only 13900 and 14900 cpus are affected
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
There is no fix for Intel’s crashing 13th and 14th Gen CPUs — any damage is permanent
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
All true RaptroLakes with above 65W voltage are affected.
14900K
13900K
14700K
13700K
13600K
The non-K i5 arent affected because they are B0 stepping and dont eat a bunch of voltage.
Dunno if the C0 non-K CPUs are affected though, havent seen any reports.
AncientOrigin
Member
No they are better and use less watt and heat.
AncientOrigin
Member
Why is it ultra level bad to change to AMD?If you buy newer motherboard your are future proof
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
AncientOrigin
Member
What a difference AMD is amazing.I wonder how the performance would be if AMD made an 300 watt CPU would smoke Intel of the charts.The 139/149K eat closer to 300 stock when fully loaded....if your motherboard profile auto unlocks the powerlimits then it gets closer to 400W.
At least one company is fucking their customersFuck Intel
Doctor Hades
Member
Glad I bought an i5-13600KF back in December 2022 rather than the i7-13700KF as an upgrade to my 2013 i7-4770K as at the time I was aware of the high power consumption of the higher end chips and the 13th gen i5 CPU was at least getting good reviews and recommendations. Not too concerned about the reported issues with 13th & 14th gen CPUs as I have never experienced any issues with mine in the 18 months I've had it and it has always been running with Enforce All Limits enabled and relatively low voltages and temperatures since I got it. I also do not overclock.
That said, my faith in Intel is damaged and unless AMD do something really wrong in the meantime then my next CPU upgrade will almost certainly not be an Intel one. Who would want to buy one of their 15th gen processors after the reports of the issues with 13th and 14th gen CPUs? If anything, it is probably a good idea not to buy ANY new technology on release but to wait a year and just buy it then at a lower price when you know there are no issues. Yearly CPU upgrades are mostly iterative anyway and with AMD you at least have an upgrade path unlike Intel.
That said, my faith in Intel is damaged and unless AMD do something really wrong in the meantime then my next CPU upgrade will almost certainly not be an Intel one. Who would want to buy one of their 15th gen processors after the reports of the issues with 13th and 14th gen CPUs? If anything, it is probably a good idea not to buy ANY new technology on release but to wait a year and just buy it then at a lower price when you know there are no issues. Yearly CPU upgrades are mostly iterative anyway and with AMD you at least have an upgrade path unlike Intel.
Frwrd
Member
What about Xbox though?At least one company is fucking their customers
Okay okay, let’s not go off-topic in this bitch!
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
The 136K is on the list of affected CPUs.
There is no 15th gen Core i.
The Core U 1st gen already came out and doesnt have this issue.
No reason to assume the Core U second gen will have this issue.
P.S 2nd Gen Core U is the one coming to desktops this year.
Unknown Soldier
Member
You mean the CrowdStrike of CPU's?
PaintTinJr
Member
What about a non-k model on a Z-mobo chipset? With an all core overclock and AVX2 ratio locked so it won't downclock are the non-K models at risk too - even if that would technically be considered an overclocker's problem?
Last edited:
It's unlikely to be true but MILD is saying that 15th gen are still affected. Just worth considering until we know for sure. When that will be is hard to know
Last edited:
PaintTinJr
Member
Absent this design/quality issue...it is a different design strategy for Intel (vs AMD) of chasing higher single core clocks to keep wins on single core limited execution workloads, which in desktop productivity might still win out by a large margin for many using MsOffice/Browsing/Email and simple photoshop/CAD/StudioMax, and still wins out in gaming lows.
Intel's asymmetric e -core setup combined with the higher clocked AVX2 P cores also position their popular midrange priced chips in a good place for gaming at 60fps and above at the expense of power draw at overclocks. Productivity software performance like photoshop/Cad/StudioMax and Blender Cycles typically show wins or are competitive enough when the e-cores are utilised in heavy workloads, so "better" won't be across the board IMO, I personally prefer AMD's symmetrical all p-core strategy, but I'm pretty sure I could buy either(Amd or Intel) and be happy enough with a new Intel (without this issue) and the power use if I was prepared to reconfigure each time my use case changed.
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
non-K i5s arent true RaptorLake so they arent affected.
GamersNexus is still doing some testing and validation.
But from their contacts these are the CPUs affected.
i.e All True RaptorLakes and their 14th gen refreshes.
