Why Are Some Core i9 13900K/14900K Chips Unstable In Gaming?
Vložit
- čas přidán 2. 05. 2024
- ► Watch the FULL Video: • DF Direct Weekly #160:...
► Support us on Patreon! bit.ly/3jEGjvx
► Digital Foundry CZcams: / digitalfoundry
► Digital Foundry Merch: store.digitalfoundry.net
► Digital Foundry at Eurogamer: eurogamer.net/digitalfoundry
► Follow on X/Twitter: / digitalfoundry
Putting the entirety of the blame on the Motherboard manufacturers is misleading.
All the OC settings adopted by the Motherboard manufacturers were deemed "in-spec" by Intel themselves. If you absolutely had to blame one party here, you would have to blame Intel.
I'm blaming user error right now. I've yet to have a problem...
@@BluganoStudio Solid logic there. Because it didn't happen to you YET, it's user error. 👍
@@BluganoStudio You just won silicon lottery my guy. That’s it. The rest is just Intel pushing the VERY limit of their 14th gen CPU’s before moving to the next platform, and producing some VERY bad intel quality CPU’s. People were returning these DAILY in South Korea where it started. South Koreans aren’t amateurs my guy.
@@eliesercepeda686 Luck of the Irish then, I'll accept that.
Intel has the specs that is within spec and none of the motherboards have them set close to that at all like ICC max or power limits.
Of course! "I can't blame Intel because Intel is a sponsor." Intel chips are degrading more than usual. The 12th doesn't have this problem!
That's basically what it is, if it was AMD the story would be different
Intel in the past in writing confirmed that over 300 W power budget for their top processors is "within spec" (recently reported by Hardware Unboxed) plus Intel basically wanted their processors benchmarked on those settings for high scores in reviews. Benchmarking on new intel official specs gives much lower all-core benchmark scores.
I find it bizarre that you guys would blame the current situation on motherboard vendors :D
Oh, no, it's not 300W, it's actually 999W.
It's not the mobos fault, it's Intel's fault. Chips can only hit their advertised performance with these "overclocks".
We didn't know..WE DIDN'T KNOW!!
If your 13900K crashes at 253watts power limit then your CPU might have already degraded OR was shipped as defective.
Some testing reports are putting that number at 40-50% of the procs though.
@@SirSomnolent Those are just inflated numbers from a very small sample size. Not even intel has that data.
There’s ABSOLUTELY bad intel chips. It’s not Asus’s or the motherboard faults, a lot of the chips are failing STOCK settings. Specially on the 14th gen chips.
I mean, I run AMD 7900x. And someone thought putting the thermal limti at 95c out of the box. Pushing performance isnt worth it. I drop the temps 20c and keep 90 percent of the performance while greatly increasing reliability. And DDR 6 can run at 6400. But for how long.
These things shouldnt be running all out from the factory. Leave the overclocking headroom for enthusiasts.
@@ralphwarom2514 Well that’s the the thing, we are not. We are putting the exact specified wattage and amps required for MINIMUM performance and we are still having unstable conditions and BSOD. The only solution for me is downclocking -400Mhz if not more from stock performance.
@@ralphwarom2514 Zen 4 goes to 95C because it's designed to act like laptops do. Laptop chips do that normally for years without issue. Intel's thermal limit is now 100C for example. And you mean DDR5. What matters is the voltage needed to reach 6400 MT/s or whatever. That's what to look for. Either way DDR5 is still maturing and 6400 isn't very fast when 8000 speeds are available and eventually we should get to 10,000. Since DDR5 can run 2X the speed of DDR4 we can compare 6400 to DDR4-3200 (not exactly but just an estimate). DDR4-3200 is very stable for Zen 2 and Zen 3 and the sweet spot memory speed for those CPUs is faster at DDR4-3600 though some people could hit DDR4-3800 and DDR4-4000 with Zen 3. When I had Zen 2, I had a 32GB kit with Micron E-die (top alternative to Samsung B-die) that hit 3600 MT/s with 1.35V (a very safe standard voltage) though you could buy faster kits with 4000 MT/s or higher that might hit 1.45V which seemed to be fine as well though I preferred lower voltages personally. The fastest DDR4 available was 5100 MT/s last I saw but expect at least 1.5V to reach that speed and maybe active cooling to keep the dies from overheating long term.
