Intel Core i9-9900k 8c/16t, i7-9700K 8c/8t, i7-9600k 6c/6t 2nd Gen Coffee Lake CPU's + Z390

I wonder how close a super-cooled 9900K with an extreme OC can come to this? Should be interesting to see.
http://hwbot.org/submission/3958963_

http://hwbot.org/submission/3958961_

 

Hi, as a member of i9-9900K waiting for team. I'd love to know the result too. Should be cool!



Sent from my iPhone using Tapatalk

 

Do the math. At 5GHz, you get around 2200. So, under LN2, if you can hit 7.5 on all cores, you'd hit around 3300 on linear scaling. You will not be in the 4000s. That I know of, the 9900K doesn't have mesh, so memory will have less effect than on hedt platforms.

Sent from my SM-G900P using Tapatalk

 

Not close at all, you've got twice the cores (hyperthreaded too) and still at 5.1Ghz!

 

So, a product released over 6 months ago goes down 10% in cost on release of a competitors chip and you say it is temporary without proof, or from fear, etc. You do realize that is how business works, right? Oh, that's right, you are used to Intel not cutting prices even in the face of competition, without realizing that their shortage in capacity that causes a constraint on supply will, in the long run, cause them to lose market share and that they then must keep prices high to actually make the earnings needed without engaging in a price war that they currently are not equipped to fight. Great business analysis on your part. I should have you managing a stock portfolio. SMH!

They can go lower. The price per die is about $23 USD. Granted, you still have to add in the cost for the engineering, etc. But, they have tons of margin to squeeze and Intel cannot fight a price war when they have constrained capacity for manufacturing. Otherwise they lose the margins in their chip, cheapen the brand image (which many in here seem hung up on), and will eventually sell out while getting less than they would otherwise, which results in ceding that market share to the competitor otherwise. This is why Compal, one of the largest ODMs in the world, said they don't expect Intel's shortage to be resolved until middle of next year, Intel increased spending by around 7% for expanding 14nm capacity (which, if not able to convert those lines over to 10nm leaves a situation of over capacity of 14nm when Intel starts finally delivering on 10nm silicon end of next year and into 2020), etc. You've read the news, I won't belabor the point.

Either way, due to the MCM multi-die design, they know they cannot win a price war, especially on any chips that contain more than one die. Even with a single die, that die is used throughout the stack, with the dies binned to separate into different product segments. Intel, on the other hand, designs at least 3 dies, then cuts them down through binning. That is the lower core count chips, the HCC chip, and the XCC chip, usually only on Xeons, but which will be seen on the 28-core overclockable Xeon released around December. Intel stuck with Skylake-X on the 9th generation refresh on the HEDT platform, meaning the chips are the same as last release, just with solder. That was rather disappointing to hear and makes the decision for any current 7th gen HEDT owners to be wary of performance increases as they may not be enough to justify the added expenditure unless you are the type to not delid your chips to replace the old TIM with liquid metal, in which case the new soldered chips will increase your performance if using it for production, etc. (so if not overclocking the chip already because of being in a production workstation).

As to saying people are overpaying, look at Intel chips and the value calculations I put forth earlier. After you recoup the expenditures, you then have margins, which price drops at specific times are often figured into the equation long before the price drop occurs (hence dropping the price 10% on a 6 month old piece of hardware that the new model comes out in about 6 months). Considering how their supply line is set up, it makes sense. It also gives better value to customers considering alternatives on the basis of outrageous pricing on their new flagship mainstream CPU.

You really don't seem to have a grasp on their market AT ALL. Maybe you should step back and read through all those news articles to understand why they are doing this, or that analysts have changed expectations from 20% marketshare as the high end of what AMD is expected to hold on market share to 30% based on the Intel CPU shortage and capacity issues. That estimate didn't change after Intel announced the $1B additional to be spent on manufacturing, but investors bought the shares up as if it did. It really is sad.

Moreover, AMD is fine, pulling marketshare, destroyed the chance of bankruptcy, got refinanced, has been in the green and giving distributions to shareholders, something that wasn't happening before Zen, etc. They are meeting their expected margins, which have always been 30-40%, whereas Intel and Nvidia work at 60-70% margins (meaning those are the companies really screwing consumers, but who cares about facts anyways).