You can check you CPUs stepping using CPU-Z.
IF it says revision B0 then its vulnerable.
If it says C0 or H0 its not.
There is no 15th gen Core i.
If there was, it would be MeteorLake which is already out and hasnt exhibited any of the issues.
If ArrowLake and LunarLake were Core i CPUs they would be 16th Gen.
But they arent, they are second Gen Core U.
Thus they are named Intel Core U 5 2xx
MeteorLake and Arrowlake/LunarLake are on completely new nodes dont use as much voltage and LunarLake most certainly wouldnt have the issue cuz they barely go above 100Ws.
MLiD doesnt know shit.
Core U CPUs are disaggregated, thus them being a whole new naming scheme and evo no generation.
JohnnyFootball
GerAlt-Right. Ciriously.
This is NOT the case. There are reports of 700 and 600 CPUs being affected as well.
Much less to a degree as 900 series, but all 13 and 14 gen CPUs are potentially at risk.
JohnnyFootball
GerAlt-Right. Ciriously.
It's barely been out. There is no way to know if it will or will not have this issue down the line.
However, I do agree that there is no reason to assume upcoming CPUs will have it.
marquimvfs
Member
Generally speaking, CPUs don't scale this way. Doubling the power wouldn't translate to even 15% more speed, i guess. And would fuck the efficiency in the process.
winjer
Gold Member
They do. It's called Threadripper and they are productivity monsters.
And also very efficient.
SonGoku
Member
Is there a reason why AMD does not use compact cores for their multi CCD chips like the 7950X and 9950X?
A 8P CCD + 16c CCD configuration (compared to 8P + 8P) would give them better multithreaded performance while keeping the same level of single threaded permeance for tasks like gaming.
Seems like a no brainer very curious why AMD chose not to for Zen 5
Last edited:
winjer
Gold Member
Because E-cores have a major problem by not having the same instructions as the P-cores.
That is why AMD went with the C cores. It has 35% die space of the full core, but with the full compatibility.
But until now, AMD only has used the C cores in a few products, mostly mobile and server stuff.
marquimvfs
Member
winjer
explained everything, there's also the scheduler compatibility problem, not every software can adapt its workload to properly use the efficiency cores, sometimes, the performance is better with them disabled. In the end, it's just a workaround to improve the core number were the architecture doesn't allow to put enough cores without fucking (even more) the CPU energy usage. Want more cores? Put a large amout of regular ones in the package, like Threadripper, or Epyc.
Last edited:
winjer
Gold Member
X86 is already bloated with a ton of legacy instructions.
And because of the e-core problem with not having the same instructions support, we get non-sense like AVX10
chipsandcheese.com
And because of the e-core problem with not having the same instructions support, we get non-sense like AVX10
AVX10/128 is a silly idea and should be completely removed from the specification
Intro and lots of context
chipsandcheese.com
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
winjer
Gold Member
As far as I know, it's only AVX512.
But then there are the problems with having smaller caches and a less wide pipeline, meaning it can't take advantage of the same optimizations as the P-cores.
And then there is the issue with scheduling, that has caused applications and games to break or to run worse when the e-cores are enabled.
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
RaptorLake P-cores cant do AVX-512 either.
In fact lategen AlderLake couldnt do AVX-512.
If you had early AlderLake you could disable e-cores to get AVX-512 to work, but first revisions had this removed and RaptorLake just never had the option.
Cache sizes shouldnt affect anything as long as the scheduler is working properly.
But yes as you mentioned some games will erroneously send work to the e-cores when they really should be prioritizing P-cores.
Thats actually alot rarer than people make it out to be.
Most consumer products with WIndows 11 support wont make such mistakes and if they did thats generally because the CPU is actually fully loaded already, and the e-cores will actually accelerate the job.
Looking at die space Intel probably could have made a 10P CPU but the 16e cores would actually perform better than an extra 2 cores in multithreaded applications.
Last edited:
winjer
Gold Member
This is for Arrow Lake. And why Intel added AVX10, which is supposed to eventually support 512bits.
BTW, Alder Lake initially did support AVX512. But Intel decided to disable it.
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
ArrowLake is highly unlikely to support AVX-512 just like RaptorLake.
