Intel Core "Special Edition" Owners' Club

[Outback Bronze]

I only have the 13900K (not the KS) and it's in an MSI board that doesn't have that info available.

Here's my highly under-clocked 14900KS

View attachment 353627

That KS inst too bad. Looks like your P's are around 115. Your 5.9 vf is prob the pick of the bunch. This would be the clocks I would be aiming for with this particular KS.

To achieve 6Ghz+ on all P's & E's is no easy task, only the unicorn CPU's are able to this or have extremely exotic cooling like chillers etc.

What I'm finding more interesting atm is this cooler prediction and underneath that, the P-Core Light/Heavy.

Notice on my 101 on the above screen shot that my cooler prediction is 192 but on my better 105 SP KS its worse with the same setup. Please see pic attached:

Even the P-Core Light/Heavy is better and check out those temps @ 16°C. It still doesn't beat the 101 @ 36°C so something tells me there is more to testing these CPU's than just the SP.

Disregard the speed @ 4.5GHz as I have tested this, and the cooler prediction is the same whatever the clock speed is.

Edit: 101 Attaching picture here as it was on the previous page..

 

[fevgatos]

Has someone tried how well the chip undervolts at 5.5 ghz / HT off? I've found a 14900ks for a killer price but I'd like to know if on average the ks part is better at those low clockspeeds.

 

[Dr. Dro]

Has someone tried how well the chip undervolts at 5.5 ghz / HT off? I've found a 14900ks for a killer price but I'd like to know if on average the ks part is better at those low clockspeeds.

If this thread told me anything thus far, is that 14900KS chips are coming with a pretty wide variance on how well they're binned, looks like this time around Intel's approach was to get whatever chips managed 6.2 on two cores no matter what instead of the 13900KS chips that tend to be much better than the 13900K and on par/ever so slightly better than 14900K on average.

I think your chances are great, all things considered. If it's a killer price it'll be a great upgrade from your 12900K for sure

 

[fevgatos]

If this thread told me anything thus far, is that 14900KS chips are coming with a pretty wide variance on how well they're binned, looks like this time around Intel's approach was to get whatever chips managed 6.2 on two cores no matter what instead of the 13900KS chips that tend to be much better than the 13900K and on par/ever so slightly better than 14900K on average.

I think your chances are great, all things considered. If it's a killer price it'll be a great upgrade from your 12900K for sure

Yeap, that's what I'm afraid of, since the ks I'm finding for a good price are returned products. Feel like they tried them, were duds and returned them. Still 410 for a 14900ks sounds like a steal.

 

[Dr. Dro]

Yeap, that's what I'm afraid of, since the ks I'm finding for a good price are returned products. Feel like they tried them, were duds and returned them. Still 410 for a 14900ks sounds like a steal.

I'd be all over that and I chose not to flip my 13900KS and buy a 14900KS because there's no guarantee I'd get a better chip at all... and I'm really happy with mine. Just grab it brother.

 

[fevgatos]

I'd be ordering already but I'm waiting for that shop to stock a monitor I wanna buy too since it's outside my country, don't feel like paying 30 euros for transport fees twice.

 

[Dr. Dro]

I'd be ordering already but I'm waiting for that shop to stock a monitor I wanna buy too since it's outside my country, don't feel like paying 30 euros for transport fees twice.

Mmm I'd act on it man. It's 14700K money for a bloody 14900KS, even if it was the single worst 14900KS sample to ever exist it'd be a killer deal

 

[Super Firm Tofu]

Has someone tried how well the chip undervolts at 5.5 ghz / HT off? I've found a 14900ks for a killer price but I'd like to know if on average the ks part is better at those low clockspeeds.

I haven't tried with HT off, but I've been running at 13900K specs on mine (5.5 all core, boost to 5.8 on two, 253W PL1/PL2 and 320A ICCmax). I can only manage a 0.05 negative offset. Anything further is unstable. Stock speeds and power limits (320w/400A) weren't stable with any undervolt. I can't compare numbers directly because two different boards, but on my MSI board with my 13900K, I can run 0.07 neg offset, 253w PL1/PL2, and unlimited (512A) ICCmax. Comparing power consumption, the 14900KS at the same settings uses 20-50% more power than the 13900K.

