Power Temperature and Frequency
On this page, we are going to investigate the Intel Core i9-14900K power draw, temperature, and frequency. This is a stress test, which means we are pushing each CPU to its maximum potential in all-core load in the Cinebench R23 multi-core test for 10 minutes.
Power
We use HWiNFO64 sensor data to record the results. We report on the sensor data for “Package Power” in Wattage.
We really didn’t expect the peak power usage to change with the Microcode Fix 0x129 applied since the PL1/PL2 power limits are still in place at 253W. This is the case, there was no change in the maximum power in an all-core load here in Cinebench since PL1 and PL2 are locked at 253W in both the Extreme and Performance Profiles now.
Temperature
We use HWiNFO64 sensor data to record the results. We report on the sensor data for “Tcidle” in Celsius.
There were some slight temperature differences with the Microcode Fix 0x129 update, however. With the update applied, the Performance Profile’s temps dropped pretty significantly, since the CPU is peaking at less Voltage, it does run slightly cooler. In the Extreme Profile, there was basically no change.
Frequency
With the Intel Core i9-14900K in the Extreme Profile with the Microcode Fix 0x129 BIOS applied, we experienced an all-core full-load workload in Cinebench to give us about 5GHz-5.1GHz on P-Cores and 4.1GHz on E-Cores. However, only 1 or 2 P-Cores were at the 5.1GHz speed, most were at 5GHz.
If we compare this to performance before the Microcode Fix, also in the Extreme Profile, we experienced the same frequencies, however, one or two more P-Cores at a time were at 5.1GHz before the Microcode Fix update. Therefore, it seems maybe one or two P-Cores have dropped by 100MHz at a given time in a maxed-out, all-core, Cinebench workload, which would explain the slight under 1% performance differences in multi-core workload performance testing.
In the Performance Profile with the Microcode Fix 0x129 BIOS applied we experienced all the P-Cores to operate at 5GHz, except for maybe 1 P-Core at a time at 5.1GHz. All the E-Cores were at 4.1GHz.
If we compare this to previous performance before the Microcode Fix 0x129 BIOS, there were about 2 more P-Cores that would also run at 5.1GHz, for a total of 3 P-Cores at 5.1GHz, compared to just 1 P-Core at 5.1GHz with the Microcode Fix 0x129 BIOS applied. Again, it seems about 2 of the P-Cores are reduced by 100MHz with the Microcode Fix 0x129 BIOS applied, which again accounts for the under 1% loss in performance with this new fix in Performance Profile.
Discussion (18 replies)
Join Discussion →Great writeup as usual. This is "a time will tell" scenario if this was a placebo or if it indeed worked.
I can't help but wonder if this is akin to putting lipstick on a pig and declaring it is fixed. Slight changes all of a sudden will prevent degradation (if a chip has not been degraded yet)? I don't know. I get the impression that Intel has no idea what is causing this in their 13/14 series, so they put out this microcode to appease the angry mobs with pitchforks banging on their door. Right now, I do not trust anything they say or do.
good article, let's hope it works long term
The uCode update did not change the power draw, nor are the boost clocks much lower. The max voltage was lower however (1.495 V in HWinfo). A few questions:
1) What was the VCore before the update? (In HWinfo) ? Did you see the 1.55V?
2) Does this new limit on VCore have a built in margin for already degraded processors?
3) Does tracking VCore across loads tell us anything about whether a processor has degraded or not? (accounting for the effects of Silicon lottery?)
Thanks for this article. I didn't see any noticeable difference in my 14700K after the latest bios update other than my voltages going from 1.5XX down into the 1.4XX range.
I installed the update for my 13600K system and I'm sure I'll never notice a difference in performance.
I think the real comparison is between the new microcode and the initial version that was used for the day 1 13XXX and 14XXX reviews.
Just from what i seen, before the update my cpu voltage was in the 1.35 to 1.39volts range. After the update it was in the 1.45-1.49. I wasn't having any stability issues since my cpu wasn't running on that 4000watt profile for very long before i turned it off cause all issues.
Currently i have turned off unvolt protection so i can undervolt -0.1volts to be back where it was and gonna go lower to see help like did on my rtx4070s
From watching Buildzoid's video on the update, he had easily captured 1.58V+ on the VCore prior to the update, and not able to trigger the scope after the update.
Wow! That is crazy. However, it is important to ensure that such transients are not due to instrument/environment errors. Perhaps I ought to watch that video.
[embedded media]
Video form of this article is also up -
[embedded media]
Very informative. My findings were the same with both my Intel rigs.
Certainly not the terrible impact we had all feared, but I am still not convinced this was the only root cause and that it fixes the problem.
Time will tell.
Time hasn't been kind to Raptor Lake: https://www.overclock.net/posts/29521182/
I hate it when I'm right :(
I’ve found the older I get the more that I pray that I’m wrong