I feel like both Intel and AMD are not doing great in the desktop CPU stability department. I made a machine with a Ryzen 9900X a while back and it had the issue that it would freeze when idling. A few years before I had a 5950X that would regularly crash under load (luckily it was a prebuilt, so it was ultimately fixed).
When you do not have a bunch of components ready to swap out it is also really hard to debug these issues. Sometimes it’s something completely different like the PSU. After the last issues, I decided to buy a prebuilt (ThinkStation) with on-site service. The cooling is a bit worse, etc., but if issues come up, I don’t have to spend a lot of time debugging them.
Random other comment: when comparing CPUs, a sad observation was that even a passively cooled M4 is faster than a lot of desktop CPUs (typically single-threaded, sometimes also multi-threaded).
Your comment about the passively cooled M4 is misleading.
Sure, in single thread, it will be definitely faster. In multithread unless you are going for low end or older CPUs it's basically a lie.
A 10 Core M4 will score around a 14TH gen mobile i5. It will consume much less power but the argument is on performance, so that's beside the point.
And if we are talking about a passively cooled M4 (MacBook Air basically) it will quite heavily throttle relatively quickly, you lose at the very least 30%.
So, let's not misrepresent things, Apple CPUs are very power efficient but they are not magic, if you hit them hard, they still need good cooling. Plenty of people have had the experience with their M4 Max, discovering that actually, if they did use the laptop as a workstation, it will generate a good amount of fan noise, there is no other way around.
Apple stuff is good because most people actually have bursty workload (especially graphic design, video editing and some audio stuff) but if you hammer it for hours on end, it's not that good and the power efficiency point becomes a bit moot.
Have had three systems (two 5800x, one 3600x) that reboots/freezes due to WHEA errors. Started after about 3years problem free.
One of the 5800xs so frequently it was trashed.
Occasionally occurring issues are so annoying. I lived with these issues for years before becoming able to reliably reproduce them by accident and thus making a good guess on the cause:
My system would randomly freeze for ~5 seconds, usually while gaming and having a video in the browser running a the same time.
Then, it would reliably happen in Titanfall 2 and I noticed there were always AHCI errors in the Windows logs at the same time so I switched to an NVMe drive.
The system would also shut down occasionally (~ once every few hours) in certain games only.
Then, I managed to reproduce it 100% of the time by casting lightning magic in Oblivion Remastered.
I had to switch out my PSU, the old one probably couldn't handle some transient load spike, even though it was a Seasonic Prime Ultra Titanium.
I have a 5950X system that will just randomly shut down, I've RMA'd the CPU, tried swapping the RAM, GPU, PSU and the motherboard in different combinations. I cannot track down a specific issue and it just won't be stable. I've given up and decided to discard the PC of theseus and build a new one -_-.
9900x here and zero crashes since I built it 9 months ago. A lot of the stability comes down to choosing the right RAM with the right timing for Ryzen CPUs.
I haven't moved on from AM4 yet but the way XMP is advertised you'd think it was guaranteed instead of overclocking that technically voids your warranty.
It's crazy how unreliable CPUs have become in the last 5 years or so, both AMD and Intel. And it seems they're all running at their limit from the factory, whereas 10-20 years ago they usually had ample headroom for overclocking.
I have not had any issues with Intel or AMD CPUs but I have so many issues with AMD APUs, I would steer clear of them. In my experience with different models, they have many graphics issues, broken video transcoding and overall extremely unstable. If you need decent integrated graphics then Intel is the only real option.
No desktop CPU I’ve ever used has remained stable at 100 degrees.
My 14900k crashes almost immediately at that temp.
3 hours at 100 degrees is obscene.
And any CPU from the last decade will just throttle down if it gets too hot. That's how the entire "Turbo" thing works: go as fast as we can until it gets too hot, after which it throttles down.
The last 15 years, servers has gone from 3x memory channels to 12x, while desktop still only have 2x memory channels.
It is by far the biggest bottleneck today.
