SPEC2017 Multi-Threaded Results

Single-threaded performance is only one element in regard to performance on a multi-core processor, and it's time to look at multi-threaded performance in SPEC2017. Although things in the single-threaded SPEC2017 testing showed that both Zen 4 and Raptor Lake were consistently at loggerheads, let's look at data in the Rate-N multi-threaded section.

SPECint2017 Rate-N Estimated Scores

Looking at the data in our first part of SPEC2017 (int) nT testing, we're seeing similar trade-offs between Raptor Lake (13th Gen) and Zen 4 (Ryzen 7000) platforms. While Raptor Lake won in the 500.perlbench_r single-threaded test, Zen 4 has the lead by around 13% in multi-threaded performance, despite the Core i9-13900K having eight more physical cores (albeit efficiency cores).

One stand-out part of our SPECint2017 multi-threaded testing is just how far ahead the Core i9-13900K is ahead of the previous Core i9-12900K in multi-threaded tests. This comes thanks to more cores (2x the E-cores), and higher turbo frequencies. For example, in the 525.x264_r test, the Core i9-13900K is nearly 50% better than the i9-12900K; the only part where Raptor Lake failed to outperform Alder Lake was in the 502.gc_r test.

SPECfp2017 Rate-N Estimated Scores

Moving onto the second half of our SPEC2017 multi-threaded results (Floating Point), the Core i9-13900K really does show itself to be a formidable force compared to Zen 4. In the majority of SPECfp2017 tests, the Core i9-13900K is ahead in multi-threaded performance. The improvements in overall performance from Rocket Lake (11th Gen) to Alder Lake were decent, but the improvement from Rocket Lake last year in Q1 2021 to Raptor Lake today – a more useful metric for the usual 2-4 year hardware upgrade cycle – is very impressive indeed.

Summarizing the SPEC2017 multi-threaded results, in some areas Zen 4 is the winner, some areas Raptor Lake (Raptor Cove + Gracemont) is the winner. It is incredibly close in quite a few of the tests, and without sounding negative on the Zen 4 architecture here, but Intel has done a very good job bridging that initial gap to make things competitive against AMD's best.

Update 18/07/23

Following on from our initial results in our SPECint2017 Rate multi-threaded testing of the Core i9-13900K, we wanted to investigate the 502.gcc_r result, which we believed to be an anomaly of sorts. We observed a score of 49.1 on the Core i9-13900K, which, compared to the previous generation Core i9-12900K, is a regression in performance; the result was around 37% lower than the previous generation. 

To investigate further, we've re-tested the Core i9-13900K using SPEC2017 Rate to identify any issues and to see if we could further replicate the issue or, at the very least, provide a more up-to-date list of results.

SPECint2017 Rate-N Estimated Scores (i9-13900K)

Looking at our updated SPECin2017 results, we are comparing the original Core i9-13900K data to the new data for comparative purposes. Although the results are very similar in many cases, we can see some slight regression in a few results, which could be attributed to various factors, including Windows 11's scheduler, power budget, or just general variance in running.

The biggest highlight of our re-test is the 502.gcc_r result, which seems to be an anomaly for the original run. We've run SPEC2017 numerous times to confirm that the above results are exactly where they should be.

SPECfp2017 Rate-N Estimated Scores (i9-13900K)

Focusing on our latest results for SPECfp2017 Rate N, we can see a similar story, with very similar results in multi-threaded SPEC2017 performance as with our original testing. In fact, a couple of the results yielded slightly higher results, which could be simply down to scheduler maturity, OS-related improvements including the scheduler, or overall firmware maturity. The results include 527.cam4_r, 549.fotonik3d_r, and 554.roms_r, which show better gains in our latest testing, especially compared to the Core i9-12900K, which this chip replaced in the market.

The biggest takeaway from our re-testing is the updated SPECint2017 Rate-N result for 502.gcc_r, which shows that our original results were nothing more than an anomaly, and we've been unable to replicate the issue.

