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

Sounds like somebody is selling a JPG of a 6900XT with a waterblock for $15.

 

I'm one of the rare people that like it with the air cooling shroud ON.

 

$15 as In $1,500.

 

If it is a brand new GPU and a brand new high end waterblock $1,500 is not a bad price. If the parts are used, that price seems a little steep. If the seller bought when prices were more idiotic than they are today, I feel bad for them. They will be taking a big loss because they were suckered into paying too much. (Current price new is still way too high IMHO.)
https://www.newegg.com/asrock-radeon-rx-6900-xt-rx6900xt-ocf-16g/p/N82E16814930057

 

I was testing power on my 11900K, and it pulls 248 watts during R23 @5.3Ghz cores /[email protected]. I thought this was pretty cool.

 

I was thinking more or so from the overall price point. So, a brand new one is $1,549 with $10 shipping and $120 in taxes. So, around $1,680 total If you buy a new one. And there is no water block.

So, $1500 flat from a OC member that may be negotiable, and it has a waterblock sounds a whole lot better to me.


But, a brand new one in a sealed box, is also great for $1,680 too. So that’s a tough choice for anyone I think.


I know these cards are not the fastest. But they hold up very well in 1080P and in some cases even 1440P. Definitely not 4K though. But they are still great GPU’s. And half price of a 3090KP, or potentially even cheaper than that.

 

I like it too. I like the Hybrid cooler on the 3090KP as well. Few times I have considered throwing it in there just because it looks pretty cool.

 

Agreed, I absolutely love the look of the KP 3090 Hybrid. I think it will end up being my favorite look for the 3000 series.

 

The biggest issue one might encounter is ASROCK and their Nazi warranty policy. Even the original purchaser has to jump through abnormal ludicrous procedural hoops with ASROCK in the event of a problem. If you are not the original purchaser, chances are good that they will use that as an excuse to leave you high and dry and shirk their responsibility as a company. I would think twice about buying anything from ASROCK brand new as the original purchaser because their policy is so bad. You have to go through the vendor that sold you the product. You can't go directly to ASROCK if there is an issue. There is a way to seek their approval for an exception to that process, but who wants to have to deal with such idiotic nonsense? Better to buy another brand and not have to put up with that kind of crap. If I have a problem, I expect swift and immediate action from the company that sold me a defective product. If I see any indication that I am going to get half-a$$ed service, I will just look for another brand. This is one of the reasons people still tolerate and buy pathetic turdbook rubbish from Dell. I think all companies should be required by law to provide the level of support and service that EVGA does simply because they want to be awesome.

Nothing yet. No update to tracking status and no delivery according the expectation set in the last email that it would be delivered today. Still "in transit" to the destination more than 24 hours later when the destination is 1 hour away.

 

A week since I shipped it and it's in the wrong town now. Wow. They really want to screw with us don't they? Lets hope they get it figured out today and it goes out for delivery tomorrow? I may need to stop into my brand tomorrow and see what's going on. The automated teleprompt gives me nothing.

So I know most would never do this, but I have been using one of those IC thermal graphite pads since my BSOD issue started occurring and I was trouble shooting. I have to say, I am moderately impressed with it. Obviously if you're going to max overclocks, this ain't it. But for daily use, gaming, etc, this thing is pretty great. Idle temps are comparative to my Gelid Extreme paste, gaming temps also very similar. Where things diverge is when you place the CPU under heavy OC, render type loads, etc and hit high power loads. The temps are around 10-15c warmer than what the paste can sustain.

The reason I put one of these on is I was messing around with my block trying to figure out if I had a CPU pressure mounting issue or something that was causing the BSOD and didn't want to screw around with repasting. Since everything is working 100% again, I haven't been bothered to even rip it apart to do a proper paste job. I'm honestly content with it's performance for now. That will change eventually, probably when I'm done with work stuff after the new year.

 

Hey, I have that 3M pressure sensitive micro film.

You lay it on your cpu, mount the block/cooler, then remove cooler/block and check the film. It turns pink upon tinniest pressure due to micro capsules.

You can even take a photo with your phone, and upload it for free. And it will be professionally examined and you will know how much force and pressure is being applied in each corner of each specific area.


This is how I discovered that (Direct die frames) create new temps/scenario each mount during direct die usage.


This paper is great, and they’ll mail it to you for free.

 

I finally did it!



I managed to push the power consumption on my 10900K all the way down to 1 watt when idle!. This thing can be undervolted extremely far if you limit the clock speed to 800 MHz. I also enabled voltage optimization in the BIOS.