Why is the author showing a chart of room temperatures? CPU temperature is what matters here. Expecting a CPU to be stable at 100C is just asking for problems. Issue probably could have been avoided by making improvements to case airflow.
Expecting a CPU to be stable at 100C is just asking for problems.
I had an 8th-gen i7 sitting at the thermal limit (~100C) in a laptop for half a decade 24/7 with no problem. As sibling comments have noted, modern CPUs are designed to run "flat-out against the governor".
Voltage-dependent electromigration is the biggest problem and what lead to the failures in Intel CPUs not long ago, perhaps ironically caused by cooling that was "too good" --- the CPU finds that there's still plenty of thermal headroom, so it boosts frequency and accompanying voltage to reach the limit, and went too far with the voltage. If it had hit the thermal limit it would've backed off on the voltage and frequency.
I run a 13900T unlocked (meaning, it runs 35W TDP at idle, 1.1ghz, but is allowed to peak to 210W for up to a minute, with the hugest Noctua D14something I could fit on it). It runs at ~29c idle, peaks to 80ish celsius at 210W (~4.5ghz over all cores - songle core peaking to 5.3ghz).
For a time I ran it 24/7 without suspend. It's a big system, lots of disks, expansion cards, etc. If it doesn't suspend, and doesn't do anything remarkable, it uses about ~5kWh per day. Needless to say, it suspends after 60 minutes now (my daily energy usage went from ~9 to ~4 kWh).
Gaming seems to be the final stronghold of x86 and I imagine that will shrink. Clearly games are able to run well on RISC architectures despite decades of x86 optimisation in game engines. Long term, an architecture that consumes more power and is tightly locked down by licensing looks unsustainable compared to royalty-free RISC alternatives. The instability, presumably because Intel are overclocking their own chips to look OK on benchmarks will not help.
> I also double-checked if the CPU temperature of about 100 degrees celsius is too high, but no: this Tom’s Hardware article shows even higher temperatures, and Intel specifies a maximum of 110 degrees. So, running at “only” 100 degrees for a few hours should be fine.
I'd say that even crashing at max temperatures is still completely unreasonable! You should be able to run at 100C or whatever the max temperature is for a week non-stop if you well damn please. If you can't, then the value has been chosen wrong by the manufacturers. If the CPU can't handle that, the clock rates should just be dialed back accordingly to maintain stability.
It's odd to hear about Core Ultra CPUs failing like that, though - I thought that they were supposed to be more power efficient than the 13th and 14th gen, all while not having their stability issues.
That said, I currently have a Ryzen 7 5800X, OCed with PBO to hit 5 GHz with negative CO offsets per core set. There's also an AIO with two fans and the side panel is off because the case I have is horrible. While gaming the temps usually don't reach past like 82C but Prime95 or anything else that's computationally intensive can make the CPU hit and flatten out at 90C. So odd to have modern desktop class CPUs still bump into thermal limits like that. That's with a pretty decent ambient temperature between 21C to 26C (summer).
I realize this has not much to do with CPU choice per se, but I'm still gonna leave this recommendation here for people who like to build PCs to get stuff done with :) Since I've been able to afford it and the market has had them available, I've been buying desktop systems with proper ECC support.
I've been chasing flimsy but very annoying stability problems (some, of course, due to overclocking during my younger years, when it still had a tangible payoff) enough times on systems I had built that taking this one BIG potential cause out of the equation is worth the few dozens of extra bucks I have to spend on ECC-capable gear many times over.
Trying to validate an ECC-less platform's stability is surprisingly hard, because memtest and friends just aren't very reliably detecting more subtle problems. PRIME95, y-cruncher and linpack (in increasing order of effectiveness) are better than specialzied memory testing software in my experience, but they are not perfect, either.
Most AMD CPUs (but not their APUs with potent iGPUs - there, you will have to buy the "PRO" variants) these days have full support for ECC UDIMMs. If your mainboard vendor also plays ball - annoyingly, only a minority of them enables ECC support in their firmware, so always check for that before buying! - there's not much that can prevent you from having that stability enhancement and reassuring peace of mind.
