The rocky tale of the Core i9!
The announcement of the Core i9-9900K sure did raise every tech journalist and reviewer’s eyebrows, except the one who flew from India for the l(a)unch. When the original report from Principled Technologies came out, many inconsistencies were found. Naturally, it made the Intel Core i9-9900K look like the next best thing with a difference of up to 50% against Ryzen 7 2700X. This is even more serious as Intel stifled reviewers with an embargo while having this report out in the open and have its pre-order options available. Obviously, Intel was banking on using this report as a way to justify its pricing. It was trying to pull an RTX 20 series on the CPU for all of us!
But thanks to the number of details in settings and benchmarks, we were able to catch multiple discrepancies. Whether Principled Technologies’ original attempts were intentional or not was debatable until its first response. They did acknowledge some feedback and promised to provide the revised version.
Is #GameModeGate a thing?
AMD Ryzen Master utility’s Game Mode was one of the settings that low-blowed Ryzen 7 2700X since it was essentially made for Threadripper platform to half its core/thread count for optimal gaming performance. But it was used on the mainstream AMD Ryzen 2 2700X making the 8-core-16 thread CPU into a quad-core-8 thread processor. In all these game benchmarks, the comparative results looked awfully strange for an R7 2700X. Hence what started the flow of reporting and investigation.
Ideally, AMD should have disabled Game Mode on mainstream Ryzen processors since its not intended for these processors. Not really a big deal but…it can if it wants to.
After releasing the second revision of the port with the corrected AMD Ryzen 2nd generation and Threadripper benchmarks, the official statement says the following:
As you may well imagine, the last few days have been busy. I (and others at PT) have spent long hours finishing the supplemental tests, trying to keep up with the community’s feedback, and responding to some of them. We would like to have responded as quickly as everyone requested, but there are way more of you than there are of us.
Intel requested that we test several CPUs to compare game performance. We tested various generations of Intel and AMD processors as listed in our original report. Based on community feedback, we have done additional testing.
Specific to AMD CPUs, we started the testing on Game Mode for AMD Ryzen™ Threadripper™ processors. Those results did indeed show Game Mode overall yielded the best gaming performance on Threadripper. For consistency, we then used Game Mode on all of the AMD processors. We have now added results from our testing of the AMD Ryzen™ 7 2700X in its default mode (i.e., Creator mode) as well. That mode overall yielded the best gaming performance on the 2700X. We apologize for not testing both modes in the initial report.
The updated report includes both modes. You can find the report and testing results here.
Finally, I’ve left our earlier response below so you can find it easily.
Thanks for listening.
Bill Catchings, co-founder of Principled Technologies
PT’s 2nd Revision Report- the good, the bad and the ugly
The good corrections are the use of Noctua NH-14S TR4 on the Ryzen Threadripper CPUs. Principled Technologies added the all-core-all-thread Creator’s mode in the Ryzen Master utility. For Ryzen 2nd generation, it stuck with Wraith Prism CPU cooler based on AMD’s confidence in it. PT re-tested the games on AMD platform in Creator’s mode which is the default setting for keeping all the cores and threads enabled.
Two overclock profiles of the AORUS GTX 1080Ti 11G Xtreme!
Nobody noticed it. But the GV-N108TAORUS X-11GD- the AORUS GTX 1080Ti 11G graphics card has two overclocked profiles:
- Boost: 1746 MHz / Base: 1632 MHz in OC mode
- Boost: 1721 MHz / Base: 1607 MHz in Gaming mode
Here’s hoping Principled Technologies used only one of these on all CPU testing.
There is a logic behind using 2666MHz for Intel and 2933MHz for AMD as its the maximum supported memory frequency. Memory timings were never mentioned in the original report but it did include in the second revision. There’s also a real-world argument against it. But we know the timings used were the same.
Restricting the benchmarks to 1080p-only is debatable but Principled Technologies as a valid point. The main issue here is that Ryzen 7 2700X is terrible at 1080p gaming. That needs to be highlighted because its not a dying resolution- not even close. This choice of resolution cannot be dismissed entirely also for a reason that many users prefer gaming on a 1080p display with 240Hz refresh rate to enjoy the best of performance and best of visuals. The issue however here is that it could have included 1440p since there are people who use 1440p resolution for gaming with similar setups. This way, we could have known both side of the story- performance in this case.
Performance comparison with PT’s analysis
For the sake of simplicity, we took the numbers from Principled Technologies report and compared the Core i9-9900K to Core i7-8700K and Ryzen 7 2700X. We included the Game Mode and Creation Mode benchmark results. Its needed to explain how Intel tried to pull a fast one.
Principled Technologies used 64 GB Corsair Vengeance RGB DDR4 RAMs (again- unrealistic choice for a mainstream) with variable frequency for Intel (2666MHz) an AMD (2933MHz). Both used Gigabyte GTX 1080Ti graphics card.