Funnily enough, this is only achievable with the balanced power plan. For some reason, power saver makes the CPU draw MORE power (7 watts), like that makes any sense.

This is the best XTU profile for maximum battery life when I need it. Even at 800 MHz, this CPU can still do normal everyday tasks quickly.

 

Bro, I completely agree with you. Their warranty policy is very bad but they can't enforce it (all) here in the states. I can be very litigious when it comes to protecting my personal and/or company's rights. If my GPU were to break and it ISN'T due to my own stupidity or mistake, I would get my card repaired and I'm more than willing to go to court to make sure it does. Most of the time I don't modify my hardware (you know that, being that most of the hardware I've sold looks and performs like brand new), but when I do, it is within the framework of the marketing or promotion of that product. Asrock is knowingly selling an overclocked GPU and if I decide to use that GPU in an overclocked state, whether with the original cooler or with an enhanced cooler, they have to prove that I damaged it by using it beyond its overclocked state (good luck with that). End Rant...

 

It is so close, yet so far. I could have literally walked to Phoenix and back in the time it sat there

 

It’s probably your terminal/sorting center that just gets backed up. I had a 2080Ti stuck for 1-2 months. I was refunded, and then I purchased another 2080Ti. Then it too got stuck at a USPS center too. I had thousands of dollars tied up in video cards and didn’t have anything to show for it.

Then one day they both decided to show up on my door step.

USPS is great at times, and really poor at others. Chances of something actually getting lost are slim though. It’ll just get delayed for a bit.

And it’s all because companies literally lock down portions of shipping companies during the holidays just to get their orders delivered on time. Think about Apple during launch day. Apple literally has FedEx,UPS, and USPS in their back pocket for an entire day. Just to make sure everyone in the world gets their new iPhone delivered (On launch day) not a day early, and not a day late. How they do it is beyond me. But Nvidia can keep a few graphics cards in stock for $699. It’s a joke lol

 

I think I may try to ride it

I would to see nothing but copper radiators all over the card lol

Here's a lemon story for you guys

https://www.overclock.net/threads/x...orse-than-my-a-key-finger-56k-warning.783513/

 

Bought two more 12900k's. One from Newegg (arrived today) and one from Amazon (arrives tomorrow).

Opened the Newegg one a little bit ago.

Finally got a decent chip (Look at that V/F curve). On the off chance the Amazon is better I'll obviously keep it but the odds suddenly get much tighter with an SP90.

Tested 4000 right out of the gate and it passed fine so that baseline is covered. I'll see what else it can do later.

Collecting 49x -> 53x data as I go and as expected this one runs much cooler and pulls much less than the other two. I only put a thin line of test paste assuming this one was going to regress and be some garbage like an SP79 or something so after I test the last one tomorrow whichever is better (we know it's going to be this SP90) that's it for me binning for a little bit.

 

Nice!!! That looks like a solid chip. Hope your Amazon chip is even better.

 

Testing CPU’s is so fun!!

 

Amazon one will be here between 12-3 but I am content now with this SP90 on every level so if the Amazon is a dud I won't have a, "wtf...." moment. I do like testing CPUs though. I always have ever since the Celeron 300a (I'm dating myself with that one!). I'm sure over the course of the next year till Raptor I'll test a few more as they happen to come my way organically (usually with other people's upgrades and builds).

 

I spoke to USPS Customer Service and then the manager of the local USPS branch called and let me know it was routed to Glendale in error. They expect to have it from the Glendale branch sometime today. The manager for my local branch said more likely than not it will be delivered tomorrow, no later than Friday in the worst case scenario. She apologized for the delay and the misrouting to the wrong branch for final delivery.

Edit: Now it is back at the Phoenix Distro Center and will then be sent to my branch for final delivery, LOL. Sheesh. At least it is not lost, so that's good.

 

BartX is due in today for the X170SM-G. I missed it on Saturday and USPS doesn't auto redeliver now so I'll pick it up from the post office.

I've been working with the X170KM-G and already tamed the 11900k to run stock with no throttling but I'll be doing much more with it physically to dial it in even more as the default spread pattern is full but not optimal (way too many peaks and valleys and it requires a larger dollop to spread out versus what is needed). I can see the issues @Mr. Fox was talking about so I'll come at it a few ways before lopping off the support arm to cure the gangrene.

Another problem is its default voltage is INSANE.

It pulled 226w on default performance settings. You should have seen my eyes... Like "WTF?!" That is ~26w more than my MSI board, ~46w more than my Gigabyte and ~60w+ more than my Asus.