Update: 07/22/23

We are aware of potential issues with memory capacity, and as such, we are re-running the Core i9-13900K with higher capacity DDR5 memory at JEDEC settings relevant to the platform. We have done a run with 64 GB instead of the regular 32 GB, which with the Core i9-13900K is 2 GB per thread (2 GB x 32 = 64 GB). Looking at preliminary results, we aren't seeing any major variances in these results.

SPEC2017 Single-Threaded Results CPU Benchmark Performance: Power, Office And Web
POST A COMMENT

169 Comments

View All Comments

  • Pjotr - Thursday, October 20, 2022 - link

    Closing thoughts typos: Ryzen 580X3D and Ryzen 700. Reply
  • Ryan Smith - Thursday, October 20, 2022 - link

    Thanks! Reply
  • mode_13h - Thursday, October 20, 2022 - link

    Thanks for the review!

    Could you please add the aggregates, in the SPEC 2017 scores? There's usually a summary chart that has an average of the individual benchmarks, and then it often has the equivalent scores from more CPUs/configurations than the individual test graphs contain. For example, see the Alder Lake review:

    https://www.anandtech.com/show/17047/the-intel-12t...
    Reply
  • Arbie - Thursday, October 20, 2022 - link

    TechSpot / Hardware Unboxed show that to complete a Blender job the 13900K takes 50% more total system energy than does the 7950X. Intel completing a Cinebench job takes 70% more energy. Meaning heat in the room. And that's with the Intel chip thermal throttling instantly on even the best cooling.

    Looking at AT's "Power" charts here, which list the Intel chip as "125W" and AMD as "170W", many readers will get EXACTLY THE OPPOSITE impression.

    Sure, you mention the difficulties in comparing TDPs etc, and compare this gen Intel to last gen etc but none of that "un-obscures" the totally erroneous Intel vs AMD picture you've conveyed.

    ESPECIALLY when your conclusion says they're "very close in performance" !! BAD JOB, AT. The worst I've seen here in a very long time. Incomprehensibly bad.
    Reply
  • gezafisch - Thursday, October 20, 2022 - link

    Cope harder - watch Der8auer's video showing that the 13900k can beat any chip at efficiency with the right settings - https://youtu.be/H4Bm0Wr6OEQ Reply
  • Ryan Smith - Thursday, October 20, 2022 - link

    We go into the subject of power consumption at multiple points and with multiple graphs, including outlining the 13900K's high peak power consumption in the conclusion.

    https://images.anandtech.com/graphs/graph17601/130...

    Otherwise, the only place you see 125W and 170W are in the specification tables. And those values are the official specifications for those chips.
    Reply
  • boeush - Thursday, October 20, 2022 - link

    Not true. You have those insanely misleading "TDP" labels on every CPU in the legend of every performance comparison chart. This paints a very misleading picture of "competitive" performance, whereas performance at iso-power (e.g. normalized per watt, based on total system power consumption measured at the outlet) would be much more enlightening. Reply
  • boeush - Thursday, October 20, 2022 - link

    *per watt-hour (not per watt)

    [summed over the duration of the benchmark run]
    Reply
  • dgingeri - Thursday, October 20, 2022 - link

    Is it just me, or does the L1 cache arrangement seem a bit odd? 48k data and 32k instruction for the P cores and 32k data and 64k instruction on the e-cores. Seems a bit odd to me. Reply
  • Otritus - Thursday, October 20, 2022 - link

    Golden/Raptor Cove has a micro-op cache for instructions. 4096 micro-ops is about equal to 16Kb of instruction cache, which is effectively 48Kb-D + 48Kb-I. I don’t remember whether Gracemont has a micro-op cache. However, it doesn’t have hyperthreading, so maybe it just needs less data cache per core. Reply

Log in

Don't have an account? Sign up now