Intel has a Pretty Big Problem..

[Chud_Jenny]

Since owning my 13600k, I’ve never been able to run xmp on the memory and also experience instability at stock voltages.

Could the XMP issue also be attributed to the reported manufacturing problems?

Do you have the model number of the Ram kit to hand?

 

[dazzlaa]

Do you have the model number of the Ram kit to hand?

I don’t currently, but it’s a ddr4 board and it runs on Tforce extreme 4133 memory.

 

[Jay85]

Since owning my 13600k, I’ve never been able to run xmp on the memory and also experience instability at stock voltages.

Could the XMP issue also be attributed to the reported manufacturing problems?

Not being able to run XMP could just be down to a poor IMC on the cpu.

The instability could be down to a degraded chip though, i had a 13700K that couldn't run stock settings and needed 1.5v to run stable at stock.

 

[Silent_Scone]

I can't be bothered to read back for context as the poster you are replying to I have on ignore - but are you saying here 14th gen CPUs aren't hitting voltages above 1.5V? because I've seen a fair few that on W680 boards where individual cores are getting into the 1.5s and on some consumer boards where they are getting into the 1.6s.

I can't say that is connected to the issue but doing so without consequences goes against everything I understand.

If you can't be bothered to read what it is in reply to then of course it's not going to make much sense lol. He was claiming the CPUs were being exposed to > 1.5v in heavy multithreaded workloads. If having used the CPUs it should be obvious that the power dissipation level at that level of current would be completely unmanageable.

Since owning my 13600k, I’ve never been able to run xmp on the memory and also experience instability at stock voltages.

Could the XMP issue also be attributed to the reported manufacturing problems?

No, that can be regarded as overclocking instability. XMP is often considered plug-and-play overclocking when it is not

 

[Chud_Jenny]

I don’t currently, but it’s a ddr4 board and it runs on Tforce extreme 4133 memory.

4133MHz DDR4 would be very difficult to run I'd think. It's official memory specs are:
Up to DDR5 5600 MT/s
Up to DDR4 3200 MT/s

Intel® Core™ i5-13600K Processor (24M Cache, up to 5.10 GHz) - Product Specifications | Intel

Intel® Core™ i5-13600K Processor (24M Cache, up to 5.10 GHz) quick reference with specifications, features, and technologies.

www.intel.com

You might get 4000MHz to work, it's all down to the quality of the IMC (memory controller)

 

[Rroff]

If you can't be bothered to read what it is in reply to then of course it's not going to make much sense lol. He was claiming the CPUs were being exposed to > 1.5v in heavy multithreaded workloads. If having used the CPUs it should be obvious that the power dissipation level at that level of current would be completely unmanageable.

Depends how you mean it - I'm not seeing 1.5+V on all cores across the board but I've seen as above even W680 boards with stock settings with 14900s where 1 or more cores is getting over 1.5V including in MT heavy loads. As well as worse situations on some consumer boards. Intel have said there is a problem with incorrectly high voltage and frequency being held in some situations incompatible with the temperature condition.

I don't know if this is the problem but it certainly doesn't look healthy to me.

 

[jigger]

If you can't be bothered to read what it is in reply to then of course it's not going to make much sense lol. He was claiming the CPUs were being exposed to > 1.5v in heavy multithreaded workloads. If having used the CPUs it should be obvious that the power dissipation level at that level of current would be completely unmanageable.




No, that can be regarded as overclocking instability. XMP is often considered plug-and-play overclocking when it is not

Voltage, temperature, power use and frequency (in both time and speed senses) degrade silicon ICs over time. Intel are pushing all these factors very hard. In my opinion 1.5v is beyond insane for everyday use even if the other factors are at sensible levels which they are not. Rroff is correct in what he is saying.

 

[jigger]

Depends how you mean it - I'm not seeing 1.5+V across all cores board but I've seen as above even W680 boards with stock settings with 14900s where 1 or more cores is getting over 1.5V including in MT heavy loads. As well as worse situations on some consumer boards. Intel have said there is a problem with incorrectly high voltage and frequency being held in some situations incompatible with the temperature condition.

I don't know if this is the problem but it certainly doesn't look healthy to me.

Intels strategy is distinctly unhealthy IMO.

 

[eeii]

I don't have one of these chips, but from observation, mainly tech jurnoes i see them running 1.5v or even a smidge above that in heavy MT workloads, i've even commented on it here, saying its utterly insane, its mental, and was told i don't know what i'm talking about.