It needs Prema but more importantly the X170KM-G has a case of the "MSIs" and its ABT issues with the 11900k where no matter the quality of the silicon it would just arbitrarily send 1.45v to your chip no matter what for 5.1 all core. I tested 5 different 11900k's in the MSI Z590-A Pro and regardless of silicon quality "You get 1.45v! You get 1.45v! EVERY 11900k GETS 1.45v!!"

-110uv dropped that to ~183w and based on silicon quality I'll eventually drop that further.

 

Thankful it's not lost! I was getting a bit worried that I was going to have to do an insurance claim and then that chip would be lost forever.

 

AIBT is a little heavy on the voltage. It is pretty silly too. It is not too bad on my current 11900K at all, but still sends way too much voltage. Considering I can type 1.350V in to the bios and run lights out at 5.3Ghz all cores.


This is AIBT enabled (Auto voltage) It’s not a terrible profile. My cache is hitting x43, and bclk at 101.1, and my memory is gear (1) 4000 with heavy IO/SA voltages. So, the power usage is not anything special here.

AIBT varies from bios to bios as well. As does auto voltage. I like the benefits of auto voltage, but I don’t like how much power and heat it just wastes away. AIBT is generally unreliable with voltages depending on the bios.

 

Eh... it's a laptop. I accept it as a given fact that it is going to be a turd as released. None of them are worth a damn without a @Prema BIOS mod. I don't think I have ever owned a laptop that didn't suck the way it ships from the factory. Clevo is no exception, but at least you have a somewhat acceptable foundation to work with as a starting point. The rest of the options (excluding the X170) are a hopeless cause and total waste of time and money because there is nothing worth any amount of effort to salvage from the start. That may not be the case going forward for Clevo. I think the X170 might be the last of the Mohicans. I think I heard a rumor that no more LGA and MXM Clevos are planned in the future, so we will be down to a big fat zero on acceptable mobile platforms.

 

Yeah I get that. With proper bios. That laptop will run his chip well though. I’ve been on the Dark Z590 since 1.02 bios. And it did suck as well lol. It sucked bad. But it got a little better with the auto AIBT voltages. But, I gave up on it a long time. I was testing it last night, and was a little surprised to see it wasn’t murdering my CPU with the extreme voltages that @electrosoft speaks of. So, it’s really not too bad (It is a Quick stockish stable setting to run an 11900K at) it certainly helped the anemic 8/16 chip. As 4.8Ghz all cores stock is not enough. Most games just immediately run the 4.8 all cores.

 

Unfortunately, with laptops the issues are not limited to crappy firmware. They are riddled with engineering flaws, form factor and power delivery limitations and all sorts of defects that we don't have to deal with in the realm of desktops. I have been a laptop hater since around 2017 and my contempt for them grows hotter and runs deeper with every new generation of trash that surfaces. It is really disgusting how little there is left to be excited or happy about in laptops.

 

Well the 12900k from Amazon was a bust....

As soon as I opened the shipping box I could tell it had been opened before. The edge was uneven and the label crease was loose. I put my X-acto Knife down and with just the gentlest of pressure the label peeled away from the center where someone oh so nicely had finely cut the label horizontal to make it easy to fold one side back over the other and with just a quick once over it looked almost normal.

If I hadn't opened a few of them already I probably would not have noticed but the way the packaging normally is, it is VERY precise, tight and you will not be easily or gently opening it up with a simple light finger pry.

My biggest fear was there wasn't going to be a CPU in there at all but luckily the right CPU was in there that matched the container batch.

Holding it up under the light you could make out a nice, square remnant left behind from a heatsink of sorts.

Even as I was putting it in my Strix to test, I just went, "I know this CPU is going to suck.," and it did not disappoint.

It popped back an SP89 which seemed decent till I went to go run it and it clearly got that score on the strength of it's e-cores and shows sometimes SP is just an indicator and not an absolute as it ran more like the SP81 than anything. It's VID was the same as the SP82 and pulling 237w at stock hitting 80 (hottest of all the chips tested). Final insult? It wouldn't do G1 4000 like the SP81. I would take my SP82 over this chip any day of the week.

Final tally of the four tested:

 

The silicon variance is wild this gen. Thanks for the detailed results.

That voltage you quote, that that load voltage during the run or is it a BIOS voltage you are setting manually?

For example at stock settings, LLC3, and XMP 6000 CL36 set, I am seeing 1.172v load in CB23. But my power usage during that run is 220w at stock. I am stable with a -110mV offset voltage though.

