Mr. Fox

That is a good article and it explains how it works in a clear way, unlike "Rambling Bill" LOL. Thanks for the link. I wish ASUS had this feature. I had never heard of it and had not seen the article before. Is this something you remembered reading before, or did you Google search "CPU Loadline Saturation Control" and come across it that way?

@Rage Set I found something interesting that you might want to try on your MPG Z790i Edge. This actually worked very well for me, even with this HORRIBLE 14900KF that I am sending back to Amazon after the 14900K I bought from @Talon arrives tomorrow. This turd CPU cannot run the E-cores above 45x without a BSOD. I found this trick in a Buildzoid video, but to spare you from his rambling and cut to this chase, here is what he showed with his motherboard (exactly the same one). Here is the video if you are a glutton for punishment and have a half hour to spend to harvest a 5-minute nugget. MSI Z790i EDGE WIFI: Load Line Calibration and Vcore regulation Long story short, MSI has a terrible implementation of LLC regulation. There is a workaround that seems totally counter-intuitive if you are used to overclocking with ASUS and EVGA motherboards. Set LLC 8 (maximum vdroop... yes, really), set override voltage, not adaptive, then apply a high Load Line Saturation (I used 75A). I was able to clock the P-cores 100MHz higher with lower load voltage and shaved roughly 12-15°C off of my Cinebench R23 max core temps and scored about 1,000 points higher because it was no longer thermal throttling. This provides better voltage control and he proved it with his oscilloscope. Below are the BIOS settings used on this CPU illustrating what I did, but use the idea conceptually for yours based on what your CPU needs for vcore under full load. As you can see, I set the vcore very high for 56x (much higher than needed)... 1.435V override. Now, go to the DigitALL menu, set CPU and CPU AUX LLC to Mode 8, then enable Load Line Saturation Level and set it to like 75A. This is an amperage negative offset value that reduces the erratic overshoot/undershoot swing in load voltage and slightly reduces power draw (about 50W less for me) and heat. Below is the no-longer-overheating garbage bin 14900KF Cinebench R23 result. It was thermal throttling with the P-cores at 55x before this tweak. In this screenshot the middle number (indicated with green arrows) is the Cinebench R23 load voltage.

I suspect it would have fit issues as well with the Z690 and Z790 Dark because they have the massive VRM heatsink in the same place as the large NVMe heatsink on the Apex/Apex Encore.

In some cases the reviewers are just heiny-kissing shills. Maybe in most cases. Those that dare to call balls and strikes and can't be controlled don't get nearly as many freebies because they can't be bought. Short of saying they are dishonest, it is easier to overlook and forgive defects when you pay nothing. When you use your own money and it turns out to be trash, that changes things... and, you complain about it because it was your money that was wasted.

Why do people do these crazy things?

Looks like the Green Goblin is forcing PCIe 6.0 and the new revision of 12VHPR on all 5000 cards, including the lowest tier junk that is still using 8-pin PCIe power connectors.

Yes, I agree. I run my GPUs at firmware defaults except when benching. I am using the Galax HOF OC Labs 1000W vBIOS for both 4090s. The default core boost clock is like 2850 or so. With the 4090 (and probably 4080, maybe even 4070 Ti) there is no real need to overclock the GPU to have a fantastic gaming experience.

Good. Glad to hear it. I wish they had not included that option for 4090. It's stupid. Had they not included it, HWBOT would not have been able to make it mandatory for 3DMark submissions for 4090 owners. It impairs performance having it enabled.

@tps3443does the 4080S have ECC in NVCP?

If 4080 has it, you will find it in NVIDIA Control Panel. I do not know if it does or not. I am curious to know if it does or doesn't. On 4090 it is disabled by default, but the dumb-dumbs at HWBOT require it to be enabled for 3DMark. (Not older 3DMarks like 3DMark 11 and Vantage, but the newer 3DMark suite of benchmark bloatware.)

Seems like it might be good, and if it sucks it is free.... https://store.epicgames.com/en-US/p/dakar-desert-rally

What kind of game was that he said he wanted to play? He really likes words that start with F and S. It's pretty sad that his vocabulary is so limited. He's going to pop a vein one of these days. I hope his breath doesn't smell as bad as the putrid stuff coming out of his mouth. That would be really disgusting. Most of what he is saying is accurate, but it's hard to respect him because he creates an image that makes himself seem so stupid.

Yes, 4080S is an excellent GPU. It's not a 4090, nor is it a 7900XTX or 4070 Ti. If it were then nobody would have any reason whatsoever to buy a 4090. It's also very consistent. Ran this heavy ray tracing benchmark on Apex and Apex Encore with stock GPU and results are the same within a margin of error. I did not bother checking to see if they are using the same driver. (And, no, I don't care, LOL.) The point is, you can count on consistent performance. MSI 4090 Gigabyte 4090

With ECC cripple mode enabled 41.6K graphics score. Add another 3-4K with ECC cripple disabled.

Same. I mean... c'mon... it's white. But it wouldn't be for very long, LOL. It looks nice before you start using it. Not a good choice for something you touch the entire time you are using the computer. If I ever sell it I can let the next owner enjoy the white mousepad... for a day or two. Then it will be pretty skanky. Just grabbed a few screenshots with the 4090 Suprim. Ultra preset, 4K native, 165Hz, no Vsync, Gstink and no DLSS. Stock GPU clocks.

Scores in what? I can't tell what benchmark or measurement we are speaking about.

