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
POST A COMMENT

112 Comments

View All Comments

  • SkipPerk - Friday, April 8, 2016 - link

    "Anyone putting Microsoft on bare hardware these days is nuts"

    This brother is speakin the truth!
    Reply
  • warreo - Thursday, March 31, 2016 - link

    Can someone clarify this line for me?

    "The average performance increase versus the Xeon E5-2690 is 3%, and the Broadwell cores get a boost of no less than 19%."

    Does that mean IPC increase is 19% for Broadwell, offset by ~16% decline in clockspeed to get to 3% average performance increase? But that doesn't make sense to me as a 3.8ghz (E5-2690) to 3.6ghz (E5-2699 v4) is only 5% decline in max clockspeed?
    Reply
  • ShieTar - Thursday, March 31, 2016 - link

    I understood it as "the -Ofast setting boosts Broadwell by 19%", so with the -O2 setting it was actually 16% slower than the 2690.

    And I think the AT-Theory based on the original measurements is that the 3.6GHz boost are not even held for a significant amount of time, so that Broadwell in reality comes with an even worse decline in clock speed.
    Reply
  • warreo - Thursday, March 31, 2016 - link

    Your interpretation makes much more sense than mine, but still doesn't quite add up. The improvement from using -Ofast vs. -O2 is 13% on average, and the lowest improvement is 4% on the xalancbmk, well below the "no less than 19%" quoted by Johan.

    Perhaps the rest of the disparity is normalizing for sustained clock speeds as you suspect? Johan is that correct?
    Reply
  • Ryan Smith - Thursday, March 31, 2016 - link

    I've reworded that passage to make it clearer. But ShieTar's interpretation was basically correct.

    "Switching from -O2 to -Ofast improves Broadwell-EP's absolute performance by over 19%. Meanwhile the relative performance advantage versus the Xeon E5-2690 averages 3%. "
    Reply
  • JohanAnandtech - Thursday, March 31, 2016 - link

    That means that the -ofast has much more effect on the Broadwell. I mean by that that -ofast is 19% faster than -o2 on Broadwell, while it is 3% faster on Sandy Bridge. I assume that the older the architecture, the better the compiler is able to optimize it without special tricks. Reply
  • warreo - Friday, April 1, 2016 - link

    Thanks for the clarification. Loved the review, great work Johan! Reply
  • Pinn - Thursday, March 31, 2016 - link

    I'm still happy I went with the 6 core x99 over the 8 core. Massive core count is nice to see available, but I don't see the true value. Looks like you have to do the same rough math to see if the clock speed reduction is worth the core count. Reply
  • Oxford Guy - Tuesday, April 5, 2016 - link

    Why would there be "true value" for six and not for eight? Reply
  • Pinn - Wednesday, April 6, 2016 - link

    Single threaded workloads. Reply

Log in

Don't have an account? Sign up now