46 minutes ago, blaunarwal said:

Now we soon will celebrate Christmas, friends are coming. Merry Christmas Ray

And to you my friend. I'm very grateful for your help. This has been an extremely irritating problem but I'm hoping there will be a resolution when Chillblast return after Christmas.

Monitoring of the system has been key in identifying the problem.

Your higher temps are an outcome of the lockup, not the cause. You should be able to run at close to 100C almost indefinitely without lockup - the CPU will throttle but continue to function. It definitely looks like P3D is locking up, the cores all get maxed and the CPU generates too much heat for the cooling system to dissipate.

Cheers

On 12/23/2023 at 9:25 AM, Juliett Alfa Romeo said:

Completely off-topic (and I believe the error is my fault), but I'll report it anyway:

I had that exact problem yesterday when Verifying Files for ORBX scenery. Never had it before. The cure is to delete shaders and start P3D away from Australian airports. If I just selected YSSY at the scenario screen it caused that error. Starting at EGCC and then switching to YSSY cured it. Crazy eh? 😁

19 minutes ago, Luke said:

Your higher temps are an outcome of the lockup, not the cause. You should be able to run at close to 100C almost indefinitely without lockup - the CPU will throttle but continue to function. It definitely looks like P3D is locking up, the cores all get maxed and the CPU generates too much heat for the cooling system to dissipate.

Cheers

If P3D is locking up why do the three cores assigned to it continue to run? This is scenery I've had for ages including in my old computer. Never caused a problem then. Why doesn't P3D show any warning messages or even just CTD.

I've never heard of it just freezing without some form of error generation.

@Luke, it still doesn't explain why the fan doesn't boost up to 3500rpm to cool things down.

26 minutes ago, Ray Proudfoot said:

If P3D is locking up why do the three cores assigned to it continue to run? This is scenery I've had for ages including in my old computer. Never caused a problem then. Why doesn't P3D show any warning messages or even just CTD.

According to your Task Manager image, all of the P3D cores are running, not just those three, and it's consuming 97% of total CPU across all of the cores. Have you tried selectively disabling add-ons? The good news is that you are getting close to reliably triggering the hang.

Regarding the fan, you'd need to see where that fan profile is setup - probably in the UEFI settings. It may be configured to have a lower maximum limit.

Cheers

Hi @Luke, thanks for the info. I’ll investigate further on Boxing Day. The addons have been in use for years. I’m still perplexed why P3D continues to use those three primary cores as though nothing has happened to it. Why aren’t those locked to 100%?

I’m not familiar with the BIOS so will wait for Chillblast. I don’t want to change anything they’ve configured.

34 minutes ago, Luke said:

Have you tried selectively disabling add-ons?

Not yet but the obvious one would be ORBX AUS v2. If the lockups continue then that points to default scenery being possibly suspect.

I would then uninstall Scenery and reinstall it. Fly the same route and see what happens.

But if there was a corrupt file why wouldn’t P3D not register it and display a warning? Or even just CTD. Nothing is unusual and it’s never happened to me in all my years of running flight sim programs.

37 minutes ago, Ray Proudfoot said:

Not yet but the obvious one would be ORBX AUS v2. If the lockups continue then that points to default scenery being possibly suspect. I would then uninstall Scenery and reinstall it. Fly the same route and see what happens.

Let's be clear - at that point you will discover that the scenery is involved, not necessarily the cause. Add-ons may be reading scenery or facility data and crashing based on that. Right now you are still in the first stage of troubleshooting - determining what is happening. Then you'll move to the how, and only then to the why. It will likely be more complex than you first thought.

40 minutes ago, Ray Proudfoot said:

But if there was a corrupt file why wouldn’t P3D not register it and display a warning? Or even just CTD. Nothing is unusual and it’s never happened to me in all my years of running flight sim programs.

As I inferred above, P3D may consider it perfectly valid, and some other add-on does not. Additionally, P3D and its predecessors did little to no validation of BGL files. It was raw binary data that was ingested into memory unaltered for the purposes of speed. I certainly wouldn't design a file format like that today.

Cheers

There seems to be something wrong with the new hardware IMO. Temps should be nowhere near 100%. Hopefully, Chillblast will sort you out after Christmas. Very frustratiting, I feel your frustration. I have a similar setup to yours pre upgrade and this is putting me off upgrading!

I really do doubt this is a software problem.

To rule out an unstable overclock, I'd start by going into the BIOS and drop the p-core multiplier way down to something like 48 (4.8 GHz) max and maybe also impose a conservative PL1 power limit of ~200W, and see if that stops the lockups.  If it does, it's nearly certain that some exceedance related to your overclock is the culprit.

