As someone who waits 6-8 years between system upgrades (and then builds a system a gen or two behind), thank you for your service as a beta tester, Jay 😁
@@xellestar Thats 10 more than an AMD5950x! It must be pretty good. Honestly, I can't follow CPU numbering these days, it makes no sense. At least AMD's past few CPU generations have followed a familiar pattern. If they stick to AM5 for next year, what will we well? 10050x?
@@MOSMASTERING Nvidia have almost gone full circle - my Win98 machine has a 6200 AGP card in it. I really hope the RTX 6000 series has a 6200 product in there, so I can start sending bug reports on the latest games and confuse the hell out of their customer support 🤣
Overclocking has entered into it's practical limit at 10/7nm. Going from 7nm down to 5nm, electron leakage has went up -requiring the same current level to almost to where we were at 28nm. This means that we're seeing chip arch designers able to hit max freq as current being the hard limits, as a secondary, thermals. (Leaving little overclocking headroom if any) Larger node sizes would hit thermal limits before current limits when driving frequency. With it flipped, current before thermals, thermal runaway becomes an even bigger issue, to where we are seeing CPU's begin to internally throttle at 70C. This is also why we're seeing chiplet/tile designs and wider execution paths in Zen5 and ArrowLake. I'd be curious if LN2 overclocking these does. Cryogenic temps lowers drastically leakage.
Mostly True. Power consumption has not reduced as much as it did in earlier node shrinks. Leakage currents make it harder to maintain efficiency, so while newer nodes are more power-efficient than 28nm, they’ve been less efficient in shrinking power consumption relative to older node transitions. As fabrication technology shrinks (from 10nm to 7nm to 5nm and even smaller), electron leakage does indeed become a bigger problem. Smaller transistors have thinner gate oxides, and quantum tunneling causes leakage currents to rise, making it harder to push higher frequencies without significant power and heat penalties.
Considering that smaller processes aren’t actually smaller, just different configurations with different names. The 10nm takes up almost as much space as 7nm. They are just marketing names, relax
@@nerdynumen Backside and GAA will be introduced, hopefully, next year by Intel. Until such time, FinFET is seeing a higher leakage with every fin pitch shrink. Gate shrinks have been negligible since 7nm. Backside power should be a more impactful innovation to hopefully address the issues we're seeing with today's bleeding edge nodes.
On 12th gen you could BCLK overclock non-k chips to the moon and back. 5.3GHz on a 12100 is quite rad. I've got a china special 12490f on a Asrock B760 PG Riptide, stock 4GHz all cores and with a crappy cooler and some BIOS tweaks 4,6 all cores and 6800MHz ram. If you want to overclock there's always an exotic solution to be found. Imagine a Celeron G6900 2c/2t running at 5GHz... You'll basically have a 15th gen processor in you hands !
You needed to watch DerBauer's newest video. He explains all the overclocking details. You need to keep the cpu under 70C. anything over and it limits itself. He was running 5.6ghz pcores and 4.9ghz ecores and getting 47.1k R23 scores. These new cpu's don't work like the prior gen ones.
Oldskool overclocker here. Won the Celeron600mhz lotery back in the day and boosted it to 900mhz. 1200mhz was doable but i was too scared to run that 24/7. Undervolting is my new hobby now. 12700k and 4070 5% less fast and system is quiet and going stable at way less wattage. So yeah, undervolting olé olé
I get more performance from urndervolting and limiting my temperatures. I have a 9950x I think it was limiting my performance hitting 95C. By dropping 30W and limiting to 85C, I get my best scores. Higher than out the box?
7:39 PL4 is another "internal" power limit, "internal" because the implementation of Turbo Boost Technology 2.0 only requires configuring PL1, Tau and PL2. PL3 is disabled by default, but it would be "A threshold that if exceeded, the PL3 rapid power limiting algorithms will attempt to limit the duty cycle of spikes above PL3 by reactively limiting frequency", and PL4 is "A limit that will not be exceeded, the PL4 power limiting algorithms will preemptively limit frequency to prevent spikes above PL4." So it would basically react more aggressively than PL1/PL2, usually at a higher value for the limit. Because with PL1/PL2 you can still have a bit of overshoot for the actual power draw, for PL4 it would not allow any overshoot whatsoever.
I have to agree with GN Steve where he said, "these chips simply make no sense". Too expensive, requires motherboards which make them the most expensive platform intel has released in many years, suck way too much power, yet in test after test get beat not only by AM5 but often by their own cheaper previous gen and even AM4. Oh well maybe in a couple years (If they don't end up killing themselves like Raptor Lake) I'll pick up one on the used market.
I hope people didn't really want them to be faster than 14th gen out of the gate when 14th was an abomination that basically came highly overclocked out of the box
Intel needed to make inroads into the laptop market since so many people buy laptops where efficiency matters; they needed to pivot and it takes a turn to make a turn.
It was 59 in my room before I started up my 11700K/RTX 3080 machine. It's been just idling in here for about an hour and it's up to 64, I definitely use my PC as a space heater as the weather cools off!
i honestly dont understand how things work now, i used to think the one performing better will get better result now, or in here specifically, isnt the case and im waiting and earning just as much as anyone else even same as one not uploading anything
Disagree. It's the cool thing to do, a gimmick, then everything else. 😂 Please consider that in the time of 5% generational gains, that "only" 5% gain from OC starts sounding real nice. 🧐
@@71janas and what did that translate to in applications? Let me know how much of a benefit you got. Did you get a 17% increase in FPS? Did you get a 17% increase in render times?
Electricity can be compared to water: Voltage = Pressure, which is require to overcome resistance (Ohms) and is why low Voltage can't electrocute you. Not enough pressure to overcome the resistance of the human body. Amp = AKA, current is the flow rate, how many gallons a minute. V x A = Watts or total power. Ohms is what controls how much Amps flow based on how many Volts is being pushed through that resistance.
@@monk3synuts V=IR or I=V/R or R=V/I V = Voltage R = Resistance (Ohms) I = Current (Amps) Which is why 0 Resistance = runaway current because anything divided by 0 is undefined. Either way, that quote is incorrect, "Ohm's law states that the electric current through a conductor between two points is directly proportional to the voltage across the two points." That, and I was trying to keep it simple, AKA comparing it to water pressure.
@@cloudanu Voltage is like the spiciness of an onion. The stronger the onion, the more it "pushes" tears to your eyes. Similarly, a higher voltage pushes electrons more forcefully through a circuit. Current, or amperage, compares to the number of onions you chop. If you cut just one onion (low current), there won't be many tears. But chopping ten onions (high current) will release a much stronger concentration of "spice" in the air, just as a higher current means a larger flow of electrons. Resistance is like the thickness of an onion’s skin. A thick skin makes it harder to reach the core of the onion (or the electrons in a circuit). A thinner skin provides less resistance. Power represents the combined effect of spiciness (voltage) and quantity of onions (current). If you have a very strong onion in large quantity, the effect will be intense, similar to high electrical power. Conversely, a mild onion in small quantity will produce only a weak sensation. There you go brother
Overclocking has changed alot over last few years. Gone are the days of doing a static overclock and getting the absolute best performance. Key to overclocking now is to keep your CPU as cool as possible due to the parameters that are built into the microcoding with boost being temp and power driven. I've noticed as you can get around 3-5% more performance on single core activities but the real tuning of overclocking is when you start using multicore. Perfect example is 9950x. I was only able to get around 3.5% more in ST CB23 but MT jumped up from 42k to 47k about 11%. And with controlling the temperatures the boost clocks tend to stay higher for longer periods.
Jay, derbauer today released a video and gets 5.7ghz all p-core and 4.9ghx all e-core with over 47600 score in the same benchmark, maybe his video might help gain info?
@@blackbirdpctech Jay's channel caters to normal tech bros. If you are looking for unreasonably technical analysis, then go to Debauer and other similar channels. Jay is for creating entertaining content for tech people who can't stand getting their entertainment from television the way normal people do.
Dyno! As a tuner I see better results the longer you run (if the cooling system is good) Gearbox and diff oils as they get hot I gain about 7kw But yes, temperature is detrimental to power, I target 80-90°c on track, beating on it, temps. 98° I start going for the temp reducing stuff... Run it rich, pull power down.. or straight up install a water sprayer. It's surprising how little water you need to smash temps down like a beast. (Our cars won the D class Bathurst 6hr last 2 years)
Yeah I really don't care Jay, i just want to listen to you yapping in the back ground while i game. Coming from a 12900K i expected so much more, Thursday (10/24) i walked into Microcenter, got the Z890 board i wanted and they only had 265K, so I impulsively bought it. Today 10/27, i have officially returned both items, with the exception of exchanging the motherboard for a X870E chip-set, ill be waiting for 9000x3D series to officially switch back to AMD since the 2011's upset FX-8150 i bought a that time. Except this time, I'm not holding, I already swapped one major piece, now waiting for that CPU to drop. This feels almost exactly like 2011s FX-8150... its been 13 years of Intel for me, since those days and here I am... I'm going back!