I think what I've determined from all this is that my 13900K is just really good. I bought the 14900KS intending to run at lower speeds with a greater undervolt. That didn't end up working out. It's ok though, because the other goal I had was lightly tuned 8200MT/s and that worked out fine.

I actually tried to sell the Apex and the memory at a large discount, and was then going to sell the KS, but there was zero interest in the MB and RAM. I put it back together and gave it a second try and ended up where I am now.

TL/DR - If it were me, I'd probably skip a used/returned 14900KS if undervolting is your goal unless it has a no hassle return policy.

If this thread told me anything thus far, is that 14900KS chips are coming with a pretty wide variance on how well they're binned, looks like this time around Intel's approach was to get whatever chips managed 6.2 on two cores no matter what instead of the 13900KS chips that tend to be much better than the 13900K and on par/ever so slightly better than 14900K on average.

I'd agree with this, and in my opinion, I think getting to 6.2 takes way too much voltage no matter the silicon quality. I think it's just past all reasonable expectations for Intel 7. Even the voltage required for the stock 5.9 all-core isn't practical with anything other than a big custom loop and probably at least a delid, or direct die. I'm only running an AIO (although one of the better models - Arctic LFIII 420) and I'd thermal throttle before ever seeing 5.9 all core on any type of heavy load.

To be fair, Intel offered a replacement CPU if I wanted. In the end I decided to keep the original as I had no guarantee that a replacement would be any better, and I might lose the IMC that could do 8200. I guess you win some, and you lose some. Wasn't my luck this go around.

 

[Outback Bronze]

my 13900K is just really good.

Sounds like it!

I'd probably skip a used/returned 14900KS

Yep!

I guess you win some, and you lose some. Wasn't my luck this go around.

Yep!

 

[Dr. Dro]

I'd agree with this, and in my opinion, I think getting to 6.2 takes way too much voltage no matter the silicon quality. I think it's just past all reasonable expectations for Intel 7. Even the voltage required for the stock 5.9 all-core isn't practical with anything other than a big custom loop and probably at least a delid, or direct die. I'm only running an AIO (although one of the better models - Arctic LFIII 420) and I'd thermal throttle before ever seeing 5.9 all core on any type of heavy load.

I had a suspicion that the i9-13900KS is essentially the limit of what's practical on air cooling. You can see some performance degradation on the i9-14900K already, because of its more aggressive boost routines and comparatively loose binning, there's a minor improvement in single-threaded performance due to the extra 100 MHz, but also a decline in MT, particularly in sustained MT loads.

Truth be told, I reckon the entire "14th gen" is completely redundant, except for the i7 level. Even then all of those chips could be called 13650K, 13750K, 13950K and 13950KS... and it'd probably be far more appropriate than bestowing a full generation on these.

 

[ShrimpBrime]

I had a suspicion that the i9-13900KS is essentially the limit of what's practical on air cooling. You can see some performance degradation on the i9-14900K already, because of its more aggressive boost routines and comparatively loose binning, there's a minor improvement in single-threaded performance due to the extra 100 MHz, but also a decline in MT, particularly in sustained MT loads.

Truth be told, I reckon the entire "14th gen" is completely redundant, except for the i7 level. Even then all of those chips could be called 13650K, 13750K, 13950K and 13950KS... and it'd probably be far more appropriate than bestowing a full generation on these.

All the 14th gen (13th too) are not frequency or V-core limited.

I seen you mention earlier it seems like wide variations od SP ratings, truthfully it's probably the difference in cooling capability between you all (us).

I can safely argue that 5.8ghz all core on any i7 or i9 will take very adequate cooling.

In the OC world I come from, on ambient you OC to the throttle point (100c) and lean on memory performance over that.

Through experimenting on my i7 chips, I've found v-core was never the issue, but the cooling IS for sure the problem.

(I didn't buy into i9, it's just more Atom cores I don't need personally)

Back to SP, what's the negligible difference between 110 and 112SP? Probably none noticed at first glance.

I can safely argue with my i7 14700K that I need chilling to accurately achieve 6ghz all core, e-cores enabled but with pathetic overclocking.

_______

Observing the thread, p-cores is the main focus I think.... But why is everyone worried about 8 P-cores when they purchased into more E-cores? Is not the E-core performance MORE Important than P cores just for the simple fact we've bought into more area effecient cores?? Please help me understand this concept.