As to money to compete, I already laid out their roadmap to gaining marketshare and what analysts are expecting, but lets dive deeper into that, shall we. AMD, due to being out of the server market for a decade or more had less than half a percent. Test beds in companies considering large deployment take about 12 to 18 months. Considering the first Epyc chips were released at Computex 2017, the test beds and purchase orders are coming in starting in the third and fourth quarters of this year, with Epyc 2 chips being drop in replacements at 7nm, so you wouldn't have to rebuy the entire server for the deployment, simply upgrade the chips. This is why I think Q3 was supposed to be about 1.5% server market share and 4-5% market share for servers in Q4, according to the Q2 conference call with Lisa Su in July. This was before the Intel CPU shortage was known and news broke that HP and other OEMs were recommending AMD servers to customers, or that simply mentioning AMD would cause OEMs, on Intel's direction, to cut prices on their servers drastically, which amplified their shortage while at the same time reducing Intel's revenue in the sector where Intel makes its highest margins. You had the former CEO BK say that Intel's goal is holding AMD to 15-20% market share, meaning they already accept AMD will take that much in the server sector by the end of next year. That is Intel's projections, not AMDs. With the increased server market share by that much, when previously being at zero, and that being more important of a sector than consumer chips, AMD hit Intel where it hurts. But, once again, those pesky facts get in the way of your characterization.

So, to do this, AMD did take the money from Radeon that would have been used to compete on the high end. I think they made the right decision considering their balance sheet has been corrected. But, Su doesn't focus on consumers, she focuses on corporations and the commercial market. That means that consumers, by and large, are afterthoughts. They try to still put out good value, and at their prices, they are. But, the truth is the truth, consumers are not their target audience and anyone so caught up in that race that they cannot see Intel being bludgeoned in the high margin sector is taking their eye off of how AMD is making their comeback.

As to 2700X, if they tried to even market that at $400, needless to say $500, I'd be bashing AMD's choice as going crazy and literally would have a different selection of CPUs I'd recommend as better value, like the 8700K or 9700K. If anyone is dumb enough to pay a premium not justified by the performance given, regardless of the company that manufactures the part, you are a rube and the old saying of a fool and his money are soon parted is proven to be true. Intel tried to keep AMD from doing a price war, and since last year, AMD has been trying to goad Intel into one. Why? Because AMD knows they can win and watch Intel bleed out. The manufacturing capacity shortage at Intel just exacerbated the need to not compete on pricing, because if they did, they would wind up not delivering on the quarterly distribution and cause a sell off. Simply put, because they are selling all they can because their manufacturing line is messed up because they couldn't go to 10nm on CPUs, they will have the Q3 and Q4 perform fine, with bad predictions coming after that time period. So of course anyone following the market would ride those quarters out on profit until the real implications of it hit. This has been reported on in many outlets.

Intel is testing if people are rubes. I do agree there. Too bad.

 

I can still hit 4.5ghz reliably on air cooling in mine, click on my CPU in my sig to see my all-time high score

I wish AMD/Intel would REALLY push ahead and go bleeding edge, we are always getting stuff that is 2-3 years old already by the time it is released.

 

I removed a few posts in here that were political in nature. (I know there was a lot of content in those posts, so I apologize, but I saw no way to filter out the political discussion without adversely affecting the posts. Hence they are deleted.)

Let's talk about the Core i9-9900K and the like. Thanks.

Charles

 

oct 19th is the review, then on 26th we'll know just how much better it is binned over 8086k. at best maybe +1 binning.

hella expensive but 4MB extra on the L3 cache and 2 more cores probably worth it.

 

Some of the first people who preordered 9900K here in Australia are being told the ETA is mid to late November.

Paper launch.

 

Intel splitting manufacturing group into three parts after repeated 10nm delays
By Malcolm Owen
Wednesday, October 17, 2018, 12:49 pm PT (03:49 pm ET)
http://forum.notebookreview.com/thr...hree-parts-after-repeated-10nm-delays.825466/

 

then we just have to get from SL. theirs would be around oct 26th but we'll see if they also get delayed too

 

So, first post shows @Mr. Fox was correct to be worried on solder implementation.


