Severe Issues for Intel: Core i3 and Core i4 Forced to Reduce Frequency in DX12 Games for Stability

It is a problem that we have been closely monitoring for months, and we have not yet discussed it because the impact was not widespread enough to do so. However, as of today, things have changed, and we now have a more solid foundation to describe its complexity and address it. Specifically, we are discussing the issues experienced by certain Intel Core 13 and Core 14 CPUs with certain games and various software. These Core 13 and Core 14 CPU generations are having to decrease their frequency to find stability, but what is really causing this serious issue?

There are numerous sources, perhaps too many. There are many problems, which we cannot cover in detail in an in-depth article because it would be extremely long, but we will mention them. The common denominator is always the same, and it is at the heart of what we are going to explore, as the last two generations of Intel processors seem to be pushed to their limits.

Intel Core 13 and Core 14 exhibit various symptoms

The principal challenge is to find the common denominator for the affected users, but identifying the causes that trigger so many problems proves to be difficult. Consequently, there will be arguments that intersect in this article, while others will be introduced without prior warning. With this understanding, everything revolves around the Core 13 and Core 14 CPUs regarding their frequency, frame contact, and end voltage, which has resulted in a partial solution of lowering frequencies for the P-Cores.

Of course, we are talking about completely stock processors, without overclocking or undervolting, with and without XMP, and different operating systems, regardless of whether they are Linux, Windows 10, or Windows 11. In other words, it does not matter the configuration variant we include; there are similar and completely contrasting cases affected, which adds to the overall complexity.

The issues began with DX12 and games based on Unreal Engine 4 and Unreal Engine 5. Errors in games using this API and these engines, as well as processors like the i9-13900K or i9-14900K, are causing problems such as a lack of “Shader composition” or other issues like “no video memory when trying to allocate a rendering resource.”

The number of errors varies and is challenging to diagnose, as they range from blue screens, stuttering, or CTD. The common factor for all of them was the high-end processors from Intel’s last two generations, coupled with the fact that the number of problems was escalating and could begin in various ways, such as freezing or simply poor performance while gaming.

In other words, the initial diagnosis does not necessarily have to be an error in a game; it may manifest in any of the aforementioned ways, making it difficult to identify the roots (which are several) of the whole issue.

The issue particularly affects the i9-13900K

Before addressing the causes behind everything, it is essential to understand the severity and confirmation from companies like Gearbox (Remnant 2) or Fatshark, which, after numerous user complaints, issued official statements:

We have identified an issue with some Intel 13th generation CPUs where, when starting the game, a message is displayed saying there is no video memory or a crash report is generated referring to a problem decompressing a shader. If you experience this issue, you will likely see it in other DX12 games as well.

If your CPU is overclocked, try resetting it to the default values. If it is not overclocked or that does not work, try installing the Intel Extreme Tuning Utility: https://www.intel.com/content/www/us/en/download/17881/intel-extreme-tuning-utility-intel-xtu.html and reduce your “Performance Core Ratio” from 55x to 54x.

Intel is aware of the failures but has remained silent and is covering warranty changes for affected processors, making it a very serious issue that has been lingering for many months and is becoming more widespread. Time is a key factor, but what is actually causing the failures? Is it hardware or software-related? Is it only the CPU, or is there something else involved?

Potential causes of stability issues in software and games with Intel Core 13 and Core 14 CPUs

Some key insights include the most affected processors being the i7 and i9 from both generations (Core 12 appears to be unaffected), which have more cores and higher power consumption. The causes are unconfirmed and numerous, so we will only cite them and briefly discuss a select few.

First, lowering the frequency of the P-Cores can only grant partial or temporary stability because the damage appears to be physical, which is very significant. Second, it all points towards multiple problems converging and making it difficult to track.

Starting with the multicore enhancement technologies provided by motherboard manufacturers to squeeze out more MHz “out of the box,” continuing with voltage compensation systems (SVID), and ending with an incorrect frame from Intel for these motherboards, which, as we know, is mandatory for securing the CPU to the socket.

Today, we have seen Arctic release its own frame for its AIOs, arguing that Intel’s frame (which all manufacturers must include in their boards by contract) exerts 40 kg of pressure in two specific areas on the IHS (side wings), not helping to disperse the pressure across the entire processor.

A pressure, contact, and electrical problem

Here, various factors intertwine, where we gather all the information collected and present an arguments based on what we have read from forums, websites, industry companies, and engineers.

It appears that the uneven pressure, already known since LGA1700’s early days, is causing physical failure in the Core 13 and Core 14 CPUs. The existence of customized frames (Thermalright’s frame deserves special mention as it seems to be the best of all) is no coincidence.

These frames are almost entirely solving the problem, and none of the described failures (and others, as there are more) have been detected in CPUs that implement them, suggesting a long-term solution. As mentioned earlier, time is crucial. The issue of uneven pressure combined with heavy heatsinks and additional attachment systems attempting to compensate for Intel’s mistake is resulting in poor contact between the processor and the motherboard’s pins.

This is not new, and if you are up-to-date, you are likely aware. However, what is novel is that this, combined with technologies like ASUS Multicore Enhancement (MCE) and its counterparts in other manufacturers, is leading to an interesting situation.

Together with DX12 and Unreal Engine, they are driving Boost frequencies in TAU (PL2), PL1, and PL3, creating a voltage peak that curiously does not occur in stress benchmarks.