Even with the 1080p handicap performance on the Ryzen platform, the R7 2700X performance is lower than Core i7-8700K. But the performance difference between it and the Core i9-9900K is negligible for PUBG and Fortnite, two of the most popular games for streaming. For Counter-Strike Global Offensive, there is a gap of 68 FPS. This is with the Creator Mode. You won’t have this point of view if you looked only at the game mode’s performance- what Intel originally intended.
The performance difference between the Core i7-8700K and Core i9-9900K for Forza Motorsport 7 and Far Cry 5 is minuscule. While for the GTA-V is the same for both. For Ashes of the Singularity, the performance of the Ryzen 7 2700X is comparable to Core i7-8700K. While the Core i9-9900K gallops by an additional 5.3 FPS only. What about AMD Ryzen 7 2700X performance? We all know it doesn’t really hold well in 1080p gaming. Which is why 1440p would have helped to have a better understanding. This is where the analysis of an independent reviewer comes in.
But this will not be your point of view if you only looked at Game Mode.
Analysis (and change in POV) with Game Mode only
If you look at Game Mode, you can understand what’s the perception PT’s original report and Intel will create. The difference between Core i9-9900K and the game mode 2700X is 11.9%, but with the default Creator mode, its 4.9% below the core i9. Another major difference is Forza Motorsport 7, Far Cry 5, massive difference in Ashes of the Singularity and GTA V.
There are some tests which hardly makes a difference. Such as Core i9-9900K having 49.5% advantage over R7 2700X game mode and 48.4% advantage over R7 2700X Creator mode. There’s always going to be that one exception, but on the most count, the difference between Game mode and Creator mode is night and day.
Intel’s odd behaviour
Why does Intel deviously compare Core i9-9900K with Ryzen 7 2700X? Principle Technologies know the purpose of the Game Mode since they claimed to use it on Threadripper and Ryzen with an excuse to keep the settings uninformed. We appreciate your honesty in your reports but I doubt that was the intention. I am sure you guys knew Game Mode was meant for Threadripper so that most games can run easier on it. I seriously doubt if Intel plays innocent. On its own and within this limited analysis with 1080p gaming, Many would speculate the Core i9-9900K is impressive on its own because it is on a mainstream socket. Sure, it is the fastest on paper on a mainstream platform. It is also the most expensive mainstream processor. But pitching it like the best gaming processor is too over the top. The fastest mainstream CPU might be a realistic claim if it is better in all workloads against AM4+ based CPU. We’ll only know once the benchmarks are out. The definition of ‘best’ in this context changes- if you look at only the processor’s performance- or just the mainstream processor- or for value for money overall.
Intel’s odd-ball of a response
Intel’s response after the Principled Technologies release of a report is too funny and designed to hope they don’t get into trouble and just going with an act:
Given the feedback from the tech community, we are pleased that Principled Technologies ran additional tests. They’ve now published these results along with even more detail on the configurations used and the rationale. The results continue to show that the 9th Gen Intel Core i9-9900K is the world’s best gaming processor. We are thankful for Principled Technologies’ time and transparency throughout this process. We always appreciate feedback from the tech community and are looking forward to comprehensive third party reviews coming out on October 19.”
BULLSHIT! You gave them the parameters to test them on and paid for it. They delivered what you asked them to do but because of the huge backlash PT had to clarify by correcting those mistakes. It was always your intention to use the end results of those tests for downplaying Ryzen 7 2700X’s performance to show a wider gap.
Core i9-9900K VS. Ryzen 7-2700X: A terrible CPU comparison
Therefore the recommendation against pre-ordering is still valid until the reviews from independent review websites are out. The corrected report from Principled Technologies paints a clearer picture for 1080p gaming performance, but that’s it! The eyes and ears are back on Intel.
An investment of $529.99 on pre-order for Core i9-9900K over $294.99 worth AMD Ryzen 7 2700X whose Achilles heel is 1080p gaming performance is a tough cheese to bite on! I can buy an X470 motherboard-Ryzen 7 2700X CPU combo with WiFI AC at the price of the Core i9 flagship. It doesn’t make sense to compare them. The Core i9-9900K costs 56.97% more. Ideally, we would expect the natural comparison to be Core i7-9700K whose pre-order price is $409.99 and Core i5-9600K for $279.99. That did not happen. Therefore, many questions arise about gaming and non-gaming performance on the upcoming Core i7-9700K and Core i5-9600K. It also does not have Hyper-Threading instruction set, while AMD Ryzen 5 and 7 have SMT (Simultaneous Threading).
Intel’s purpose of such practice
With the exception of mentioning Principled Technologies, using standard units of measurements for performance (in this case FPS) and detailed report about its settings, Nvidia pulled off the same stunt for the RTX 20 series. We all know how that ended up. The whole exercise is to convince people to make a pre-order decision based on a single perspective. It conveniently ignored price-per-performance comparison against half-the-cost Mainstream Ryzen 7 2700X CPU. Intel never needed to do that since the Core i9-9900K would have stood on its own with no mainstream CPU comparison from the previous generation or from AMD (for now). If Intel had good intentions, it would have talked about Core i5-9600K and Core i7-9700K and even have PT include its testing.
— Hardware BBQ (@HardwareBBQ) October 13, 2018