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

It seemed like it had potential with single rank DDR4. The memory I was using was Patriot 4000Mh CL21. It was Samsung B-Die. But not good at all. I managed to get 4000 gear 1 CL17-17-17 to boot in windows. My other 11900K wouldn’t get in to
The bios with more than 3,900Mhz Gear 1 with single rank, no matter the timings.

After running dual rank though it seems to be a whole new ball game here.

Maybe it’s just me. I know dual rank likes 2 command rate. And a few other timings need to be set properly as well.

I’m gonna check out some profiles and learn a little more about dual rank memory.

It can do 3733 and 3800 even with just 1.250V VCCIO, and 1.1V Memory Aux.

This is 3733 CL13-13-13-30-1T


36.2NS latency. The Z590 Dark helps that latency some. Very impressive either way. Oh, and the cache is at 4700 with the 11900K at 5.4Ghz it’s still sub 320 watts in R15. This is very stable, however anymore than that 3800 and it’s gonna freeze while loading Windows 10. So yeah maybe the IMC just isn’t as good as I was initially thinking while using single rank with higher timings.


It could also be my overclock as well. I’m gonna start fresh on it tomorrow, maybe try the CPU stock and see how it goes. If it turns out to only do 3,800 I’m still happy with it!! I have yet to max out the cache on it, but it must getting close.

 

swapped to Win 7 and new gold in PCmark 04

https://hwbot.org/submission/4867004_papusan_pcmark04_core_i3_6100u_7959_marks?recalculate=true

 

Great video!

Yeah dual rank memory is amazing. This Corsair 2x16GB set seems to be pretty good so far. The Corsair 3600 CL14 sticks have plenty of room in them for overclocking. I always thought the Dominator Platinum line was a little overpriced, and overrated. But these are great. If only my IMC was a little better. I reset the bios, and started fresh. Looks like about 3,863Mhz CL14 on the IMC. (It seems to clap out around that point)

Gonna shoot for 3848 Gear 1 with CL13-14-30. That’d be great.

 

My 10900KF is SP 65.

The 10900KF requires 1.450V for Cinebench at 53x all core. I've dropped it to 1.400V on the 10850K and may be able to go lower yet. Just starting testing now.

Dual rank generally won't clock as high, but it will perform better and having 32GB will also help in many games and benchmarks. Even though you are not maxing out 16GB, something about Windows behaves differently with a higher percentage of memory resources are "available" to the OS.

You will generally need to feed more voltage to dual rank sticks as well. So, if you were trying the same voltage and settings as you ran on the single rank, it is probably not enough DIMM voltage.

 

I was more curious about your original 11900k to see while your newer 11900k could at least boot @ 4000 and your old 11900k could not I was wondering if they equaled out running dual rank.

I couldn't get 1T to work on the MSI or Gigabyte even stepping down to 3600 and looser timings with the 11900k and on the MSI I couldn't get 1T to run testing the other 4 11900k's either that was running 2x8 SR. That is a testament to the Dark Z590.

I'm curious to see how the dual rank stacks up at stock vs the single rank results you have at stock on the 11900k.



For any dual rank testing, I keep these 2x16GB mid grade 3600 SK blah sticks around with loose timings to push 3800+. I used them with the 5000 and 11th gen for testing. They are nowhere near as good as B-die obviously but with looser timings I can at least get an idea.


(I don't know if I would call them, "Fast like a Bolt") This is my 1 of 6 parts drawers in my main computer room.

 

Ack, that is a case of that being a not so great 10900kf and a 10850k punching above its weight and running like a decently binned 10900k.

 

Looks like 1.390V for 53x all core. Nice drop from 1.450V. If I go to 1.385V I get a BSOD. I rebooted and set it in the BIOS to confirm the actual voltage settings after tinkering with ELEET X1 to get a ballpark.

I can still bench the 10900KF at 5.5GHz so it is not "horrible" per se.

 

https://valid.x86.fr/y1j7jb

 

I’m glad to see that chip is running well. I had so much fun testing that CPU once I delidded it. It was better at overclocking than my 10900K was.

