Is my i7 13700KF faulty?

[BarryB]

I bought some bits to build a new system just over 1 year ago: ASUS TUF Gaming Z790 Plus D4 mobo with an i7 13700KF CPU, 64GB Corsair Vengeance RAM, ASUS TUF Gaming RTX 4070-Ti SUPER OC GPU, 4 x M.2 NVME SSD's and Kraken X63 AIO all in a HAF932 case with a Corsair AX1200 PSU.

Initially the system was unstable with XMP enabled, crashing/hanging regularly so disabled it, this was with whatever BIOS shipped at that time and before the instability issues were reported so CPU was probably running outside of intel defaults. A few BIOS updates later XMP seemed stable but some games would crash so disabled it again!

Today i'm on BIOS 1802 with the intel defaults (have applied all the previous BIOS updates and left Intel defaults as soon as they were released) and the RAM seems stable with XMP but it isn't on the QVL but seems stable now with BIOS updates.

Problem is I have a game: Talos Principle Reawakened that simply crashes after about 20 minutes of running? I've redone the CPU paste with Kryonaut and added a faster exhaust side fan 1200RPM over 700RPM and added bottom fans to the radiator so the top fans now pull air in from the top and push through the rad then pull from the bottom and push into the case. Exhaust fan is 140mm running at 1200RPM and the intake is a 230mm fan running at 700RPM.

It still crashes and the CPU temp doesn't hit more then 70°C according to HWINFO and the NZXT CAM software so it's not getting dangerously hot, GPU runs about 65°C, even disabled XMP and it still crashes when the CPU hits about 70°C? Idle temps are about 28-30°C ambient temp is about 18°C

It's a UE5 engine game so is this indicative of the instability issues of the 13th/14th Gen CPU? I've set fan profiles for the radiator fans to ramp up slowly at 50°C, faster at 65°C and then quite fast at 75°C but only just hear them ramp up so the CPU can't be getting that hot to activate the faster fan profiles!

Any advice on what to try next or any more info required like fan profile / BIOS settings or just RMA the CPU?

[Nate152]

Hello BarryB

Does Device Manager show any missing drivers?

After your pc crashes, check Event Viewer, it can tell you what is causing the crashing.

[inge70]

So, basically, a fan arrangement should look like this:

Front and bottom: cool air in the PC
Top and back: warm air out of the PC

This is called the chimney effect and provides optimal airflow, as long as there aren't any figures or tangled cables in the case that get in the way.

Regarding the problem of UEF5 games crashing on an Intel 1st, 3rd, and 14th generation processor, it can now be said that if the game plays but crashes later, it's probably not your CPU's fault.

The Intel 13th and 14th generation processor problems already occurred when shader compilation failed the first time the game was launched. The compilation aborted, and that was it.

This was one of the main indicators that something was wrong with the CPU, and that's how it all came to light.

A UE5 game is therefore, in itself, a good test for such a CPU.

If the shader compilation runs smoothly and the game starts, there's nothing wrong with the CPU for early degradation or corruption.

Another test would be the Intel Diagnostic Tool or a run of Cinebench R23 or R24.

If all of that works, it's not your CPU, or it's probably not damaged, etc.

Especially since these corruptions and early degradations mostly only affected 13th and 14th generation I9 CPUs, less so the smaller I7 CPUs, etc.

You mentioned XMP and RAM that's not on the QVL list.

Please tell us the exact name of your RAM kit.

Unfortunately, the QVL lists aren't necessarily maintained, so newer RAM isn't added.

But if your XMP occasionally becomes unstable, depending on the UEFI update, your RAM could be one of the causes of the game crashes.

Your CPU officially supports RAM up to 5600 MT/s, and anything above that CAN run, but it doesn't have to.
This usually also depends on the quality of the CPU itself, as to how much higher clock speed it can manage without errors with the RAM controller.
Another note:
If you use two RAM kits with two modules each, i.e., four sticks in total, things can go wrong with XMP, as Asus explicitly doesn't guarantee error-free operation with two different kits (i.e., two sets of two sticks).

I personally use four RAM sticks from two kits, but 1. they're the same series, and 2. my RAM can only run at a maximum of 5600 MT/s, so even Memtest86, which is included in the UEFI, runs error-free to this day with the current UEFI on my board.

So, proceed as follows:
1) Disable XMP in the UEFI
2) Play your game for a while and see if it crashes again

-> If it crashes again, check the reliability history and the event log for any error entries from the time of the crash.

