*Official* NBR Desktop Overclocker's Lounge [laptop owners welcome, too]

I've seen that before. No idea what causes that. Almost like it's some sort of oxidation. Usually don't see that if there is a healthy amount of LM left on the die when you remove the heatsink / IHS.
That color looks suspiciously the same as the color you get when trying to remove LM from a copper HS once the liquid part comes off...

 

For me, it was the iGPU temps idled at around 85-95C, and if I ran something a bit intensive, like loading a webpage, it would hit 99 and cause the CPU to thermal throttle (and this is the ONLY time I had seen the 4930 thermal throttle, before that whole issue, it would go all the way to 100C at whatever speed I told it to until thermal shutdown.) And then if I loaded many webpages or loaded any game, a thermal shutdown would occur. All of this was only for the iGPU sensor, the highest the cores would go was like 80ish C

 

Doesn't running the Alienware Ranger in dGPU only mode make it ignore the iGPU? I don't even see an option in Intel XTU to monitor the iGPU's temperature, voltage, or clockspeed when running in dGPU only mode.

 

The LM was done right. I always check to confirm it does not wipe away like water on the silicon or contact surface. The nickel Optimus SIG V2 block was difficult to obtain this my old SIG V2 was copper, so the LM would soak in better.


I usually apply the LM, let it soak and keep going to make sure it is fully coated. Then just run the Qtip around random areas to make sure it doesn’t wipe away.

Im very very very funny and particular over temps. I will re-apply and re-mount something a gazilioon times.

 

This cpu was removed and re-mounted with the same LM a few times though. After it died on me, I was troubleshooting a lot, removing, reinstalling etc.Being direct die, I naturally figured something was up with the socket pins contacting properly with the die frame.

I didn’t care to re-do the LM after each mount. (Something I would normally do on a functioning CPU) but the chip was dead.. I was just trying to get it to work.

 

Could very well be a burn mark from a hot spot then. I didn't know that it was dead. Sorry.

 

Since my 10900K is dead, I spent too long deciding which new CPU to run… I wanted a really good 10900K, or a really good 11900K.

It’s very difficult to obtain either one of these in a binned form.

So I just purchased a retail 11900K from Newegg. Hopefully I get a good chip! I am curious to really see what it does.

Also, the real deciding factor to even consider buying a cpu as ridiculous and silly as the 11900K came down to one thing. I like to play the game rust a lot! And, this game absolutely sucks at using a CPU properly (Just like most games do) usually only 3-8 cores will see really high utilization. So the better IPC of the 11900K made way more sense. It isn’t often game titles will push our whole CPU to 100%. Most games will push a few cores to very high utilization, and the end result is, this starts eating away at our GPU usage. So, I can definitely spare a few cores for 15% faster IPC.

And if for any reason I don’t like it, then I’ll just go buy a 10900K, and sale the 11900K.



show me credit union







Also, a word of wisdom. (Do not buy cheap
Memory) I wasted $350 on a 2x16GB set of dual rank DDR4 OLOY. This ram is 4000Mhz [email protected] XMP. (It’s total crap!)

It won’t even post on my Z490 Dark KP. I managed to get one stick working. On an older bios. And then I went in and started dumping supplemental voltages, and memory voltages, cpu voltages, anything that could potentially assist. And it won’t post anything, and that’s with just this one stick..

I’m guessing this is a memory QVL support issue. I have never had a problem like that with memory before though. I gave up, and just went back to my 2x8GB G.Skill 4000CL15.


These belong in the bin.. The trash bin, not Newegg’s bargain bin. To think that Newegg was originally asking $500 dollars for these two sticks of ram! And now they still want $330 plus taxes. That kind of money isn’t far off from buying brand name Samsung B-Die. Does Oloy just say a prayer and ship them lol, they don’t these sticks out?


EDIT:

I tried them on my brothers motherboard too. Same results. No post.

 

You can't change the sensor, but you can adjust the reading in some software. In fact, that is how I get HWiNFO64 to show 4000 instead of 2000 on my memory clock. Right click on the sensor and choose "Customize values" in the menu. I think AIDA64 also has some similar sensor adjustment capabilities.

 

It will be fun to see where you take the 11900k on that big ole CL and rad.