Intel pretty much pooped the bed by setting on their hands from Haswell till first gen ryzen. Imagine if they would have done something different for five generations aside from telling use four cores and eight threads was plenty.
Question, why did you even buy the 265k on release day? I assume you didn't wait for the reviews to go online to watch them first because then you would've known how bad these CPUs are before going through the hassle of buying one and testing for yourself. Could've saved yourself a lot of time there. I'm eyeing 9000X3D as well, but even though I have a good feeling on it, I will be waiting for those reviews and then will probably wait another few weeks after that to ensure no really nasty issues are discovered in that time.
@@123Suffering456 The simple answer is, this is why I wait 3+ years to upgrade, naturally it should be better. Sure productivity apps was a nice uplift, but I'm here to game on PC, and there the difference was negligible.
@@Mystikalrush I doubt it, while not much faster than previous gen the 265 is quite the improvement over a 12900K, which doesn't even keep up with the 13600K most of the time. You're surely gonna get a faster gaming CPU with the new X3Ds but paying basically 500 bucks for an 8 core in this day and age is stupider than buying an Ultra 200 on release day imo
@ekifi Hahaha your all wrong. Go tell this to Jay, all hundreds of reviewers and thousands of customers, you are hilarious. Anyways, I'm already in my 9800X3D setup, absolutely crushing everything that shirty 265K couldn't and even my 12900k. Obviously I'm here to game, not compute.
The Board enhancements for this intel series was what I was interested in this time around, before even the CPU speeds. I was really curious how well the chipsets will manage the expanded options and bottlenecks that come from the consumer grade CPUs. You can fit a lot in a compact z890 mini-ITX build with adequate thermals more easily. Which bring it closer to what's needed for +90% workstation related work. Having a smaller, quieter, and colder system close to you while you work is something I appreciate. Less in the server and less I need to off load to another system to render or do sims on a workstation. Less money spent on creating/cooling/maintenance/space on multiple systems/workstation in a way that makes it quiet enough to have close to me is a good thing for me at least
It's obvious those CPU get out of factory already close to the limit (I mean, last gen was destroying itself for going too fast). The days of 2.6GHz oc'd to 4GHz are decades old.
Nah, the CPUs just overclock (aka "turbo") themselves automatically now. Clarksfield (mobile version of Nehalem) was the first chip that really showed what it could do beyond a measly couple hundred mhz. A 1.6ghz quad core with the maximum clocks of 2.8ghz previously only seen in dual cores, dynamically adjusted according to load. And it always hitting either a temperature or power limit because it was a desktop chip in a laptop and stupid Asus blocked off the intake of the ONE fan (G51J, along with its desktop derived "GTX 260M" which was really a 55nm 9800GT, the thing was a stupidly hot fireball of a gaming laptop)
Overclocking my 7800X3D was like juicing sand. Overclocking my 7900XTX, on the other hand, was like the 2nd best athlete in the world took ROIDS! A Windows update reset my overclock, and I spent forever tuning it😭 Does anyone know if there's a way to recover settings?
No way to recover but you can at least save settings. Overclock in AMD's own software - you can save your settings as an XML file. Their software is quite conservative and it'll disable OC settings due to any unexpected shut-off including power loss, holding the power button down for a hard shutdown, etc. so it's good practice to save some good settings if you come across them.
Answer to your power limit issue Power Limit 4 (PL4): A limit that will not be exceeded, the PL4 power limiting algorithms will preemptively limit frequency to prevent spikes above PL4. Turbo Time Parameter (Tau): An averaging constant used for PL1 exponential weighted moving average (EWMA) power calculation.
Falcon NW is straight up ripping people off at 7500$ with that system. The exact same system costs roughly 4500-5000$ to build. If they charge 500$-850$ because of their warranty and such okay, but its 2000-3000$ (depending on the exact components as they are fluid and do change from availability). There is absolutely 0 reason to be overcharging THAT much and expect people not to be mad at it.
Go try to buy a new 4090 these days and see how much you pay. The system cost of goods could possibly be closer to $5500. Yes, the warranty does have a cost, but let's assume their total cost of parts, labor, marketing, testing and support/maintenance is $6k, that's about 20 points of profit margin per unit, not an insane profit margin for a boutique low volume system. Low volume means higher risk of not moving enough units to break even. Company like Falcon Northwest shouldn't even bother selling a system with a profit margin below 15%. Dell on the other hand can still build a business on selling low end laptops at 12% margin because they know the volume will make up for lower margin.
The M.2 slots are impressive. I had the first one on my MSI Z170 board and one on my C232 workstation board by HP. Ever since then I have two on all my mainboards (Z370, B550 & X570). But yeah - M.2 drives are becoming more and more a serious option. Like in 2019 I paid as much for a 1TB PCIe 4.0 M.2 drive as I paid for a 4TB one this year (as well as a 14TB HDD). As soon as you get like 8TB ones for the money I can see myself slowly ditching my HDDs.
I have always had a mid tier system, like 3070, 2060 super, etc, i finally have a high end build. 4080 super 7950x3d 64gb ram 1000 psu. Build costed me 3500 bucks, I have many extras in it as well. Not nessasary but i love it!
I do kinda find it hard to get interested in new tech stuff when I know it's something I won't be able to ever remotely afford or make use of. Its like, how many supercars can you look at before you just don't care anymore, they're all effectively unobtanium and just kinda blur together. I paid less than 7500 for my car, I would never spend that much on any computer
At least when it comes to CPUs prices aren't creeping up that much gen over gen, sure these 16/24 core 9 series units aren't cheap but they're part of lineups that start from the low end while still offering the latest archtectures and silicon. It's not like every new CPU is 600 bucks
The sad thing is, the 13900k/14900k voltage spikes seem to be finally be under control and temps are under control now which makes benchmarks more predictable than bouncing against thermal limits. The best Intel chip now is the 14900k (I won't say KS because that may be touching the limits of this generation and I had to RMA 2 of them, admittedly before the latest microcode and BIOS updates). Certainly won't be upgrading my 14900k, my family 13900k machine or my work office 13900k system.
I can't blame falcon northwest for making the silly system, this is what some people want. This is like the 200mph automobile, you can never go that fast unless you are insane or on a track or a very few special places in the world which are most likely thousands of miles from the owners home. Yet people have them. You can poke the company that makes the components, but they probably do have a niche case user here and there. Some folks just have money to throw at things so they feel better about whatever purchase, or don't care as they are not paying for it themselves like in a business scenario and the entire expense is a tax write off for someone.
@@reptilespantoso People deserve good products for how overpriced it is. It'd be like buying a lambo but the motor falls to pieces if you press the gas pedal more than halfway.
These parts and build are like 4500 5000 grand max, so bling means you get to rip people off? What is the extra 2500 bucks for? The actual build?? Lol yeah okay
P cores probably only make sense for 1-2 thread boosts, otherwise giving power budget to the E cores seems to be the way to go. also the DLVR is burning A TON of power in high load scenarios, so the next priority would be reducing the switching losses of the DLVR by optimizing the voltage levels so the input voltage can be closer to the output voltages. you probably want all the voltages fed by the DLVR to be as close as possible - so pvcore, evcore, bus and another thing i forgot. check skatterbencher, he has a great breakdown. also, TSMC silicon seems to be scaling more towards lower voltage instead of higher voltage like intel silicon. you could hammer raptorlake with voltage and it would just go up in clockspeed, but it seems like the TSMC manufacturing plateaues off at mid 5 ghz. but i have seen quite some remarks about undervolting. i think arrowlake actually gets better the lower the power is, not only more efficient, but also faster, when you turn the right knobs.
Try the direct-die cooling from der8auer. The CPU is 30°C cooler and gives a lot of headroom for OC. He got >47000 points in cinebench while overclocking P and E cores.
Delidding is a different level of OCing what jay is doing is what most people would do that is interested in tweaking, Delidding is up there with LN2 shit thats extreme crap.