-> If it doesn't crash again when XMP is disabled, then you've 99% sure you've found the cause: your RAM is overclocking too much with XMP, and the CPU's RAM controller is no longer cooperating, even though it worked with earlier UEFI versions.

UEFI updates usually also bring changes to how XMP is handled to further stabilize the system. It can also happen that RAM that used to run stably with XMP is unfortunately no longer as stable because the CPU's RAM controller no longer supports this high load to protect the CPU from premature failure.

Keep in mind that XMP is also an OC, even if it doesn't affect the CPU cores themselves.

[BarryB]

@Nate152: Nothing missing in Device Manager and nothing in the Event Viewer regarding Talos crashes, it's the UE Crash Reporter that pops up so might not log an event?

@inge70: The rad being at the top (optimal place from what i've read) has cooler air pulling in from the top so it goes through the rad, I see pulling warm air from the case and pushing it through the rad and outside to achieve the chimney effect would make the liquid in the cooler warmer wouldn't it?

RAM is the same, 2 x 32GB kits of Corsair Vengeance CMK32GX4M2D3600C16 and XMP seems stable now on BIOS 1802, 1805 just said 'Modify strings in BIOS settings' so did not apply that one! I'll disable XMP and leave it off for now and see what happens!

I have Stalker 2 and shader compilation completes in that game and you can hear the fans ramp up while it does it but hasn't crashed so far and that's UE5 I believe and shader compilation is set to 10GB in nvidia control panel!

As a footnote, one of the Developers of Croteam replied on a steam post to downclock the CPU/GPU and see if that helps, but why do that with stock settings?

[inge70]

1) Wrong fan setting at the top!

The fan at the top should push warm air out of the PC into the room, not cool air into the PC.

Cool, fresh air comes from the front and/or bottom.

Warm air always rises!

What you're doing is counterproductive and pushes the warm air that's already inside the PC back into the PC, or it stays there longer than necessary when fresh air is drawn in from above.

The chimney effect means:
cold air enters the tower from the front and bottom, and warm air exits the top and back, because warm air always rises UP and not down or back.

Your fan setting could well cause the CPU and other hardware to get warmer than it should or cool down more slowly.

2) Regarding your RAM:

You only set XMP for the RAM, and then you get this message:
"Modify strings in BIOS settings"
Very strange. It looks more like you're manually changing various RAM OC settings.

I've never had anything like this before.
I have XMP enabled; when I exit the UEFI, a summary of the XMP changes is displayed, and then I reboot.
Now the RAM training takes place via the UEFI. This checks whether the settings are stable. If they aren't, XMP is automatically deactivated.

Do you only have XMP or also XMP2, etc., available to choose from?
For some RAM, there are multiple XMP settings in the UEFI.

3) Shader Compiling for Your Game:

So, in the Nvidia driver, you don't set the size of your game's shader compiler, but rather the shader cache size available for the game compiler.

You shouldn't really tinker with this, just leave it at "Driver Default Value."

Furthermore, a crashing game doesn't necessarily mean defective hardware. It could also be due to the GPU driver or the game itself.
Many games have not yet been patched correctly, either because they are console ports for the PC and the developers don't want to continue improving them, or because they were simply only half-baked for the PC from the start.

There are enough examples of games that crash regularly, like Cyberpank2077, Fortnite, etc., which, despite the high number of users, still crash regularly today and drive users to despair because they think it's their hardware.

