The Intel Skylake-X Review: Core i9 7900X, i7 7820X and i7 7800X Tested
by Ian Cutress on June 19, 2017 9:01 AM ESTComparing Skylake-S and Skylake-X/SP Performance Clock-for-Clock
If you’ve read through the full review up to this point (and kudos), there should be three things that stick in the back of your mind about the new Skylake-SP cores: Cache, Mesh and AVX512. These are the three main features that separate the consumer grade Skylake-S core from this new core, and all three can have an impact in clock-for-clock performance. Even though the Skylake-S and the Skylake-SP are not competing in the same markets, it is still poignant to gather how much the changes affect the regular benchmark suite.
For this test, we took the Skylake-S based Core i5-6600 and the Skylake-SP based Core i9-7900X and ran them both with only 4 cores, no hyperthreading, and 3 GHz on all cores with no Turbo active. Both CPUs were run in high performance modes in the OS to restrict any time-to-idle, so it is worth noting here that we are not measuring power. This is just raw throughput.
Both of these cores support different DRAM frequencies, however: the i5-6600 lists DDR4-2133 as its maximum supported frequency, whereas the i9-7900X will run at DDR4-2400 at 2DPC. I queried a few colleagues as to what I should do here – technically the memory support is an extended element of the microarchitecture, and the caches/uncore/untile will be running at different frequencies, so how much of the system support should be chipped away for parity. The general consensus was to test with the supported frequencies, given this is how the parts ship.
For this analysis, each test was broken down in two ways: what sort of benchmark (single thread, multi-thread, mixed) and what category of benchmark (web, office, encode).
For the single threaded tests, results were generally positive. Kraken enjoyed the L2, and Dolphin emulation had a good gain as well. The legacy tests did not fair that great: 3DPM v1 has false sharing, which is likely taking a hit due to the increased L2 latency.
On the multithreaded tests, the big winner here was Corona. Corona is a high-performance renderer for Autodesk 3ds Max, showing that the larger L2 does a good job with its code base. The step back was in Handbrake – our testing does not implement any AVX512 code, but the L3 victim cache might be at play here over the L3 inclusive cache in SKL-S.
The mixed results are surprising: these tests vary with ST and MT parts to their computation, some being cache sensitive as well. The big outlier here is the compile test, indicating that the Skylake-SP might not be (clock for clock) a great compilation core. This is a result we can trace back to the L3 again, being a smaller non-inclusive cache. In our results database, we can see similar results when comparing a Ryzen 7 1700X, an 8-core 95W CPU with 16MB of L3 victim cache, is easily beaten by a Core i7-7700T, with 4 cores at 35W but has 8MB of inclusive L3 cache.
If we treat each of these tests with equal weighting, the overall result will offer a +0.5% gain to the new Skylake-SP core, which is with the margin of error. Nothing too much to be concerned about for most users (except perhaps people who compile all day), although again, these two cores are not in chips that directly compete. The 10-core SKL-SP chip still does the business on compiling:
If all these changes (minus AVX512) offer a +0.5% gain over the standard Skylake-S core, then one question worth asking is what was the point? The answer is usually simple, and I suspect involves scaling (moving to chips with more cores), but also customer related. Intel’s big money comes from the enterprise, and no doubt some of Intel’s internal metrics (as well as customer requests) point to a sizeable chunk of enterprise compute being L2 size limited. I’ll be looking forward to Johan’s review on the enterprise side when the time comes.
Facebook
Twitter
RSS
264 Comments
View All Comments
rascalion - Monday, June 19, 2017 - link
Are the Ryzen numbers in the charts retests using the last round of bios and software updates?Ian Cutress - Monday, June 19, 2017 - link
As much as possible, the latest BIOSes are used.Our CPU testing suite is locked in for software versions as of March 2017. This is because testing 30/50/100+ CPUs can't be done overnight, we have to have rigid points where versions are locked in. My cycle is usually 12-18 months. (Note I'm only one person doing all this data.)
FreckledTrout - Monday, June 19, 2017 - link
Ian any chance once there are a few BIOS tweaks you could say do a mini updated review on the 7820x vs Ryzen 1800x. With Ryzen having latest BIOS as well plus 3200Mhz memory. I'm just curious really how the 8-core guys line up when some of the dust settles and I think a lot of people will be.Ian Cutress - Monday, June 19, 2017 - link
Any reason why 3200? I'll have Intel people saying it is pushing the Ryzen out of specjjj - Monday, June 19, 2017 - link
You could do a memory subsystem scaling review for all platforms, Skylake X, Threadripper, Ryzen (Summit Ridge) and Coffee Lake. Cache, interconnect, DRAM. See where they are, how they scale, where the bottlenecks are, single rank vs dual rank modules and perf impact in practice.Why not even impact on power and efficiency.In any case, you'll need to update Ryzen 5 and 7 results when Ryzen 3 arrives , isn't it?
For DRAM at 3200 it might be out of spec - overclocking the core is out of spec too but that has never stopped anyone from overclocking the memory. Right now 3200 is what a lot of folks buy , at least for higher end mainstream Ofc some will argue that Ryzen scales better with memory and that's why it is unfair but it's a hell of a lot more reasonable than testing 1080p gaming with a 1080 TI since it's a popular real world scenario.
At the end of the day the goal should be to inform, not to watch out for Intel's or AMD''s feelings.
vanilla_gorilla - Monday, June 19, 2017 - link
>For DRAM at 3200 it might be out of spec - overclocking the core is out of spec too but that has never stopped anyone from overclocking the memory.This. Exactly. We're enthusiasts and we always push the envelope. No one cares what the specs are all we care is about what these processors are capable of in the right hands.
And Ian I think you guys do an awesome job, there's no other place I look for CPU benchmarks. Keep up what you do, we all appreciate it, as well as your willingness to have a dialog with us about the process. Really cannot say how impressed I am by how open and engaged you are, it's really commendable.
Ian Cutress - Tuesday, June 20, 2017 - link
Thanks for the comments :)Though on your comments about pushing things out of spec. We have a good deal of readers who want plain stock for their businesses - AT isn't solely a consumer focused site. Otherwise I'd just jack all the CPUs and just post OC results :D Our base testing will always be at stock, and for comparison testing there has to be an element of consistency - testing an OC'ed part against a stock part in a direct A vs B comparison is only going to end up with a barrage of emails being rammed down my throat. There has to be some planning involved.
Ian Cutress - Tuesday, June 20, 2017 - link
I've been planning a memory scaling article, I just haven't had the time (this article was around 6 weeks of prep with all the events going on that I had to attend).Note we don't retest stuff every review. With our new 2017 test suite, I've been going through regression testing. Usually regression testing is done once for the full segment until the benchmarks are changed again. I'll look at my next few months (still stupidly busy) and look at the priorities here.
FreckledTrout - Monday, June 19, 2017 - link
Most people can easily buy a 3200 kit for not a lot of extra money. It doesn't take a lot tweaking(well not anymore on AGESA 1.0.0.6) or silicone lottery like an OC, just a bit more cash. From what I have seen with Ryzen it is the sweet spot on price and performance. I would assume Its likely the most chosen configuration on the R7's. To make it fair use 3200 on the 7820x as well. I only ask because Ryzen did way better than I would have thought and would like to see it with 3200Mhz memory and latest updates to see really how close Intel and AMD are on 8-core systems. Then im going to build :)tipoo - Monday, June 19, 2017 - link
Launch review! Nice work dude(s).