A new Beta BIOS has been released for ASUS motherboards that introduces microcode update 0x129 to fix the dreaded Intel error 13 and 14. 14th Generation Self-Immolation ProblemThe problem is that it seems to reduce CPU performance a lot under multi-core workloads. Voltages stay more or less the same, but there is a slight overall decrease. And core speeds seem to be lower.
We tested both configurations with only XMP enabled to allow our RAM to reach 6800 MT/s. We wanted to emulate the “standard user,” so rather than fully optimizing core speeds and making tweaks in the BIOS, we opted to leave it as is. The BIOS we tested first was version 1002, which introduced microcode 0x123. One thing we should mention is that we have largely unused our 14900K since the initial revision at launch, so it’s unlikely that it’s already corrupted, but you never know.
If you want to get ASUS BETA BIOS for yourself, VideoCardz has compiled a list of all of them. for each affected ASUS motherboard.
Testing and results of the latest microcode update
We used a Core i9-14900K in a Maximus Z790 Dark Hero for our test rig, along with a set of Corsair Dominator Platinums running at 6800 MT/s.
Microcode 0x123 (BIOS version 1002) Synthetic results
These are the results we got from the synthetic tests we ran. We found that the voltage did exceed 1.4V, but that's generally fine for very fast spikes. It seems our CPU may not be as affected by the issues as others.
Proof | Maximum core voltage (V) | Average core voltage (V) | Maximum core speed (MHz) | Average core speed (MHz) | Average CPU temperature | Score |
---|---|---|---|---|---|---|
Idle | 1.405 | 1.245 | 1.065 | 666 | 38 | |
Cinebench Multi | 1.414 | 1.232 | 5,512 | 5,114 | 86 | 39,783 points |
Cinebench Simple | 1.431 | 1.39 | 5,528 | 5,490 | 56 | 2,226 points |
Blender Monster | 1.393 | 1.277 | 5.150 | 5,477 | 97 | 238.583 SPS |
Zip 7 Unzip | 1.38 | 1.332 | 5.627 | 5,107 | 95 | 53,422 seconds |
Average | 1.4046 V | 1.2952 V | 4576.4 MHz | 4370.8 MHz | 74.4 °C |
Microcode 0x123 (BIOS version 1002) Gaming results
Game | Maximum core voltage (V) | Average core voltage (V) | Maximum core speed (MHz) | Average core speed (MHz) | Average CPU temperature | FPS |
---|---|---|---|---|---|---|
Days gone by | 1.41 | 1.374 | 4.075 | 2.803 | 66 | 283 FPS |
Cyberpunk 2077 | 1.446 | 1.381 | 3.971 | 3.229 | 77 | 210 FPS |
CS2 | 1.402 | 1.401 | 3.768 | 3.127 | 57 | 360 FPS |
Average | 1.4193 V | 1.3853 V | 3938 MHz | 3053 MHz | 66.7 °C |
Performance is pretty solid across the board and is what you'd expect the 14900K to achieve. We tried to make our gaming benchmark results as repeatable as possible, but as always, there are a lot of variables in games, especially in CS2.
What we discovered when we switched to the new microcode was surprising: we saw performance on multi-core workloads drop dramatically.
Microcode 0x129 (BIOS version 1503) Synthetic results
Proof | Maximum core voltage (V) | Average core voltage (V) | Maximum core speed (MHz) | Average core speed (MHz) | Average CPU temperature | Score |
---|---|---|---|---|---|---|
Idle | 1.304 | 1.232 | 1.322 | 643 | 36 | |
Cinebench Multi | 1.419 | 1.303 | 3.657 | 3.199 | 79 | 30,660 points |
Cinebench Simple | 1.376 | 1.367 | 5.804 | 2.286 | 58 | 2,276 points |
Blender Monster | 1.415 | 1.3 | 3.481 | 3.206 | 79 | 186.150 SPS |
Zip 7 Unzip | 1.412 | 1.308 | 4.577 | 3.878 | 82 | 57,766 seconds |
Average | 1.3852 V | 1.302 V | 3768.2 MHz | 2642.4 MHz | 66.8 °C |
As you can see – and yes, we did run these tests multiple times over multiple reboots just to make doubly sure everything was correct – multi-core performance after the update to microcode 0x129 takes a massive hit. We went from 39,000 points in Cinebench multi to 30,000 points, which is a performance loss of close to 25%.
From what we can see, voltages and clock speeds have been capped quite substantially, and it looks like the thermal overclocking is not that appealing. On the individual core side, we saw increases above the pre-update 0x123 microcode. We have to let everyone know that this is a beta BIOS release for now, and a full release is on the way. Presumably, the thermal overclocking is not working in this release, but the good news is that games were not affected at all.
Microcode 0x129 (BIOS version 1503) Gaming results
Game | Maximum core voltage (V) | Average core voltage (V) | Maximum core speed (MHz) | Average core speed (MHz) | Average CPU temperature | FPS |
---|---|---|---|---|---|---|
Days gone by | 1.429 | 1.387 | 4.376 | 2.698 | 68 | 282 FPS |
Cyberpunk 2077 | 1.405 | 1.363 | 3.939 | 3.152 | 71 | 209 FPS |
CS2 | 1.442 | 1.406 | 3.976 | 3.274 | 58 | 361 FPS |
Average | 1.4253 V | 1.3853 V | 4097 MHz | 3041.3 MHz | 65.67 °C |
As you can see, gaming performance hasn’t changed much. A big advantage of having reduced voltages and clock speeds is that the CPU runs much cooler on average. We’ve always had throttling issues with the 14900K when pushing it to the max, even with 420mm AIO coolers. But when we updated to the new microcode version that comes with the ASUS BETA BIOS, we saw little to no throttling issues. The ring bus did get throttled during more intensive workloads, though.
So what's going on here?
I think the thermal speed boost isn’t working properly, as we didn’t see the 14900K hit the advertised speeds even once. Firstly, that was because the 14900K was too hot. As we mentioned before, we always have trouble cooling it down. TVB may have worked properly the first time, but we never met the parameters.
I believe TVB has been disabled or corrupted in this BIOS version, and perhaps it is the temporary fix ASUS felt was necessary to stop the damage to the 13th and 14th gen CPUs. It seems like the CPU is severely throttled and the clock speeds reflect that. As I said, this BIOS is just a BETA version, so the real version may be a bit more usable.
JayzTwoCents made a video which showed virtually unchanged performance save for one or two hits, but seemed to fix the issues and boost to 1.6V. The difference is that J2C is running a full version of the BIOS update that is already available on MSI motherboards. Thankfully, we didn’t have that issue right off the bat, but we still saw an overall reduction in voltage, performance, and clock speeds in multi-core scenarios.
Leave feedback about this