 

It’s a great CPU! Especially for the value alone. It was $379.99 brand new. And this was in January of 2021. I sent me brother a link through a YouTube video, and he bought it right away.

My brother always proclaimed 5Ghz at a really really low voltage with that chip. My brother Brandon is not much on stability testing. So I didn’t really believe him when he said he had a magical 10850K that was better than a 10900K. (Sounds pretty unbelievable I suppose lol)

But it really is a one off 10850K.

 

I noticed that I was having to juice it to 1.600V on my new DR modules. My SR modules never needed much more than 1.500V. However, these DR sticks handle low timings so much easier. I couldn’t stabilize 3733 13-13-13-33 1T. These sticks are pushing upwards up 3844 at 13-13-13-29-1T.

I can push them in to 35NS which is phenomenal!! This is great Samsung B-Die.

 

@electrosoft


I went back at memory overclocking again. And turns out, I just had to juice up my Corsair Dominator Platinums even further than I thought (DR sticks really like voltage) I added some more dimm voltage, and +35 Dimm VTT.

I’ve got 4,000 Gear (1) CL14

The IMC is looking good so Far!! That’s why it was freezing loading windows, but still posting fine. The memory just wants 1.630V.


Looks like a golden cpu all around!!

 

@Papusan check out my above edited post. This chip is absolutely magical!!!

4000 CL14 gear 1!

 

If I would have known this 11900K was some kind of miracle chip, I would have bought some faster dual rank DDR4 sticks lol. Who woulda thought?


4,000Mhz Gear 1. No bclk overclocking at all.

I am sending 1.650V to my dual rank sticks. They seem to like that lol.

 

The 11900K running 5.4Ghz with Gear (1) 4000Mhz CL14 1T Dual rank. After sending 1.65V to my memory, it fixed any freezing issues.

The IMC has more room. But I feel like I’m running short on my memory capability.

I/O S/A voltages are also reasonably low. I’m sending 1.400V to each.

EDIT: I managed to get the IO/SA voltages down to 1.375V range for VCCIO and MEM AUX. This seems unbelievable compared to my other 11900K which was pegging 1.550+ on both and struggling with just 3,882Mhz.. The IMC really was good after all, and all this freezing while loading windows 10, was just the dual rank memory that was having a hard time with only 1.450-1.500V not being enough.. My ram is only 3600Mhz CL14 after all, so it just needed some extra juice. And 4000 CL14 in (Gear 1) worked just like that.

 

Very nice bro congratz! Have you run these settings vs CB23 loop and RB 2.56 stress?

 

This was a pretty shocking discovery. Gonna run some more testing after work, and try see where the IMC stops. I feel like I’m hitting a memory limit on my sticks though.

 

I can't wait for the delid. If you experience even a fraction of what you extracted from your previous 11900k with its delid, the results will be bonkers and 5.5 all core most likely incoming.

 

Probably so. As a general rule, 4000-4200 is an upper limit on dual rank modules. It kind of makes sense though. You can apply similar logic somewhat loosely to CPU with very high core counts. They generally do not overclock as high, but perform better because there are more core/threads being processed. Even more loosely than that, a 4 cylinder engine is generally capable of revving up to a much higher RPM, but can't compete with the performance of a V8, unless it is a really pathetic/anemic V8.

Brother @tps3443 really scored a winner based on my demoralized state of mind with crap technology. Had I known I had a silicon lottery winner on my hands, I would have either kept it for myself or charged him double.
(I am just kidding. I am happy it turned out to be a good sample.)

 

Yeah who woulda thought. What a ridiculous sample of a 11900K. This thing is just unbelievable. I’m gonna end up having to buy that super high end dual rank memory.

What is the highest a 11900K has gone in gear 1?

 

Dual rank modules need watercooling. You ever tried watercooling memory? It seems like it would be super helpful getting long term stability with high dimm voltage. Once memory hits like 45C-50C it starts spitting out errors like crazy. I have not tested this memory yet. But, I’m hoping it hold out with fans on it.


