New Intel i9 9900K CPU benchmarks come with updated AMD results … and an excuse
Principled Technologies has now published an updated version of its controversial processor benchmarks. The company was initially commissioned by Intel to validate its requirements for the Core i9 9900K as "the world's best game processor", but competing Ryzen 7 2700X failed leading to skewed reference results.
In the first tests, Principled Technologies had set up the Ryzen Threadripper chips in game mode to test their game performance accurately. This is in line with AMD's recommendations, as the pseudo-inherited mode allows the mega kernel pieces to allocate memory resources more efficiently to gaming and does not interfere with multi-thread game engines than they can handle.
Unfortunately, "for consistency" Principled Technologies had left game mode enabled when testing standard Ryzen 7 2700X. Similarly, 2700X core scores cut in half, leading to lower game performance almost across the board. The company has now apologized together with updated reference numbers using standard Creator Mode for all AMD processors it has tested.
This new version of the Principled Technologies report can be viewed online right now, showing a shrunk performance management for Intel's new Core i9 9900K, although it can still hold it up as the fastest game processor around right now. That was the whole goal of the test, as Intel told me this week. It would be lawfully called 9900K to be called "The World's Best Game Processor" in the New York launch event, with reference data to back up it.
 Although I'm not absolutely sure that there is necessarily a legal requirement to support such marketing blasts, especially given that the term "best" can be quite subjective. ..
"We have now added results from our testing of AMD Ryzen 7 2700X in standard mode (ie Creator mode)," said Bill Catchings, co-founder of Principled Technologies. "The overall mode gave the best gaming performance at 2700X. We're sorry to not test both modes in the original report. "
Intel has also offered a new statement, repeatedly repeating its 9900K performance boasting.
" Given feedback from the technological community, "said the Intel response," We are pleased that Principled Technologies drove further tests. They have now published these results together with even more detailed information about the configurations used and the reasons. The results continue to show that the 9th Intel Core i9-9900K is the world's best game processor.
"We are grateful for Principled Technologies' time and openness throughout this process. We appreciate technical community feedback and look forward to comprehensive third-party audits coming out on October 19."
| AMD Ryzen 7 2700X ” width=”900″ height=”507″ srcset=”https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-900×507.jpg 900w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-213×120.jpg 213w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-740×416.jpg 740w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-768×432.jpg 768w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-580×326.jpg 580w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-297×167.jpg 297w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-183×103.jpg 183w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-280×158.jpg 280w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-263×148.jpg 263w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-233×132.jpg 233w, https://www.pcgamesn.com/wp-content/uploads/2018/04/AMD-Ryzen-7-2700X-verdict-124×70.jpg 124w” sizes=”(max-width: 900px) 100vw, 900px”/>
Core i9 9900K vs Ryzen 7 2700X fps delta
|| Original fps delta
|| Updated fps delta
| Warhammer 2
|| 24 %
|| CS: GO
|| Ashes of the Singularity
| Forza 7
|| 19659016] Assassin's Creed Origins
| Far Cry 5
| WoW: Camp for Azeroth
|| 12%  5%
| Rise of the Tomb Raider
It's not quite as comprehensive a win for Intel provided non-Game Mode results now published since the game performance of AMD Ryzen 7 2700X, when paired with a Nvidia GTX 1080 Ti, has gone up quite consistently. In many cases, performance data between Intel and AMD octa kernel CPUs has decreased by more than half.
But it is still clear that the Intel octave core has its edge over its competitor, but given that 9900K is likely to cost twice as much as 2700X, you will have the kind of hope it would. On the kind of high resolutions that people who use on both an eight-core CPU and a GTX 1080 Ti will play, the framework package will shrink even more.
Interestingly, not only the AMD results have changed from the original report to this latest updated report. It seems that Principled Technologies has gone back and tested some of the Intel chips as well, as a few of the references have changed. For Fortnite, there is a big difference, with the performance of the i9 9900K which is down by around 24%.
We do not yet know why Intel results have been changed, but we've reached Intel for an answer.
Together with the actual results, Principled Technologies has also provided more transparency about the actual test method and platform layouts of the different processors. Most importantly, it has been detailed exactly how motherboards and memory are set up, with one of the most important concerns people had – memory magazines – were cleaned up. On the Intel page, it used the XMP configurations, and on AMD used the corresponding DOCP settings to provide optimal relative memory times.
Of course, these are still the results of a paid benchmarking session commissioned by Intel – and only 1080p performance difference – but it is in Principled Technologies' interests to make the results as transparent and impartial as possible. It will not burn bridges with AMD, who it has also worked for before. Although it is likely that it is likely to find missions from the red team in the future, it probably looks like it is unlikely right now.
We have our own independent reference results when the revision embargo promises October 19 next week.