my ONT box to my router for my internet connection is on Cat5e and it stretches about 25 meters, but near the line from my main panel to my sub-panel that actually provides most of the power to my house and it also runs for a distance next to both the telephone line (which is no longer in use though), some coax for cable (also not in use) and a couple of other network cables for some portion of it's run.
But 2.5GbE should be no sweat. In fact, it likely could support 5GbE over that distance and installation.
Use wireless all the time of course. I also have everything that can be possibly wired, wired.
Most of my stuff, 1GbE is frankly overkill. I don't need even a 100MbE connection for my airplay speakers that are wired in my network. My desktop and server on the other hand I saturate dual 1GbE links doing SMB multichannel for transferring files. My laptop that 1GbE NIC is a serious hindrance when the SSD in the laptop is easily capable of >500MB/sec reads and >300MB/sec writes...and no way to do SMB multichannel on it, so I can't even take advantage of my server's dual GbE NICs.
So I'd love to see even 2.5GbE switches hit the market sometime soon. They would likely be a LOT cheaper than 10GbE switches, which seem to be hovering around $75 a port for the cheapest I can find. NICs for desktops or on motherboards are even more than that per port.
I assume 5GbE at this point is likely to also be expensive, but still less. That higher end NIC mentioned I am sure, since it can do 10GbE, is going to hover in the $70-80 range for just the network interface module itself, not including board prices and mark-ups. However, it may promise slightly cheaper per port prices in something like a switch if the company also has larger switch modules in the pipeline. The lower end max 5GbE capable network interface might well be less than $50 and built as a 4 port switch module to go in a router/switch could promise fairly cheap per port prices.
A $30-50 price added on to the BOM of a high end laptop isn't ridiculous. On a high end desktop motherboard, that is reasonably affordable (much more so than 10GbE). A stand alone NIC, that is down right cheap. Instead of an new Intel GbE adapter at $80, you might be paying $120 for a 5GbE adapter. Again, on a switch, you might have an 8 port 5GbE switch that only costs $200-300, which is still expensive, but much more home user friendly than $600+ for 10GbE.
A more cut down 2.5GbE max network interface would probably a little cheaper.
Just saying, I can see a LOT of use cases for 2.5GbE and 5GbE. Ideally I'd love to see 10GbE, but the fact of the matter is, prices just aren't going down very fast on it and there will always be physical limitations to it. Even as power usage comes down on it, the same goes for 2.5GbE and 5GbE. We see a drop from 5w to 3w? You'd likely see 5GbE drop from 3w to 2w.
bill.rookard - Thursday, December 15, 2016 - linkIt's pretty funny that everyone goes around saying 'I want a 10GbE network' and think they can just drop a card into their system, hook up a 10GbE switch and it'll give them that speed. Not that simple at ALL.
There is something else to remember, and that's the simple limit of your average amount of bandwidth between your CPU and peripherals. Your average Haswell/Skylake build has 16pcie lanes off the CPU (graphics), and if that's in use, and your motherboard doesn't provide for an x8/x8 split set, plugging in a 10gb controller into the PCIe x4 off the PCH creates a potential bottleneck.
In fact, they don't recommend plugging in a high bandwidth device like that into the PCH because many of the lanes are hardwired into other motherboard resources (audio/GbE/USB/SATA etc) and share aggregate bandwidth and are still only connected to the CPU by the DMI interface (DMI 2.0 equivalent to PCIe x4 2.0/ DMI 3.0 equivalent to PCIe x4 3.0). You would actually need a high end board (SLI capable) to be able to utilize 10GbE and a high end GPU, which is yet another expense to adoption.
With this 5GbE option in a chip it would make life a little bit easier on adoption as it wouldn't suck up all the available bandwidth in the DMI connection to the CPU, but we're quickly hitting the point where the actual physical CPU design is becoming a limiting factor.
bobbozzo - Wednesday, December 14, 2016 - linkHi Ian, I'm not clear on this:
2.5GHz and 5GHz will work on Cat 5e cables, or they need Cat 6?
yuhong - Wednesday, December 14, 2016 - link5Gbit/s is more likely to require Cat 6 I think.
johnthacker - Wednesday, December 14, 2016 - linkIt's a similar situation to how 10GBASET can run on Cat 6, but only sort of supported up to 55m. IEEE 802.3bz promises 2.5GBASET on Cat 5e up to 100 m runs, 5GBASET on Cat 6 up to 100 m runs, and 5GBASET on Cat 5e "on defined use cases and deployment configurations" somewhat shorter than 100m.
Kamal Dalmia - Wednesday, December 14, 2016 - linkAccording to IEEE 802.3bz, both 2.5G and 5G are specified to run "up to" 100 meters of CAT5e or better cabling. We expect users to be able to achieve 5G over 100 meters of CAT5e over most SMB and consumer use cases.
Kamal Dalmia - Aquantia
Kamal Dalmia - Wednesday, December 14, 2016 - link2.5G and 5G Base-T do not require Cat 6 cabling. They are specified to run up to 100 meters of CAT 5e or better cabling.
Kamal Dalmia - Aquantia
cygnus1 - Wednesday, December 14, 2016 - link"10GBase-T also requires Cat 7 cabling for any serious distance requirements."
wat? Heard of Cat6a??? Both will do 100m for 10g
mariush - Wednesday, December 14, 2016 - linkAll we need now are wireless routers with up to 1866 mbps on wireless, 1gbps input for WAN (from ISP) , a couple or more 5 gbps ports and a bunch (4-8) of regular 1 gbps network ports.
This way you could cheaply connect a NAS device / computer / file server to the wireless router, and all the devices in your house (tv , phones, wired computers) can read media from the NAS at up to 5 gbps speed.
Someone needs to make a cheap 3-4 port 5gbps switch IC ... perfect to offer 2-3 x 5 gbps ports and use the other one to connect a 4-8 port regular gigabit switch IC to it on the same PCB.