Edit: Also good news for those that still want to delid, the Der8auer Die Mate still works to delid without having to heat up the chip.

 

I think I'd wait for a larger pool of Max OC results of OC'd 9900k's before deciding to waste time delidding and grinding, as even though the temps went down he didn't get any higher OC than before - der8auer said this was a bad sample 9900k which only reached 4.8ghz all core, but he also said he checked 8 9900k's (01:50) - so what were the highest OC's / voltages for all those samples?

Intel seems to have "jumped the shark" with the 9900k / 9600k thick die - I wouldn't have thought that it is thicker to support STIM - it could have been another manufacturering choice such as the process equipment Intel were forced to use to expand manufacturing.

I wonder if the paper launch quantities of the 9900k would allow Intel to pull back and re-release the 9900k with thinner dies and LM TIM later this year or next year before ramping up for full production?

 

So far looks good to me.

 

i9-9900K at -230 °C - LIQUID HELIUM


i9-9900K Review & Delid: Solder vs. Paste, Game Streaming Benchmarks

 

So, he was also talking about 4.8 and 4.9 OCs before delidding. He said he found 8C difference, and said per core difference of just delid and LM was 6-12C per core. He mentioned delidding prevented a situation of a single core spiking 5C then returning back.

Now, I'm not recommending people grind for an extra 4-6C, as that also exposes the unguarded silicon to electron migration. NOT SOMETHING TO BE DONE LIGHTLY, needless to say the issues that can arise with uneven sanding, etc. But, there was a sizeable number concerned about Intel's solder, and it shows that those fears were well founded. Der8auer found 1-4C difference for delidding AMD CPUs. He found 8C on delidding Intel. But, Intel changed other aspects to go to solder, like the silicon layer thickness, using a 0.5mm solder layer (which, from what I've looked at, 0.1mm would give drastically more heat transfer with less thermal resistance). This temp drop is still half of what is seen with their other TIM, so that is something that is nice.

Edit: @Dr. AMK - I ninja'd you. Check the prior page.

 

Intel's HOT now!!

 

6 cores @ 4.5ghz = basically no difference except the 9900K is a few degrees cooler than stock 8700K

 

Did you see the 13% more power than the 2950X at stock? Laughing so hard, makes me think that interview meme is coming back!

Edit: 5.1GHz was hard to hit and required massive liquid cooling. So my 5.0-5.2GHz estimate seems to be about right (if you can cool it). 1.375V

 

Looks like it gets warm but this thing absolutely beats on the 2700X at stock.

 

So long as you saw his part on cooling to get 5.0 or 5.1GHz.


When overclocked to 5.1GHz, it consumed 299W while an OCed 2950X overclocked to 4.1GHz consumed 349W in blender.


Hardware unboxed showed 5.1 failed on CLC AIO and 5.0 was pegged at 100C using a Corsair H100i, same with using Noctua NH D15
a 360mm Open loop custom could do 5.1GHz at 98C

 

https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/

https://www.tweaktown.com/reviews/8768/intel-core-i9-9900k-9th-gen-coffee-lake-review/index.html

 

1, 2, even 10 CPUs are not enough of a sample size to see if they got a low binned chip or not. I will wait for SL or other reliable results to see how far these chips can go. Regardless 5.0-5.2Ghz is simply impressive for an 8 core Intel chip. Ryzen clocked or overclocked if it could get that far lol would get just as hot and consume just as much power. Clockspeed makes a huge difference in power draw.

 

All of the reviews confirmed my predictions, except I never expected that high of power draw or VRM overheating on the budget Z390 boards. Turns out they need those beefy 12-16 phase boards.

Meanwhile, watch all the reviews. So far, I'm finishing the Hardware Unboxed after having watched Der8auer's video. But I think, considering seeing some other titles, that I nailed it.

 

The 9900k is 11% faster in 1080p games than the 2700x, and only 3% faster in 1080p games than an 8700k, and as @aaronne 's chart shows the resulting effect on the Graphics score even at 1080p is very small.