@@Aaron-zl5gq What? No. Delidding is as easy as putting on the paste with the right tools. der8auer/Thermalgrizzly sells the delidding tools and the fitting waterblock. The last time I did, it was a 7700k and it was a matter of putting it in, twisting a screw and putting on liquid metal, done. (it was not soldered) It was 30°C cooler after that, WITH the stock IHS back on. It was 50$ for the delidder, liquid metal and special glue to put the IHS back on. But even soldered, you just need to heat it up before doing it, the rest is the same.
so technically it isn't heat that causes degradation. The melting point of silicon is a lot higher than 100c. The issue is electron migration which was always going to be the thing that killed the Moors law. We are reaching or have reached the limits of how small we can make pathways with the given materials. Eventually they get too small the slightest heat change will cause a short or get close enough together to allow electron migration to occur.
Not surprised by the Cinebench oddness. With 14900K the E-cores offered 50% of the thread count. Now, with the 285K and no hyperthreading to make optimal usage of the P-cores, the E-cores now handle 66% of all threads. So obviously they're more important to overclock than the P cores which suck down too much power and hold back the E cores. For gaming it'll be the opposite, crank the P cores. Maybe even try disabling a few or all E cores to get the power budget back.
"Zero side thinkers" Yeah that applies to so much of society today.. really insane how people just can't think for themselves and never see things unless deliberately laid out for them (and even then can still miss it). Is critical thinking even taught in school anymore?
Critical thinking threatens traditions and the oldschool ways of doing, thinking, and saying, so people in positions of power who adhere to them benefit from curbing the spread of critical thinking. It's the same reason why civics isn't taught in schools.
It wasn't much better historically. Human ability to process information is limited and the amount of information is much greater than it was in the past.
It is sad that they put 6 or 7 M.2 slots on a board and in most cases you are lucky to use Half of them or end up with X8 instead of X16 PCI-e or you loose the PCI-e slot all together as they are shared lanes.
@JayzTwoCents - Get Romans 285K delid / heater and try again. He got 20-30 degrees savings, it would be cool to see these tests your doing with the complete removal of temps from the equation. And ya ... even if you only delid one CPU, even though you only have a test pool of one, it would be cool to see hyow it affects your wanderings thru bios voltage settings and ITU tweaks.
I still think $7500 is way too much for what you’re getting. I’m pretty sure people had more of a problem that it seemed lacking in features, add-ons, and higher end hardware for what you’re paying.
It's silly. You can increase the price of a PC by stuffing it with a lot of unnecessary stuff like multiple drives, crazy amounts of RAM, an ultra-premium MOBO with features you'll never need. But why? Except for a very few niche users, it's just for bragging rights and people who have more money than they know what to do with.
@@BobBobson Did this build in PCPartpicker except for the case, its roughly 4500-5000$ depending on specific choices. You are paying 2000$ minimum overcharge for the system.
Yeah this price is goofy. It's $7500 for maybe $5000 in hardware, and there's nothing custom about it. It's got a paperweight cpu, no custom loop, and even a boring case. I get that every business prices their products to make a profit, but this is a massive markup for a very basic setup with expensive parts... Lowkey annoys me that Jay is actually trying to defend this insane price tag.
Gamers are a strange tribe from a parallel Universum, strictly believing that more RGB makes the PC faster, and the difference between 102 and 105 fps helps them winning
LOL we don't care about RGB (although it is pretty), we just like to "tweak" things to maximize performance per dollar. There's something about getting an 11 when you pay for a 10.
@@NJ-wb1cz And they will try to support that logic like what happens if the building is on fire and that 3 seconds matters cause you know I'm a music producer or visual effects artist but I am really a 13 year old getting my ass wiped by mommy.
@@waynetuttle6872, yeah. It’s highly software and architecture dependent. My workflow would probably be a fair fight between a 14900k or X3D and a dual Xeon workstation. You’d be surprised how much larger CAD and CAM projects can benefit from high core counts and lots of cache with more recent software updates. I was tempted to try a desktop platform again, then AMD had its issues with scheduling, Intel started burning up ring busses, and ECC support was questionable etc. I’m not going to fight with my tech to get a marginally useful single threaded improvement and questionable longevity/stability.
@jayztwocents As far as I understand, Raising the simple view P Core multiplier does nothing if the TVB is set to limit 8 core to 5.4; TVB overrides the general setting. That explains the initial weirdness.
I've found that on my 10th gen, TVB is just all around bad when it comes to hitting desired clockspeeds. For some reason the Intel utility will do everything it can to prevent it from hitting the target frequency when TVB is enabled.
They had to make it harder I’m sure sooo many people were killing their components cuz they didn’t know what they were doing and just threw max voltage at it and prolly instantly killed the parts.. that’s why they have made stuff harder cuz humans are stupid and will fuck stuff up so they have to save us from ourselves! lol it’s sad I know
@@Pasi123 absolutely - my trusty celeron 300a was my first (and only) cpu to run stable at 100% oc. (it was still just a celeron but i was a happy camper)
@@AtomskTheGreat the thing is, modern CPUs are overclocking themselves. The same Celeron 300 could've automatically boosted to 600 or even more if we had modern algorithms back then
Jay, make no mistake, your audience very much appreciate you burning hours of your life to over clock out a hot turd so none of us have to Your sacrifice does not go unnoticed!
I mean Intel literally said in the announcement that pushing E cores are gonna result in far more advantages that the P cores. In fact even you said that same thing in your first video covering the 285K
IMHO I think the loss of hyperthreading is why there's such a massive loss of gains with OC now on P cores. There's simply a lot more E cores now than P cores, so pushing those hard their gains add up together to be more real performance than pushing P cores slightly, even though P cores at the base level are far more powerful.
This "backward" behaviour is the same as Ryzen Zen 4: My 7700x doesn't like high voltages. There's a sweetspot. If I want more performance, I need to cool it more. Yes it's counterintuïtive, but also logical if you think about it. These chips are now so dense and small they can't handle high voltage, high heat at the same time. Sure.. you can cool it to the extreme, and then blast it. But I need this PC for work. 🤷♂
@@JohnWalsh2019it has nothing to do with DEI. Pull your head outta your arse. Intel has been laying people off in the 10s of thousands as of late. Further, the leads of the long term CPU core team (the likes of the legendary Jim Keller) were let go earlier than anticipated so those projects got interrupted or cancelled altogether. Quit trying to blame DEI for something that it has absolutely nothing to do with.
@@TheKazragore can't accept DEI hiring practices are based on certain "check boxes" rather than talent can have an impact on a company....okay bro. If you don't think this is Intel, just check their website, they have fully embraced it. Also, I worked at Intel (not on the CPU side) for two years. It's easy to see when a company is taking a turn for the worse. Good luck!
When you switched to the Asus OC profile what was the cache OC set at? Did you put it back to 3.8 before you ran the first Asus OC profile cinebench run?
@@NJ-wb1cz Yeah, and if you watch Hardware Unboxed Intel review its all about how much they think they suck as opposed to just telling you what tests have shown Same goes for many others
@@gingerbread6967 HUB are well known Intel simps. If you need people to be even more biased towards Intel, you would probably benefit the most from searching your own feelings towards corporate brands
Jay, you need to use the shutdown command with the /fw switch (fw for firmware). No more smashing ESC, F10, or whatever. The shutdow command will boot you into the BIOS. e.g. C:\> shutdown /r /f /t 0 /fw
@@outlet6989wasting so much time for a potential stable oc is not worth it, you most likely make more $ per worked hour to buy a better CPU/Mobo or faster ram instead.
I am letting the video run J. I..... for the first time in my life, am considering a full AMD build. Intel is YEARS of "not being dicks" away from me considering their chips. Nvidia is most likely having multiple meetings TO NOT make the joke: "50 series will start at $5k ha ha" a reality. I am doing a full system, set-up, peripherals, and furniture upgrade.
