MSI Kombo Strike Lowers Power Draw, Temps On Ryzen 7 5800X3D
But, its real-world performance improvements are lackluster.
A Reddit user named Mannekino shared a post yesterday about his experience using MSI's new "Kombo Strike" beta BIOS on his Ryzen 7 5800X3D and MEG B550 Unify motherboard. For Mannekino, the Kombo Strike feature was a success on his chip, giving his Ryzen 7 5800X3D an 11 degrees Celsius drop in peak temperatures, and 19W less power draw while adding slightly more performance simultaneously.
Kombo Strike is a new MSI feature released two weeks ago for AMD's Ryzen 7 5800X3D. We still don't know how Kombo Strike works officially, but MSI says it will improve the CPU performance of the Ryzen 7 5800X3D. There are three levels to choose from, with the third option being the most aggressive setting. Right now, this new feature is only attainable in beta BIOS updates for specific MSI 500-series motherboards.
However, according to Mannekino's report, the user believes Kombo Strike is an automated version of AMD's Curve Optimizer, which is an undervolting utility designed for Ryzen 5000 CPUs. It would certainly make sense considering the MSI exclusive feature does improve the Ryzen 7 5800X3D's performance. Plus, Mannekino's temperature and power results show very similar behavior to Curve Optimizer undervolting. But do take this with a grain of salt since we don't have official confirmation on this theory.
Mannekino's setup consists of an MSI MEG B550 Unify equipped with a Ryzen 7 5800X3D CPU and a high-speed G.Skill Trident Z Neo DDR4 3600 CL14 4x8GB memory, and an MSI GeForce RTX 2080 Super Gaming X Trio. Cooling is about as overkill as it gets with dual 360mm radiators, but it does make Mannekino's temperature results very surprising.
In Cinebench R23 with Kombo Strike disabled, the Ryzen 7 5800X3D averaged 14,641 points after a second run to heat up the system. With Kombo set to level 1, that score goes up to 14,732 points, level 2 to 14,961, and level 3 to 15,049 points. However, the chip netted only a 2.7% increase in performance at the highest level.
Other benchmarks, including Superposition, 3DMark Time Spy, Horizon Zero Dawn, and Red Dead Redemption 2, all show very similar behavior. But the performance results become much more interesting if we look at the power and temperature results.
With Cinebench R23 on a 10-minute loop, Mannekino's Ryzen 7 5800X3D averaged 81.3 degrees Celsius and 112.2W power consumption. But when Kombo Strike is on level 3, the temperature and power reduced quite drastically to 72.3 degrees Celsius and 98.2W. It is fantastic to see, considering we are getting a performance boost simultaneously, even if it's small. The power and thermal results are around a 12% to 14% reduction in wattage and temperatures.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
On paper, these numbers are cool to see from one of AMD's hottest chips to date -- thanks to the 3D V-Cache. However, in the real world, the efficiency gains you get from Kombo Strike won't make a noticeable difference in gaming and day-to-day use, with less than a 3% on average performance bump.
Heat and power are also non-concerning issues since the Ryzen 7 5800X3D is known to hit 80 degrees to 85 degrees Celsius under all-core workloads due to the additional 3D V-Cache hindering heat transfer from the die to the IHS. So these temperatures are very average and safe for this chip.
The CPU is also heavily influenced by a locked 1.35V voltage limit only found on the Ryzen 7 5800X3D (compared to 1.5V on a regular Ryzen 5000 chip.) As a result, Precision Boost can't achieve higher boost frequencies at 80 degrees Celsius than if CPU temps are much lower. It is also why the Ryzen 7 5800X3D pulls less power than the Ryzen 7 5800X despite its higher thermals. As a result, you would have to use a tiny CPU cooler or run the chip inside a thermal-constrained environment to make MSI's Kombo Strike feature a viable option for improving CPU performance. If you are a hardware enthusiast and like the lower numbers for the sake of it, it's a cool feature to have. But for real-world performance gains, MSI's Kombo Strike seems to be more gimmicky than helpful.
Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.
-
hannibal Well just reducing temperatures is good update that help the chip last longer!Reply
The cache is locked in many ways so any speed updates are not likely unless they start breaking those locks and that can cause some really bad behavior... -
pclaughton Reducing temps and power draw while maintaining performance is useful. Doing it with a performance gain as well? Seems odd to label that "gimmicky".Reply -
TechyInAZ pclaughton said:Reducing temps and power draw while maintaining performance is useful. Doing it with a performance gain as well? Seems odd to label that "gimmicky".
It's gimmicky because the efficiency gains are not good enough. A 19W decrease, while great, is pretty much nothing when we're talking about a 112W average power consumption. Good and even average CPU coolers and motherboards today, can handle well over 112W, especially with Intel chips consuming as much as 240W or even more.
The temps are similar, 80-85C is perfectly safe for Ryzen, and is still 5C below the maximum rated spec of 90C. Granted the temperature difference is definitely nice to get more wiggle room since a 5C delta isn't all that great. But, temps like this have been found to be normal on the 5800X3D in general, due to the heat transfer issues generated by the 3D Vcache.
Another is that MSI says this is a CPU performance enhancer specifically, but 2.4%is laughable. -
BillyBuerger Hey, if I can get a power reduction without loosing any performance, sign me up. I realize the trend these days is to use as much power as possible to get the highest numbers, most of the time, just average numbers are sufficient and any savings in power and heat are welcome to me.Reply -
evdjj3j Why doesn't Tom's do their own testing instead of stealing the work of someone on Reddit and writing an article about it. So sad to see what this site and Anandtech have become after 25 years.Reply -
KyaraM
The tester was mentioned by name at the start of the article and even the thumbnail states it's "a redditor" who found it out. How is that stealing if everything is clearly labeled as someone else's work?evdjj3j said:Why doesn't Tom's do their own testing instead of stealing the work of someone on Reddit and writing an article about it. So sad to see what this site and Anandtech have become after 25 years.