So I'm trying to learn more about this stuff, but aren't there multiple ECC flavors and the AMD consumer CPUs only support one of them (not the one you'd have on servers?)
Does anyone maintain a list with de-facto support of amd chips and mainboards? That partlist site only shows official support IIRC, so it won't give you any results.
The big problem with ECC for me is that the sticks are so much more expensive. You'd expect ECC UDIMMs to have a price premium of just over 12.5% (because there are 9 chips instead of 8), but it's usually at least 100%. I don't mind paying reasonable premium for ECC, but paying double is too hard to swallow.
Do you live at a very high altitude with a significant amount of solar radiation, or at an underfunded radiology lab or perhaps near a uranium deposit or a melted down nuclear reactor? Because the average machine should never see a memory bit flip error at all during its entire lifetime.
Now where can I get 64GB ECC UDIMM DDR5 modules so that my X870E board can have 256GB RAM? The largest I found were just 48GB ECC UDIMMs or 64GB non-ECC UDIMMs.
Any specific recommendations? I am having random, OS agnostic lockups on my ryzen 1xxx build and thought DDR5 will be enough, but true ECC sounds good.
edit: Looks like a lot of Asus motherboards work, and the thing to look for is "unbuffered" ECC. Kingston has some, I see 32GB module for $190 on Newegg.
Seems like failure in choosing cooling solutions. These high-end chips have obscene cooling needs. My guess would be using something that was not designed for TDP in question.
Sufficient cooler, with sufficient airflow is always needed.
I generally prefer AMD Zen5 to Intel due to AVX512 not being gimped by crippled E-cores that really don't belong on a desktop system, SMT (hyperthreading) that actually works and using TSMC processes, but they've also had their issues recently:
If OP's CPU cooler (Noctua NH-D15 G2) wasn't able to cool down his CPU below 100C, he must have been (intentionally or unintentionally with Asus multi core enhancement) overclocked his CPU. Or he didn't apply thermal paste properly or didn't remove the cooler plastic sticker?
I have followed his blog for years and hold him in high respect so I am surprised he has done that and expected stability at 100C regardless of what Intel claim is okay.
Not to mention that you rapidly hit diminishing returns pass 200W with current gen Intel CPUs, although he mentions caring able idle power usage. Why go from 150W to 300W for a 20% performance increase?
He did have the Fractal Define 7 Compact case, and the pictures[1] only show a single 140mm case fan. From personal experience the Fractal Define cases are great at sound reduction due to the thermal padding, but those pads also insulates well.
Given the motherboard and RAM will also generate quite some heat, if the case fan profile was conservative (he does mention he likes low noise), could be the insides got quite toasty.
Back when I got my 2080 Ti, I had this issue when gaming. The internal temps would get so hot due to the blanket effect of the padding I couldn't touch the components after a gaming session. Had to significantly tweak my fan profiles. His CPU at peak would generate about the same amount of heat as my 2080 Ti + CPU I had then, and I had the non-Compact case with two case fans.
> […] so I am surprised he has done that and expected stability at 100C regardless of what Intel claim is okay.
Intel specifies a max operating temperature of 105°C for the 285K [1]. Also modern CPUs aren't supposed to die when run with inadequate cooling, but instead clock down to stay within their thermal envelope.
The cpu temps are one thing, but if - as you said - even a beast like the D15 G2 has it pegged at 100C, this very much sounds like bad ventilation and other parts of the system being toasted as well - VRMs in particular, for which the "PRIME" (actually being the low-end series) mainboards from Asus, as used here, don't exactly have a stellar reputation.
And yeah, having Arrow Lake running at its defaults is just a waste of energy. Even halving your TDP just loses you roughly 15% performance in highly MT scenarios...
> If OP's CPU cooler (Noctua NH-D15 G2) wasn't able to cool down his CPU below 100C, he must have been (intentionally or unintentionally with Asus multi core enhancement) overclocked his CPU. Or he didn't apply thermal paste properly or didn't remove the cooler plastic sticker?