13:30 got the same thing on 5700X. Tried UV it to save some power and thermals, but despite higher clocks with lower voltage, CB score would be lesser than stock. AMD calls it clock stretching, so even *if* core boosts higher, internal governor won't let it because it doesn't have enough power and would crash. Or at least that's what I think it is, just on Intel's side
It's interesting seeing the difference between the two generations. It's also a little disappointing that the 285K isn't much of a jump on the 14900K here. A 14900K is around 40,000 on Chinebench too, so we're really in the same ballpark. However, what's interesting is how the new gen gets there because it seems to be all on E-core here, and it's not something you can duplicate on a 14900K. A 14900K won't run its E-core up to 5GHZ. I can push to 4.6GHz on my own, but it's not happy about 4.7Ghz. It does matter the voltage profile either. What you'll also notice directly against a 14900K is that the higher performance cores aren't gaining you anything. Because it's still a heat bottleneck, you're still only up to around 5.6GHz on the P-cores, or less, with a pretty hard thermal limit on any current off the shelf AIO you can buy. So the 285K seems to give up on peak GHz on the performance cores (for long term stability maybe), and then ramps up those E-cores a good bit to get them to do a bit more heavy lifting. An interesting note on the power limit. The 14900K will also do this despite having limits disabled. I'll generally see this once I go north of 385W for an extended period of time. It'll hard lock it down to 252W for a little bit. It'll just trigger in the background. I'm curious what the 285K is using for its definition. Too bad it's not accurately showing power, so...who knows. GamerNexus is plowing into that whole mess, so it'll be interesting to see how these compare on actual wattage. GamersNexus does seem to indicate an ok improvement on efficiency overall meaning the watts are at least being utilized better to some extent. One thing you can do is tune the individual cores a little. There can be some hot cores, and you can even them out some. What I've found is this lets the other cores get a little more, and I especially saw a bump up in what it was allowing for peaks overall on both P and E cores. You can use any monitor like HWiNFO64 or whatever to see the individual cores. Bump them up and down to even out the whole thing. As part of this, I'd be curious to see if you can bump up one or two P cores to north of 6Ghz stably. Since one major hit the 285K has is single or low core count activities, and it starting at a mild 5.7GHz isn't doing it favors. I'd be curious to see some individual core tuning to see how high it goes with stability and without thermal overloading on that core. The per core fine tuning and dialing in the single (or couple) core peak, this might bring the gaming performance back inline with the 14900K and others.
Has it ever occurred to you that the errors you get are user error? Every other reviewer was able to run CUDIMM but not Jay. Same as the memory problems you had on AM5. Suggest you put the bios on Basic instead of Advanced next time.
E-core favoring is probably cos physical layout for the cores are in 1P-4E-2P-4E-1P per row... if it were 2E-2P-4E-2P-2E instead... might have been able to push P-cores more...
I was wondering when this would start to happen... It's LGA 775 all over again. The mobile /laptop stuff was super overclockable but the desktop stuff was so hot that it couldn't really do anything. Now the ecores are the new overclocking kings
I think the 1W thing happens because the variable in the code are set the watt to 1W without updating the UI. When you then make a change and press apply this 1W becomes applied. If this is true then you should see the watt change to 1W by changing any variable but the watt in the UI and press apply.
It's funny Jay knows all too well that most of us will just watch this out of curiosity. I personally have a hard time justify a CPU upgrade unless there's at least a 15-20% performance uplift. I'm on 13th Gen intel, and the only upgrade that might make sense is 9800X3D when it comes out. One thing I am interested in is the number of NVME slots available, at least for now with 4TB and 8TB drives carrying a premium price tag still. I bought Ryzen first gen and went through all the teething pains with slow bios and other weird issues but I find thats the best way to wrap my head around a new design, but with Arrow Lake it seems like it's more than just teething pains, but actual design flaws they need to work. If they do a refresh that reduces latency and improves temperature sensitivity then I'd be interested, if I haven't already jumped back to AMD before then.
![[#2024MAMA] BIGBANG (빅뱅) - 뱅뱅뱅 (BANG BANG BANG) + FANTASTIC BABY | Mnet 241123 방송](http://i.ytimg.com/vi/EQsYfiPR9uM/mqdefault.jpg)
As someone who waits 6-8 years between system upgrades (and then builds a system a gen or two behind), thank you for your service as a beta tester, Jay 😁
As a 5960x user, those are rookie numbers 😏
@@xellestar now thats a number i havent heard in a while
Can't hear you over the fan noise of my 3600x
@@xellestar Thats 10 more than an AMD5950x! It must be pretty good.
Honestly, I can't follow CPU numbering these days, it makes no sense. At least AMD's past few CPU generations have followed a familiar pattern. If they stick to AM5 for next year, what will we well? 10050x?
@@MOSMASTERING Nvidia have almost gone full circle - my Win98 machine has a 6200 AGP card in it. I really hope the RTX 6000 series has a 6200 product in there, so I can start sending bug reports on the latest games and confuse the hell out of their customer support 🤣
Overclocking has entered into it's practical limit at 10/7nm.
Going from 7nm down to 5nm, electron leakage has went up -requiring the same current level to almost to where we were at 28nm.
This means that we're seeing chip arch designers able to hit max freq as current being the hard limits, as a secondary, thermals. (Leaving little overclocking headroom if any)
Larger node sizes would hit thermal limits before current limits when driving frequency.
With it flipped, current before thermals, thermal runaway becomes an even bigger issue, to where we are seeing CPU's begin to internally throttle at 70C.
This is also why we're seeing chiplet/tile designs and wider execution paths in Zen5 and ArrowLake.
I'd be curious if LN2 overclocking these does. Cryogenic temps lowers drastically leakage.
Mostly True. Power consumption has not reduced as much as it did in earlier node shrinks. Leakage currents make it harder to maintain efficiency, so while newer nodes are more power-efficient than 28nm, they’ve been less efficient in shrinking power consumption relative to older node transitions. As fabrication technology shrinks (from 10nm to 7nm to 5nm and even smaller), electron leakage does indeed become a bigger problem. Smaller transistors have thinner gate oxides, and quantum tunneling causes leakage currents to rise, making it harder to push higher frequencies without significant power and heat penalties.
Considering that smaller processes aren’t actually smaller, just different configurations with different names. The 10nm takes up almost as much space as 7nm. They are just marketing names, relax
Backside Power Delivery & GAA kinda throws you're entire point out the window, Shoutout Intels 1.8AA and TSMC 2nd gen 2nm. GG's
@@nerdynumen Backside and GAA will be introduced, hopefully, next year by Intel. Until such time, FinFET is seeing a higher leakage with every fin pitch shrink. Gate shrinks have been negligible since 7nm. Backside power should be a more impactful innovation to hopefully address the issues we're seeing with today's bleeding edge nodes.
On 12th gen you could BCLK overclock non-k chips to the moon and back. 5.3GHz on a 12100 is quite rad. I've got a china special 12490f on a Asrock B760 PG Riptide, stock 4GHz all cores and with a crappy cooler and some BIOS tweaks 4,6 all cores and 6800MHz ram.
If you want to overclock there's always an exotic solution to be found.
Imagine a Celeron G6900 2c/2t running at 5GHz... You'll basically have a 15th gen processor in you hands !
You needed to watch DerBauer's newest video. He explains all the overclocking details. You need to keep the cpu under 70C. anything over and it limits itself. He was running 5.6ghz pcores and 4.9ghz ecores and getting 47.1k R23 scores. These new cpu's don't work like the prior gen ones.
Delid LM needed
he delidded the damn thing dude TOTALLy different then this.
He's a creator not a consumer 😂
What a twist lol coming from Intel's long running stretch of hitting hundred c and trying to convince people that was not to be worried about lol.
I get more performance from my 9950x by dropping it 30watts and limiting to 88C.
I can get 45800 in R23
Oldskool overclocker here. Won the Celeron600mhz lotery back in the day and boosted it to 900mhz. 1200mhz was doable but i was too scared to run that 24/7.
Undervolting is my new hobby now. 12700k and 4070 5% less fast and system is quiet and going stable at way less wattage. So yeah, undervolting olé olé
I get more performance from urndervolting and limiting my temperatures. I have a 9950x I think it was limiting my performance hitting 95C.
By dropping 30W and limiting to 85C, I get my best scores. Higher than out the box?
Intel went full power on 13/14th gen
@@aliabdallah102 feiner
Way better for the war with heating or cooling however you look at it in the modern Cooling wars.
I had an SL2QG at 448mhz then a 300a at 512mhz. Those were the days. Faster than a P3.
7:39 PL4 is another "internal" power limit, "internal" because the implementation of Turbo Boost Technology 2.0 only requires configuring PL1, Tau and PL2. PL3 is disabled by default, but it would be "A threshold that if exceeded, the PL3 rapid power limiting algorithms will attempt to limit the duty cycle of spikes above PL3 by reactively limiting frequency", and PL4 is "A limit that will not be exceeded, the PL4 power limiting algorithms will preemptively limit frequency to prevent spikes above PL4." So it would basically react more aggressively than PL1/PL2, usually at a higher value for the limit. Because with PL1/PL2 you can still have a bit of overshoot for the actual power draw, for PL4 it would not allow any overshoot whatsoever.
Once OC and BIOS are all figured out, can you do an all in overclocking head to head? Like max OC results between 285k, 14900k 9950X?
I second this idea.
Jay doesn't know enough to do max OC comparisons. He doesnt even tune the RAM, watch Framechasers for those comparisons.
