SPEC - Single-Threaded Performance

Starting off with SPECint2017, we’re using the single-instance runs of the rate variants of the benchmarks.

As usual, because there are not officially submitted scores to SPEC, we’re labelling the results as “estimates” as per the SPEC rules and license.

We compile the binaries with GCC 10.2 on their respective platforms, with simple -Ofast optimisation flags and relevant architecture and machine tuning flags (-march/-mtune=Neoverse-n1 ; -march/-mtune=skylake-avx512 ; -march/-mtune=znver2).

While single-threaded performance in such large enterprise systems isn’t a very meaningful or relevant measure, given that the sockets will rarely ever be used with just 1 thread being loaded on them, it’s still an interesting figure academically, and for the few use-cases which would have such performance bottlenecks. It’s to be remembered that the EPYC and Xeon systems will clock up to respectively 3.4GHz and 4GHz under such situations, while the Ampere Altra still maintains its 3.3GHz maximum speed.

SPECint2017 Rate-1 Estimated Scores

In SPECint2017, the Altra system is performing admirably and is able to generally match the performance of its counterparts, winning some workloads, while losing some others.

SPECfp2017 Rate-1 Estimated Scores

In SPECfp2017 the Neoverse-N1 cores seem to more generally fall behind their x86 counterparts. Particularly what’s odd to see is the vast discrepancy in 507.cactuBSSN_r where the Altra posts less than half the performance of the x86 cores. This is actually quite odd as the Graviton2 had scored 3.81 in the test. The workload has the highest L1D miss rate amongst the SPEC suite, so it’s possible that the neutered prefetchers on the Altra system might in some way play a more substantial role in this workload.

SPEC2017 Rate-1 Estimated Total

The Altra Q80-33 ends up performing extremely well and competitively against the AMD EPYC 7742 and Intel Xeon 8280, actually beating the EPYC in SPECint, although it loses by a larger margin in SPECfp. The Xeon 8280 still holds the crown here in this test due to its ability to boost up to 4GHz across two cores, clocking down to 3.8, 3.7, 3.5 and 3.3GHz beyond 2, 4, 8 and 20 cores active.

The Altra showcases a massive 52% performance lead over the Graviton2 in SPECint, which is actually beyond the expected 32% difference due to clock frequencies being at 3.3GHz versus 2.5GHz. On the other hand, the SPECfp figures are only ahead of 15% for the Altra. The prefetchers are really amongst the only thing that come to mind in regards to these differences, as the only other difference being that the Graviton2 figures were from earlier in the year on GCC 9.3. The Altra figures are definitely more reliable as we actually have our hands on the system here.

While on the AMD system the move from NPS1 to NPS4 hardly changes performance, limiting the Altra Q80-33 from a monolithic setup to a quadrant setup does incur a small performance penalty, which is unsurprising as we’re cutting the L3 into a quarter of its size for single-threaded workloads. That in itself is actually a very interesting experiment as we haven’t been able to do such a change on any prior system before.

Test Bed and Setup - Compiler Options SPEC - Multi-Threaded Performance
Comments Locked

148 Comments

View All Comments

  • Silver5urfer - Friday, December 18, 2020 - link

    25% more cores for Zen2 7742 class. If paired with multi socket and then Milan drop in this is not going to be any major breakthrough.

    "The Arm server dream is no longer a dream, it’s here today, and it’s real." lol so until today all the articles on the ARM are not real I guess.

    Anyways I will wait for market penetration of this with server share and then see how great ARM is and how bad x86 is going to be as from AT's narrative recently.
  • Spunjji - Monday, December 21, 2020 - link

    Are you this mopey every time there's a paradigm-shift in the tech industry? Feel free to keep looking for metrics that "prove" you right, but eventually it's going to be a very hard search.
  • eastcoast_pete - Friday, December 18, 2020 - link

    Thanks Andrei! Maybe I am barking up the wrong tree here, but I find the "baby" server chip in that lineup particularly interesting. Nowhere near as fast as this, of course, but for $ 800, it might make for a nice CPU for a basic server setup; nothing fancy, but low TdP, and would probably get the job done. The question here is how expensive the MB for those would be.
    Lastly, if Ampere sends you one of those $ 800 ones, could/would you test it?
  • Wilco1 - Friday, December 18, 2020 - link

    They will likely sell desktops using these just like the previous generation, but they are not cheap as it is high-end server gear using expensive ECC memory (and lots of it since there are 8 channels). If you don't need the fastest then there is eg. NVIDIA Xavier or LX2160A (16x A72) boards for around $500.
  • Spunjji - Monday, December 21, 2020 - link

    I think those are probably most useful for workloads that are pathologically memory and/or I/O limited - 4TB per socket, save ~$3000 over the faster CPU, benefit from power savings over the life of the server.
  • twtech - Friday, December 18, 2020 - link

    Ironically, AMD's opportunity to win might turn into an ultimate loss - Intel's manufacturing advantage kept x86 relevant, and with access to the x86 instruction set limited by ownership of the IP, AMD lived alongside Intel in that walled garden.

    With the manufacturing advantage gone however, Apple has left the garden, and maybe other personal computers won't be far behind - software compatibility I think is actually less of an issue in the era of SaaS and continuous updates. Ie. you were going to have to download new versions of the software you use as time went on anyway.
  • FunBunny2 - Friday, December 18, 2020 - link

    "you were going to have to download new versions of the software you use as time went on anyway."

    Solar Wind? :)
  • lorribot - Friday, December 18, 2020 - link

    This is all great but when all licencing is per core it limits the usage scenarios or benefits of these developments as they can really only be used with open source type licences.
    For the rest of us on Windows, Oracle, Java, Apple, IBM, etc licencing it doesn't bring anything to the table.
  • The_Assimilator - Friday, December 18, 2020 - link

    Just in time to be obsoleted by Milan.
  • Spunjji - Monday, December 21, 2020 - link

    For a given definition of "obsoleted", where it means "still more than competitive in performance per dollar at a lower price of entry".

Log in

Don't have an account? Sign up now