And yeah, never skimp on memory depending on criteria. When I am setting up 3200 systems (which is the lowest I go) price is a consideration, but when I'm looking for better tiered memory? I tend to stick with brands and results I know especially if they aren't on the QVL list.

I tend to stick with G.Skill as my first choice for upper tier use ever since the MSI 1761 days as they just worked and were superior and the few times I had bad sticks they exchanged them fast and no problems.

Have you seen exactly what they're capable of? Do they post at SPD? Maybe see where they cap out on each w/ up to 1.7v just as a test.

 

Try running ThrottleStop.

https://www.techpowerup.com/download/techpowerup-throttlestop/

While you are at TechPowerUp, download GPU-Z.

https://www.techpowerup.com/download/techpowerup-gpu-z/

When you are having a throttling problem, show me a screenshot that includes ThrottleStop with the Limit Reasons window open and include GPU-Z with the Sensors tab open. Do this while the CPU is loaded and throttling and show me a second screenshot when your computer is idle at the desktop.

I have never seen an iGPU temperature sensor crap out but anything is possible. Have you ever replaced the thermal paste? Are your CPU idle temps normal?

 

Well I still have my G.Skill 4000Mhz CL15 4x8 matched set that I previously used on X299. However, with my Z490 Dark I can only use 2x8GB of it. And this set is just stupid good. It will run anything you type in. I haven’t even needed to send crazy high voltages to it. This set would probably do over 5,000Mhz with the right CPU and person.


I wanted 32GB though, that’s what led me to purchase that Oloy 2x16GB set.

Dual rank performs better, and I figured this set would work. It would not work at all. The 1 stick would not re-post with any settings at all. I spent a few hours messing with it. Them again on a different motherboard. I eventually gave up.

Should have just grabbed these lol. they’ll post lol.

https://www.newegg.com/g-skill-32gb-288-pin-ddr4-sdram/p/N82E16820374249?Description=2x16GB Gskill&cm_re=2x16GB_Gskill-_-20-374-249-_-Product

 

Grabbed that kit a week ago

Stupid expensive though.

 

Holy cow those are pricey and blingy.....

@Talon Do you have a pic of them installed and lit up?

 

https://imgur.com/a/rMHn307

 

Dual rank performed measurably better for me on Intel, but not so much on AMD. Latency was higher (if you can even imagine that) and overclocking capacity is lower with 16GB sticks. I am not sure spending gobs of money on high spec memory with Ryzen makes good sense because of its buggy architecture and ludicrous latency issues.

I have this kit coming and I am going to flash the XMP profile to the dual rank 2*16GB 4000 CL18 sticks that I have which look identical and have the same B-die IC to see if they are stable at CL15. If not on the 5950X I am pretty sure it will be with the 10900K.

I am considering a 5950X delid to see if that helps the volcanic temps.

 

That's a huge gamble bro. If you succeed, you will be one of the few in the world that successfully delidded a 5950X.

 

As long as they are responsible for replacing it if they break it... https://rockitcool.myshopify.com/products/delid-relid-service-amd-ryzen-cpus then I say why not? It might help make my decision to give AMD another chance less regrettable.

 

I was thinking the same thing.

5600x delidded:

https://www.pcgamer.com/heres-an-amd-ryzen-5-5600x-delidded-and-destroyed-you-know-for-science/

That would be a nice swing and hit if it works out especially as you say if they are responsible for a bad delid.

I didn't even know they offered 5000 series delidding, nice.

 

Very interesting. They have no disclaimer either. They must be highly confident in their work.

At the very least, if you choose to sell your Ryzen setup later on, you can get more for it due to the delid.

 

I would really like to make it work. I am tired of spending and losing money hand over fist. Buy high, sell low... that needs to stop, LOL. I should have known better. But, I was bored with reaching the end of the fun and games with 10900K/KF. There was nowhere left to go because everything newer that Intel offers is a downgrade from 10900K/KF. I love the Cinebench and 3DMark physics scores, but there is not much of anything else I can find to say nice about Ryzen. Other than those two things, Ryzen sucks. I would never recommend it to anyone. Too damned flakey and unreliable unless you want to run it stock or with an unimpressive/modest overclock like a zombie gamer-kid. The boots fine with nice BIOS overclock settings for a few days and then suddenly doesn't and having to clear CMOS has made the thought of using a hammer to "fix it" seem attractive to me. The nicest thing I can say is they are consistent. Their products were flakey 12 years ago and they haven't changed that at all. It has made me appreciate my other desktop more, even though it has a chintzy Strix mobo.