Stalker 2 is still relatively new on the market, and I think there are one or two issues with the game or the GPU driver when using a multi-monitor setup or a monitor with 240 Hz or higher that also uses CSM (Compatibility Software Mode).

``` All Nvidia drivers have problems with this since the Blackwell series was released, and these issues are even affecting the older GPUs in the 4000 and 3000 series. Users sometimes have to revert to older drivers from 2024 to get everything running reasonably well again, but then they have to forgo new features with new drivers.
A vicious circle.

4) Regarding your footnote and the downclocking:
The tip probably came because the UE 5 caused the Intel scandal with the Intel 13th and 14th generation K-CPUs, or rather, because that's how the full extent of the problem first became public.
Mostly, the high-clocked I9s were affected, and the clock speed was then slightly reduced to ensure they continued to run stably.
However, a clock speed reduction for these components more likely means that they are no longer stable at the standard clock speed because they have already degenerated or are internally damaged.
This was temporarily circumvented by reducing the clock speed, because Intel didn't want to do mass RMAs either.

Regarding the GPU, I think this tip is complete nonsense as long as the GPU OC has not been manually adjusted by the user via Afterburner etc.
If you have a GPU that's already overclocked from the factory, or like me, an Asus TUF RTX 5070 Ti OC that's already overclocked by Asus, then I'm not going to mess with it again and risk something going wrong.
At least, that's my opinion.

Edit:

As long as the shader compilation of a UE5 game runs smoothly and the game is then playable, a test with the Intel Diagnostic Tool completes with a "PASS," and one or two load benchmarks with CB23 or CB24 10-minute or 30-minute tests run without errors (I always only used the 10-minute throttling test), your CPU is most likely not damaged or faulty in my opinion.

But only Intel can say that with 100% certainty, if they can and want to.

But those were the tests I've done so far since the Intel chaos started, to give me some peace of mind that I'm not obviously affected with my i7-13700K, since unfortunately, Intel doesn't have a test tool to detect precisely this kind of damage.

They apparently originally wanted to release a test tool (at least there was a post online about it), but then decided not to. The risk of a mass RMA was probably too great 🙂

The tests mentioned here place a strain on the CPU that's probably not what you'd expect from normal PC use, whether gaming, encoding, or CAD.

If everything else on your PC is running smoothly, but the game crashes every now and then, in my opinion, it's not due to a damaged CPU. The game crash isn't the only cause.

[BarryB]

Have now disabled XMP and put a more aggressive fan curve on the rad fans so they speed up quicker at lower temps so 55°C fans run at 60% and 65°C+ fans run 80%, this combo has allowed me to play Talos for much longer and without a crash! So either it's XMP or better cooling?

I do have a program called ProShow Gold that's a multimedia slideshow maker and when I import a video file into it the CPU hits 80°C - 90°C while importing according to the CAM software and the fans do ramp up noticeably, this only lasts as much as 30 seconds but doesn't crash the system, this was with XMP enabled but probably not long enough for XMP instability to show?

Mounting the radiator somewhere else isn't an option in this case so has to be mounted at the top! I can replace the front 230mm intake fan with a Noctua NF-A20 200mm fan that runs a bit faster, 800RPM vs 700RPM, so that's an option but the radiator location seems to be the problem. This case only has 1 intake fan and it has 6 x 5.25" drive bays in the front so only 1 intake fan can be fitted at the bottom front of the case.

[Nate152]

In your 2nd post you say you're using two 32GB memory kits. Combining memory kits (even of the same specs) is not recommended.

Try with one memory kit in the A2-B2 slots (2nd & 4th slots from the cpu), clear the cmos then boot up and try XMP.

[BarryB]

I've now put the radiator fans in Push/Pull so the bottom fans are pushing through the radiator and top fans pulling and exhausting out of the top of the case and CPU temps average 60°C, Liquid Temp about 34°C BUT XMP is now enabled and I can happily play Talos Reawakened without any crashes!

Can only think the internal temp in the case was getting too high and affecting the RAM when XMP was enabled for this particular game only but now it seems OK! The front intake is obstructed by the 3.5" drive bay frame so nothing I can do about that so might swap the side fan around and make that an intke rather than an exhaust, it's a 200mm Noctua NF-A20 PWM.

[inge70]

---

@BarryB wrote:

Can only think the internal temp in the case was getting too high and affecting the RAM when XMP was enabled for this particular game only but now it seems OK! The front intake is obstructed by the 3.5" drive bay frame so nothing I can do about that so might swap the side fan around and make that an intke rather than an exhaust, it's a 200mm Noctua NF-A20 PWM.

---

It doesn't matter that one fan is obstructed by a 3.5" hard drive cage.😉
Let the front fan still draw air INTO the case.
The hard drive cage doesn't completely block the airflow.

Otherwise, you can already see from your current temperatures that you used to have a heat buildup, and now the warm air is being cleanly expelled upwards and out of the back of the case.

Is your case a Coolermaster HAF 932?
Then you have a large fan in the front and even on the side, if I read it correctly online.
See if you can turn the fan around so that it draws air into the tower. So remove it, turn it around, and reinstall it.
It should work with most fans, and you don't have to spend any more money.

Let both fans draw INTO the case, then the airflow will be fine, and the temperatures should drop a bit.
Add a good fan curve to that, and it's perfect.😊

[BarryB]

Yes, it's a HAF932, not the X or Advanced model!

I replaced the stock 230mm side fan with the faster RPM/CFM 200mm Noctua fan as it's 4pin PWM if needed!  Will make it an intake and it's already set to full speed in BIOS so will see if that makes temps better!!