 

[Dr. Dro]

All the 14th gen (13th too) are not frequency or V-core limited.

I seen you mention earlier it seems like wide variations od SP ratings, truthfully it's probably the difference in cooling capability between you all (us).

I can safely argue that 5.8ghz all core on any i7 or i9 will take very adequate cooling.

In the OC world I come from, on ambient you OC to the throttle point (100c) and lean on memory performance over that.

Through experimenting on my i7 chips, I've found v-core was never the issue, but the cooling IS for sure the problem.

(I didn't buy into i9, it's just more Atom cores I don't need personally)

Back to SP, what's the negligible difference between 110 and 112SP? Probably none noticed at first glance.

I can safely argue with my i7 14700K that I need chilling to accurately achieve 6ghz all core, e-cores enabled but with pathetic overclocking.

_______

Observing the thread, p-cores is the main focus I think.... But why is everyone worried about 8 P-cores when they purchased into more E-cores? Is not the E-core performance MORE Important than P cores just for the simple fact we've bought into more area effecient cores?? Please help me understand this concept.

Compared to the Core i7, the i9 has a complicating factor - TVB. TVB is this extended range that reaches 6 GHz+ frequencies (going beyond the turbo target). Think of a turbo beyond the turbo, and that's what it is. It changes things considerably to that regard. 5.8 all-core is beyond the capabilities of all but the very best i9-13900K chips, around the ballpark of what the i9-13900KS and i9-14900K can generally do.

The 14900KS kicks it up a notch further with a 5.9 all-core TBT and 6.2 TVB target, and by that point, you are asking a LOT out of the CPU IMHO. As for SP, I don't really know. It's beyond my area of expertise, since I have an MSI board. I did some reading, mostly articles previously posted in this thread, but that'd be the extent of what I know.

 

[Outback Bronze]

Please help me understand this concept.

It's basically better silicon on the P core i9's is the main reason I purchase them. You can run them with lower volts at any given clock speed.

I just managed to this, this morning with 10°C ambient. Personal record for me I did try with my old 13KS and couldn't get past 6Ghz on all 8 Cores little alone 6.3ghz. Did try for 6.4Ghz but windows was crashing

This is really for testing OC purposes only. A realistic speed with this type of CPU would be between 5.6/5.9 all cores depending on cooling and silicon.

 

[Dr. Dro]

It's basically better silicon on the P core i9's is the main reason I purchase them. You can run them with lower volts at any given clock speed.

I just managed to this, this morning with 10°C ambient. Personal record for me I did try with my old 13KS and couldn't get past 6Ghz on all 8 Cores little alone 6.3ghz. Did try for 6.4Ghz but windows was crashing

This is really for testing OC purposes only. A realistic speed with this type of CPU would be between 5.6/5.9 all cores depending on cooling and silicon.

View attachment 353743 View attachment 353744

Oof, had to disable HT and E-cores to bench 6300 on CPU-Z... insane and utterly unrealistic, I love it!

I'm running on air right now, would not dare attempt this feat

 

[ShrimpBrime]

It's basically better silicon on the P core i9's is the main reason I purchase them. You can run them with lower volts at any given clock speed.

I just managed to this, this morning with 10°C ambient. Personal record for me I did try with my old 13KS and couldn't get past 6Ghz on all 8 Cores little alone 6.3ghz. Did try for 6.4Ghz but windows was crashing

This is really for testing OC purposes only. A realistic speed with this type of CPU would be between 5.6/5.9 all cores depending on cooling and silicon.

View attachment 353743 View attachment 353744

Yes, these numbers seem very similar to mine actually.

P-cores only But something a little bit heavier than cpu-z benchmark though.

Are you SURE that the binning is better between i7 and i9 and not the cooling???

Compared to the Core i7, the i9 has a complicating factor - TVB. TVB is this extended range that reaches 6 GHz+ frequencies (going beyond the turbo target). Think of a turbo beyond the turbo, and that's what it is. It changes things considerably to that regard. 5.8 all-core is beyond the capabilities of all but the very best i9-13900K chips, around the ballpark of what the i9-13900KS and i9-14900K can generally do.