For such a high price premium for the CPU - and for the requisite high end z390 motherboard - not to mention a heat premium requiring expanded cooling + the delidding / shaving of the die to get the most headroom - I don't think it's worth all that expense to get a 9900k.

We haven't seen enough samples for sure, but at those temps without delidding and shaving down the die, it's gonna be running too hot to push the CPU past 5.0ghz all core, so maybe that's a good working OC to shoot for.

I wonder how the z370 motherboards are going to handle that power / VRM thermal load - even at stock, not to mention 9900k OC's near 5.0ghz all core.

 

-- Kit Guru seems to report the same-ish overclock as well. 5.0Ghz All cores and the CPU runs 4.8Ghz all cores out of box. He also shows 5.2Ghz all cores with no HT.

5.0Ghz all cores in blender showing 205w CPU draw with 270w total power consumption. So 75w more draw than a 4.0Ghz 2700X isn't that far off the mark considering it's running a full 1Ghz above that CPU.

 

The good ones will be fine, the budget tier ones nah unless you select override the VRM limits in the BIOS and wait for failure. A z370 Maximus board or Taichi will most likely be fine even when OC.

 

Intel Core i9 9900K processor review Guru3d.com

BIOS overclocking
Overclocking, tweaking the Core 9000 series is fairly simple. Any and all CPUs we tested reached 5200 MHz on all cores (this includes the six and eight-core parts). The voltage will vary per processor, but you should sit in the 1.30~1.35 Volt range. The good news is that on most motherboards you can leave voltage settings at auto. Even newer motherboards have AI overclocking available in the BIOS, by algorithm it'll learn the best stable tweak. For now, we'll do it manually though. If you plan to overclock, always invest in good hardware by the way (MOBO/PSU/Memory/Cooling), the cheaper motherboards often are not well tuned or do not have broad-spectrum features for enthusiast overclocking. Also get yourself a good power supply and proper processor cooling. Overclocking with a many core processor (doesn't matter if that is Intel or AMD) is more complicated than you expect it to be. Overclocking multiple cores to a high clock frequency is a relatively easy to do job, but is managed best from the BIOS. The Guru3D reader-base overclocks mostly from the BIOS to try and find the maximum stable limit. The generic overclock procedure for multiplier based overclocking is as follows:

Your reachable target for Coffee Lake-S in the 9000 series is 5 GHz to 5.2 GHz on all cores, with a good processor. Results may vary, as we are using ES samples.

Final words
Normally I'd say that the Core i9 9900K is the new king of desktop processors. However, price versus performance wise the Ryzen 2700X does hold ground. Also, we cannot shy away from the fact that from 4 to 32 cores, AMD has some sort of Ryzen (Threadripper) processor available. Strictly speaking, from a gaming point of view, if you have that GeForce RTX 2080 (Ti), then there's no better CPU than what Intel released today, and that is a simple truth. Whether or not you need 8 cores for that is something only you can decide, as really I am liking that Core i5 9600K just for what you can achieve with it (gaming wise). If you want the best of all worlds, then the Core i9 9900K clearly is the winner here today, but at a significant price. The generic Core 9000 series picture is that any of these procs overall simply delivers fantastic results there due to their high clock frequency. Six cores is the new four cores, but eight cores are more future proof. The 9900K does come with SMT/hyper-threading enabled and it did show the benefits of that immediately, but that makes this purchase more difficult to recommend. In closing, stability and functionality wise anything and all ran as expected with even better performance, if you are just gaming, the 9600K or 9700K are fantastic. If you need that extra bite that is close to the HEDT level, well, then there is no other choice than the Core i9 9900K as it'll bring enthusiast class level performance onto the mainstream desktop PC. Highly recommended, and let's hope that Intel will have good volume availability as this processor cannot afford to be even a slight bit more expensive than it already is.

 

I'm more interested in the 9700K at this point. I haven't had a chance to watch it yet, but Hardware Unboxed just put its review up.



And then there's this. $600 motherboard? Wew lads.



Edit: Looks like prices of Intel CPUs have been creeping up in recent weeks/months, which makes Ryzen 2 look like an even better value. Makes me glad I got my 8086K when I did.

 

This guy is also reporting similar power draw to a 2700X for total power consumption. Again someone that dives deep into technical and hardware testing.