With DDR5 I've heard that voltages below 1.5 are safe. DDR5-5600 is easy to hit with only 1.25V stock from the factory. I overclocked a 32GB 5600 CL40 kit with hynix M-die (the best DDR5 at the time I bought it last year, A-die is the new best I think) to 6000 MT/s at a stable 1.35V. I might be able to get 6200 without raising the voltage but not sure on stability there (not that going up to 1.4V would be any issue whatsoever). There were Zen 4 CPUs that were stable with DDR4-6400 from day in 2022 before BIOS updates but not enough were so AMD deemed 6000 as the sweet spot. That sweet spot is at least 6400 with the Zen 4 APUs which can also do at least DDR5-7000 from what I can tell and of course Intel has been stable at 7200 since 13th gen came out. People used to run DDR4 at like 2133 when it was new so imagine if they looked at 3200 as being an unsafe speed when it was ready with compatible CPUs by 2018 or so (it might have worked with Intel earlier but I know it was compatible with AMD's Zen+ eventually but not 1st gen Ryzen for example, I had a Ryzen 5 2600 in 2020 in my first build and I got DDR4-3200 RAM for it because I planned on getting a R5 3600 but it worked fine anyway).
It’s not a “rumour” lmao
Intel have straight up acknowledged the issue and released new BIOS settings for it.
Which is what they said in the clip
Intel isn't a motherboard manufacturer. They themselves aren't releasing BIOS updates.
Actually, even at stock Intel recommended settings, some are still unstable. Intel is basically selling lower-binned CPUs as higher bins.
These are the same people who can’t plug in a 4090, and don’t realize XMP is overclocking. Kids these days
@@a36538 Right because plugging in a cable is that hard and people forgot how to do it as soon as the 4090 was released. Also, the 12V-2x6 was released because 12VHPWR worked perfectly right? And those repair shops who found that 4090s (and even 4080s) burned anyway with the plug seated properly welded itself to the card. Do you have stock in Nvidia and Intel?
think this is what happens when your cpu start reaching 400 watts and 100c like its nothing
that's more power than a 4080/super.
@@mgk-metalgearkelly5054 even more than a 4090fe in its normal use
Intel's ultimately at fault here by not enforcing stock settings by default. If the user wants to overclock then fine but it should never be overclocked automatically out of the box when some chips may not even be able to sustain it, not to mention they have no idea what cooling solution a user will be using. AMD just had a similar issue with the 7000 series because motherboard manufactures were shoving way to much voltage into the CPU and literally cooking them to death and now this with Intel. Both companies need to force board companies to use default safe stock settings out of the box.
The problem about the Intel chips is the fact that people think running them so hard has caused early degradation making them unstable even running at very low baseline configs after taking that beating. So even if you correct the power regulation, if you let the motherboard run it too hard for months there is a good chance it may be ruined. THAT is much worse than if you just needed to correct the settings and the problem would vanish.
The reduction of max power consumption leading to stability is because the clocks are reduced to keep within the power budget.
I had the same problem with my 13900KF not being able to run at stock which is a big problem as a game dev.
I had to reduce the clocks or increase voltage to dumb amounts (1.5v+) for stock operation, until I RMA'd it with Intel and got a new 13900KF.
The new CPU however can only run at stock, any OC requires dumb voltages (near 1.5v for 4.6 P-Cores) to be stable.
I feel this for me at least, adds credibility to how close to the red line Intel are creating these CPUs.
Wow so they are auto overclocked to the point that they are STANDARD unstable.
GREAT idea!!!
Just give me a STABLE basis to work with. Something I can trust to WORK. As basis!
I have a i7 13k series and its literally the worst purchase ive ever made. I have to tinker and change settings using underclocking and overclocking affinity checking etc depending on what game i wanna play. Its awful. Before i "upgraded" i had an intel i5 8400k and NEVER had a single problem
user issue
@@fcukugimmeausername cool comment bro
@@tedahwooga user issue 😆😆
Asus tried to fry my Ryzen 5 3600 by running it at 1.475v out of the box.
No overclocking settings selected or applied, and they never fixed it. I had to manually "undervolt" it to make it stable and achieve base clock speeds.
This is not a new problem.
yep gigabyte z97x gaming 7 did the same with the 4790k back in 2014, out of the box on auto it was more than 1.4 volts when it would run just fine if you manually limited it to 1.17 volts.
Intel knew they motherboard makers were doing this and allowed it. Intel shares some of the blame too.
They ignored it when the results favored them. Intel shouldnt 'share' the blame. Should take the whole blame.