Edit: I emailed James at RockitCool to get more info about it.

 

That memory works really good.

I am considering delidding my 11900K. Honestly, I think I am gonna go for the baking method. This looks very safe and easy to me for some reason.


I will bake at 338F for 15-20 minutes, then lift the IHS right off. Scrap all the adhesive off, coat the little SMD’s with some sealant, and boom direct die!!

Der8auer did the exact same thing. I really don’t see how it could go wrong.

As long as the 338F doesn’t damage the CPU then I’m ready to move forward with this right away!

 

They apparently have a proprietary tool and process. They don't sell a Ryzen delid tool, only the service. Being a reputable business, I can't imagine them not being accountable. And, as @Rage Set noted, there is no disclaimer associated with the service.

No need for scraping. Please don't. Use the stuff from RockitCool to dissolve it. It's faster, easier, does a cleaner job and there is no risk involved with gouging the die.

https://rockitcool.myshopify.com/collections/10th-gen/products/quicksilver-solder-remover

Remember, the reviewers on YouTube get stuff sent to them for free. If they destroy something it costs them nothing. Intel actually sponsors and endorses things like this, even when they do not publicly admit it, because it makes their products shine. If you or I do it and screw it up in the processs the loss is significant. I can't really afford most of what I do and I make sacrifices in other areas to pay for it, like driving an older car and not having car payments, not eating out, etc. I generally pay cash or pay it off in 90 days, but it is painful.

 

I think I mentioned a long time ago in this thread the one thing about AMD was the natural boosting algorithms really gave each user a large chunk of whatever overclocking potential each chip contained. That doesn't leave much on the table for overclockers who are used to fairly juicy margins but for the vast majority of end users, they are getting most of what their CPU can deliver which is good in its own way. We do not represent normal end users....not even close.

It's flaky for those looking for the devil in the details but for most users it will be a set and forget situation. My brother has stumbled upon STILL existing USB problems since he is using his 5600x system daily and that was with the newest BIOS back in June. I still had USB problems with my X570 Tomahawk and he is using a Gigabyte. He's not comfortable getting "too far under the hood", so I'll be taking a trip up to his abode with a dedicated USB 3.2 PCIe card and giving his system a complete checkup.

XOC really pushes AMD nicely past 5ghz for benchers but that is that "next next" level.

I see rockitcool also does 11th gen delids now too for anyone who wants to get to that next level.



https://rockitcool.myshopify.com/collections/11th-gen-intel/products/delid-relid-service-intel-cpus

 

I think it’s because it’s easier than we make it out to be. Bake the cpu and lift the IHS right off. The only thing I worry about is the 170C it takes to melt everything real good.

 

Wow! They lift the IHS directly off the substrate. It just goes up! That is a cool delid tool. It has a wedge on each side.

Now, why can’t they just sale the tool or loan it out. I’d rather do it my self.

 

I think with 11th gen and 5000 series there are just too many potential DOA scenarios for end user deliddings. I've seen some of their reviews from end users complaining how their 6th-10th gen kits killed or ripped their chips up and its definitely a pebkac issue. 11th gen and 5000 series are a recipe for a lot of disgruntled end users trying to delid their own CPUs.

 

I used quicksilver on my last delid (9900ks). Previously delids I used a plastic razor (not metal) to shave down the solder as low as possible then liquid metal to agitate and break it up then metal polish to bring it home.

I will say the quicksilver was MUCH easier.

 

The tool is fairly elaborate and the cost to purchase one for delidding one CPU is probably not cost-effective for the end user. It would likely be a whole lot less expensive to pay them to do it. Loaning it out or renting it is a great way to lose money for a business (due to damage, theft and slow returns) and they only need one or two of them for their own use. They are a fabrication company, so they can make another one, or make a new component for one, if they need to. They would have to produce more of them for rent or customers would be on a waiting list. I think they would be totally nuts to even consider that option as a business.

