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

I finally reached the point with CPUs and GPUs that I am going to risk losing the warranty to own the product that I want. So far, that has never let me down. Sooner or later I will most likely have to suck it up and pay for it when a part dies that I have altered, even though the alteration may have absolutely nothing to do with the failure. Just because most OEMs and ODMs look for excuses to shirk their responsibilities.

Asrock stands out as a deplorable example of the worst the industry has to offer with respect to customer service, warranty, and doing the right thing. Their policies are unacceptable on face value.

But, AMD's policy is every bit as despicable and degenerate as Asrock when you start reading the fine print. It's written in such a way that you're just gonna be screwed if they get a wild hair up their butt and decide to shirk their responsibilities. I am going to delid the 5950X, and that will certainly void the warranty. But, so does any form of overclocking according to AMD. Even when you do it using their software or firmware. It makes me angry every time I read their disclaimers. They are not an honorable company, and they put the evidence of that in writing.

Contrast that with Intel... until recently offered a warranty with free replacement (because they know overclocking almost never kills their CPUs... or used to almost never) and they have replaced two delidded CPUs for me under warranty. That includes the 10900KF that the ASUS mobo killed when the mobo failed. They had no stock for replacement, so they rimbursed me based on my invoice from NewEgg. The first was an 8700K that suddenly stopped working without explanation. Turned it off and went to bed and the next morning I tried to turn it on and it was dead. (That 8700K was also installed in an ASUS mobo at the time. Gee... wonder if that is a trend.) They didn't deflect, blame me, ASUS or the delid. In neither case did they care about the delid, and in both cases they required that I submit photographs. They apologized for it and took care of it, exactly as they should. Much like EVGA only very slow warranty service, whereas EVGA is very fast.

 

You're the third owner of the OC Formula I've asked and all three said no coil whine. Maybe (hopefully) a byproduct of a superior PCB design. I'd hate to be the outlier from a small sample and get one and it sounds like the 6900xt's on youtube that are nails on a chalkboard.

I still think (hope) they will come down a touch more. Newegg has had them on eBay all week at $1699.99 and haven't sold not even one. You are able to order them in quantities of 10 from Newegg directly versus 5 or less for others. Here's to fingers crossed. I'd pick one up at $1400. I wonder if it is the dismal warranty?

 

So before cracking open (aka breaking the seal) the Gigabyte 3070ti and reading that Diablo II resurrected @ 4k actually can be taxing and performance bad, I decided to burgle the wife's Aorus 3070 and install it into my system to try it out at 4k.

Wow, I was NOT expecting that drop in performance. I mean serious drop. I've been playing it everyday since launch and everything has just been silky smooth so far. The KPE3090 (yet again) has really spoiled me
quietly purring along and handling 4k. I quickly forget how good it is till I try a lesser card trying to be frugal.

4k max details and there is definite frame drops into the 50's and lower and it feels even chunkier in some areas with a lot of action going on with the 3070.

(Quietly puts KPE3090 box back on display shelf)

I did reroute my AIO tubing to the left instead of the right for cleaner aesthetics when putting it back in though.

 

Excellent evolution of AMD CPU gaming performance and an ode to how Intel absolutely trounced them badly till AMD started making rapid gains and eventually at least overall drew even and then wins some and loses some. The massive ascent of AMD with their CPUs and GPUs is pretty amazing:

 

Well, shoot. The delid is going to have to wait. That crappy little Athlon that I purchased is on its way back to NewEgg for a refund. Even though it's on the list of supported and tested CPUs published by ASUS, it would fail to POST with code Q-code 4d (CPU) and a red CPU fault LED. So, it is either a DOA sample or not compatible with X570. I don't want to be without the ability to use the system, so I will have to figure out how I am going to manage that. It makes no sense for me to purchase a more expensive CPU to use it only a week or so.

While I had it apart to install the Athlon I put a better thermal pad on the PCH and now this ROG piece of junk idles with PCH temps of 86°C. So, I guess I better take it apart again. The times like this make me extra-glad I am running an open bench rather than an enclosed case. That will be three times today. Once to install the CPU that was no good. Then to put back in the 5950X and now a third time to try to fix what should have made the lousy PCH temps better, not worse.

 

I had to open a can of whoop-ass on the chintzy PCH cooler that the retards at ASUS put on this thing. Temps are back down to 70°C at idle. I have a Gelid pad on the core, with KPx on both sides of it and thick thermal pads on each side to wick heat off the heat plate to the PCB. If the temps go back up again, I am going to pack that entire area under that mickey mouse junk heat sink with a thick layer of K5 Pro and be done with it. That fan is about as worthless as tits on a boar with the GPU plugging up the intake.