I have to agree with GN Steve where he said, "these chips simply make no sense". Too expensive, requires motherboards which make them the most expensive platform intel has released in many years, suck way too much power, yet in test after test get beat not only by AM5 but often by their own cheaper previous gen and even AM4. Oh well maybe in a couple years (If they don't end up killing themselves like Raptor Lake) I'll pick up one on the used market.
They can simply drop the prices faster than usual. As always, it's all about the price
I hope people didn't really want them to be faster than 14th gen out of the gate when 14th was an abomination that basically came highly overclocked out of the box
Intel needed to make inroads into the laptop market since so many people buy laptops where efficiency matters; they needed to pivot and it takes a turn to make a turn.
@@LiveErrors yeah, a true Intel fan would want their processors slower
GN is a smart guy he always speaks the truth
We care about YOU Jay. We care about you 💙
Exactly!
Go through all that stuff for a 2-5% gain in certain software is a sign of too much time on our hands
@@ddzang precisely.
You obviously don't enjoy tinkering and see how far you can push the silicon.
ASUS advanced OC profile for increasing your room temp, useful winter feature.
It was 59 in my room before I started up my 11700K/RTX 3080 machine. It's been just idling in here for about an hour and it's up to 64, I definitely use my PC as a space heater as the weather cools off!
"There's an equation there depending on which way you go with the equation there." Science Jay - 2024.
It's true! Probably.
but I digress!
I remember Falcon NW being the absolute GOAT even back in the day. I'm so happy they have survived this long!
As a dude who's been doing this for a long time, this must be very frustrating to have to cover. I sympathize.
Others are getting great results btw.
hey just because i have 0.00% interest in the new intel stuff for myself doesn't mean I don't wanna keep up with what modern overclocking is like!
i honestly dont understand how things work now, i used to think the one performing better will get better result
now, or in here specifically, isnt the case and im waiting and earning just as much as anyone else even same as one not uploading anything
Please add chapters
Why? Not needed
Chapter one Jay curse. Chapter two the end..😂. Credit goes to 1k MOTHERfooker Motherboard. What?
It's over 26 minutes! @@AphillyatedYT
@@AphillyatedYT It's half an hour long
@@auturgicflosculator2183 huh? When did I say it bothered me? I think it is unnecessary work for the team for not much in return.
Chips are so fast and powerful these days that over clocking is more of a gimmick then anything.
then anything what?
Disagree.
It's the cool thing to do, a gimmick, then everything else. 😂
Please consider that in the time of 5% generational gains, that "only" 5% gain from OC starts sounding real nice. 🧐
😂. So the extra 17% I got out of my 7900x is nothing but a gimmick . Okay 😂
Binning is generally so good that overclocking is generally not as fruitful as it used to be.
@@71janas and what did that translate to in applications? Let me know how much of a benefit you got. Did you get a 17% increase in FPS? Did you get a 17% increase in render times?
Electricity can be compared to water:
Voltage = Pressure, which is require to overcome resistance (Ohms) and is why low Voltage can't electrocute you. Not enough pressure to overcome the resistance of the human body.
Amp = AKA, current is the flow rate, how many gallons a minute.
V x A = Watts or total power.
Ohms is what controls how much Amps flow based on how many Volts is being pushed through that resistance.
Can you explain it in onions? I'll find it easier to understand in onions.
Voltage is proportional to resistance times voltage;
v = I x R.
@@monk3synuts
V=IR
or
I=V/R
or
R=V/I
V = Voltage
R = Resistance (Ohms)
I = Current (Amps)
Which is why 0 Resistance = runaway current because anything divided by 0 is undefined.
Either way, that quote is incorrect, "Ohm's law states that the electric current through a conductor between two points is directly proportional to the voltage across the two points." That, and I was trying to keep it simple, AKA comparing it to water pressure.
@@cloudanu
Voltage is like the spiciness of an onion. The stronger the onion, the more it "pushes" tears to your eyes. Similarly, a higher voltage pushes electrons more forcefully through a circuit.
Current, or amperage, compares to the number of onions you chop. If you cut just one onion (low current), there won't be many tears. But chopping ten onions (high current) will release a much stronger concentration of "spice" in the air, just as a higher current means a larger flow of electrons.
Resistance is like the thickness of an onion’s skin. A thick skin makes it harder to reach the core of the onion (or the electrons in a circuit). A thinner skin provides less resistance.
Power represents the combined effect of spiciness (voltage) and quantity of onions (current). If you have a very strong onion in large quantity, the effect will be intense, similar to high electrical power. Conversely, a mild onion in small quantity will produce only a weak sensation.
There you go brother
Alot of info here (current)
Overclocking has changed alot over last few years. Gone are the days of doing a static overclock and getting the absolute best performance. Key to overclocking now is to keep your CPU as cool as possible due to the parameters that are built into the microcoding with boost being temp and power driven. I've noticed as you can get around 3-5% more performance on single core activities but the real tuning of overclocking is when you start using multicore. Perfect example is 9950x. I was only able to get around 3.5% more in ST CB23 but MT jumped up from 42k to 47k about 11%. And with controlling the temperatures the boost clocks tend to stay higher for longer periods.
Sounds similar to what happened to gpus, at least on nvidia with their 'boost' tech (not sure if amd does the same)
Jay, derbauer today released a video and gets 5.7ghz all p-core and 4.9ghx all e-core with over 47600 score in the same benchmark, maybe his video might help gain info?
He delidded it. But there's no good cooling solution then? hmm. Check out Skatterbencher too.
Agreed, Jay should watch someone that actually knows what they are doing to learn how to do the things that he has based his entire channel on.
Yep it was naked naked!!
At least Jay-Z has funny body 😂😊
@@blackbirdpctech Jay's channel caters to normal tech bros. If you are looking for unreasonably technical analysis, then go to Debauer and other similar channels. Jay is for creating entertaining content for tech people who can't stand getting their entertainment from television the way normal people do.
Dyno! As a tuner I see better results the longer you run (if the cooling system is good)
Gearbox and diff oils as they get hot I gain about 7kw
But yes, temperature is detrimental to power, I target 80-90°c on track, beating on it, temps.
98° I start going for the temp reducing stuff... Run it rich, pull power down.. or straight up install a water sprayer.
It's surprising how little water you need to smash temps down like a beast.
(Our cars won the D class Bathurst 6hr last 2 years)
Yeah I really don't care Jay, i just want to listen to you yapping in the back ground while i game. Coming from a 12900K i expected so much more, Thursday (10/24) i walked into Microcenter, got the Z890 board i wanted and they only had 265K, so I impulsively bought it. Today 10/27, i have officially returned both items, with the exception of exchanging the motherboard for a X870E chip-set, ill be waiting for 9000x3D series to officially switch back to AMD since the 2011's upset FX-8150 i bought a that time. Except this time, I'm not holding, I already swapped one major piece, now waiting for that CPU to drop. This feels almost exactly like 2011s FX-8150... its been 13 years of Intel for me, since those days and here I am... I'm going back!
Intel pretty much pooped the bed by setting on their hands from Haswell till first gen ryzen. Imagine if they would have done something different for five generations aside from telling use four cores and eight threads was plenty.
Question, why did you even buy the 265k on release day? I assume you didn't wait for the reviews to go online to watch them first because then you would've known how bad these CPUs are before going through the hassle of buying one and testing for yourself. Could've saved yourself a lot of time there. I'm eyeing 9000X3D as well, but even though I have a good feeling on it, I will be waiting for those reviews and then will probably wait another few weeks after that to ensure no really nasty issues are discovered in that time.
@@123Suffering456 The simple answer is, this is why I wait 3+ years to upgrade, naturally it should be better. Sure productivity apps was a nice uplift, but I'm here to game on PC, and there the difference was negligible.
@@Mystikalrush I doubt it, while not much faster than previous gen the 265 is quite the improvement over a 12900K, which doesn't even keep up with the 13600K most of the time. You're surely gonna get a faster gaming CPU with the new X3Ds but paying basically 500 bucks for an 8 core in this day and age is stupider than buying an Ultra 200 on release day imo
@ekifi Hahaha your all wrong. Go tell this to Jay, all hundreds of reviewers and thousands of customers, you are hilarious. Anyways, I'm already in my 9800X3D setup, absolutely crushing everything that shirty 265K couldn't and even my 12900k. Obviously I'm here to game, not compute.
The Board enhancements for this intel series was what I was interested in this time around, before even the CPU speeds. I was really curious how well the chipsets will manage the expanded options and bottlenecks that come from the consumer grade CPUs.
