Mr. Fox

14900KF - P 60x / E 48x / Cache 52x - NOT AI overclocked.

Yes, thank God. Hopefully, ASUS doesn't kill it as well. The more hours I spent trying to make it usuable the more frustrated I got. It's not worth keeping. It's a piece of trash. Even trying to de-tune it to run at 13900KS/14900K/KF clocks is a no go. It guzzles voltage and power beyond reason functionality in Cinebench. Trying to use settings that @tps3443 was suggesting (L4 and Auto) was seeing spike up to my 1.700V cap and idle voltage at 1.650V, load voltage in the 1.450V range, which cannot be managed effectively unless you are using LN2 or cascade cooling. AMD will never be open for consideration again for me. They don't offer any products that I am interested in buying. So, I am wondering if maybe the systems I own now will be my last. I'm not liking anything at all in the computing technology realm. Windows sucks, gaming sucks more than it ever has before. It's no longer any fun to bench on HWBOT because of how things are. Winners are predestined and receive parts and firmware nobody else has access to. Everything is grossly overpriced and the OEMs are becoming anti-overclock. So, really there is nothing left for me at this point. It was fun while it lasted, but my patience is wearing thin with the way that everything is designed by idiots for idiots. I am starting to hate all of the companies involved (Intel, AMD, NVIDIA, ASUS, Microsoft... all run by imbeciles with the endoresements and blessings of their retarded fanboy possees.) In my futuile effort to make a silk purse of the 14900KS sow's ear, I booted into my Windoze 11 Cancer OS and fired up XTU. I guess there is now an AI overclocking feature. Super stupid. I had not seen it before. It tuned mulltple cores for 6.3GHz and the rest at 6.2GHz, about 1.625V on VCore. It ran and it was stable, but was not usable due to overheating and thermal throttling with my water chilled to 16°C. Cinebench pulled over 650W before it collapsed from heat, at which point the core clocks dropped to about 53x on the P-cores and 41x on E-cores and under 40K Cinebench score. Utterly worthless. It might not have overheated playing games. But, yeah... like I give a rat's butt about that.

Back to the better CPU...

Yes, I tried the latest from ASUS and other betas, even old BIOS. I am not trying any more. Already wasted too many hours on this poorly binned piece of trash. I will just be content with the excellent 13900KS and 14900KF CPUs that I already own. The voltage needs are what they are and it will always be too hot unless I detune it to match 13900KS or 14900K and if that has to be done the KS is meaningless and serves no purpose. It is going back and I am not buying another one. It is cleaned up and back in the box. I will call them Monday for an RMA number. I even booted into my cancer OS 11 and ran XTU AI tuning and it tuned 63x on several cores and 62x on the rest and it throttles down to 5.6GHz in Cinebench and reaches 107°C. It would probably be OK for a gamer where the CPU never has to work, but it is unsuitable for the purpose for which I bought it. Gaming is not that important to me. If it were then I would be buying an i7 and running it stock.

The reason it is slower is because of excessive voltage, overheating and thermal throttling. It doesn't perform as advertised because nothing normal is capable of cooling it. You would ne need LN2 for daily driver cooling to run it at the rated clock speeds. In order for it to be a usable product you have to de-tune it and make it function like a K so there is no point in having a KS unless you are the 1 in 1000 that gets a god tier CPU.

I cannot get it to run Cinebench R23. It errors out. I do not think the firmware is right for KS. Something is off. Yeah, the e-cores are trash, but they were also trash on all of the 14900K/KF that I returned. 13900KS e-cores were almost always better than 14th Gen CPUs for me. I am seriously thinking about calling them on Monday and ask them to accept a return as a defective product. You would need cascade or LN2 for this to be usable at advertised clock speeds. I am not happy with it. It is a voltage pig and runs too hot to be usable. If I delid it and go bare die that might fix it, but then it would no longer be returnable if that doesn't fix it. With BIOS defaults it idles at 45°C and sits at 1.527V. If you have to tune it down to 14900K clock ratios then owning a KS is totally pointless because it can't be used with default clocks.

Yes, they are very close. I'll probably use it with the IHS this weekend then move it to bare die next weekend. So far it seems very close to the same silicon quality as my 14900KF. So, now I will sell the two dead 13900K/KS and my good SP108 13900KS. And, I will keep the 14900K that I bought from you for an emergency backup. Once I get this 14900KS on bare die and can do better head-to-head comparisons with the 14900KF I will keep the better one on the Apex and move the other one to Muchkin. I'm going to leave my hands on the Apex Encore and the 13900KS in that setup because it's the best I have had before and I don't want to disturb anything to upset the apple cart. The performance regressions are already evident to me. All 14900K/KF seem to have slightly lower performance than my 13900KS and Cinebench scores seeem to creep a little lower with each ME and Microcode update. Memory latency also seems to grow a little with each update. Memory latency with this new 14900KS is about 1 ns worse than the 14900KF and 13900KS using the same memory profile. As usual, updates are actually downgrades. Only good for those obsessed with "security" and suck for everyone else.