Yeah that definitely makes sense on the whole power thing. I like power to weight ratio. Super light weight, with high rpm and high HP. Kinda like my Suzuki GSX-R 1000, it broke 172WHP. I am guessing at least 205HP at the motor. Not bad for a little 1 liter 4 cylinder. But the power to weight is everything. You won’t come across a car on the road nearly as fast. Every now, and again I will see a really expensive exotic car traveling down 40HWY. Hehe.


Anyways, I’d like to think the 11900K is a high revving small block V8 (Or 4 cylinder lol). Which is why my previous V10850K could out pull it.

 

The 11900K is a V8 C8 and the 10850K a V10 C10.

 

I really like the 1UZ in the AE86

 

One really cool motor is the 2JZGTE by Toyota. 3.5 liters, Iron block, inline 6 cylinder, with Twin Turbo’s, these factory Toyota blocks could hold 800WHP just fine.

It is a legendary engine. Probably the most reliable “performance” engine ever built.

 

Too rich for my blood lol Slowly making my way towards the K24A/K24A2 for my RSX, I dont need any more than 300 tops in fact probably going to keep it in the 250-275hp area. Its my DD though so I would like to retain some gas mileage.

 

My brother just did his RSX Type S. The K20 Type S spun a bearing. And he put in a motor from a JDM Honda Odyssey Absolute Thats a JDM K24 full vtec Type S motor. Few tasteful Bolt on’s and he has Ktuner and a full dyno done on the car. It put down right at 238ish WHP.


^ It is absolutely ridiculous. The torque is huge on it, it’s making like 180’s WTQ. Car is super super strong. Jmills in NC tuned it.

He had to do a few things like Honda CRV passenger motor mount, throttle body adapter plate, and a few other things to make the JDM K24 work in a RSX/RSX Type S

 

How do you guys feel about 1T VS. 2T, and dual rank memory?

1T seems to cause BSOD during heavy heavy AVX loads like RB256 or R23. If I jack up the voltage to 1.655V to the memory. (It fixes it) and it is stable/good to go. Or if I just run 2T I can get by with just 1.605V to my memory.

1T helps memory bandwidth a lot as well.

I’m thinking I really do need better Samsung B-Die.

 

Yep not that difficult for a K24A/A2 to install in a RSX. Im going to see about keeping it Auto though, sticks are fun but I am often in stop and go traffic and I like the Manual mode quite a bit for how it bridges that gap. Should be fine up to 300hp but I will likely have it bulletproofed regardless. I dont even mind the K20A3 that much but no top end power makes things difficult at times when other drivers try to cut you off or truckers that dont see you, not too keen on putting in a turbo but I supposed I could since these engines make lower compression. JDM K24 is just K24A while the USD variant is the K24A2 iirc

Its a long ways off yet, for now fixing up some issues to get it back to baseline. Tomorrow ill be replacing the Intake Manifold Runner Control and cleaning up the throttle body / IACV and just generally looking over what I can. I like the compact nature of this car quite a bit, and parts are relatively cheap.

 

If you can keep your memory relatively cool and you get the gains I'd stick with 1T. The fact you got a Z590 Dark to run 1T with 11th gen at those frequencies and timing is pretty righteous. 1.655 vs 1.605? Meh....

 

Hey, you should look in to the VTEC Killer mod for K20A3. I had one of the funky looking 2005 Civic SI’s it only had half Vtec type deal. And this was years ago, but I performed the Vtec killer cam mod. The car was tuned out making over 200WHP. Considering this motor only makes 160HP and about 145 to the wheels stock. This is a great modification!! They sound really loapy during idle like blop blop blop blop blop. Lol.

This mod absolutely dirt cheap. Just modification. You can get 200HP without a motor swap.

 

I am going to go ahead with DELID tonight. Gonna do it just like my first one. Only I have learned a little more this time around.