I knew/know about early gen AlderLake.......as I said in my post, if you disabled the e-cores you could activate AVX-512, Intel eventually disabled this with lategen AlderLakes and RaptorLake just never had it.
winjer
Gold Member
It won't support AVX512. It's going to support AVX10.
But supposedly, AVX10.2 will support 512 bits. But we don't know when that is coming out.
SonGoku
Member
I mean why doesnt AMD used their already existent C cores for say the 9950x's second CCD. That way the P cores CCD can be used for tasks that dont benefit from more than 8 cores and their second CCD can pack more cores per mm2 by using C cores making it perfect for multithreaded tasks.
My point is, if all you care about is lightly threaded performance you would get a single CCD CPU like the 9700X but for those also interested in heavy MT performance a P Core CCD + C core CCD would be the best of both worlds
Last edited:
winjer
Gold Member
That could be one possibility. One ccd with 4 full cores. And another ccd with 6 c cores.
It could be an interesting CPU for productivity.
nkarafo
Member
Thanks for giving me another reason to not like e-cores.X86 is already bloated with a ton of legacy instructions.
And because of the e-core problem with not having the same instructions support, we get non-sense like AVX10
AVX10/128 is a silly idea and should be completely removed from the specification
Intro and lots of context
Kamina
Golden Boy
Okay, i have an i7 14700f, which says B0non-K i5s arent true RaptorLake so they arent affected.
GamersNexus is still doing some testing and validation.
But from their contacts these are the CPUs affected.
i.e All True RaptorLakes and their 14th gen refreshes.
You can check you CPUs stepping using CPU-Z.
IF it says revision B0 then its vulnerable.
If it says C0 or H0 its not.
There is no 15th gen Core i.
If there was, it would be MeteorLake which is already out and hasnt exhibited any of the issues.
If ArrowLake and LunarLake were Core i CPUs they would be 16th Gen.
But they arent, they are second Gen Core U.
Thus they are named Intel Core U 5 2xx
MeteorLake and Arrowlake/LunarLake are on completely new nodes dont use as much voltage and LunarLake most certainly wouldnt have the issue cuz they barely go above 100Ws.
MLiD doesnt know shit.
Core U CPUs are disaggregated, thus them being a whole new naming scheme and evo no generation.
No issues so far, so I thought perhaps some preventive measures are in order.
I’ve read setting CPU Loadline Cibration to Mode 3 and AC Loadline as well as DC Loadline to 1 in Bios (MSI) is a good idea until Bios updates are released.
Any opinion on that?
Edit: e-cores are disabled
Last edited:
marquimvfs
Member
The point still stands. If you need REAL mt performance, get a CPU with a lot of symmetrical cores.
Edit: sorry, I was meant to quote
SonGoku
post that you were answering.
Last edited:
Ironbunny
Member
MLID's sources said that the "leading theory inside Intel right now" is that the Ring Bus is "getting cooked because it's fed the by the same rail as the P-Cores and E-Cores".
www.tweaktown.com
Intel cooking the Ring Bus on dying 13th and 14th Gen Core CPUs, says leaker
Intel 13th Gen and 14th Gen Core CPU issues: the ring bus is getting cooked because it's fed the same rail as the P-Cores and E-Cores, says leaker.
www.tweaktown.com
Last edited:
marquimvfs
Member
Makes sense, it would take a new socket to separate the Ring Bus rail.MLID's sources said that the "leading theory inside Intel right now" is that the Ring Bus is "getting cooked because it's fed the by the same rail as the P-Cores and E-Cores".
Intel cooking the Ring Bus on dying 13th and 14th Gen Core CPUs, says leaker
Intel 13th Gen and 14th Gen Core CPU issues: the ring bus is getting cooked because it's fed the same rail as the P-Cores and E-Cores, says leaker.
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
Probably a good call to set it up like that.
From my understanding at this point.
The CPUs that regularly sit above 65W are in 13/14th gen are cooking their RingBus with Voltage, so an undervolt is likely to help keep them for longer, im guessing the BIOS update is also going to be doing something similar.
ohh I forgot about LGA-1700 BartlettLake........what a shit show if its affected by this and Intel was hoping for an LGA-1700 victory lap.
People had been asking for P core only CPUs, Intel was finally delivering their wishes, and its a ticking timebomb?
In a timeline that favored Intel this would have been such a win cuz LGA-1700 would have lasted so long and AlderLake users would have had 3 "generations" of upgrades so the whole Intel wasting sockets narrative could have been hushed atleast for a while.