You can fit a lot in a compact z890 mini-ITX build with adequate thermals more easily. Which bring it closer to what's needed for +90% workstation related work.
Having a smaller, quieter, and colder system close to you while you work is something I appreciate.
Less in the server and less I need to off load to another system to render or do sims on a workstation. Less money spent on creating/cooling/maintenance/space on multiple systems/workstation in a way that makes it quiet enough to have close to me is a good thing for me at least
Poor chair always gets the blame. 14:14
Now it's Intel's turn to go through the medieval age.
It's obvious those CPU get out of factory already close to the limit (I mean, last gen was destroying itself for going too fast).
The days of 2.6GHz oc'd to 4GHz are decades old.
Nah, the CPUs just overclock (aka "turbo") themselves automatically now.
Clarksfield (mobile version of Nehalem) was the first chip that really showed what it could do beyond a measly couple hundred mhz. A 1.6ghz quad core with the maximum clocks of 2.8ghz previously only seen in dual cores, dynamically adjusted according to load.
And it always hitting either a temperature or power limit because it was a desktop chip in a laptop and stupid Asus blocked off the intake of the ONE fan (G51J, along with its desktop derived "GTX 260M" which was really a 55nm 9800GT, the thing was a stupidly hot fireball of a gaming laptop)
Intel's got Jay so frazzled he's performing elbow drops on the furniture.
Overclocking my 7800X3D was like juicing sand.
Overclocking my 7900XTX, on the other hand, was like the 2nd best athlete in the world took ROIDS!
A Windows update reset my overclock, and I spent forever tuning it😭
Does anyone know if there's a way to recover settings?
No way to recover but you can at least save settings. Overclock in AMD's own software - you can save your settings as an XML file. Their software is quite conservative and it'll disable OC settings due to any unexpected shut-off including power loss, holding the power button down for a hard shutdown, etc. so it's good practice to save some good settings if you come across them.
Answer to your power limit issue
Power Limit 4 (PL4): A limit that will not be exceeded, the PL4 power limiting algorithms will preemptively limit frequency to prevent spikes above PL4. Turbo Time Parameter (Tau): An averaging constant used for PL1 exponential weighted moving average (EWMA) power calculation.
Falcon NW is straight up ripping people off at 7500$ with that system. The exact same system costs roughly 4500-5000$ to build. If they charge 500$-850$ because of their warranty and such okay, but its 2000-3000$ (depending on the exact components as they are fluid and do change from availability). There is absolutely 0 reason to be overcharging THAT much and expect people not to be mad at it.
So, what looks better, a Cadillac or a Chevrolet car badge?
@@outlet6989 In that case, it's not just the badge, it's the comfort, appearance and features as well.
Go try to buy a new 4090 these days and see how much you pay.
The system cost of goods could possibly be closer to $5500. Yes, the warranty does have a cost, but let's assume their total cost of parts, labor, marketing, testing and support/maintenance is $6k, that's about 20 points of profit margin per unit, not an insane profit margin for a boutique low volume system. Low volume means higher risk of not moving enough units to break even.
Company like Falcon Northwest shouldn't even bother selling a system with a profit margin below 15%. Dell on the other hand can still build a business on selling low end laptops at 12% margin because they know the volume will make up for lower margin.
@@racerex340 They're not that expensive, why are you acting like 4090s are suddenly worth $2k+?
Who are you to say what they should and shouldn’t charge for their product? Or what people should and shouldn’t do with their money?
The M.2 slots are impressive. I had the first one on my MSI Z170 board and one on my C232 workstation board by HP. Ever since then I have two on all my mainboards (Z370, B550 & X570). But yeah - M.2 drives are becoming more and more a serious option. Like in 2019 I paid as much for a 1TB PCIe 4.0 M.2 drive as I paid for a 4TB one this year (as well as a 14TB HDD). As soon as you get like 8TB ones for the money I can see myself slowly ditching my HDDs.
It is not stable at baseline why try to overclock it?
I have always had a mid tier system, like 3070, 2060 super, etc, i finally have a high end build. 4080 super 7950x3d 64gb ram 1000 psu. Build costed me 3500 bucks, I have many extras in it as well. Not nessasary but i love it!
I do kinda find it hard to get interested in new tech stuff when I know it's something I won't be able to ever remotely afford or make use of. Its like, how many supercars can you look at before you just don't care anymore, they're all effectively unobtanium and just kinda blur together. I paid less than 7500 for my car, I would never spend that much on any computer
At least when it comes to CPUs prices aren't creeping up that much gen over gen, sure these 16/24 core 9 series units aren't cheap but they're part of lineups that start from the low end while still offering the latest archtectures and silicon. It's not like every new CPU is 600 bucks
@@ekifi Granted, yea
The sad thing is, the 13900k/14900k voltage spikes seem to be finally be under control and temps are under control now which makes benchmarks more predictable than bouncing against thermal limits. The best Intel chip now is the 14900k (I won't say KS because that may be touching the limits of this generation and I had to RMA 2 of them, admittedly before the latest microcode and BIOS updates). Certainly won't be upgrading my 14900k, my family 13900k machine or my work office 13900k system.
I can't blame falcon northwest for making the silly system, this is what some people want. This is like the 200mph automobile, you can never go that fast unless you are insane or on a track or a very few special places in the world which are most likely thousands of miles from the owners home. Yet people have them. You can poke the company that makes the components, but they probably do have a niche case user here and there. Some folks just have money to throw at things so they feel better about whatever purchase, or don't care as they are not paying for it themselves like in a business scenario and the entire expense is a tax write off for someone.
It's like having a 200mph car with covers over all the radiator cooling openings
Agree. It's the bling, and some ppl. don't want to assemble and tune something themselves. Why be angry? Just don't buy it. No one has to be angry.
@@reptilespantoso People deserve good products for how overpriced it is. It'd be like buying a lambo but the motor falls to pieces if you press the gas pedal more than halfway.
These parts and build are like 4500 5000 grand max, so bling means you get to rip people off? What is the extra 2500 bucks for? The actual build?? Lol yeah okay
P cores probably only make sense for 1-2 thread boosts, otherwise giving power budget to the E cores seems to be the way to go. also the DLVR is burning A TON of power in high load scenarios, so the next priority would be reducing the switching losses of the DLVR by optimizing the voltage levels so the input voltage can be closer to the output voltages. you probably want all the voltages fed by the DLVR to be as close as possible - so pvcore, evcore, bus and another thing i forgot. check skatterbencher, he has a great breakdown.
also, TSMC silicon seems to be scaling more towards lower voltage instead of higher voltage like intel silicon. you could hammer raptorlake with voltage and it would just go up in clockspeed, but it seems like the TSMC manufacturing plateaues off at mid 5 ghz. but i have seen quite some remarks about undervolting.
i think arrowlake actually gets better the lower the power is, not only more efficient, but also faster, when you turn the right knobs.
Try the direct-die cooling from der8auer. The CPU is 30°C cooler and gives a lot of headroom for OC. He got >47000 points in cinebench while overclocking P and E cores.
This, derbauer and others got over 47k score
It's not even released yet, it's still in development.
Delidding is a different level of OCing what jay is doing is what most people would do that is interested in tweaking, Delidding is up there with LN2 shit thats extreme crap.
@@Aaron-zl5gq What? No. Delidding is as easy as putting on the paste with the right tools.
der8auer/Thermalgrizzly sells the delidding tools and the fitting waterblock.
The last time I did, it was a 7700k and it was a matter of putting it in, twisting a screw and putting on liquid metal, done. (it was not soldered)
It was 30°C cooler after that, WITH the stock IHS back on.
It was 50$ for the delidder, liquid metal and special glue to put the IHS back on.
But even soldered, you just need to heat it up before doing it, the rest is the same.
@@Shadowdane He released a video today saying it will be in the shop on monday.
And he explained how to use the thing including showing results.
so technically it isn't heat that causes degradation. The melting point of silicon is a lot higher than 100c. The issue is electron migration which was always going to be the thing that killed the Moors law. We are reaching or have reached the limits of how small we can make pathways with the given materials. Eventually they get too small the slightest heat change will cause a short or get close enough together to allow electron migration to occur.
Lmao, now I get it.. I genuinely thought 7500 was just for the case.. :'D
Not surprised by the Cinebench oddness. With 14900K the E-cores offered 50% of the thread count. Now, with the 285K and no hyperthreading to make optimal usage of the P-cores, the E-cores now handle 66% of all threads. So obviously they're more important to overclock than the P cores which suck down too much power and hold back the E cores.
For gaming it'll be the opposite, crank the P cores. Maybe even try disabling a few or all E cores to get the power budget back.