No, the Gigabyte 4090 I purchased was like that on the air cooler as well. It ran freakishly cool on the air cooler. Unlike anything I had seen before. Installing the water block had little effect on temperatures. The benefit will only be obvious using chilled water. Having a vapor chamber with lots of heat pipes and a good fin stack does a fantastic job as long as the ambient temperatures are low. If you are in a hot environment it might not be adequate. Both of my 4090s have excellent memory overclocking. The Strix 4090 was trash. The Gigabyte 4090 easily handles +1500 (maybe more, haven't tried yet) and the 4090 Suprim +1700.

Awesome. Glad to hear you got a solid sample. I am not sure what is going on with your text. Did you copy and paste from somewhere else? The color of the font is almost the same as the background and I have to highlight the text to read it. This happens with things @Papusanposts often as well. I have to change to the ugly white eyeball-searing default unisex theme to read the text or highlight it with my mouse. In the post quoted above I cut the text from the post and pasted it again as plain text. Screenshot below shows what I see with the original formatting using the forum Dark Bravo 6 theme.

Correct. That is the part we do not know yet. Because it is made by EK I expect the build quality to be superior, but whether it works well is to be determined. Under most circumstances it would be safe to assume if it is made by EK it is going to be excellent (fittings, radiators, etc.) but things like this require very precise fit, with some degree of flexibility for variance between installation scenarios without losing any performance. It is more difficult to execute. And, it has to be idiot-proof, easy to install and with good instructions so that an average user can install it without breaking anything. Alienware provides a good example of excellent build quality with poor performance and flawed design.

As you may recall, I had one. The issue was not build quality. The problem was the block not making proper contact with the die. If I had to guess what went wrong, they were probably exercising more caution than they should have to avoid the block putting too much pressure on the die and breaking it, and that caused contact to be insufficient in some installations. Even Kit Guru had to mill down their test prototype to obtain proper die contact because of that. I could see the fear of inexperienced users breaking their CPUs by improper installation might be a concern for them, but it needs to work right and working right requires solid contact and enough pressure to compress the CPU into the socket firmly, not barely making contact or not at all. Some people had good results with theirs and others (like me) had terrible results. They released an updated CPU frame and cold plate that I understand corrected the problem. I did not want to deal with the nonsense of them releasing a product that was not ready and needed more work, so I opted for a refund instead of playing the part of guinea pig. Time will tell if they learned what they needed to change and corrected their mistakes on the AIO. If they did, then this direct die AIO should be a great product. If they did not, it will be evident right away. This is a very much-needed product and I hope they made it right so that the block makes good contact with the die. If it works well, this will be a real game-changer for a lot of people and copycat products will likely follow. The release of a direct die AIO is long overdue. I am happy to see them making the effort. I'm sure that an AM5 version would be appreciated by our AMD friends as well.

I am not sure how our wives would feel about that. I'm sure she is a wonderful woman, but I'm probably as attached to mine as you are to yours. Hard to beat 38 years of familiarity and friendship, even if they prefer different beverages than we do. I wonder if EK ever figured out what they did wrong on the bare die thing. I am wondering if the newly launched EK Nucleus CR360 Bare Die AIO would work well in the midget monster. That is also an interesting prototype LGA 1700 heat spreader made like the one Roman made for AM5. It replaces the ILM and IHS. Looks like between 15-20°C temperature reduction with the bare die AIO.

My wife loves coffee and she can drink it black all day long. I do not care for the smell or taste of it. If I drink coffee it has to be loaded with flavoring additives, about half cream and lots of artificial sweetener. Or, with melted caramel and vanilla ice cream added. Then it is like liquid candy and doesn't resemble coffee any more. But, I drink almost exclusively cold beverages, even when I am in a cold climate. When I was in Chicago a few weeks ago and it was sub-zero temperatures I was drinking iced tea, diet Dr. Pepper and diet Coke, even at breakfast.

The Unify-X was an excellent motherboard. It's sad they never released a newer Z790 version of it. The Z690 Unify-X and Gigabyte Z690 Tachyon were their best motherboards, but finding either of them for sale is rare. Even more rare than the Z790 Apex (white). I don't understand why. They could easily sell every one they made if they made more of them. This tiny little MSI MPG Z790i Edge is better at memory overclocking than more expensive 4-DIMM motherboards more than twice its size.

I see what you did there. Mine are not crocs, but still black. Other than tubing, both of us like most things in black. Even though I live in Arizona, my skin stays very white all year, just like my PC lighting. But not as white as my socks. I don't like black coffee though. Come to think of it, I just don't like coffee, period. So, doing a little bit of analyzing, the SP96 14900K that I purchased from @Talonis extremely close to my SP108 13900KS on both P-cores and E-cores. I am not confident that the MCSP actually has much bearing on things as it seems to be a somewhat unreliable measurement with variance. That 13900KS has an MCSP83 while the 14900K is MCSP78, so that remains to be seen. (I don't actually have it yet, just adjusting my Apex BIOS settings to match the BIOS screenshots he provided and then comparing the numbers on the AI and VF screens.) Give or take about 5mV... 13900KS P120 = 14900K P104 13900KS E84 = 14900K E80 Tinkering the the MSI BIOS has been interesting. The "Dynamic" CPU clock feature disables Turbo Boost and EIST and locks the cores at a given clock. The CPU runs cooler and performs better with "Dynamic" enabled. I discovered that purely by accident because the settings are on different menus and seeing that they are directly related was accidental. I was trying to figure out why Turbo Boost and EIST were disabled and greyed out. When I changed the CPU clock behavior from "Fixed" to "Dynamic" those became "Enabled" (still greyed out). My temperatures went down and Cinebench R23 score went up. And, my VRM MOS and Stage temps went down.

I used one just like that for many years in my work that required taking 8 or 10 dozen photos every day. The floppy disk went into a plastic sleeve in each file.