PaintTinJr
Member
I guess it would depend on what the energy and heat savings gave back in return, Intel have only gone the route they've gone because the headroom allows(or did before this flaw) them to get higher clocks on the primary core and downclock other P-cores proportionally less - from the single core peak - as they try to push all p-cores at the highest possible clock.
AMDs P cores CCD might be design bottlenecked elsewhere, like signalling making the chase for higher clocks unviable and would explain why they went with a monster 3D cache solution to eliminate MT bottlenecks at the existing clocks.
Last edited:
nkarafo
Member
When did this happen? I didn't notice...
I'm thinking of getting a 6-P core 12400 but i would wait for a 8-P core only CPU.
Shame they are ditching hyperthreading though
Yeah that it what I was referring to with my previous post. Like you say a shitshow. Lets hope they can address these issues if they are real.Probably a good call to set it up like that.
From my understanding at this point.
The CPUs that regularly sit above 65W are in 13/14th gen are cooking their RingBus with Voltage, so an undervolt is likely to help keep them for longer, im guessing the BIOS update is also going to be doing something similar.
ohh I forgot about LGA-1700 BartlettLake........what a shit show if its affected by this and Intel was hoping for an LGA-1700 victory lap.
People had been asking for P core only CPUs, Intel was finally delivering their wishes, and its a ticking timebomb?
In a timeline that favored Intel this would have been such a win cuz LGA-1700 would have lasted so long and AlderLake users would have had 3 "generations" of upgrades so the whole Intel wasting sockets narrative could have been hushed atleast for a while.
winjer
Gold Member
Intel's crashing CPU crisis deepens as more models are affected than originally thought
The news coming out of Intel about its crashing 13th- and 14th-generation CPUs is not getting any better, even after they said they had finally "solved" the...
www.techspot.com
Why would any consumer even consider getting an intel cpu right now? Boggles the mind.
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
Im on a 12400 right now, and ive been waiting to upgrade on LGA-1700 and right when I was expecting the price drops to happen (August/September) we are getting this nonesense.
Granted the only reason I hadnt upgraded yet is actually because the 12400 just kept on trucking along and I never felt like it was causing me any issues.
From a stop-gap CPU im actually really impressed by it.
If trhe issue isnt solved by the time ArrowLake is out and ArrowLake doesnt have this problem I might just move to LGA-1851.
That user is planning on getting an AlderLake CPU.
AlderLake doesnt suffer the same issues as RaptorLake, so no reason NOT to jump on one especially this late in the cycle they are probably going for pocket change.
And if BartletLake doesnt actually have the same issues as RPL then said user can even upgrade to an 8, 10, or 12P core CPU down the line on the same socket.
We also have no reason to assume ArrowLake is gonna suffer the same issues as RPL so Intel is still viable.
YeulEmeralda
Linux User
proandrad
Member
As a consumer you have plenty reason to be wary. Two generations of cpus suffering from the same hardware fault is a major issue and is a red flag. How did intel’s QA testing not catch this, or worst what if they did and intel ignored the issue to pushed out an unstable chips just not to fall behind AMD. I wouldn’t even consider intel until they have a couple of generation chips into 7nm before I would feel good about purchasing another intel chip. If anything this smells of class action lawsuit.
BWJinxing
Member
I remberer when my cable company would only let you use cable modems with the flawed PUMA7 chipset, that drop packets underload, still till this day.
Then there was the half assed cellular modems in the iPhone that got less performance over Qualcomm.
Oh and the floating math bug on the pentimum CPUs in the 90s.
I had a i7 2600k that was solid till the spectre meltdown patches.
Personally affected by the puna 7 and 2600k, I have no faith in the Intel brand in sometime.
The ringbus frying itself is on par for the course for Intel. Glad Pat got that big fat raise after cutting departments that could have prevented this fuckup.
Then there was the half assed cellular modems in the iPhone that got less performance over Qualcomm.
Oh and the floating math bug on the pentimum CPUs in the 90s.
I had a i7 2600k that was solid till the spectre meltdown patches.
Personally affected by the puna 7 and 2600k, I have no faith in the Intel brand in sometime.
The ringbus frying itself is on par for the course for Intel. Glad Pat got that big fat raise after cutting departments that could have prevented this fuckup.