If they sell it, then someone can use the tool to compete with them for delid service, so that also makes less sense for them as a company. If they charge a reasonable price for delid service they will ultimately make more money than selling the tool, and won't have much competition.

If I were in their shoes I would do exactly what they are doing. The average PC enthusiast is not going to do it, especially if they have never done it before. It is (in their mind) too risky and the CPU is too expensive for novice experimentation. The first couple of delids I did were on 6700K. I used a razor blade and was sweating bullets. And, I am more of a risk-taker than most PC enthusiast and gamer-boys.

 

Oh I was referring to the black adhesive.

 

I think it is not by accident that they do not have any photos or videos demonstrating their Ryzen delid tool or delid process. I would not if I were them. I think it is smart on their part to keep that much under wraps.

Yeah, other than the USB issues some folks are having, the 5950X is fine for the average user/gamer that doesn't overclock the CPU and RAM, or does a very modest overclock. I am sure it is nice for content creators as well. I was hopeful that using the chiller would yield better results for me. I expected it to suck with ambient water cooling. The boosting algorithms are unimpressive in terms of both clock speed and benchmark results. Manual overclocking is the only thing that comes even close to producing scores high enough to take note of. The PBO overclocking tends to use more voltage and produce more heat as well. The only advantage I can see that it offers is that c-states still function and that lower idle temps by a few degrees. But, I also don't like c-states. I want the CPU to operate at a fixed clock speed and voltage at all times, just like I have always done with Intel.

The 7960X and 7980XE were high core count and have a MUCH lower base clock and MUCH higher overclocking potential. I had lots of reservations about rolling the dice on Ryzen due to a less than satisfying experience with AMD products. There are some things about it that are extremely impressive, but they are few and overall I find it rather disappointing. My bar was set too high as a long-term Intel overclocker. All that said, I am almost positive I would be disappointed with 11th Gen Intel based on everything I have seen.

I am not giving up yet. But, it is not as rewarding and a helluva lot more frustrating than I anticipated due to the bugs I didn't expect. Overclocking challenges are one thing. Those are fun. Dealing with nonsense is another. That's why I basically have nothing to do with laptops anymore and loathe them. I have a very low tolerance for nonsense.

This overclocking hobby has been a real joy for a long time, but I am seeing the writing on the wall. The industry is circling the drain and selling trash, so I am probably going to be forced to find another hobby because I do not plan to adjust my expectations. Hopefully, it will be a less expensive one. Easy come, easy go... que sera, sera. If I do that, I probably won't upgrade anything for at least a decade because, apart from overclocking, my expectations are only that it works whenever I press the power button. I have mostly lost interest in gaming. I find a few FPS titles enjoyable as long as they are linear, single-player campaign and work offline. Those are becoming more rare and I have no interest in other game genres. But, gaming is also too much of a time-sucker. I tend to be a binge gamer and I can waste an entire weekend on it if I get started. Since gaming isn't high on my list of PC priorities, I don't need new tech or updated drivers.

 

They could charge half the cost of a tool and people would probably still use it. And if it’s lost or stolen, there is a charge like Performance PC. I remember my first delid a 6600K with the same method, a razer blade. I sealed it back together with some white stuff called (Toilet sealer) lol it was the ugliest cpu that ever walked the earth. I can probably dig up a picture of it on google somewhere. It’s actually pretty funny

 

But, I think it still makes more sense for them as a business to not show the tool, demo the process or rent the tool. They basically have a monopoly, and if I were in their shoes I would milk that cow for as long as I could. Offering a unique service with no competition is a business owner's dream. If they sell it or rent it there is nothing to prevent others from stealing the design and offering a similar product, patent or no patent. I think it is brilliant on their part.

Now, if Ryzen delid tools start to surface and other people develop an idea that works well, then yeah, sell the tool for the DIY people, and continue offering the delid service for the faint of heart.

 

I dont disagree but it would be very interesting just as an experiment to see how renting would work.

Instead of sending it back and forth, you could simply send it to customer 1 who is "renting" the tool, then Customer 1 sends it to #2 to eliminate the back and forth cost of shipping. Take pictures before and after for documentation and such.

Terrible for holding the IP for a particular function but the above hypothetical for some reason that I cant explain seems interesting to me.