As I read it, your CPU pump will go into 3500 rpm high speed mode when the coolant temp reaches 60 deg C.  Your BIOS doesn't have any way of reading the coolant temp, so I would surmise that the cooling system itself has some kind of a thermal failsafe built into it.  It's likely that the coolant temp isn't getting hot enough to trigger the high speed mode--the only way to tell is to measure the coolant inlet temp at the radiator, e.g. shoot the radiator inlet area with an IR thermometer.  Regardless, increasing the pump speed only marginally increases the thermal capacity of the system--the system is still limited by the surface area of the radiator, airflow (radiator fan speed), and the ambient temp of the radiator cooling air.  I quite frankly am highly sceptical that any 240mm AIO (especially one internally mounted and drawing warm air from inside the case) is up to the task of keeping an overclocked 13900K below the thermal throttling threshold with a prolonged all-core load. 

Also, there are some other temps that are important to monitor--VRMs and PCH (chipset) temp come immediately to mind.  That EC error you mentioned above makes me wonder if the chipset isn't getting properly cooled.  My initial 13900KS build had a serious issue with excessive PCH temps because the monster-sized RTX4090 GPU was partitioning off the case and blocking airflow to the components in the lower half of the case, including the motherboard chipset and the SSDs.

100% core utilization only means that the core is tasked 100% of the time--it isn't a measure of how much work the core is doing.  I recommend watching the CPU package power consumption when the lockups happen, as it's a better indication of workload.

I know you don't want to mess with all this, but it appears necessity may demand that you do.

1 hour ago, Luke said:

Let's be clear - at that point you will discover that the scenery is involved, not necessarily the cause. Add-ons may be reading scenery or facility data and crashing based on that.

Can you clarify what you mean by addons? Do you consider LittleNavConnect, Aivlasoft's EFB and Pollypot GIT as potential suspects? They run alongside P3D but in a read-only state bar the last one which interfaces with my GoFlight hardware. It has no concept of where I am. If not those then what? In that part of Oz I have AUS v2 and Canberra and FlyTampa airports. Do you consider those add-ons?

The lockups have occurred as early as 4 minutes out of Canberra, half-way to Sydney and 7 miles out of YSSY. Never at the same place.

And if it is scenery how can I successfully fly FS Labs Concorde from Heathrow with 150 Ai on a third party airport and not have an issue. The sim freezes over the Australian bush but not in my most complex and demanding aircraft at my busiest airport. It makes no sense.

All the software I've installed was present on my old computer from the sane highly respected company, Chillblast.

Why does it never happen with the aircraft on the ground? Only when airborne. I did update the Nvidia drivers last weekend and went 7 flights without an issue until I switched to Australia earlier this week.

It's not that I don't want to believe you. It's just that logically it's hard to believe the scenery or an addon is responsible.

59 minutes ago, Bob Scott said:

I really do doubt this is a software problem.

To rule out an unstable overclock, I'd start by going into the BIOS and

Me too. But as this is a computer bought with a 2 year on-site warranty plus 3 years RTB I'm reluctant to change anything before talking to Chillblast. If we can't resolve the problem over the phone with them talking me through various BIOS settings I'll insist on a home visit by someone from the company who handles the support. They can do only so much with TeamViewer so a home visit is essential.

Is it overclocked though Bob? Isn't the 5500 speed just a multiplier of the base board speed? No overclocking was mentioned in the order or on the receipt. I thought itvwas just a bog-standard i9-13900K.

1 hour ago, Bob Scott said:

I quite frankly am highly sceptical that any 240mm AIO (especially one internally mounted and drawing warm air from inside the case) is up to the task of keeping an overclocked 13900K below the thermal throttling threshold with a prolonged all-core load. 

But when I talked through the items with Chillblast they never raised any concerns about the case not being large enough or the cooling insufficient for the use i was putting it to. They supply computers to FS Labs so are well aware of the demands the software will make.

Also, the recorded temps show the CPU temp stays around 60-70 most of the time. Look at the charts i've posted proving that. That suggests to me the cooling is fine right up to the point where the sim locks.

There's no gradual throttling as heat builds. One moment it's happily producing 30fps, the next it's locked and all VPs are running at 100% bar the 3 assigned to P3D in the JobScheduler section.

I feel only a visit from Chillblast support will get to the root of the problem and I'll just have to be patient. Thank god I bought from a reputable company with a warranty rather than building it myself. I used to until 10 years ago when they started to get very complicated with overclocking at which point i was happy to pay a company to do it for me.

I remain grateful for all your advice and knowledge. Have a happy Christmas everyone.