Physical damages to specific cores in Core 13 and Core 14 processors

The problem reaches such an extent that PCs are failing in basic tasks like playing a YouTube video and in many of the ways described above. The reasons for the overvolt are causing physical damage to one or more P-Cores.

This has been discovered by isolating the cores and assigning different affinities, revealing in each case which core is affected, although in some cases, there are two or four. This perfectly aligns with Intel’s Boost distribution, so what we see here is a severe issue.

On the one hand, excess pressure and Intel’s poor frame design are causing pins not to make contact over time and thermal cycles (bending the CPU’s PCB), while on the other hand, some pins make too much contact.

It is speculated that this, along with multicore enhancement technologies, is physically forcing the processor to experience voltage (SVID) and frequency peaks over the months that it cannot withstand. Furthermore, even if we install a frame and disable these technologies, lower frequencies, and adjust LLC in BIOS to compensate for the positive Vdroop, there are cases where the problem cannot be resolved.

That is, there are CPUs that were already severely physically damaged, and an RMA has had to be requested from Intel and even the motherboard manufacturer. Others, however, have temporarily resolved the issue by lowering frequencies and adjusting LLC, although problems with the frame persist.

A custom frame is mandatory; motherboard performance policies are a mistake, SVID must be disabled

In summary, knowing that Intel is aware and has not yet made a statement, if you are experiencing this or another problem that you cannot diagnose despite your efforts, try the suggestions we have mentioned (more information below for assistance), and if you cannot achieve stability or if games continue to cause errors, it may be time to process an RMA for the motherboard and CPU and purchase a frame for reassembly.

Also, you will need to disable the multicore enhancement technology from the motherboard manufacturer to prevent your CPU from experiencing this issue. Finally, monitor frequencies and leave everything in the hands of Intel and Microsoft’s control, management, and allocation policies, which are producing the best results, rather than the board’s guidelines.

Will there be a solution from Intel and manufacturers? We do not know, but we hope that this compilation of arguments, errors, and solutions can at least help many people overcome stability issues with Core 13 and Core 14 CPUs, or if they are assembling a new PC, know what to buy and what to do to prevent problems over time.

The curious thing about all this is that in the first few months, all users experienced success, but after three to six months, problems began, undoubtedly due to thermal cycles and the degradation that CPUs undergo due to exposure to the aforementioned conditions, eventually causing irreversible damage. Lastly, no deaths have been detected in these CPUs, but it would not be surprising if some cores eventually fail entirely due to continued exposure to these conditions.

In response to the socket and frame issue, Intel responded when they launched LGA1700:

“We have not received reports of 12th generation Intel Core processors operating outside of specifications due to changes in the integrated heat sink (IHS). Our internal data shows that the IHS on the 12th generation desktop processors may have a slight deviation after installation in the socket.

Such a small deviation is expected and does not cause the processor to operate outside of specifications. We strongly recommend not making modifications to the socket or the independent loading mechanism. Such modifications would result in the processor operating outside of specifications and may void any product warranties.”

If there were already problems with Core 12 on this matter, and they were not as stressed, what we have seen with Core 13 and Core 14 in terms of frequencies and consumption could only lead to a breaking point.

Partial solutions for ASUS, GIGABYTE, and MSI for LGA1700

It goes without saying that these solutions are for extreme cases, but they may work for some. Although there are more motherboard brands, it is a matter of researching BIOS options to find the same setting with a different name and change it, as instructions have been provided for ASUS, MSI, and GIGABYTE.

For ASUS:

In BIOS: Advanced (F7) -> SVID Behavior -> Change to “Intel’s Fail Safe”

Restart and run XTU in AVX2 to see if it can pass. Run games and check if the problem occurs again.

For GIGABYTE:

Solution A): In BIOS, select “ADVANCED MODE,” go to the Tweaker tab, locate CPU Vcore and choose the “Normal” option, select “Dynamic Vcore (DVID),” change it from “Auto” to “+0.005V”. Increase DVID by +0.005 until the game issue disappears, and the system operates stably.

Solution B): In BIOS, select “Tweaker,” choose “Advanced Voltage Setting,” select “CPU/VRAM Configuration,” adjust “CPU Vcore Load Line Calibration,” start with “Low” to “Medium” until the system is stable.

After implementing Solution A or B, run the XTU benchmark again and check if the AVX2 test passes. Run games and checks if the problem reoccurs.

For MSI:

Solution A): In BIOS, select “OC,” choose “CPU Core Voltage Mode,” select “Offset Mode,” select “+ (by PWM),” adjust the voltage until the system is stable, do not exceed 0.025V for a single increase.

Solution B): In BIOS, select “OC,” choose “DigitALL Power,” change “CPU Vcore Load Line Calibration Control,” start with “Mode 7” and lower values until the system is stable.

After implementing Solution A or B, run the XTU test again and check if the AVX2 test passes. Run games and see if the problem reoccurs.

This is the most we can offer on this issue at the moment. As you can see, it is not easy to solve if you already have the system and have not made the necessary adjustments regarding the frame and BIOS policies on motherboards. Therefore, these are all palliative measures for a problem that has no solution beyond an RMA and starting from scratch with a new processor and, if necessary, a new motherboard.

The above article first appeared in El Chapuzas Informático.