 

Your right. It’s just crazy how different CPU’s are. I’m gonna take a gamble on a couple 12900K’s in a few months. I might try Walmart.com lol.

Also, as for temps I feel like this is largely in part of the factory soldering application, and how flat the IHS and die are, if the voltage is equal of course.

My dud 11900K was a hot dog at 5.2Ghz 85C on some cores. But, after the delid it fixed it completely. But no matter how cool it ran, it could not stabilize [email protected]. (Physically impossible)

That 10850K I traded @Mr. Fox was also a hotdog at 5.2Ghz. After the delid, day and night different. It’s unfortunate the CPU doesn’t pair well with a Z590 Dark it seems.

Anyways, that’s great you’ve got a good IMC and a 12900K that uses low voltage. No better feeling than when you realize you’ve got a really good chip in the socket.

 

All my testing is cmos clear, load optimized defaults, establish G1 max (goal = 4000) @ 1.3 SA and then run tests pure auto for fairness across all chips. That means LLC4 so my numbers could be lower across the board.

The Vcore after temp on the first table is under load at pure stock which is LLC4 (I think).

I did amass some data using my parameters to show you the variance at each bin:



That SP81 (first chip) was just such a dog but I can see someone buying that SP89 and going, "12th gen runs SO HOT!" I repasted it 3x as I couldn't believe how hot it was running then double checked all my fittings down to the posts. Finally, I swapped my SP90 back in and it just let me know, yep, that SP89 is a bleeder and carries that high SP rating due to its E-cores.

....I'm sure the first buyer came to the same conclusion as well.

 

It's hard to gauge two chips unless all testing parameters are the same to gauge the soldering job but yeah my first 11900k had my AIO winding up even on the desktop but when you correlate the temps with the Vcore and pull that adds to the picture. But on the other hand the massive drops you experience delidding your first one (maybe plowing the smd's off helped with that? ) was much greater than delidding your second one.

11th gen obviously pulled more as it was 14nm, but that massive die also provided a nice, large surface for better cooling.

I'm imagining the SP81 or SP89 in a board that pulls much more than is required anyhow at stock like the MSI Z690-A Pro and even at stock they would be in the 90s depending on conditions/cooler throttling even with some AIOs.

I was watching De8auer with a pre-built 12900k on his channel and it was throttling under runs. I usually find watching 12900k reviews or videos when they run CB23 the bulk of them throttle.

 

It’s nice when you have a great score. And, it’s all done on your “Easy Overclock”

This is 5,[email protected] set in the bios. Breaking 2,800 in R15.


This 5.4 is achieved so easily it is truly incredible. Especially on 11th Gen Intel that doesn’t really like high frequencies very much.

 

Fingers (and toes) crossed.

 

Showing delivered . Nothing like that speedy 9 day express service.

 

Based on the video @Talon made that CPU it a SUB 1.2v 5Ghz chip which is phenomenally phenomenal for a 10900K. My 10900K could not run 5Ghz at even 1.300

@Mr. Fox is probably skipping to the door in his jammies like a kid on Christmas haha. (ITS HERE) (it’s here) (it’s here!!!)


I mean, that’s what’d I’d do lol.


One thing about life is, you never grow up. The toys just get wayyyyy cooler. And a whole lot more expensive.

 

USPS: When you absolutely need something tomorrow in 10 days....

This is the absolute truth. I get just as excited about new high tech toys now as I did as a kid with my generation's microcomputers and consoles.

 

As soon as I can break loose from work I will go grab it. It's in a locked box, so it is safe now. Super happy it finally got delivered.

USPS = Unacceptably Slow Postal Service
But, in all seriousness, my experience with USPS has been at least 80% of the time there is no issue with speed or on-time delivery. About 15% of the times delivery is not on time, it has only been delayed by one or two days. More than that is rare. I can only identify a couple of times that I can remember in my lifetime that a package was never delivered.

I have had numerous First Class letters that were never delivered/received. Having the tracking number seems to create an added layer of security and accountability. Speaking of letters, Brother @tps3443 still no envelope with screws and torx tool so far. It may still show up. They might be overloaded/understaffed with the added load of Christmas volume and giving snail mail very low priority.

 

Hey @Mr. Fox I can certainly pay for the tool and screws, If you wanna take a look at the pricing and cost of what you need, shoot me a text. And I will send it over via PayPal.


At this rate, there is no telling when that stamped envelope will be delivered. So, please order on Amazon and I will pay for it friend.