The 14900KS kicks it up a notch further with a 5.9 all-core TBT and 6.2 TVB target, and by that point, you are asking a LOT out of the CPU IMHO. As for SP, I don't really know. It's beyond my area of expertise, since I have an MSI board. I did some reading, mostly articles previously posted in this thread, but that'd be the extent of what I know.

5.9ghz all p-core and 4.4/4.5ghz E-Core is about all I can muster on water. Seems legit!
For me 14700K, 5.8ghz p-core and 4.7ghz e-core is actually better All core package performance. At least from what I've observed from any messing around.
Mind you I am on a full OS W11 install ready to game no special tweaks or stripped OS. Most benchers take this approach, I refrain and use my bench rig as a daily driver. Seems I can push the best of both worlds from here. I do have an Ice res. Dice and LN pots. Don't do much LN2 because of pricing these days. But 7-7.2ghz on Dry-Ice 2 or 3cores seems to be about the limit. Get the same exact blue screen at 6.4ghz on chilled liquid. Same exact blue screen 6ghz air. 6.2ghz on ambient water. All the same blue screen. Clock watchdog timeout. In the past meant more v-core, I've decided it means get colder these days.


Also have MSI boards. My Asus B660-G doesn't seem to display SP figures. (I'll have to take a look, but they don't mean anything if only Asus owners can observe them) But I think the E-cores on ambient cooling at 4.7ghz to be pretty darn stable as long as I don't breach the thermal wall too hard. Can get 5.8ghz P-cores in conjunction. Unfortunately my DDR5 Z790 MSI board is out for RMA. It crashed during a CB20 run and never turned on again. Luckily everything else is fine and in working order.

Edit: Let me toss in that am using the HeatKiller IV waterblock.

 

[Outback Bronze]

Are you SURE that the binning is better between i7 and i9 and not the cooling???

Yeah, pretty sure. The SP ratings of 14900K R batches are pretty good little alone top 14KS's.

Gees what cooling you have? I was only trying to boot up @ 6.3 + little alone get through a cpuz. You managed to get through a cinebench. I'm only running 1x 480 rad in push

 

[ShrimpBrime]

Yeah, pretty sure. The SP ratings of 14000K R batches are pretty good little alone top 14KS's.

Gees what cooling you have? I was only trying to boot up @ 6.3 + little alone get through a cpuz. You managed to get through a cinebench. I'm only running 1x 480 rad in pull.

Yeah, a lot of the 14900K/KS chips seem to clock maybe 100 or 200mhz higher than 14700 given similar cooling. Doesn't seem to be a huge difference.

For my 14700K to actually PASS that benchmark, it took DryIce. Same settings on chilled water loop just wasn't cold enough for the amount of v-core to remain stable and maintain the performance at the same time. CBR24 is extremely difficult to pass. Many more failed attempts than successful ones I can tell you that In fact I have grown to actually dislike it. More challenge that I need haha.
_____
5.9ghz P-core only HT enabled ambient. This is the line where e-cores need to be shut off because wattage. Can chill 6ghz all P and E cores. Maybe 6.1ghz. But the E-cores would need to be 4,4ghz. Seems counter productive. Them Atom cores though. Why not just some more P-cores for the i9's (speculation inquiry, I know it's not going to happen). I believe 6ghz P-cores hold much higher value to everyone, that's why 14900K sells I suppose.

Did you know that I've experienced benchmarking 3D where turning off E-cores raised the GPU score quite a bit. I'm not sure exactly why, but not always does it seem to benefit having e-cores enabled (or at all for that matter-Opinion)

 

[Outback Bronze]

5.9ghz P-core only HT enabled ambient.

This is pretty much what I run my cpu max at also, except with HT off. 6Ghz+ is really just going for gold and I'm not too keen on degrading my KS just yet with more than 1.6v.

I would love to see more P cores too

Did you know that I've experienced benchmarking 3D where turning off E-cores raised the GPU score quite a bit. I'm not sure exactly why, but not always does it seem to benefit having e-cores enabled (or at all for that matter-Opinion)

This could be because I have heard that if you turn all the E cores off that the P core's get all the L3 Cache to themselves, or that the threading on different benchmarks might prefer less cores.
I'm sorry, I can't answer that one correctly as I really have no idea, just speculating.

 

[Vayra86]

OMG. Why you people torture the chips this way.

You can run it for 5 minutes, if it is stable lower settings somewhat and you are stable with minimal chip stress and degradation.

