There are two important trends in the server market: it is growing and it is evolving fast. It is growing fast as the number of client devices is exploding. Only one third of the world population has access to the internet, but the number of internet users is increasing by 8 to 12% each year. 

Most of the processing happens now on the server side (“in the cloud”), so the server market is evolving fast as the more efficient an enterprise can deliver IT services  to all those smartphones, tablets and pcs, the higher the profit margins and the survival chances.

And that is why there is so much interest in the new star, the “micro server”.  

The Demand for Micro-Servers

The demand for very low power server is, although it is high on the hype curve, certainly not imaginary. When you run heterogeneous workloads, for example a mailserver, an OLAP database and several web and fileservers, some of these workloads will be demanding heavy processing power, others will be close to idle. In that case it is best to buy a dual "large core" CPU (or better) server, install a hypervisor and cut the machine up in resource pools. As one workload demands high processing power, your hardware and hypervisor will deliver it. Single thread performance will determine for a large part whether a complex SQL query is responded in a fraction of a second or several seconds.  The virtualization layer gives you extra bonuses such as high availability, no more unplanned downtime etc.

If your web based workloads are very homogenous and you know how much horsepower your webapplication needs, things are very different. The virtualization layer is just adding complexity and cost now. In this case it is a lot more efficient to scale out than to divide your heavy server into virtual machines.  The single thread performance has to be good enough to respond to a request quickly enough. But throughput demands can be handled by adding a load balancer in front of low power servers. It is much easier to scale this way.

The problem is that the your average server is not well suited for these kind of homogenous workloads. Yes, servers have become a lot more efficient by including advanced power management features. The introduction of CPU C-states and more efficient PSUs are among the top of technologies that saved a lot of power. However, even the most modern servers are still needlessly complex with lots of disk, network and other interfaces. A useless interface wastes a few hundred of mwatt and a less efficient PHY (Copper 10 Gbit Ethernet for example) wastes a few Watt, but in the end it adds up. 

Low Power Server CPUs

When you read about micro servers, the spotlights are on very low power servers with Atom and ARM servers. The reality in the server market is however very different.  But as you probably understand from the reasoning above, a server with an Atom CPU or ARM CPU is hardly a good solution for the homogenous webfarms.

In fact, with the exception of Seamicro, many micro servers have utterly failed in that market. Quite a few server vendors have offered dense Atom based solutions, but those solutions were poorly received. Simply cramming tens of Atom based servers in a small chassis is a pretty bad idea:

1.       The single threaded performance of the Atom is even for webfarms too low

2.       The power gains that you make by using lower power CPUs are negated by the cable management costs, the higher amount of PCBs and PHYs. 

To sum it up: the performance per watt of those servers did not and will not thrill anyone. Seamicro, the pioneer of this niche market, was successful despite and not thanks to the Atom processor inside it’s micro servers.  When Seamicro offered Low Power Xeons instead of Atoms inside their Borg Cube inspired server, the demand for their products really took off.

And even Calxeda, the champion of the low power servers admits that the current “ultra low power but low performance” microserver market is a small one.  Calxeda has high expectations for its next generation of servers as they will be based on the more powerful Cortex A15 and A57 CPUs.

Seamicro and Calxeda succeeded where others fails as they understood that an optimized PCB and network fabric was necessary to make the concept of micro server work.  Seamicro reduced the serverboard to its bare minimum (“credit card size”), turned unnecessary features off and connected all I/O via a high performance 3D torus interconnect.

Calxeda integrated several servers on one PCB and connected them together with a 2D torus network fabric. The result was a low power draw per server, not just per CPU.  Once you add a CPU with good enough single threaded performance, things get very interesting.

Time for New Server CPUs!