Well, yours is better than mine, but considering some of the horror stories I am seeing I think I did better than some folks. I was hoping it would be better, but it could certainly be a lot worse. It's not bad enough that I can get away with sending it back, and trying to exchange it for a better one might end badly. I think ASUS needs to make some changes to firmware because my LLC tuning isn't having the expected effect. VID stays high no matter where I set the DC_LL. Not sure if it is relevant or not, but mine is also made in the USA.

If this is printed on the box and an "S" is tacked on the end of the product name because it is "special" then it needs to be special. If it is a KS and it is not better than a K, then it's not true and you were sold a bill of goods. You deserve to get your money back. There will always be duds, and duds should be exchanged at no cost to the purchaser. But, the average KS should always (100% of the time, not randomly) be better than the average K or you're getting screwed. If that is not illegal, it should be.

Yeah that is unacceptable. And, the return police wouldn't have a job if they were not throwing turds at the wall to see how many stick. It's actually shameful to think that they would expect people to say nothing and be content with defective samples. They should expect people smart enough to know better to demand something acceptable, send back the trash and ask for an exchange or a refund.

I would be satisfied to have one like that SP105 @Talon got. It is almost the same as the SP102 14900KF I have with SP114 P-cores, but the E-cores are better. I mean, who would not like having a god-tier chip? But, I don't need one to be content. I just need it not to be a boiling hot piece of crap that takes an absurd amount of voltage for stock or modest overclock and I think that I'm seeing too many examples of defective trash that will overheat running stock clocks.

Yes. I have 0.1mm and 0.2mm and the 0.2mm is too thick and doesn't work well, even for temporary testing. It can't transfeer the heat as well as the 0.1mm foil. I haven't tried 0.05mm but I think it would too thin and fragile to be reusable. The 0.1mm works fantastic for short term testing. I think I could use it permanently on my work computer and not have any issues with thermals. Wow, that must be something new. I have not seen an Intel CPU before that was labeled as made in USA. Most have been Taiwan (better ones) and China (often the worst). I wonder if it came from the plant here in AZ?

The indium sheets? I got them from a place in Mesa, AZ (through Amazon) that sells industrial thermal products. Totally reusable. I've only cut square from a 100x100mm sheet that I have used about a half-dozen times. It never gets hot enough to melt or stick to the IHS or waterblock. It works about as well as KPx or MX-6 does on a CPU that has not been delidded as long as you cinch down the waterblock good and tight. You'll still be pushing 100°C in Cinebench, but it's totally acceptable for just about anything else. Works better than TG Kryosheet and isn't a fragile single-use option. https://www.amazon.com/gp/product/B092Z8QDS8 I guess now we know why so little has been said by Intel about the 14900KS release, apart from the recycled press releases a few weeks ago to confirm the leak was true, there has been next to nothing. Interesting observations in the conclusion. Things are looking kind of sucky on both sides of the fence. Overclocking enthusiasts: Intel: higher power draw with minimal performance gain over K CPU, higher price tag Gaming enthusiasts: AMD: half the power draw, equally high temperatures, lower gaming performance, poor overclocking, cheaper

The three that I returned to NewEgg and Amazon last month were tested with a sheet of indium sandwiched between the waterblock and heatsink and were not tested beyond a view of the VF curve and an F12 screenshot saved as evidence of unacceptability. No cleanup whatsoever. The indium sheet leaves no visible traces of installation and keeps the CPU as cool as cheap thermal paste.

At the very least I'll be doing my 1-minute silicon check on the one coming from Central with the sheet of indium and CPU frame so it will still look like a virgin sample if it has to go back. If the VF curve looks like that it will take about 10 seconds in the BIOS to find out if I am keeping it or sending it back to them. But, now I am wondering if I shouldn't just cancel my order and refuse delivery from UPS. This is shaping up to look like a scam product launch. If these are what cherry-picked KS CPUs look like, the only way a person can get an acceptable sample is purely by accident and the chances of getting a 14900K that is actually worth keeping is close to zero percent.

Well, good that you've got the ability to buy from Micro Center and giving them the defective trash back is little more than a minor inconvenience.

Sometimes the SP ratings are jacked up and I have to flash down a version and then back up before I can see what the actual SP ratings are. Reflashing the same version on top often doesn't correct the erroneous numbers. I recently had a P-core rating of over 1,000 due to the firmware being jacked up. Looks like just another scam from the dumb-dumbs at the ASUS Chinese Clown Possee. They've figured out the zombie sheeple will rush to waste their money for anything with RGB or "Gamer" on the label, no matter how crappy it is. Well, that didn' take long. Changed to Monday already. Not sure what the deal is with UPS and USPS, but it seems like they've been failing hard with delivery delays lately. I think maybe one of the last half-dozen deliveries have happened on time.