"Zero side thinkers"
Yeah that applies to so much of society today.. really insane how people just can't think for themselves and never see things unless deliberately laid out for them (and even then can still miss it). Is critical thinking even taught in school anymore?
lol no, Schools actively discourage critical thought and reinforce copy-paste learning, imo
Nope.
Critical thinking threatens traditions and the oldschool ways of doing, thinking, and saying, so people in positions of power who adhere to them benefit from curbing the spread of critical thinking. It's the same reason why civics isn't taught in schools.
It wasn't much better historically. Human ability to process information is limited and the amount of information is much greater than it was in the past.
@@theodentherenewed4785 uh what? Sure.
It is sad that they put 6 or 7 M.2 slots on a board and in most cases you are lucky to use Half of them or end up with X8 instead of X16 PCI-e or you loose the PCI-e slot all together as they are shared lanes.
Did they drill holes in that 4090 for the support bracket?
Nah, the fe like most 4090 comes with holes for mounting support brackets. On the FE you remove a little cover to expose it
@@Jzwiz Ah ye your right, just checked my 4090 cooler and it also have it. Never noticed it. I'm using a waterblock though.
The 4090 founders edition has already holes for a bracket.
@JayzTwoCents - Get Romans 285K delid / heater and try again.
He got 20-30 degrees savings, it would be cool to see these tests your doing with the complete removal of temps from the equation.
And ya ... even if you only delid one CPU, even though you only have a test pool of one, it would be cool to see hyow it affects your wanderings thru bios voltage settings and ITU tweaks.
I still think $7500 is way too much for what you’re getting. I’m pretty sure people had more of a problem that it seemed lacking in features, add-ons, and higher end hardware for what you’re paying.
It's a halo build. It's going to be stupid. That's how halo builds work. The motherboard alone costs $1k retail.
It's silly. You can increase the price of a PC by stuffing it with a lot of unnecessary stuff like multiple drives, crazy amounts of RAM, an ultra-premium MOBO with features you'll never need. But why? Except for a very few niche users, it's just for bragging rights and people who have more money than they know what to do with.
@@BobBobson While I still think $7500 is a tad much for a i9 system, I do understand what you and Jay are saying.
@@BobBobson Did this build in PCPartpicker except for the case, its roughly 4500-5000$ depending on specific choices. You are paying 2000$ minimum overcharge for the system.
Yeah this price is goofy. It's $7500 for maybe $5000 in hardware, and there's nothing custom about it. It's got a paperweight cpu, no custom loop, and even a boring case. I get that every business prices their products to make a profit, but this is a massive markup for a very basic setup with expensive parts... Lowkey annoys me that Jay is actually trying to defend this insane price tag.
what i understand from this is, That e cores are working like accelerators to uplift the scores. They wont increase performance everywhere
Gamers are a strange tribe from a parallel Universum, strictly believing that more RGB makes the PC faster, and the difference between 102 and 105 fps helps them winning
LOL we don't care about RGB (although it is pretty), we just like to "tweak" things to maximize performance per dollar. There's something about getting an 11 when you pay for a 10.
As opposed to other users who think that archiving in 7-zip in 102 seconds instead of 105 seconds will make them more productive
@@NJ-wb1cz And they will try to support that logic like what happens if the building is on fire and that 3 seconds matters cause you know I'm a music producer or visual effects artist but I am really a 13 year old getting my ass wiped by mommy.
I have yet to meet a person who unironically thinks RGB makes it faster. It's a bad joke that's gone on for too long
i only got a new gpu when the old one couldn't do what i needed or just went bad. i refuse to downgrade but 102 to 105 is literally margin of error
Jay just FYI you don't have to spam the Del key, holding it works just the same.
Water cool it and send it through the gauntlet
Heck, put a phase change cooler on it and send it to the moon
I always wanted a Falcon NorthWest, Such a old and Quality company
Wow! If I’m spending $1k on a motherboard, it better be dual socket.
Dual socket doesn’t give any benefit to the types of performance you use in a HEDT platform.
@@waynetuttle6872, yeah. It’s highly software and architecture dependent. My workflow would probably be a fair fight between a 14900k or X3D and a dual Xeon workstation. You’d be surprised how much larger CAD and CAM projects can benefit from high core counts and lots of cache with more recent software updates.
I was tempted to try a desktop platform again, then AMD had its issues with scheduling, Intel started burning up ring busses, and ECC support was questionable etc. I’m not going to fight with my tech to get a marginally useful single threaded improvement and questionable longevity/stability.
@jayztwocents As far as I understand, Raising the simple view P Core multiplier does nothing if the TVB is set to limit 8 core to 5.4; TVB overrides the general setting. That explains the initial weirdness.
It's nice to you realize that overclocking is basically a waist of time that will only result In a dead system. Focus on customization instead.
I've found that on my 10th gen, TVB is just all around bad when it comes to hitting desired clockspeeds. For some reason the Intel utility will do everything it can to prevent it from hitting the target frequency when TVB is enabled.
Looks like LGA1700 the goat for people who actually know how to tune them 😂😂💯
12th gen series seems to be the sweet spot
“It was the chair guys”. “I swear.” And that is what puts your content over the top for me! lol you didn’t even miss a beat!
tbh, i miss the time where overclocking was pretty much only fsb, multi and vcore.
They had to make it harder I’m sure sooo many people were killing their components cuz they didn’t know what they were doing and just threw max voltage at it and prolly instantly killed the parts.. that’s why they have made stuff harder cuz humans are stupid and will fuck stuff up so they have to save us from ourselves! lol it’s sad I know
Take me back to 2000 pls. UT99 voodoo2 and cable internet. All i needed 😂
And even budget CPUs could be overclocked to match or even beat much more expensive CPUs depending on the task.
@@Pasi123 absolutely - my trusty celeron 300a was my first (and only) cpu to run stable at 100% oc. (it was still just a celeron but i was a happy camper)
@@AtomskTheGreat the thing is, modern CPUs are overclocking themselves. The same Celeron 300 could've automatically boosted to 600 or even more if we had modern algorithms back then
Thanks for this. I'm quite satisfied with my 12400F DDR5 6000 GTX 1080ti and 1tb 980 Pro SSD combo.
I thought you dumped Asus? Are they doing what they should be now? I thought at one time they were ripping people off with their RMA stuff.
This is just a system that flacon northwest sent him, Jay may not have had any say as to what components they put in.
@@insignia406 Good to know.
@@jonathandykstra9652 fair enough
@@insignia406 Man that sucks. sorry to hear.
Didn’t Jay say that this is like the highest spec one they sent him tho? Yes he did say that I guess you missed that part
Thanks for the new error lake info
Jay, make no mistake, your audience very much appreciate you burning hours of your life to over clock out a hot turd so none of us have to
Your sacrifice does not go unnoticed!
Wow, you are so cool and original.
I mean Intel literally said in the announcement that pushing E cores are gonna result in far more advantages that the P cores. In fact even you said that same thing in your first video covering the 285K
Bypassing the internal voltage regulator should be the first step.
Less heat and much more predictable behavior from what I've seen in other videos.
IMHO I think the loss of hyperthreading is why there's such a massive loss of gains with OC now on P cores. There's simply a lot more E cores now than P cores, so pushing those hard their gains add up together to be more real performance than pushing P cores slightly, even though P cores at the base level are far more powerful.
And we thought Zen 5 launch was bad, this is just embarrassing for Intel
This "backward" behaviour is the same as Ryzen Zen 4: My 7700x doesn't like high voltages. There's a sweetspot. If I want more performance, I need to cool it more.
Yes it's counterintuïtive, but also logical if you think about it. These chips are now so dense and small they can't handle high voltage, high heat at the same time.
Sure.. you can cool it to the extreme, and then blast it. But I need this PC for work. 🤷♂
It feels like the same people don't work at Intel anymore and they lost key engineers.
DEI hiring practices. The virus also force retired people and they were replaced with......less skilled.
@@JohnWalsh2019, also new blood they can pay less compared to the graybeards I think the term is.
@@JohnWalsh2019it has nothing to do with DEI. Pull your head outta your arse. Intel has been laying people off in the 10s of thousands as of late. Further, the leads of the long term CPU core team (the likes of the legendary Jim Keller) were let go earlier than anticipated so those projects got interrupted or cancelled altogether. Quit trying to blame DEI for something that it has absolutely nothing to do with.
@@TheKazragore can't accept DEI hiring practices are based on certain "check boxes" rather than talent can have an impact on a company....okay bro. If you don't think this is Intel, just check their website, they have fully embraced it. Also, I worked at Intel (not on the CPU side) for two years. It's easy to see when a company is taking a turn for the worse. Good luck!