5.0Ghz OC on all cores under blender and AIO 79C. Not bad.

 

I posted that review a couple pages back, getting to it after the GN review

 

Yup, the 9700k is coming out now at the lowest from B&H for $420:
https://www.bhphotovideo.com/c/product/1435918-REG/intel_bx80684i79700k_core_i7_9700k_3_6_ghz.html

And, even though the recently discounted price for the 2700x has dropped to $295 the lowest price I can find it for sale right now is $305 from B&H:
https://www.bhphotovideo.com/c/product/1404638-REG/amd_yd270xbgafbox_amd_ryzen_7_2700x.html

So the 9700k has the same problem as the 9900k against the 2700x, hot and expensive in comparison. With the differences in performance not enough to bump up the value in the Intel choice over the 2700x.

I hope AMD comes out with a nice 2800x priced accordingly, match the performance of the 9900k a bit closer while maintaining a low price.

Intel i7 9700K vs Ryzen 7 2700X Benchmarks | Test Review | Comparison | Gaming | 10 Tests
Benchmark PC Tech - MultiTechnopark
Published on Oct 19, 2018

Spoiler: Click to See Video

Intel i7-9700K vs Ryzen 7 2700X BENCHMARK
Included tests: Hitman, Total War, Tomb Raider, Far Cry | 7-ZIP | Truecrypt | X264 | Playerunknown's battlegrounds PUBG | FAR CRY 5 | WITCHER 3 4K| GTX 1080 and GTX 1080 Ti and 4K test included – Gaming and Productivity | Encoding | Transcoding | Compression Tests Review on Windows 10

Intel i7 9700K vs Ryzen TR 1920X Benchmarks | Test Review | Comparison | Gaming | 10 Tests
Benchmark PC Tech - MultiTechnopark
Published on Oct 19, 2018

Spoiler: Click to See Video

Intel i7-9700K vs Ryzen TR 1920X BENCHMARK
Included tests: Hitman, Total War, Tomb Raider, Far Cry | 7-ZIP | Truecrypt | X264 | Playerunknown's battlegrounds PUBG | FAR CRY 5 | WITCHER 3 4K| GTX 1080 and GTX 1080 Ti and 4K test included – Gaming and Productivity | Encoding | Transcoding | Compression Tests Review on Windows 10

 

Yeah, I think I'll be sticking with my cherry 8086K. Sure, it has two fewer physical cores than the 9700K, but it has HT and can run at 4.4 GHz across the board with a -150mV undervolt. Real-world load temperatures average around 70 degrees or less on a laptop that has a vastly inferior cooling setup compared to other machines that are capable of using this processor.

 

I think that is a great idea haha. That 8086K in a laptop is beast mode already.

 

Intel i7 9700K vs i7 8086K Benchmarks | Test Review | Comparison | Gaming | 10 Tests
Benchmark PC Tech - MultiTechnopark
Published on Oct 19, 2018

Spoiler: Click to See Video

Intel i7-9700K vs i7 8086K BENCHMARK
Included tests: Hitman, Total War, Tomb Raider, Far Cry | 7-ZIP | Truecrypt | X264 | Playerunknown's battlegrounds PUBG | FAR CRY 5 | WITCHER 3 4K| GTX 1080 and GTX 1080 Ti and 4K test included – Gaming and Productivity | Encoding | Transcoding | Compression Tests Review on Windows 10

 

So, after getting through almost all of OC3d's video, I like the review, but I disagree with his advice at the end. It isn't worth the value.

I'd recommend the 8700K or 9700K for gaming, 2700X for productivity, and the 9900K doesn't make sense until prices drop A LOT.

I also think I figured out the difference in the amounts on watts on the review from HU and others. I'm wondering if most were referencing Core draw and HU was using package draw, which includes other than just cores. Just an idea, because I have core draw on my 1950X overclocked to 4.05GHz at 225W, but the value of the package draw is much higher. As an example.

Edit: This is where I hope some go through build value and also compare the 9900K to the value of the HEDT platforms out there, not just sticking to mainstream builds, which the cost blows the other costs for mainstream out of the water. It may very well cause people to consider HEDT options from either major player in the market.