 

I totally get it from a business stand point. I guess I’m thinking of it as the consumer standpoint I am Impatient, and want everything right now lol.

 

edit:


Yeah I certainly would not use one of my old delid tools. But I think the baking method is very safe. But, just like anything, extreme care and caution must be used.

Skylake X delid was pretty easy to mess things up. People would put the cpu in backwards, and the IHS would just snow plow the caps off. Or even in the correct orientation, if you went just a hair too far, it would plow off a cap/Smd on the top right of the cpu.


The Intel 11th Gen has a ton of parallel parked cars all around the IHS, and there is only one way to go, and that’s up.


I’m nervous about it, but I think I can pull of the baking delid method.

 

@electrosoft


What is considered good for the 11900K, you have tested a few of these right? I am confused by these voltages I see online. There is always a different voltage listed everywhere I read.

Did you just type in an override fixed voltage? What stability testing did you perform, how long? AVX offsets?


My 11900K will be here Monday. I am actually very excited to test this CPU. But these little 8 cores actually do very well. You know, I discounted them like most people as a terrible CPU in the beginning.

 

Easiest is to just set it to 5.3 all core, leave it on auto run CB23 and see what it pulls Vcore and watts wise.

The slow and plodding method is to start with just auto settings LLC stock and do some CB23 single multi runs 4.8, 4.9, etc.. till you crash to get an idea of what type of voltage it wants to pull on auto. That will give you a baseline to see where it wants to go and then you can start to dial in. You will also see where your 11900k "falls off the bridge" as usually there is a point where the next 100mhz bin up the pull increases pretty substantially.

If you want to jump right out the gates and go YOLO, try 5.3 @ 1.425 and give CB23 all core a run and see if it will pass and at what temps and what it is pulling (watts and Vcore). If it borks, climb as always till it doesn't or you realize it's just not going to happen. One of my tested 11900k's needed 1.55v to do it and was a furnace. This one does it at 1.4 80c or lower. The variance is insane.

Leave memory and cache at stock and get a raw assessment of where your 11900k can go.

Then after that, you'll go back to stock and see where your IMC taps out if you're going gear 1 initially.

Anything > 4500 cache (assuming your CPU can do it) is when the heat and power starts creeping in on top of everything else.

I found (with my weak cooling lol) that once I hit >85c under load, that is when cache and OC inconsistencies started creeping in. You should have much better luck especially if you delid. 11th gen is very temp sensitive or at least it is for me. I would watch my temps creep up and as soon as I was over 85c pushing cache here comes a crash like clockwork.

Some balk at weak IMCs even if it is a great chip overall. If it bonks out at >3733 and can't hit 3866, they're ready to toss it. I'll take a tightly optimized 3733 @ 5.3+ all core anyday.

Even though you don't have an Asus board, you can still load up XTU and get a look at your V/F curve up to 4.8.

After all is said and done and you find that IMC limit and OC limit, dial in those timings. Many MBs can't run G1 at 1N 3733 for some reason. I know the MSI Z590-A Pro, Gigabyte Z590 Aorus Pro AX and Asus B560M Prime all couldn't no matter what I did across several chips. I know Shamino and the Asus team got it going on some of their boards. I will be curious to see where you go with yours.

 

This is all very very helpful! I usually go full yolo with stuff.

What kind of VSA and VCCIO voltages are you running when your testing the IMC and cache OC? And what’s up with the VCCIO2?

 

After that one, decent 11900k died on me, I stay 1.5 or lower now even for benching which I'm done with on this one my desktop is keeping warm. The sample I have now runs crazy cool even on an AIO and does 3866 no problem. When I got a chance to plug it into my friend's Asus board it said SP94 and after flashing the bios to update the MC, it turned into an SP95 so it made sense why it was so much better than the previous 4 11900k's I tested. I don't want to bork it before it gets a chance to spread its wings in an X170KM down the road. In an X170KM with memory limited to 3200ish, it will run very nicely. If Prema is out by then for it, I can see where these 3800 CL18 sticks will go I have in the drawer:

 

That’s really good laptop memory!! I considered buying some, when I found a X299 Mini ITX motherboard for a good price. Asrock managed to maintain quad channel on X299 mini itx, but they used DDR4 SO-DIMMS. Pretty smart idea actually. Asrock is pretty good with their odd ball Mini ITX design motherboards, but when ASrock launched the first HEDT ITX board, the X99 Mini ITX in 2015-2016 ish, it only had dual channel memory due to space constraints. Anyways, that’s really good Samsung Bdie!! Expensive stuff.