 

Speaking of hot PCH's, I've seen quite a few performance/gaming laptops that don't have any cooling for them, just the exposed die. I've never bothered to really look into it, but is it bad that said laptops don't have any cooling for the PCH, or is cooling not a big deal.

 

If they get hot is it bad. If they do not it does not matter. When they cram it between the mobo and chassis on the keyboard side and it gets no air circulation they tend to get hotter than the area on the bottom that has the fans and at least some air circulation.

Ryzen's PCH seems to get a LOT hotter than I have seen an Intel PCH get in a long time. I'm not sure why it would be different, but it certainly seems to be. I have seen this one approach 90°C even with the heat sink and fan, so I can't imagine how hot it would get with nothing on it. I'm guessing thermal shutdown level of hotness.

 

Isn't the PCH chips for Intel PCIe 3 and Amd PCIe4 ? The PCIe-4 for Intel is directly from the Cpu. Aka the PCH chips in AMD run hotter.

Edit:
Intel Z590 Chipset, PCIe 4.0 kinda! - Laurent's Choice

 

That could be why. It could also just be another "special" thing about AMD that isn't that special, LOL.

 

HaHa.

I think there is an variation in the bin. I have seen some Jokebooks with overheating PCH chips while other run slightly colder. And from what I mean have seen. More models come with PCH heat-shield nowadays. Many newer Aw Jokebooks offered throttling as a Dell introduced feature due overheating PCH chips

https://hwbot.org/submission/4828867_papusan_3dmark___ice_storm_geforce_gt_705_39294_marks


https://hwbot.org/submission/4828864_papusan_3dmark___ice_storm_extreme_geforce_gt_705_25227_marks


https://hwbot.org/submission/4828866_papusan_3dmark___ice_storm_unlimited_geforce_gt_705_45469_marks

 

That seems to have fixed it. No issues with PCH temps now, even after 10 minutes of AIDA64 stress test. Before it would get to 85°C in just a couple of minutes.

 

Yeah the PCH on my 11800H GE76 sits above the GPU heatsink and below the battery and it's bare die (I put a little aluminum fin heatsink on it but no idea if it made any difference) , but there are direct vent holes in the bottom cover and I have it sitting on a beast of a laptop cooler so I've never once seen PCH temps go anywhere near throttling. But even prior to purchasing the laptop cooler or mini heatsink, the PCH was always good no matter what torture tests I put the CPU/GPU through.



Sent from my SM-G970U using Tapatalk

 

My new thread about tomorrow's tech for notebooks is up. And it looks damn black/dark for tomorrow's (gaming) laptops.

Intel will follow Nvidia regarding castration of the amount cores for their coming Cpus in notebooks...
http://forum.notebookreview.com/thr...es-for-their-coming-cpus-in-notebooks.836895/

I have also added a poll. Be free to vote. Or post your thoughts if you want

 

I found a BIOS mod for my mobo that unlocks the PCH fan controls in the BIOS that the idiots at ASUS decided to hide. Now I can adjust it based on temperature, or just leave the fan maxed out all the time. [MOD BIOS ASUS X570] ROG Crosshair VIII Hero (WI-FI) & Formula - PCH fan control

 

How odd, even my ITX x570 board has that feature and I believe Im still on release version.

 

Is it an ASUS motherboard? ASUS have really become a bunch of retards over the past few years. All of the boards (Intel) I have owned before now had that BIOS option. Not sure why they dropped it on X570 Crosshair boards. Very stupid.

 

Its the Aorus/Gigabyte ITX x570 board, link below:

https://www.amazon.com/Gigabyte-X570-AORUS-DisplayPort-Motherboard/dp/B07T9PC9ZZ

 

Problem solved. I got this cheap enough. I'll sell it on Mercari (or here in the Marketplace) when the delid is done. This also won't be as painful to tolerate as the Athlon.

I didn't like my Z490 Aorus Master too much, but all things considered, I am of the opinion that Gigabyte's Aorus product line are better than ASUS ROG products. I knew better than to spend my money again on ASUS trash and I did it anyway. I can only fault myself for being stupid. I paid $379.99 for the Crosshair VIII Hero. For that kind of money, the flimsy PCB and abundance of cheap plastic trim that is deceptively similar to cast aluminum in appearance is deplorable. Overall quality is poor.

 

Yeah one of my first boards from Asus was the Crosshair III that worked fine for basic configurations but Crossfire and CPU overclocking didnt do well. I later had the Asus III Extreme in which one of the memory channels didnt work so I had to resort back to Dual channel memory.