30 minutes ago, Ray Proudfoot said:

It's not that I don't want to believe you. It's just that logically it's hard to believe the scenery or an addon is responsible.

What may be hindering you is that you're making assumptions about what it could be. The secret is to as much as possible assume everything could be the cause and start ruling things out. All of those add-ons could be contributing. Disable them one at a time and see if you can reproduce. If you cannot, disable more and more of them. If it goes away with several disabled, re-enable one and see. It is incredibly tedious work by the way, but it gives you valuable data.

My vote is the opposite of Bob. If I was a betting man, I'd put down a dollar or a pound on software, or some interaction between several components.

Cheers

1 hour ago, Ray Proudfoot said:

Is it overclocked though Bob? Isn't the 5500 speed just a multiplier of the base board speed? No overclocking was mentioned in the order or on the receipt. I thought itvwas just a bog-standard i9-13900K.

But when I talked through the items with Chillblast they never raised any concerns about the case not being large enough or the cooling insufficient for the use i was putting it to. They supply computers to FS Labs so are well aware of the demands the software will make.

Also, the recorded temps show the CPU temp stays around 60-70 most of the time. Look at the charts i've posted proving that. That suggests to me the cooling is fine right up to the point where the sim locks.

There's no gradual throttling as heat builds. One moment it's happily producing 30fps, the next it's locked and all VPs are running at 100% bar the 3 assigned to P3D in the JobScheduler section.

I feel only a visit from Chillblast support will get to the root of the problem and I'll just have to be patient. Thank god I bought from a reputable company with a warranty rather than building it myself. I used to until 10 years ago when they started to get very complicated with overclocking at which point i was happy to pay a company to do it for me.

I remain grateful for all your advice and knowledge. Have a happy Christmas everyone.

Throttling doesn't occur gradually--when the temp hits the limit, it kicks in, and does so immediately.

I'd have to look at your BIOS settings to tell if you're running overclocked, or at least with some of the limits overridden.

I always run MemTest86+ when I suspect a hardware problem, as memory problems have been the most common hardware failures I've experienced over many years.  Your motherboard has MemTest86+ built right into the tools section of the BIOS.  I'd try booting into the BIOS and running that for at least a couple hours to see if the RAM or the IMC is dodgy.

Also, have you run a stress test of any kind...Prime95, 3DMark etc?   MemTest 86+ and a stress test run are things you can do ahead of a service visit to help the vendor troubleshoot, as they won't be able to spend enough time to run a prolonged test (unless maybe they take it to their shop for that kind of thing).

The pattern of the core loads when it locks up might be telling...as I understand it, P3D fences off those three explicitly-assigned VPs from other P3D threads, especially texture loading, which is the cause of the normal and intermittent spikes in processor utilization on the other processors during a P3D session.  Maybe something's going wrong while slamming the chipset with massive transfers from the SSD to memory when loading textures.  That EC error referenced above is definitely not normal, and still strikes me as a haunting clue.  How's your PCH temp with the sim running?

Sorry you're enduring this aggravation...it's surely not one what hopes for when laying out a wad of cash for a new machine.

Merry Christmas, and I send my heartfelt wishes for a quick cure to this...indigestion!

2 hours ago, Bob Scott said:

Also, have you run a stress test of any kind...Prime95, 3DMark etc?   MemTest 86+ and a stress test run are things you can do ahead of a service visit to help the vendor troubleshoot, as they won't be able to spend enough time to run a prolonged test (unless maybe they take it to their shop for that kind of thing).

I ran MemTest86 a couple of weeks ago but stopped it after 30 mins. I'll run it fully tomorrow morning. I can search for the others and try those over the next few days. I doubt the service engineer will come before 1 January.

2 hours ago, Bob Scott said:

That EC error referenced above is definitely not normal,

Could you explain what EC is please?

2 hours ago, Bob Scott said:

How's your PCH temp with the sim running?

If that's logged by HWInfo i can include that in my analysis.

2 hours ago, Bob Scott said:

Throttling doesn't occur gradually--when the temp hits the limit, it kicks in, and does so immediately.

I haven't seen it go above 100°C and then only very briefly using RTSS. But during a lock I was able to display the HWInfo screen and it showed the temp in red - 100°C continuously. Not good.

3 hours ago, Bob Scott said:

Merry Christmas, and I send my heartfelt wishes for a quick cure to this...indigestion!

Thanks Bob, my best to you and yours. It sure is frustrating but I'm hopeful it will be fixed. Everyone I know who has Chillblast PCs says their support is excellent. Just the tine of the year makes it a nuisance.