 

Nah, it's not the end of the world. Don't worry about it, bro. I can buy screws cheap enough at Ace Hardware or online, and I have a torx bit that I can use instead of the tool that it came with. I might even have some screws that are the correct thread size in my spare parts. They probably have a Philips head, but that doesn't matter. There is a fairly decent chance they will show in a week or two if the envelope did not get jammed up in a mail sorting machine. I know a guy that works for USPS and the only reason I know better is he told me to never mail anything in an envelope that isn't flat because envelopes that have bulky items in them that will not lay down flat are quite often destroyed in transit by their sorting machines. The envelope can be thick, but can't have a bump in the middle of it (if that makes sense). Had he not told me that I wouldn't know not to do it. He actually caught me doing it one time and said, "Hey man, don't mail that if you want it to reach its destination" LOL.

 

The eagle has landed. All safe and as snug as a bug in a rug.

 

You know you have way too many computers and devices when moving your pfsense firewall from a VM to a physical box causes all kinds of mayhem. You CAN have way too many VLANs, don't let anyone tell you otherwise, hahaha.

 

@electrosoft I know of a retail open box SP97 11900K for $415 bucks OBO.


I’m tempted to buy it just because it’s an amazing price for a potentially solid CPU lol. And it’s fun testing CPU’s of course lol

 

Does the seller have some screen shots snd sample runs at least at 5.3 CB23? Wonder what the pull and temps are like.

 

Unfortunately no. He just has a screenshot of the Asus bios, and it says SP97 . He was asking $415 OBO. Looks like he has the memory at 3867 with 2x16GB no bclk overclocking at all. So, the IMC probably does at least 3867.

Not really a gamble. Even at $415 it’s totally worth it, because it’s a working 11900K that’ll probably be better than average at least. Which is about the going rate today for one at around $400 bucks.

I’ve noticed a lot of people buying 11th Gen since 12th Gen came out.

11900K wasn’t a good value until after 12th Gen haha. And the price went down about $150 bucks.

 

That looks like a Z490. Hope it's not jon's SP97 which bonks out at 5.2. I know he ran an SP97 on a Z490. One of the rare SP95+ that has a dud core at 5.3 all core. Rare, but it happens.

Still could be fun though.

I'm super content with my SP93 on the X170KM-G as cooling will just get better and better as I get in there and fine tune the physical cooling before dialing down the uv even more.

 

Before I begin tinkering with the 10900K from Brother @Talon, I figured I had better test with 10700K before I put it back into its box. I won't have any interest in it as soon as it comes out of the socket.

I seldom bench my work desktop. This is actually a superior sample. I already knew it was decent, but didn't really tinker with it. I just arbitrarily set it at 52x all core, 48x cache and 1.350V and used it for work. It can actually go lower than that and with 1.285V 50x, 46x cache the CB23 is higher than it was with the arbitrary values that I had previously set and forget at a higher clock speed. I was feeding it more voltage than it needed, but didn't really give a rat since I didn't care about benching it. At 50x core, 46x cache it pulls only 1.205V under load with a 0 AVX offset. It also has a stout memory controller. It runs the RAM at 4000 CL15-16-16-32 all day, every day... not bad for an i7 CPU that costs so little brand new.

https://hwbot.org/submission/4878404_

 

I am going to initially do my testing in the Strix board before putting the 10900K into the Z590 Dark. With its weaker cooling system and crummier power delivery system, it will be interesting to see what it can do with the cheap gamer-kid mobo before moving it to the benching rig.

 

OK, this is pretty sweet. I really owe one to Brother @Talon.
I scraped the paste off the 10700K and respread it on the water block, then slapped that onto the 10900K in the Strix mobo. Set the BIOS to auto core, cache and voltage, memory at 4000 CL15-16-16-32 20 tFAW, 296 tREFI. I did some down and dirty testing using ThrottleStop. I stopped testing at 52x all core, 48x cache, 1.2256V core and 1.2002V cache. Totally stable with crazy low voltage. Good enough. Time to put it into the good mobo. At least I know it will run like a bat out of hell even on the Strix mobo with that SP103 rating.

 

Maybe I will delid this weekend and go bare die so everything is proper. But, looking real good in the Dark. 5.3GHz all core, 4.8GHz cache, 1.325V. This 10900KF needed 1.450V for the same BIOS settings, so a 0.125V reduction. Huge. Considering it is not delidded and the chiller isn't running, a peak temperature of 73°C in Cinebench R23 is pretty awesome. I can't imagine how cool it will run bare die, with and without chilled water.



Even the EVGA OC Robot agreed 5.3GHz all core was a gamer-boy daily driver with more voltage than necessary.