They seem to be one of those companies where you find many if not all the features you want, but never seem to measure up in which you find yourself making compromises. I do still credit them with showing that there is an expensive market for ITX though. At the time it was still quite common for enthusiasts to want a sound card and back at that time no one wanted to compromise between sound and/or Graphics.

Gigabyte had faulty Vega 56/64 cards so I have been skeptical of them as well lately but the board seems to work well and I have 2x 5700 XT cards from them that do well. Its another one of those scenarios where its easier to go with brands you have good experiences with, but like anything its best to look at products individually. I think recently they had some bad PR over the PSU's? Not sure why they felt the need to double down, that could've easily been a PR win if they had chose to simply eat the L and climb onto the moral high ground....but stupid is what stupid does lol

 

I don't hate Gigabyte but I did have two Z390 boards arrive DOA: an Aorus Pro and an Aorus Pro Wifi. The current Z590 Aorus Pro AX WiFi I'm using has the Realtek sound bug that every other Z590 maker has seemed to fix via BIOS update except Gigabyte.

Aorus 3070 Master is working rock solid back in the wife's system. I still have a brand new Gigabyte 3070ti Eagle sitting here that will probably end up on eBay unless I hold onto it for a future build.

 

Havent had that experience yet as I have been out of the desktop game until only recently. I would be really pissed if the usb dropout issue was experienced on my motherboard as well as I use one of the ports for my USB DAC that I have had from when I was using laptops as my DD. Even with that being said, its now one of my miners so I am using the P750FM primarily.

 

Is that AMD 2600X gonna be something to practice the delid with?

That’s a pretty good idea!

 

DDR4 5,000 CL17-17-17-36-285-1T (Gear 2)

@1.599V dimm voltage.

This is a really good set of memory..

 

Gonna break that 80GBPS bandwidth, and sub 40NS memory latency on my Z590 Dark while on Gear 2 mode.

Really loving these memory modules.


This platform stabilizes memory so very easily! (Once you have the supporting voltages in place of course)

 

Which ram modules you using?

 

Using my G.Skill Royal-Z gold 4x8GB 4000Mhz CL15 1.5V modules.


I’m only running half the kit, so they’re certainly good with only 2x8GB.



https://www.newegg.com/g-skill-32gb-288-pin-ddr4-sdram/p/N82E16820232982?Item=9SIA4YUD3K7430

 

Cool, thank you. Was going to get some gold tridents but I think a little to blingy (maybe still deciding) and not so sure how well it’ll be with the copper look wise. Need 16gbx2 minimum for this build but some of the 16x2 is as much or more than 32gbx2… lol I have never been so indecisive in my life on parts for this one. Ultimate Staley want some good looking ram with this board that runs stable a reliable.

 

I didn’t really buy them for looks. I owned an Evga X299 Dark at the time. And I bought these in June 2020. It was really the best ram for quad channel. Because they were CL15 at 4,000Mhz, and quad channel.

It just so happens that they are incredible in 2x8GB.

Testing 5,200Mhz right now.

 

More Gold for @Prema Team

https://hwbot.org/submission/4830116_papusan_pifast_core_i7_4710hq_18sec_430ms


https://hwbot.org/submission/4830111_papusan_pifast_with_benchmate_core_i7_4710hq_18sec_430ms


https://hwbot.org/submission/483008..._with_benchmate_core_i7_4710hq_9min_29sec_8ms


https://hwbot.org/submission/4830103_papusan_wprime___32m_with_benchmate_core_i7_4710hq_6sec_726ms

 

I have discovered something interesting with my 11900K.

The cache frequency greatly increases the auto CPU voltage demand. If I leave the cache default, my CPU in R23 at 5.2Ghz only pulls 1.409V (with No AVX offsets)

I may just get 5.3Ghz 100% stable. My 11900K calls for 1.476V-1.481V at 5.3Ghz all core with cache on auto.


I’m thinking this chip might be worthwhile to delid.

 

I think I mentioned that before that pushing cache can get the 11th gen hot and hungry. Over on the OC forums, you see most of them just settle for 45/51 so they can just have a love fest with memory. Some will drop it even further for stability. You'll need to find that balance between cache gains vs OC gains as always.

Your 11900k is a decent sample. It could be MUCH worse....trust me. I usually run mine at 43x cache. The worst was still that 5.2 all core requiring 1.518 and pulling 368w for CB23.

Have you tried to manually dial in the voltage to see how much it requires or did EVGA get it right and manual is close to board auto?