This is not how you OC proper though. 5 minutes is great for the ignorance-is-bliss-stability on a chip at best. You said it yourself, minimal chip stress. But you do a stress test exactly to simulate maximum chip stress. Which includes full and sustained heat / power cycles.

What you are describing is the approach of 'YOLO-Stability'. The kinds of overclocks that fall apart a few years down the line, or when they suddenly encounter a prolonged AVX load, or during summer, etc. Its great to epeen with on a chip you'll never even stress proper while gaming, sure. But a real OC it is not.

 

[ShrimpBrime]

This is not how you OC proper though. 5 minutes is great for the ignorance-is-bliss-stability on a chip at best. You said it yourself, minimal chip stress. But you do a stress test exactly to simulate maximum chip stress. Which includes full and sustained heat / power cycles.

What you are describing is the approach of 'YOLO-Stability'. The kinds of overclocks that fall apart a few years down the line, or when they suddenly encounter a prolonged AVX load, or during summer, etc. Its great to epeen with on a chip you'll never even stress proper while gaming, sure. But a real OC it is not.

Generally when I run 6.4ghz all core and higher on better than AIO cooling, it's much longer than a mere 5 minutes. Typically 2 to 4 hours at a time.

Going from base clocks to max all cores to the turbo all core boost clock shouldn't be too bad on the cpu in most cases. Probably could still run a slight under-volt and keep longevity in the picture.

My 14700K being up around 1.7ghz on a couple of occasions hasn't really shown any significant degradation, no doubt I'm slightly more abusive than most users. Not saying degradation isn't possible, I just haven't encountered it noticeably yet.

 

[Dr. Dro]

Yeah, the chip won't degrade from sustained load unless it's operating way out of its electrical and thermal specification. Pretty insightful stuff trying these out on exotic cooling. I got mine because I like having some CPU grunt but at the same time wanted something that was well-behaved and could undervolt well, to that end, it delivered all I wanted

 

[ShrimpBrime]

Yeah, the chip won't degrade from sustained load unless it's operating way out of its electrical and thermal specification. Pretty insightful stuff trying these out on exotic cooling. I got mine because I like having some CPU grunt but at the same time wanted something that was well-behaved and could undervolt well, to that end, it delivered all I wanted

So what about that e-core clocking? How do your guys chips go? 4.6ghz was about all I could pull from a 13700K and was unstable at times.

 

[Outback Bronze]

insane and utterly unrealistic, I love it!

Managed to boot 6.4GHz & 6.5Ghz on 2 cores this morning. Got through some cpuz too

Notice the 6.4 scores are a lot better than the 6.5 scores, so not sure whats going on there..

At close to 1.6 volts this is where my OC adventures stop on custom water.

 

[Dr. Dro]

Managed to boot 6.4GHz & 6.5Ghz on 2 cores this morning. Got through some cpuz too

Notice the 6.4 scores are a lot better than the 6.5 scores, so not sure whats going on there..

At close to 1.6 volts this is where my OC adventures stop on custom water.

View attachment 353911View attachment 353912

View attachment 353913View attachment 353914

It's probably clock stretching by that point. It's pushed very far, could be unstable or some protection mechanism kicking in.

So what about that e-core clocking? How do your guys chips go? 4.6ghz was about all I could pull from a 13700K and was unstable at times.

I haven't tried overclocking my processor at all... the 13900KS's E-cores run at 4.3 stock.

If I could get like a ~5.8 P and 4.5 E i'd be happy, but truth being told, I am already happy with my chip as is

 

[ShrimpBrime]

I haven't tried overclocking my processor at all... the 13900KS's E-cores run at 4.3 stock.

If I could get like a ~5.8 P and 4.5 E i'd be happy, but truth being told, I am already happy with my chip as is

Until a faster one is produced anyways lol.
I've been happy with at least 150 cpus and 150 gpus documented through the years. All overclocked XD. I just enjoy the cooling modification part of the hobby. The benchmark score is the results. I have mixed gaming in the past at 7ghz on FX processor running LN2. Just to see if CSS benchmark was all good and if gaming would actually be possible at this speed. It was. Seemed just as stable as running stock actually. Was quite thrilling in some small sense. Had to keep LN in the pot as there's no cold bug to worry about.