Nice to see you, Brother John. I hope that everything is wonderful for you and your family. Yeah, that looks like a horrible sample. Especially the E-core and MC. Looks like the Cache is also a bit weak. Needs a bit more for voltage for 50x than ideal.

Yes, there is a 15% restocking fee. This will be my first customer experience with them and I hope to not have to entertain a return by getting an acceptable sample. If it is a terrible sample I will make an arguement that I don't actually want to return it and have no regrets in purchasing it except for the fact that the quality is unacceptable; with that being the sole reason for the return and ask for the option of receving an acceptable quality replacement instead of a return. But, even with the stocking fee that is better than selling it at a loss as an open box on eBay and paying ~15% to them in scumbag fees on the sale.

Yup. Hope for the best. Be prepared for the worst. Never expect goodness, but be pleased when good comes your way. (If we all got what we deserved, most of us would be dead already, LOL.) I just got a text from UPS and the expected delivery is Saturday.

Whose CPU is this? Is it from the Skatterbencher video? (I am at work and cannot watch for now.) Wow, 6.2GHz with only 1.418V is crazy good. Central Computer has them for $699. I ordered one. They have a 14-day return policy (does not cover shipping cost) so if it is not equal or better than my SP117 13900KS or SP102 14900KF it is going back for a refund. There is no point in owning it if it is not better than what I have. If it is worth keeping then I will sell my SP108 13900KS and the two dead 13900K/KS and cover most of the cost. If it is not an upgrade, I will get my money back. It would be stupid to pay extra for something less. It is interesting that people pay money for dead CPUs on eBay. I sold two dead 10th Gen CPUs on eBay for over $100 each a couple of years go. I don't know what they do with them, nor do I care. It's puzzling though.

@Papusan I do not know why the XMP profile runs slower, but after manually tuning with the same settings the performance is essentially the same with new beta BIOS. Maybe ASUS changed some default values for XMP that made it slower if 14900KS MC strength is not better across the board than non-KS. This is fine. Still need to put these modules on water. Still getting too hot with TM5 stability testing using only fans to blow on the RAM. Prior BIOS New Beta No, I did not see any reason to because some were complaining their VID/Vcore was higher and previous memory tuning no longer stable. Since I was stable I decided to skip it rather than invite unwanted drama. New BIOS (like most things) is often worse, not better. Best to only update firmware when hardware compatibility is added or something broken got fixed. Otherwise, it is only for people that are scared of the boggie man coming for them without "security" patch crap. (You already know this, so I am only answering your question if I tested the previous firmware with the same micro code.) I am going to begin testing when selecting the older micro code options in the BIOS soon to see if 14900K is slower than 13900KS in Cinebench at same overclock due to cancer micro code mitigations that I do not care about having mitigated. They should be the same, but 14900K/KF are always just a little bit slower in Cinebench and wPrime with the same P/E/Cache and RAM overclock values. ASUS always providing only "Improves performance" for release notes is just a lie to entice people into flashing newer UEFI filth. These were my BIOS settings on the Encore with 0071. Since we are warned against applying prior version CMO profiles, I set everything the same manually on the new beta BIOS and I cannot tell any difference. Seems the same to me so far.

Apex and Apex Encore share the same EC and ME version number, and the ME version did not change from 0071. The only changes (again the same version numbers for Apex and Apex Encore) are updates to EC 0142 and Intel MC updated to version 123.

I think the issue with them pulling arbitary numbers out of their butt like that is the problem because there is nothing to substantiate or quantify it. I agree with you in terms of how many times on a specific build, but many PSUs have a long warranty (10 years) and even those that do not have a long warranty keep working beyond their warranty period. They could get used for many years on multiple builds. So, depending on what cable one is looking at, 50 times over a span of years with multiple builds, motherboard swaps, GPU upgrades, etc. might not be too big of a stretch. The vast majority of end users won't plug and unplug cables 50+ times. If they do, it doesn't mean the useful life of the connector has arbitrarily ended. It might be less than 50 times depending on the quality of the part and the way the user handles them. It's subtle nuance, but there is a difference presenting something as fact and suggesting it is a given that cannot be questioned rather than presenting information as an educated rough estimate based on napkin math and simple logic.

It may also be the "not wholly profitable" applies to the product provider as well. If Amazon is merely distributing the CPUs for Intel and taking a commission, there is little profit for them. But, it still doesn't explain why the 14900KF got processed within about three days. That suggests it could be little more like a breakdown in process rather than a process running its course.