Gigabyte dropped a new BIOS update (F7) finally. Still doesn't fix the audio bug, but IMC voltage requirements dropped a bit.
I was testing the IMC on my 11th and it will do 3866 @ 1.35, 3733 @ 1.3. This is down from 1.375 for both on F6. I might go back in and see if 3900 is able to run now.

And yeah, a delid introduces much more headroom with the 11th gen being so temp sensitive.

 

Nice early Windows 11 vs 10 gaming fps shootout:

 

Thanks for the vote of confidence....
https://www.3dmark.com/3dm11/14632404
https://www.3dmark.com/3dm11/14582933
https://www.3dmark.com/3dm11/14582923
4748mhz
https://www.3dmark.com/3dm11/14632401

 

I can manage 5.2Ghz all core at barely SUB 300 watts in R23. (No AVX offset) and it’ll score right at (16,880-16,900) range.

5.2Ghz all core very very low 1.400 voltage range. (This is adaptive voltage)

5.3Ghz all core is hit or miss in R23. The heat gets to the 11900K sometimes in about 3 out of 5 runs in R23.

The IMC is fantastic, and the silicon seems to be good enough.

I can run my cache at 4.7Ghz stable. (While still maintaining 5.2Ghz all core) this puts a slightly higher voltage demand, but it still hangs around the 1.432-1.439v range. And power jumps up a little

 

https://hwbot.org/submission/4830142_papusan_hwbot_x265_benchmark___1080p_core_i7_4710hq_22.989_fps


https://hwbot.org/submission/4830149_papusan_wprime___1024m_core_i7_4710hq_3min_36sec_251ms


https://hwbot.org/submission/483013...ith_benchmate_core_i7_4710hq_9min_27sec_163ms


https://hwbot.org/submission/4830146_papusan_wprime___32m_core_i7_4710hq_6sec_700ms


https://hwbot.org/submission/4830143_papusan_wprime___32m_with_benchmate_core_i7_4710hq_6sec_700ms


https://hwbot.org/submission/4830125_papusan_cinebench___r11.5_with_benchmate_core_i7_4710hq_7.33_cb


https://hwbot.org/submission/4830129_papusan_cinebench___r15_core_i7_4710hq_677_cb


-----------------------------------------------------------------------

Intel CEO says AMD’s time is ‘over’ pcworld.com

Shots fired: Intel believes its Alder Lake and Sapphire Rapids CPUs will put an end to AMD's time in the spotlight.

Hope he know what he talk about or it will fire back

----------------------------------------------------------------------------

LOOL have fun

Not Even Dell Uses This Cooler Like This...

 

[

I can post and stabilize 3,867Mhz quite easily. However, that is near the wall! I just confirmed again on the latest Evga bios, and my absolute limit for gear 1 is 3,892Mhz.

Also, something about bclk overclocking changes the voltage curves within these CPU’s I can’t really explain it. If you set an all
Core of 5.1Ghz and set bclk to 104Mhz then the cpu is much easier to maintain stability VS setting a straight 5.3Ghz with 100bclk.


It’s something I can’t really explain, and totally changes the CPU. It’s weird! Give it a try.


I wouldn’t mind a Gear 1 11900K lol. Apparently only 1 in 4 can do 4,000Mhz Gear 1.

Considering how beastly just 3600Mhz Gear 1 is. I imagine it’s great.

I’m happy with 3,890 though. It could be worse.

 

New Golds. Near 5% faster than 2nd place in Wprime 1024M

https://hwbot.org/submission/483068...e_i7_4710hq_3min_25sec_637ms?recalculate=true


https://hwbot.org/submission/483068...ith_benchmate_core_i7_4710hq_3min_25sec_637ms


https://hwbot.org/submission/4830692_papusan_cinebench___r15_core_i7_4710hq_711_cb?recalculate=true


https://hwbot.org/submission/4830691_papusan_cinebench___r15_with_benchmate_core_i7_4710hq_711_cb


https://hwbot.org/submission/483069...r11.5_core_i7_4710hq_7.63_cb?recalculate=true


https://hwbot.org/submission/4830695_papusan_cinebench___r11.5_with_benchmate_core_i7_4710hq_7.63_cb


https://hwbot.org/submission/483072..._2003_core_i7_4710hq_3225_cb?recalculate=true


https://hwbot.org/submission/483071...0p_core_i7_4710hq_24.496_fps?recalculate=true


https://hwbot.org/submission/4830727_papusan_7_zip_core_i7_4710hq_26362_mips

 

Sounds like a good IMC, good cache but can't quite do 5.3 without turning into a furnace / slight guzzler or pass Realbench 15 min stress and CB23 30 min?

Time to update your sig!

 

Hey brother John great to see you again. I hope you've been doing well. I have really missed you, friend.

