Broadwell-EP: A 10,000 Foot View

What are the building blocks of a 22-core Xeon? The short answer: 24 cores, 2.5 MB L3-cache per core, 2 rings connected by 2 bridges (s-boxes) and several PCIe/QPI/home "agents". 

The fact that only 22 of those 24 cores are activated in the top Xeon E5 SKU is purely a product differentiation decision. The 18 core Xeon E5 v3 used exactly the same die as the Xeon E7, and this has not changed in the new "Broadwell" generation.  

The largest die (+/- 454 mm²), highest core (HCC) count SKUs still work with a two ring configuration connected by two bridges. The rings move data in opposite directions (clockwise/counter-clockwise) in order to reduce latency by allowing data to take the shortest path to the destination. The blue points indicate where data can jump onto the ring buses. Physical addresses are evenly distributed over the different cache slices (each 2.5 MB) to make sure that L3-cache accesses are also distributed, as a "hotspot" on one L3-cache slice would lower performance significantly. The L3-cache latency is rather variable: if the core is lucky enough to find the data in its own cache slice, only one extra cycle is needed (on top of the normal L1-L2-L3 latency). Getting a cacheline of another slice can cost up to 12 cycles, with an average cost of 6 cycles..

Meanwhile rings and other entities of the uncore work on a separate voltage plane and frequency. Power can be dynamically allocated to these entities, although the uncore parts are limited to 3 GHz.

Just like Haswell-EP, the Broadwell-EP Xeon E5 has three different die configurations. The second configuration supports 12 to 15 cores and is a smaller version (306mm²) of the third die configuration that we described above. These dies still have two memory controllers.

Otherwise the smallest 10 core die uses only one dual ring, two columns of cores, and only one memory controller. However, the memory controller drives 4 channels instead of 2, so there is a very small bandwidth penalty (5-10%) compared to the larger dies (HCC+MCC) with two memory controllers. The smaller die has a smaller L3-cache of course (25 MB max.). As the L3-cache gets smaller, latency is also a bit lower.

Cache Coherency

As the core count goes up, it gets increasingly complex to keep cache coherency. Intel uses the MESIF (Modified, Exclusive, shared, Invalid and Forward) protocol for cache coherency. The Home Agents inside the memory controller and the caching agents inside the L3-cache slice implement the cache coherency. To maintain consistency, a snoop mechanism is necessary. There are now no less than 4 different snoop methods.

The first, Early Snoop, was available starting with Sandy Bridge-EP models. With early snoop, caching agents broadcast snoop requests in the event of an L3-cache miss. Early snoop mode offers low latency, but it generates massive broadcasting traffic. As a result, it is not a good match for high core count dies running bandwidth intensive applications.

The second mode, Home Snoop, was introduced with Ivy Bridge. Cache line requests are no longer broadcasted but forwarded to the home agent in the home node. This adds a bit of latency, but significantly reduces the amount of cache coherency traffic.

Haswell-EP added a third mode, Cluster on Die (CoD). Each home agent has 14 KB directory cache. This directory cache keeps track of the contested cache lines to lower cache-to-cache transfer latencies. In the event of a request, it is checked first, and the directory cache returns a hit, snoops are only sent to indicated (by the directory cache) agents.

On Broadwell-EP, the dice are indeed split along the rings: all cores on one ring are one NUMA node, all other cores on the other ring make the second NUMA node. On Haswell-EP, the split was weirder, with one core of the second ring being a member of the first cluster. On top of that, CoD splits the processor in two NUMA nodes, more or less one node per ring.


The fourth mode, introduced with Broadwell EP, is the "home snoop" method, but improved with the use of the directory cache and yet another refinement called opportunistic snoop broadcast. This mode already starts snoops to the remote socket early and does the read of the memory directory in parallel instead of waiting for the latter to be done. This is the default snoop method on Broadwell EP. 

This opportunistic snooping lowers the latency to remote memory.

These snoop modes can be set in the BIOS as you can see above.

Broadwell Reaches Xeon E5 Broadwell Architecture Improvements
Comments Locked


View All Comments

  • ltcommanderdata - Friday, April 1, 2016 - link

    Does anyone know the Windows support situation for Broadwell-EP for workstation use? Microsoft said Broadwell is the last fully supported processor for Windows 7/8.1 with Skylake getting transitional support and Kaby Lake will not be supported. So how does Broadwell-EP fit in? Is it lumped in with Broadwell and is fully supported or will it be treated like Skylake with temporary support until 2018 and only critical security updates after that? And following on will Skylake-EP see any Windows 7/8.1 support at all or will it not be supported since it'll presumably be released after Kaby Lake?
  • extide - Friday, April 1, 2016 - link

    When MS says they are not supporting Skylake on Windows 7 DOES NOT MEAN it won't work. It just means they are not going to add any specific support for that processor in the older OS's. They are not adding in the speed shift support, essentially.

    For some reason the press has not made this very clear, and many people are freaking out thinking that there will be a hard break here will stuff will straight up not work. That is not the case.

    Broadwell has no new OS level features over Haswell (unlike Skylake with speed shift) so there is nothing special about Broadwell to the OS. As the poster above mentions, they are all x86 cpu's and will all still work with x86 OS's.

    The difference here is between "Fully Supported" and Compatible. Skylake and even Kaby Lake will be compatible with WIndows 7/8/8.1.
  • aryonoco - Friday, April 1, 2016 - link

    Johan, this is yet again by far the best Enterprise CPU benchmark that's available anywhere on the net.

    Thank you for your detailed, scientific and well documented work. Works like this are not easy, I can only imagine how many man hours (weeks?) compiling this article must have taken. I just want you to know that it's hugely appreciated.
  • JohanAnandtech - Friday, April 1, 2016 - link

    Great to read this after weeks of hard work! :-D
  • fsdjmellisse - Friday, April 1, 2016 - link

    hello, i want to buy E5-2630L v4
    any one can give me website for buy it ?

    Best regards
  • HrD - Friday, April 1, 2016 - link

    I'm confused by the following:

    "The following compiler switches were used on icc:

    -fast -openmp -parallel

    The results are expressed in GB per second. The following compiler switches were used on icc:

    -O3 –fopenmp –static"

    Shouldn't one of these refer to icc and the other to gcc?
  • JohanAnandtech - Friday, April 1, 2016 - link

    Pretty sure I did not mix them up. "-fast" does not work on gcc neither does -fopenmp work on icc.
  • patrickjp93 - Friday, April 1, 2016 - link

    Um, wrong and wrong. -Ofast works with GCC 4.9 and later for sure. And -fopenmp is a valid ICC flag post-ICC 13.
  • JohanAnandtech - Saturday, April 2, 2016 - link

    "-fast" is a typical icc flag. (I did not write -"Ofast" that works on gcc 4.8 too)
  • extide - Friday, April 1, 2016 - link

    Johan, if you read the comment, you can see that you mention icc for BOTH.

Log in

Don't have an account? Sign up now