I haven't had any issues with my 13900k with Asus z690 with all limits removed for over an year. The only way it would exit Cinebench is if I undervolt it too much.
I have a 13900kf. It's really funny mine is exactly like the guy it's only some workflows that matter. All core ones didn't have much issues. Mainly it's gaming. Something that pushes a few cores harder than all cores. It's exactly that. It's been like that for months. I had already dropped mine to 253 wats when i got it though that' s different than most. I thought the default power was insane. I doubt that's common though. Most people i'd think would just trust the board to some extent. Plus some of the settings are like msi turn but it's named turbo it's hard to identify. It never occurred to me it was a bad chip until this happened. I was replacing ram reimaging windows doing all this stuff. I'm 2 years in and also worry about warranty.
Just curious if this intel drama apply to my i7 14700k or not... :)
Lock your cores speeds in bios, that will solve it
I have R16 with 4090 & i9 1400k with 64 gigs of ram. I'm so glad I haven't had this problem. 🙏
This is the Spiderman meme, as if mobo manufacturers don't know what happens when you set it to run 1.5v through a piece of silicon BY DEFAULT
still so glad I got a 10900K
nothing but problems/disappointment with later i9s
12900K was pretty good though but yeah the 10900K is better than the 11900K for sure. The older CPU actually performs better in many multi-threading cases due to having 10 cores and 20 threads vs the 8/16 in the 11900K.
It’s simple, AMD is back baby and just get a Ryzen cpu.
13900k on a poor AIO will cause those crashes you need to set your powerlimit to the thermal limit of your cooling system so it doesnt ever overheat or become unstable. 160w is pretty low.
How is running your 13900k at 150w which rated by Intel at 253W just fine? You guys are doing damage control for Intel, and it’s obvious.
Intel going with e-cores design was a terrible decision. So many stability and performance issues as a result of this hybrid design, you're better off disabling the e-cores
Love the channel but this is such a misinformed and braindead take. Look into it properly and you'll see there's a lot more to the story. It stems back to Intel both not outlining proper power spec and equally being reluctant to outlining it as it may reduce performance test results. They only care about performance at launch which makes sales.
The chips weren't designed to run unlimited power. Keep the intel specs and you won't have a problem. I have had a 13900k since launch and haven't had any issues.
Just like an engine, don't run so much boost
Problem is, what is the "Intel specs"? Seems like not even Intel knows.
@@deathdoor 253watts 400amp max
@@deathdoor if you look it up you will find it
@@jonathonschremp1182 In spec to Intel didn't involve power limits until recently. They encouraged their MB partners to do what they wanted out of the box as long as clock multipliers weren't adjusted. According to Intel it could be running at 1000 watts and still be in spec (I forget the exact unlimited wattage but it was in the thousands).
I switched from 14900k to a 7800x3d. Intel can oc sure. but fluidity of frames is worse than amd x3d..
Intel is pure *junk* these days...😮
I was talking to someone who noticed his performance in Lord's Of The Fallen was a lot worse than it used to be. He was running a 13900k. I also have the same issue running a 12900k.
I'm sure this is the game, yet others who aren't running I9 CPU'S haven't reported this issue.
I have 13900kf for over 1 year now and i didnt have single isue with it . I game on 540Hz and 280Hz monitors. But i have Gigabyte mamaboard Aorus Master Z790 ,maybe thats why
Intel is dead
Mid range zen 5 cpu ryzen 7 9800x 499$ will be on par with high end cpu intel
9800x3d will probably destroy reality
Looking forward to the R9 9900X3D personally.
@@Deathscythe919950X3D for me ;)
Intel is trash and Asus is even worse.
Tekken Gate
what the heck does 139000k even mean, it that the number of RAMS its has?
First 2 digits tell us which cpu generation is
900 tells the performance tier (i9,i7,i5,i3 etc).
K suffix says that is fully overclockable (given enough cooling).
The number of RAMS? What? You add the RAM yourself or the PC builder does. It's a CPU. It doesn't come with RAM. And it's 13900K not 139000K.
@@vigilant_1934
Anyway it depends on the motherboard
And now... intel with less performances with new bios... AMD is a lot better! (I had a 13900k and a 14900kf and now I m using 7950x3d) what a pity!!!
The Asus/Gigabyte solution is terrible, they reduced wattage but highly overvolt the chip to make it stable. Intel need to padronize the spec with its partners.