While the network fabric and optimized motherboard designs were the key features of the micro server, it is clear that there is room for a “specialized micro server CPU” between the current anemic Atoms and the low power Xeons and Opterons.  AMD announced last week the quadcore Opteron-X series, based upon the Jaguar core.   The quadcore X1150 is claimed to perform twice as fast as the current Atoms at 2 GHz, but needs 17W to achieve this. You can lower the power usage by fiddling with a p-state cap in the BIOS. Unfortunately the 9W number that so many publications talked about without further commentary is only achieved at 1 GHz. At that clockspeed the performance per Watt advantage will be negligible compared to the 2 GHz Atom S1260 at 8.5W. The best performance/watt will be achieved somewhere between 1.5 and 2 GHz, but the advantage that the new Opteron-X has over the Atom is not as large as many people thought. Unless of course you can make use of the floating point processing power of the integrated Radeon core in the Opteron X-1250 APU.

But although the new Opteron-X can not offer a massive performance/watt improvement over the Atom, it is a much more attractive micro server chip. It can deliver the “good enough” horsepower that an Atom can not deliver. But then again, the current low power Ivy Bridge based Xeon, the Xeon E3-1220LV2 was already a very good micro server chip.  But those who think that AMD will be contend with an underdog role once again, are wrong.  AMD has a pretty ambitious roadmap to attack this market.

Meet AMD's Berlin: Radeon + Steamroller

The current Opteron 4310 EE (2 modules, 4 cores at 2.2-3 GHz, 40W TDP) and Opteron 4376 HE (4 modules, 8 cores at 2.6-3.6 GHz, 65W TDP) are about the best AMD can deliver for low power servers that need more processing power. These will be replaced by AMD’s Berlin CPU, the first Opteron with the new Steamroller cores.

The Quadcore chip integrates a next generation Radeon core, PCI-e 2.0/3.0, LPC, USB 2.0/3.0 and Serial-ATA controller.

AMD says that using the graphics core for the heavy scalar floating point will get as easy as C++ programming and as a result, Berlin should make a few heads turn in the HPC world. It even looks like SSE-x will get less and less important over time in that market. Andrew Feldman told us that Berlin will offer at least two time much CPU processing performance as the Opteron X-series. Although the TDP will be higher, the performance per Watt will be quite a bit better.  Berlin will use the same 28 nm process technology as the Opteron x1150 and x2150. This chip will be available in the first half of 2014, and might experience some stiff competition from Haswell based Xeons.

Seattle: The Atom Killer?

The chip that has the potential to give Intel some real headaches is “Seattle”. It is a pretty revolutionary design for being an AMD CPU.  No less than 8 or 16 ARM Cortex A57 are inside this new AMD lower power server SoC.  The 28 nm SoC also integrates a 10Gbe controller, a SATA controller (high port count), encryption and compression module. But the real kicker is that this SoC will integrate some of the best Seamicro technology such as TIO (Turn It Off, reducing power by shutting down unnecessary interfaces) and the high performance Seamicro Freedom Fabric.

Single threaded performance will be similar to the Opteron X1150, but throughput should up to 4 times higher. There is little doubt in our minds that this might well be one of the best micro server CPU of 2014 (based upon the paper specs). It looks like the Intel Avoton will have a very potent challenger in Q1 2014.

Better Piledriver

It is clear that the micro server market gets the lion's share of AMD’s attention. However, the current piledriver based Opteron 6300 gets a small facelift in Q1 of 2014. Apparantly both the core and uncore have received quite a few minor tweaks, resulting in lower TDPs and a better performance/watt. This CPU with 12 or 16 “Piledriver Enhanced” cores is called “Warsaw”.

Conclusion

The Opteron-X, Opteron 6300 and “Berlin” CPU will all face stiff competition from the Intel alternatives. The integrated GPU of Berlin will make it very attractive for the HPC market, but it looks like Intel will probably have the upper hand in most of the traditional server markets.

However,  the combined AMD, ARM and Seamicro technology inside AMD’s new Seattle CPU look extremely promising: these are probably the best specs of a micro server CPU we have seen so far. And since all the right components are now in place, it looks like the micro server is ready for prime time. There is little doubt that Seamicro servers will continue to thrive in their niche market while HP's Moonshot and Dell's Viking will make the market much more popular. So there is good chance that AMD will make a big comeback in 2014 in the server market.