I also would hope people picked up their 8700Ks back when the prices were $370 or less, as that was the best time to pick one up, and the current $390 price point is a bit high, although frame rates at 5GHz on those clearly excel by large percentages over the 2700X (hence builds primarily focused on gaming).

Edit 2: I was wrong, total power consumption of the system was used. Here is the techspot article.
https://www.techspot.com/review/1730-intel-core-i9-9900k-core-i7-9700k/page3.html

 

The 8086K is on sale at MicroCenter right now for $379.99, which is only $10 more than it was at the bottom of the market a couple months ago. If you want an Intel CPU, this is far and away the best value right now.

The uptick in pricing is probably why I've also seen a surge in folks on the hardwareswap subreddit looking for second-hand 8th-generation CPUs.

 

So, finishing up Paul's review from Paul's hardware. 18.4% better than the 2700X at CPU workloads, 11% better at gaming workloads at 1080p. Above 1080p, that 11% margin drops. That is one reviewer and in the specific workloads tested at stock. Not likely people would run either the 2700X or the 9900K at stock. But, it does beg some questions.

We did see more comprehensive game testing in other reviews. As I said, take the values of the games you play from trusted reviewers, average only those, not the ones you don't play (you can also, to a degree, do this based on game engine performance if there is an upcoming title on that game engine and you want an estimate), work out the performance delta between the CPUs you are considering, and then compare the performance delta to the price premium for the product. This makes the choice more personal for your specific use cases, something everyone should consider. That is where you can find value, and put the savings toward other components in your build's budget.

 

It beats the 2950x in power consumption... it should be illegal to put a 95w rating on these monstrosities, overheating and insane power usage... 8700k is the only way to go lol

 

Fixed that for you.

 

Cancelled my Newegg, and BH preorders for my 9900K. The chip is delayed or is at least a paper launch today for the most part.

I am keeping my $499.99 order at Amazon because that for me is where I feel I’m willing to pay. Beyond that I’m kinda meh on it. I don’t need the multi core and was mostly interested in log term viability of my platform. Might just wait for 10nm+ though. My 8700K offers almost same gaming performmace and at its overclock better Multi than the 9700K. It’s nice to see the 9900K on top but are we seeing diminishing returns of just slapping cores on the CPU. We need larger steps forward in IPC to really see any significant gains. 10nm+ and 7nm+ will likely bring this.

 

Depends on how you look at it... https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/16.html

Up to the owner/user increase the power limits

See also Relative performance (Cpu tests) on top of page... A good allrounder https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/19.html

 

I disagree on the price point. Half of all 8700Ks got around 5.1GHz. 65% of 8086Ks got 5.2 GHz with the top 15% at 5.3GHz (top 4% of 8700Ks). Considering the price points of closer to $370 on the 8700K and $420 on the 8086K, it begs the question whether the 100MHz difference is worth the $50 premium, to play devil's advocate, something I'm good at.

This is what I thought might happen. Now, if the information out there from Intel on 10nm+ is correct (which came before the news of the gutting of parts of the 10nm process to get yields stable to get it out the door), we will be looking at the 10nm+ Ice Lake being a side grade for transistor performance, but if you pack more instead of pure die shrink, you can improve IPC, hence getting back to your point.

I don't have much faith in Intel's 10nm design. It will bring some advancements, but as I've said for over a year now, this is Intel's dark times. But, the IPC stayed the same since Skylake because these are all practically skylake refreshes on 14nm nodes of different iterations. Ice lake will bring uarch changes, which in turn can bring their own benefits. Intel has been doing uarch refinements on skylake this entire time, but primarily relied on process refinements to get the higher clocks.

Now, we see that the IPC is roughly the same on the 9900K as the 8700K. That means the 13% IPC number on the new Ryzen 3000 chips will have the 9% IPC advantage in certain workloads. Intel will likely have an IPC gain on Ice Lake, but it really has been awhile since we have seen an 8-9% IPC jump from Intel, if being honest. But, even though expected frequency of Intel's 10nm process will be lower than 14nm++, adding the speed and the IPC gains, Intel expected 25% additional performance on cannon lake (skylake uarch on 10nm) over the 14nm chips on the market at that time (Kaby was released, coffee was not when those numbers came out, and once again, that was before the process changes were announced in August of this year, versus this other information coming from March 29th, 2017 on an Intel Press Day).