I did not overclock this CPU. I pay attention to what I change in the BIOS/UEFI firmware, and I never select any overclocking options.
Also, I have applied thermal paste properly: Noctua-supplied paste, following Noctua’s instructions for this CPU socket.
I also recently swapped to amd and my biggest surprise was how awful their platform is. With intel, getting sensor data just worked without anything special. With amd, it looks like each platform, perhaps model, requires special support. My mobo is a used godlike, and just has zero sensor support.
74 comments
[ 4.5 ms ] story [ 75.6 ms ] threadWhen you do not have a bunch of components ready to swap out it is also really hard to debug these issues. Sometimes it’s something completely different like the PSU. After the last issues, I decided to buy a prebuilt (ThinkStation) with on-site service. The cooling is a bit worse, etc., but if issues come up, I don’t have to spend a lot of time debugging them.
Random other comment: when comparing CPUs, a sad observation was that even a passively cooled M4 is faster than a lot of desktop CPUs (typically single-threaded, sometimes also multi-threaded).
And if we are talking about a passively cooled M4 (MacBook Air basically) it will quite heavily throttle relatively quickly, you lose at the very least 30%.
So, let's not misrepresent things, Apple CPUs are very power efficient but they are not magic, if you hit them hard, they still need good cooling. Plenty of people have had the experience with their M4 Max, discovering that actually, if they did use the laptop as a workstation, it will generate a good amount of fan noise, there is no other way around.
Apple stuff is good because most people actually have bursty workload (especially graphic design, video editing and some audio stuff) but if you hammer it for hours on end, it's not that good and the power efficiency point becomes a bit moot.
I also have this issue.
Twice the memory bandwidth, twice the CPU core count... It's really wacky how they've decided to name things
My system would randomly freeze for ~5 seconds, usually while gaming and having a video in the browser running a the same time. Then, it would reliably happen in Titanfall 2 and I noticed there were always AHCI errors in the Windows logs at the same time so I switched to an NVMe drive.
The system would also shut down occasionally (~ once every few hours) in certain games only. Then, I managed to reproduce it 100% of the time by casting lightning magic in Oblivion Remastered. I had to switch out my PSU, the old one probably couldn't handle some transient load spike, even though it was a Seasonic Prime Ultra Titanium.
It is cheaper and more stable. Performance difference doesn’t matter that much too
An ideal ambient (room) temperature for running a computer is 15-25 celcius (60-77 Fahrenheit)
Source: https://www.techtarget.com/searchdatacenter/definition/ambie...
Pass -fuse=mold when building.
Max Operating Temperature: 105 °C
14900k: https://www.intel.com/content/www/us/en/products/sku/236773/...
Max Operating Temperature: 100 °C
Different CPUs, different specs.
And any CPU from the last decade will just throttle down if it gets too hot. That's how the entire "Turbo" thing works: go as fast as we can until it gets too hot, after which it throttles down.
I had an 8th-gen i7 sitting at the thermal limit (~100C) in a laptop for half a decade 24/7 with no problem. As sibling comments have noted, modern CPUs are designed to run "flat-out against the governor".
Voltage-dependent electromigration is the biggest problem and what lead to the failures in Intel CPUs not long ago, perhaps ironically caused by cooling that was "too good" --- the CPU finds that there's still plenty of thermal headroom, so it boosts frequency and accompanying voltage to reach the limit, and went too far with the voltage. If it had hit the thermal limit it would've backed off on the voltage and frequency.
> After switching my PC from Intel to AMD, I end up at 10-11 kWh per day.
It's kind of impressive to increase household electricity consumption by 10% by just switching one CPU.
For a time I ran it 24/7 without suspend. It's a big system, lots of disks, expansion cards, etc. If it doesn't suspend, and doesn't do anything remarkable, it uses about ~5kWh per day. Needless to say, it suspends after 60 minutes now (my daily energy usage went from ~9 to ~4 kWh).