Look, my CPU in low stress workloads will run up to 4.9 Ghz at about 1.5v, its low stress boosting, its done that for 4 years its obviously not hurting it, but if i put the MT stress on it the voltage drops all the way down to about 1.3v and the clocks are about 4.6 Ghz, there are options in the BIOS that allow me to override that behaviour, even on a granular level i can set by how much to override that behaviour right up to and including run the same 1.5v and 4.9 Ghz no matter what, effectively that Asus thing that on Intel CPU's turns the low core stress boost to all boost always, MCE... that's it... i forgot until i typed that out...

Now i've seen that in my BIOS, you have to unlock it behind a fat warning... and i've played with it slightly out of curiosity, only enough to see a noticeable difference and at that point i was already no.... no no no this is idiotic, i need my CPU to last more than 6 weeks!
-----------

Anyway, here is Windell, again, talking about the fat elephant in the room everyone trying to avoid noticing, Memory Speeds...

when I first saw that vid I thought you had shared a gentleman’s film from the 70s

 

[Silent_Scone]

Depends how you mean it - I'm not seeing 1.5+V on all cores across the board but I've seen as above even W680 boards with stock settings with 14900s where 1 or more cores is getting over 1.5V including in MT heavy loads.

Please share with us where you're seeing > 1.5v in all core heavy workloads as that's not how the boost algorithm or loadline works.

 

[Tetras]

Please share with us where you're seeing > 1.5v in all core heavy workloads as that's not how the boost algorithm or loadline works.

Apparently even Intel didn't know how their algorithm works

 

[Silent_Scone]

Apparently even Intel didn't know how their algorithm works

Apparently so, but that doesn't stop people from actually learning something. Such as how unfathomable that much voltage is in loads such as Cinebench.

 

[Rroff]

Please share with us where you're seeing > 1.5v in all core heavy workloads as that's not how the boost algorithm or loadline works.

I can't easily as I don't have ready access - but here is an example of one from Buildzoid https://youtu.be/yYfBxmBfq7k?t=1320

I've seen it still happening on some cores, on some CPUs, when under full amperage loads.

 

[Silent_Scone]

I can't easily as I don't have ready access - but here is an example of one from Buildzoid https://youtu.be/yYfBxmBfq7k?t=1320

I've seen it still happening on some cores, on some CPUs, when under full amperage loads.

Did you mean to timestamp it showing single-threaded? If you do have access, I would implore you to try setting a maximum VRM loadline setting whilst setting global SVID to 1.5v and see if you're able to sustain that voltage in Cinebench MT lol

 

[Rroff]

Did you mean to timestamp it showing single-threaded? If you do have access, I would implore you to try setting a maximum VRM loadline setting whilst setting global SVID to 1.5v and see if you're able to sustain that voltage in Cinebench MT lol

Can't find MT quickly but it has been in his videos before. Settings don't matter as such this is about what happens even with "stock" settings on some CPUs, where even under heavy MT some cores are at times hitting voltages above 1.5V.

 

[G J]

I don’t currently, but it’s a ddr4 board and it runs on Tforce extreme 4133 memory.

I was running a set of 16GB 3200 C14 8pack DDR4 at 3733 c16 on my 12600K and now running it at 4000mhz c17 on my 13600K and I couldnt get it to go any higher/tighter timings.

I think one needs a very good bin CPU and RAM to run above 4000mhz with DDR4 on LGA1700.

 

[jigger]

Apparently even Intel didn't know how their algorithm works

Thinking about it, many of the people that originally designed and worked on these two architectures likely no longer work at Intel. Intel have had several rounds of redundancies and cost cuttings over the last decade or so. Plus these designs have resurrected, clockbumped and rehashed on nodes they were never intended for.

 

[Tetras]

Thinking about it, many of the people that originally designed and worked on these two architectures likely no longer work at Intel. Intel have had several rounds of redundancies and cost cuttings over the last decade or so. Plus these designs have resurrected, clockbumped and rehashed on nodes they were never intended for.

Yeah, that would make sense. There's a rumour they even fired their validation team, which if true, clearly worked out great for them!

 

[gavinh87]

I've had a 4 year break from the CPU forum.

Glad to see the same people arguing about the same tripe

 

[KompuKare]

Yeah, that would make sense. There's a rumour they even fired their validation team, which if true, clearly worked out great for them!

What I heard it was the server validation team and they didn't fire them as such but that they had multiple sites, narrowed that down to one and tons of people left. The way I heard it phrased is that they expected most of the old team to quit rather than move across the country. And that that teams was the one with most of the experience etc.