 

Amazon now charging $579.99, unreal.

Brian coming out with another great video. Just as I was thinking, Intel went full AMD here. They basically slapped 2 cores on the die, and clocked it like AMD to almost full potential leaving only 300-400mhz on the table for overclocking. Maybe a tad more if you got a cherry chip with a full closed loop with huge rad. This is great as it shows the real potential of the chip in review videos but leaves less toying around for fun.

Edit: 115C TjMax

Edit 2: Nvidia is adding GDDR5X to ... a 1060? What is happening in the world. 1060 3gb (with less cores), 1060 5gb (smaller memory bus), 1060 6gb, 1060 6gb GDDR5X. Nah, that's not confusing at all.

 

A fair point, but how are more recent 8700K models stacking up in terms of overclocking and/or undervolting? It's been my understanding that launch samples were fantastic, but those available post-8086K haven't fared as well. Using the stats you provided, you have a roughly 30 percent greater chance of getting a good-to-great overclocker in the 8086K versus the 8700K. The price premium, though, is only about 13 percent. That still makes the 8086 the better buy, IMO.

 

It looks like a water 360mm radiator is a must for overclocking. Noctua gives stock cooling as will a smaller water 240mm radiator barely allow a 5.0ghz OC - but you couldn't daily drive with either @ 100c+ :



For anyone that missed the der8auer 9900k delidding video a few pages back, it's worth watching to get some perspective on the cooling issues with the new STIM on the 9900k / 9700k:
http://forum.notebookreview.com/thr...e-lake-cpus-z390.811225/page-62#post-10810919

 

That is why they went to solder, to clock it higher for the reviews, and as I mentioned, no process change so no real scaling over the prior gen. Just clocking closer to max. This isn't to denigrate its performance at all. I'll watch that review fully here shortly (thanks for the share).

That is a fair point. I forgot where I saw it, but there was a video somewhere discussing the SL %s over time. I need to track that down, because I thought it dipped then came back up, but you may be correct and Intel is scraping all the higher binned chips for the 8086K, thereby dropping the chance of getting a good 8700K now. Fair point. But, then comes in the math of how much that extra 100MHz is giving in raw performance. By my math, just on paper, that is about a 2% difference in performance (around 4% if looking at the 5GHz vs 5.2GHz), so the question is whether that amount is worth the $50. That is an individual question I leave to the purchaser.

And @hmscott thanks for sharing those. It puts into graphs what I mentioned in another comment. Much appreciated. Also, it contradicts the findings at OC3D where he said you wouldn't need those types of cooling solutions. But, that opens up a question of variation by chip on solder application, which we won't have a full answer for until a larger sample is seen, as @Talon pointed out, aptly.

 

It looks like Intel "jumped the shark" with the 9900k / 9700k STIM and fat die thickness, as the temps are much worse with solder than with paste / LM, and get much better when thinning the die thickness to reduce the thermal distance between the IHS and circuitry:

For anyone that missed the der8auer 9900k delidding video a few pages back, it's worth watching to get some perspective on the cooling issues requiring delidding / shaving the die with the new STIM / extra thick die on the 9900k / 9700k:
http://forum.notebookreview.com/thr...e-lake-cpus-z390.811225/page-62#post-10810919

 

I wouldn't say much worse. Delidding give 8C, which is half of what delidding used to give. Sanding the die down by 1/4th the total silicon height on the die gave like 4 or 6C. That brings it to being about the same as if they just didn't solder at all. Now, they do the protective layer differently on soldered vs not, which means if you wanted to use LM, delid and not soldering is actually better for the chip longevity with the electron migration issues that can occur. But, there may be a chance Intel reviews his video and looks for a better implementation or refinement in the future. It's something I am going to keep an eye on, considering it is Intel's patent that is used by AMD and AMD saw around 4C from delidding. I am going to get on his video and ask Der8auer about the differences he has noticed in silicon die height between the AMD chips and Intel chips, etc. Either way, I will encourage Intel to continue soldering for the masses.