I'd say that even crashing at max temperatures is still completely unreasonable! You should be able to run at 100C or whatever the max temperature is for a week non-stop if you well damn please. If you can't, then the value has been chosen wrong by the manufacturers. If the CPU can't handle that, the clock rates should just be dialed back accordingly to maintain stability.
It's odd to hear about Core Ultra CPUs failing like that, though - I thought that they were supposed to be more power efficient than the 13th and 14th gen, all while not having their stability issues.
That said, I currently have a Ryzen 7 5800X, OCed with PBO to hit 5 GHz with negative CO offsets per core set. There's also an AIO with two fans and the side panel is off because the case I have is horrible. While gaming the temps usually don't reach past like 82C but Prime95 or anything else that's computationally intensive can make the CPU hit and flatten out at 90C. So odd to have modern desktop class CPUs still bump into thermal limits like that. That's with a pretty decent ambient temperature between 21C to 26C (summer).
I've been chasing flimsy but very annoying stability problems (some, of course, due to overclocking during my younger years, when it still had a tangible payoff) enough times on systems I had built that taking this one BIG potential cause out of the equation is worth the few dozens of extra bucks I have to spend on ECC-capable gear many times over.
Trying to validate an ECC-less platform's stability is surprisingly hard, because memtest and friends just aren't very reliably detecting more subtle problems. PRIME95, y-cruncher and linpack (in increasing order of effectiveness) are better than specialzied memory testing software in my experience, but they are not perfect, either.
Most AMD CPUs (but not their APUs with potent iGPUs - there, you will have to buy the "PRO" variants) these days have full support for ECC UDIMMs. If your mainboard vendor also plays ball - annoyingly, only a minority of them enables ECC support in their firmware, so always check for that before buying! - there's not much that can prevent you from having that stability enhancement and reassuring peace of mind.
Quoth DJB (around the very start of this millenium): https://cr.yp.to/hardware/ecc.html :)
Does anyone maintain a list with de-facto support of amd chips and mainboards? That partlist site only shows official support IIRC, so it won't give you any results.
edit: Looks like a lot of Asus motherboards work, and the thing to look for is "unbuffered" ECC. Kingston has some, I see 32GB module for $190 on Newegg.
Sufficient cooler, with sufficient airflow is always needed.
https://www.theregister.com/2025/08/29/amd_ryzen_twice_fails...
I have followed his blog for years and hold him in high respect so I am surprised he has done that and expected stability at 100C regardless of what Intel claim is okay.
Not to mention that you rapidly hit diminishing returns pass 200W with current gen Intel CPUs, although he mentions caring able idle power usage. Why go from 150W to 300W for a 20% performance increase?
Given the motherboard and RAM will also generate quite some heat, if the case fan profile was conservative (he does mention he likes low noise), could be the insides got quite toasty.
Back when I got my 2080 Ti, I had this issue when gaming. The internal temps would get so hot due to the blanket effect of the padding I couldn't touch the components after a gaming session. Had to significantly tweak my fan profiles. His CPU at peak would generate about the same amount of heat as my 2080 Ti + CPU I had then, and I had the non-Compact case with two case fans.
[1]: https://michael.stapelberg.ch/posts/2025-05-15-my-2025-high-...
Intel specifies a max operating temperature of 105°C for the 285K [1]. Also modern CPUs aren't supposed to die when run with inadequate cooling, but instead clock down to stay within their thermal envelope.
[1]: https://www.intel.com/content/www/us/en/products/sku/241060/...
And yeah, having Arrow Lake running at its defaults is just a waste of energy. Even halving your TDP just loses you roughly 15% performance in highly MT scenarios...
I did not overclock this CPU. I pay attention to what I change in the BIOS/UEFI firmware, and I never select any overclocking options.
Also, I have applied thermal paste properly: Noctua-supplied paste, following Noctua’s instructions for this CPU socket.
https://www.techpowerup.com/review/intel-core-ultra-9-285k/2... lists maximum temperature as 88.2C with the previous gen NH-D15 cooler.