I think it’s the only high end DDR4 option for SODIMM’s.

Pretty sure @jc_denton had some a couple years ago on a P870TM1.

 

https://hwbot.org/submission/4820223_ | https://www.3dmark.com/3dm11/14609428

 

Who do you think helped me train those puppies in my P870TM1 to push them to 3466? JC is da man.

 

I've also tested 1 11700K and 2 11900k now. My 11700k was a pre release sample I got my hands on early, was a decent chip for the "i7" and did 5Ghz all core no offsets at 1.39v LLC5 on Asus board or something like that.

My i9 that I got on launch day is SP83 and does 5Ghz all core at 1.32v LLC5. I can bench 5.3Ghz in CB23 no offsets but for daily use I don't need more than this for 4K gaming as I'm not generally CPU limited. Unfortunately my i9 doesn't like 3866 G1 with tight timings. I stick with 3733 CL14 (can do 13 but saw no advantage). I am currently at 1.35v VCCIO Memory and 1.25 VCCSA with 1.5v VDIMM.

I picked up another 11900K from Best Buy recently when I thought my chip was going after random BSODs. Turns out it was me not setting things correctly after getting my new RAM. I did a clean OS wipe, retuned BIOS and things have been solid again. That new 11900K was pure crap. SP50 (I am seeing a lot of reports of these chips) and it was logging WHEA errors at 5Ghz and 1.375v BIOS. I took it out and boxed it up. It's going back to Best Buy. There is no excuse for Intel selling these crap left over bins as i9s. I highly doubt that chip could have done 5.1Ghz all cores as is required by Intel's own ABT which is considered "stock". 5.3Ghz had a VID of something like 1.7v on the V/F curve. It was hilarious. I hope you don't get one of these chips.

 

/me cries in air-cooled stock i7 9700K

 

/same squared in an air cooled 9700F

xD

 

https://hwbot.org/submission/4820332_ | https://www.3dmark.com/3dm11/14609725

https://hwbot.org/submission/4820344_ | https://www.3dmark.com/3dm11/14609751

https://www.3dmark.com/hall-of-fame...core+performance+preset/version+1.0.132/1+gpu
3 of the top 5 are members of this forum and Team @Prema

https://www.3dmark.com/hall-of-fame-2/3dmark+11+3dmark+score+extreme+preset/version+1.0.132/1+gpu

 

It sucks that x170 can't push high memory speeds, but are timings locked to JEDEC specs as well?

The 4000Mhz sodimm sticks are completely overkill for these laptops, but allow for some nice low timings with bdie I'll probably part this kit out when I switch over to desktop.

 

Yep, my 11700k did 5ghz all core but required 1.45v to get there and that was that. 5.1 wasn't happening.

ABT is a variable tale of hot and hotter. Clicking on ABT on my previous 11900k's was watching 300w to 330w being pulled under CB23 multicore. Basically, on the MSI, it was default to 1.45v and let the furnace blow. I feel sorry for someone who bought one, flips on ABT and wonders why they're throttling like crazy if they happen to get a poor sample, poor cooling (or both).

5 11900k's under my belt. 4th died and the odds of getting better than this SP95 are slim to none so I'm content.

I would go all Peyton Manning on that SP50 and launch it like a rocket at the return counter while screaming, "OMAHA!"

But yeah @tps3443 you will find out REAL quick if you got a stinker for OC and/or for IMC quality.

I'm sure that 3080ti will help dry your tears.

 

Right now the X170KM, without @Prema , is stuck. I do wonder how they would do in a X170SM which has Prema. Those 3200 CL20+ sticks are pretty hard to look at.

 

CL20 and CL22 are unacceptable if you are running under like 4600-4800 MHz, LOL. See attached. Those are custom memory profiles I created for @Donald@zTecpc for the X170. Once I get the green light on it from Brother Donald, I will share the mods I did for the X170, including the thermal mods.