I don't know how that's even possible. It certainly doesn't seem mathematically possible at 47x, so I'm going to have to find out what you've tweaked to make that happen. But I'm really glad that I was wrong because it means that it's not just a random glitch and there's something special that can be changed to get higher physics scores like that. Awesome results. And, whatever that something is may also improve benchmark scores in other things. I would love to have at least a few good reasons to be glad I spent money on a 5950x LOL.

@Prema - happy birthday, bro.

 

I hope so too. It's a real shame that they're having to work hard to retake the lead when the product being sold by their competition is so crappy. It really shows how lazy and complacent they got when there was no competition. They got caught with their pants down and now it's time to make things right again.

This screenshot caught my eye and I paused to see what kind of crazy CPU was getting such a high single thread benchmark score. Then I saw it was all core, LOL. But, still great results considering the hardware.

 

That sounds like my CPU for sure.

I can sip though 5.2Ghz at 300 watts in R23 forever. But it does kinda run abnormally hot for such low power consumption. (Silicon variation I suppose)

Not sure what I want to do with it.

Exchanging would literally transform it lol.

And delidding would probably transform
It too.

 

You run the chance of getting a subpar IMC as having them crap out at 3733 is common (even though 3733 is nice IMHO). There are plenty that won't even do 5.2 all core even though they run cool.
I think I mentioned it before, but your chip runs a lot like the best buy 11900k I had that died when I was stuck on trying to get it to 3933-3966. It did ~3900 IMC, did 5.2 all core and could do 5.3 benches but was a bit thirsty and warm.
I know a delid would have helped a lot.

With your cooling setup, delidding would make a potentially significant difference.

 

I agree. Almost looks like magic. I would like to know as well.

 

Brother @Johnksss is going to have to be kind enough to share with us what the special ingredient is. It might be higher memory clock speed and higher IF clock, but I am not going to try to guess because it might be something more obscure. I've closed the gap enough to rank right next to him on the leaderboard, but there is still a fairly big gap in the score that is not accounted for in running the same core clock speed using the same CPU on the same OS. That is why I (wrongly) assumed it was a glitch. It might even be the fact that the Crosshair VIII mobo and firmware is junk (which I already believe to be true) and the Godlike board and firmware just works better.

 

I already killed a bios on this Z590 Dark lol. Oops. (This 11900K had the exact same behavior as my 10900K did, right before it went full psycho, and committed cpuAcide)

When the 11900K did this, I was pushing memory hard trying to get beyond 5,200Mhz Gear 2 CL19, heavy VCCIO, VCCIO AUX voltages, and hard on the VSA too.

^ It started acting just like my 10900K did. The 11900K was acting funny in the bios. So I backed off real fast.

Reset the bios, and bios 1 is just done for. (Reflash will most likely fix it) fortunately I have (3) bios. And I was able to catch those early warning symptoms.

 

In case you are not aware, as long as the chip itself is unharmed, the corrupted data on the chip can be corrected. This happened to me maybe a dozen times on the X299 Dark and two or three times on the Z490 Dark. What you were doing with the extreme memory overclocking can store messed up values in the firmware if there is any kind of glitch rendering it unbootable. It only happens if the messed up settings are commited to NVRAM. The Alienware Viking and Ranger laptops used to get soft-bricked all the time due to corrupted settings stored to NVRAM.

If a BIOS Recovery flash does not work, do this:
- Boot from a bootable BIOS chip and enter the UEFI environment
- Move the BIOS selection switch to the position of the corrupted BIOS chip
- Flash the corrupted BIOS chip after moving the switch to that position
- When the flash operation is finished, completely power off the machine
- Turn off the PSU switch for a few seconds, then power on the machine

It should function normally now, if the only problem was corrupted firmware stored on the IC.

 

Yep I’m sure it’s fixable, I just moved on to another bios position. Too lazy to flash it back.

 

https://hwbot.org/submission/483114...gle_core_with_benchmate_core_i7_4710hq_902_cb


https://hwbot.org/submission/483114...ti_core_with_benchmate_core_i7_4710hq_3939_cb


https://hwbot.org/submission/4831152_papusan_y_cruncher___pi_25m_core_i7_4710hq_2sec_78ms


https://hwbot.org/submission/4831155_papusan_y_cruncher___pi_1b_core_i7_4710hq_2min_24sec_732ms


https://hwbot.org/submission/4831180_papusan_gpupi_for_cpu___100m_core_i7_4710hq_24sec_564ms



https://hwbot.org/submission/4831181_papusan_gpupi_v3.3_for_cpu___100m_core_i7_4710hq_24sec_649ms