Gonna run a few cycles in the oven at 300F, and slowly move the IHS back and forward about 2MM back and forward. Then I will bake and pry off.

Everyone wish me luck!!

 

This was totally different, but I get it... Would be hard to be a YouTube tech content creator right now.

 

Don't pry it off. If the solder is melted the IHS should easily lift off with no effort.

I hope it goes well. Besides being 8 cores, one of the reasons I decided I did not want an 11900K is because of the failed delid attempt. I was afraid to ruin another one and don't want one that is not delidded.

 

His last one went well except for the extra "electric slide" down over the SMDs.

 

Yeah I saw that on one of the RSX forums, for now I need more stability so just getting her back to baseline. The next thing is probably HID conversion or at the very least swapping the headlight assembly as mine are a bit damaged. Gaining HP is akin to testosterone for most blokes including myself however, what I desire first is more granular control and some weight reduction. As someone who has been hit by 2 cars if I were to do the same I know that I will not recover from it. Right now because the IMRC isnt working correctly the idle point is out of wack, sticking to 1600rpm at a stop and increasing up to 3k on its own. Due to this stopping is quite difficult and I drive through 4 school zones.

 

300F is still too low. The solder won't melt at that temp. You need higher temps more up to...

Exactly. Only lift it off.

 

Baking now. Temp is between 165-190C. Going for 20 minutes the first run.

 

Sounds like a great chip, don't turn it into toast!

 

I’m trying my best to not ruin it. Very tedious stuff. Especially with such a good chip.

 

If you have one, you can use one of those extra-thin floppy double-edged razer blades to slip between the IHS and PCB to cut the rubber seal. Kind of like we did in the early days before delidding tools existed. If you cut through that it should not require any effort if the solder is melted fully.

 

Still working on it. The IHS is being stubborn just like my first one. I’m at 325F which with my temp gun is between 170-190C.

This is my 3rd delid on a soldered CPU. And I still have yet to see an IHS that just lifts away lol. My goodness I wish it would do that.


I popped the 10850K cold with delid tool. Moved back and forward until it fell off.

 

Popped it off. Ok, no broken caps! So far so good. Time to scrape solder off the die.

 

I have done probably 15 or 20 delids. All have been cold and most of them have been on soldered CPUs. The one and only attempt with using heat to melt the solder is the one that destroyed the 11900K. If the solder is actually melted and in a liquid state, it should not require force. If it is only softened, it might require some. A cold delid would work fine for the 11900K except for the SMDs they placed in harm's way.

Anyhoo, it looks like it turned out OK. Congratulations. How many times did you push it back and forth with the delid tool in the process of heating it up?

 

That is awesome bro, nice!

can’t wait to see the new results!

 

So here is what’s happening with the melting process. I will set 300F in the oven, and my CPU reads 170’s Celsius. If I crank up to 325F. It reads 190’s-200C. And it scares me.

Maybe my temp gun is wrong.. I’m not sure, but I’m scared to death to push temp higher. So, I get the solder right before melting where it’s softer I guess, and just push back and forward in the tool long ways on the die. You cannot put the arrow in the normal spot. Then the die would be sideways, and much weaker I think.

It eventually breaks the solder. And I
Can slowly pry off. (Very lightly) if it doesn’t feel right It goes back in the oven, or tool.


Also, I heard a faint crunch sound. My goodness. I was panicking. I was shaking the cpu to see if I heard rattling in there.

It looks ok though!!

Gonna apply LM, and fire it up. Enough shaking and nerve racking fear for one night.

This is so terrifying delidding these 11900K’s!


Oh and, I decided to leave the (16) caps on there this time.

 

Either the one I de-die'd had a defect where the die was not properly attached, or pushing it sideways instead of only back and forth against the long direction is what separated the die from the PCB. We will never know for sure, but probably pushing it laterally against the narrow width because it would not have as much strength. That is when I heard the crunch sound that made my heart sink.

 

Great new everyone! The CPU works just fine.


No IHS, No die frame, and no CPU locking mechanism as usual.

 

Yes, that is wonderful news.