I don't know why people waste their time with Arrow Lake, considering Panther Lake is expected to come in H1 2025.
Honestly, I care more about the 7800X3D and the 9800X3D
7800X3D and a 4080 Super, jack of all trades build with 64GB of DDR5 6000.
@nightshadelenar same except a 7900xtx here :).
When you switched to the Asus OC profile what was the cache OC set at? Did you put it back to 3.8 before you ran the first Asus OC profile cinebench run?
It great to have a content creator that doesn't bash Intel all day long.
@@gingerbread6967 it's called reviewing latest Intel's processors. Results vary depending on the processors
@@NJ-wb1cz Yeah, and if you watch Hardware Unboxed Intel review its all about how much they think they suck as opposed to just telling you what tests have shown Same goes for many others
@@gingerbread6967 HUB are well known Intel simps. If you need people to be even more biased towards Intel, you would probably benefit the most from searching your own feelings towards corporate brands
Yeah... Shilling is much more profitable. 😂😂
Jay, you need to use the shutdown command with the /fw switch (fw for firmware). No more smashing ESC, F10, or whatever. The shutdow command will boot you into the BIOS. e.g. C:\> shutdown /r /f /t 0 /fw
Last time I was this early. Intel still actually produced CPUs that people wanted to buy
2013?
Took 20+ years for people to buy AMD. Not too bad.
@BruhMomentPatrick
10th gen was a beast. The exact date was 4/20/22 if I'm not mistaken. Anyone know why this day was so important?
Savage... lmfao.
Intel hasn't responsed because they heard Jay likes making explosions.
I dont know anyone who overclocks anymore
Ya, seems more undervolt than overclock these days.
Then you do know a lot of intelligent people.
@@outlet6989wasting so much time for a potential stable oc is not worth it, you most likely make more $ per worked hour to buy a better CPU/Mobo or faster ram instead.
You have double the eCores so an increase to their clock should have a bigger effect on Cinabench score than an increase to pCores.
I am letting the video run J. I..... for the first time in my life, am considering a full AMD build. Intel is YEARS of "not being dicks" away from me considering their chips. Nvidia is most likely having multiple meetings TO NOT make the joke: "50 series will start at $5k ha ha" a reality. I am doing a full system, set-up, peripherals, and furniture upgrade.
Would love to see specifically how you test for true stability when OC’ing. Good stuff guys!
Give that pc 7 months and the value will decrease by 25% 😂😂😂
More like in 3 months.
13:30 got the same thing on 5700X. Tried UV it to save some power and thermals, but despite higher clocks with lower voltage, CB score would be lesser than stock. AMD calls it clock stretching, so even *if* core boosts higher, internal governor won't let it because it doesn't have enough power and would crash. Or at least that's what I think it is, just on Intel's side
don't burn the new cpu
Its intel, it’ll do it itself.
@MrMarrok657 nah its ring bus is so much lower this time.
Its also worse than last gen performance so take that as you will
@@lilkidsuave Ring bus was NOT the issue.
overvoltage in general, but Mpores law is dead mentioned that the rig bus clocks were lowered to avoid the issue
@@lilkidsuave When Derbauer overclocked his, he upped the ring bus. sooooo yea thats kinda null.
It's interesting seeing the difference between the two generations. It's also a little disappointing that the 285K isn't much of a jump on the 14900K here. A 14900K is around 40,000 on Chinebench too, so we're really in the same ballpark.
However, what's interesting is how the new gen gets there because it seems to be all on E-core here, and it's not something you can duplicate on a 14900K. A 14900K won't run its E-core up to 5GHZ. I can push to 4.6GHz on my own, but it's not happy about 4.7Ghz. It does matter the voltage profile either. What you'll also notice directly against a 14900K is that the higher performance cores aren't gaining you anything. Because it's still a heat bottleneck, you're still only up to around 5.6GHz on the P-cores, or less, with a pretty hard thermal limit on any current off the shelf AIO you can buy. So the 285K seems to give up on peak GHz on the performance cores (for long term stability maybe), and then ramps up those E-cores a good bit to get them to do a bit more heavy lifting.
An interesting note on the power limit. The 14900K will also do this despite having limits disabled. I'll generally see this once I go north of 385W for an extended period of time. It'll hard lock it down to 252W for a little bit. It'll just trigger in the background. I'm curious what the 285K is using for its definition. Too bad it's not accurately showing power, so...who knows. GamerNexus is plowing into that whole mess, so it'll be interesting to see how these compare on actual wattage. GamersNexus does seem to indicate an ok improvement on efficiency overall meaning the watts are at least being utilized better to some extent.
One thing you can do is tune the individual cores a little. There can be some hot cores, and you can even them out some. What I've found is this lets the other cores get a little more, and I especially saw a bump up in what it was allowing for peaks overall on both P and E cores. You can use any monitor like HWiNFO64 or whatever to see the individual cores. Bump them up and down to even out the whole thing.
As part of this, I'd be curious to see if you can bump up one or two P cores to north of 6Ghz stably. Since one major hit the 285K has is single or low core count activities, and it starting at a mild 5.7GHz isn't doing it favors. I'd be curious to see some individual core tuning to see how high it goes with stability and without thermal overloading on that core.
The per core fine tuning and dialing in the single (or couple) core peak, this might bring the gaming performance back inline with the 14900K and others.
Essay
@@Chicken-o5e side effect of college is "write a 10 page report at 10pm that'll still get you a decent grade," wordy over concise.
dudes gotta shut up about being first
First
Lol First to Copium 😎
If you ain't first, your last.
*trademarked, not to be used without written consent of Ricky Bobby incorporated
First
I like the early BIOS stuff- how at the POST/Splash screen Asus has F3 to Toggle for EZ Flash 3 Utility, but when in BIOS its EZ Flash 4.
Raid 0 - who´s honestly still using that BS in a prosumer setting?
With today's SSD speeds I question if there are any advantages running raid 0. Seems like it's not worth it but someone can educate me if I'm wrong.
@@JohnWalsh2019 maybe in a Caching server, but in a workstation? Nope :D
In a workstation that´s pure liability.
@@axelhorneff5077 agreed!
Jay, you don't need to keep tapping the button to get into the bios, just hold it down. Don't know why so many ppl haven't figured that out yet.
Has it ever occurred to you that the errors you get are user error? Every other reviewer was able to run CUDIMM but not Jay. Same as the memory problems you had on AM5. Suggest you put the bios on Basic instead of Advanced next time.
E-core favoring is probably cos physical layout for the cores are in 1P-4E-2P-4E-1P per row... if it were 2E-2P-4E-2P-2E instead... might have been able to push P-cores more...
No one cares about cinebench test a game ffs
Try overclocking the SoC fabric! That'll see huge uplifts in memory-bound scenarios (which the 285K is in a lot of places)
First
4 seconds late
We went from Intel fighting with AMD to Intel fighting itself and losing bruh 😵💫
Jay you were right in the other video, these Core Ultra 285K chips are all locked to a +400Mhz OC on the P-cores with any cooling other than LN2.
Even though I'm not interested in overclocking, I come to watch the hilarity that ensues whenever Jay and crew attempt something new.
I was wondering when this would start to happen... It's LGA 775 all over again. The mobile /laptop stuff was super overclockable but the desktop stuff was so hot that it couldn't really do anything.
Now the ecores are the new overclocking kings
For that much money, they should have a, “Air Flow.” Box to be checked. Money should get you more than something that looks good.
I think the 1W thing happens because the variable in the code are set the watt to 1W without updating the UI. When you then make a change and press apply this 1W becomes applied. If this is true then you should see the watt change to 1W by changing any variable but the watt in the UI and press apply.
It's funny Jay knows all too well that most of us will just watch this out of curiosity. I personally have a hard time justify a CPU upgrade unless there's at least a 15-20% performance uplift. I'm on 13th Gen intel, and the only upgrade that might make sense is 9800X3D when it comes out. One thing I am interested in is the number of NVME slots available, at least for now with 4TB and 8TB drives carrying a premium price tag still. I bought Ryzen first gen and went through all the teething pains with slow bios and other weird issues but I find thats the best way to wrap my head around a new design, but with Arrow Lake it seems like it's more than just teething pains, but actual design flaws they need to work. If they do a refresh that reduces latency and improves temperature sensitivity then I'd be interested, if I haven't already jumped back to AMD before then.
Great video Jay, looking forward after you get the Intel sheet. Have a great rest of your day. 🤛
Question for the audience. Where were you looking when he was spamming the delete button?