Mr. Fox

Check the W7 physics score, with NO performance tweaks, at a lower clock speed compared to Winduhz 10/11 filth. Hmm... that forum name is... unacceptable. I am surprised it was allowed to exist.

Maybe my suggestion above will work for you. I hope so. Things are as they should be again... Now, out of sheer curiosity, I need to boot into Windows 7 with the 3090 KPE on the other computer, confirm it still works, then install the 6900 XT and see if it immediately fails.

This is common in a multiboot with Windows 7/10/11 because one uses "LFS" and the other does not. They do not need to and it is not necessary. Try this and it may stop the problem. Create this reg file and merge in in Windows 10/11 registry. Windows Registry Editor Version 5.00 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem] "NtfsDisableLfsDowngrade"=dword:00000000 "NtfsDisableLfsUpgrade"=dword:00000001

No, I cannot flash it. Still has the stock vBIOS. It runs out of core voltage to go higher that 2800 (1.200V is not enough). It is not a 100% stable product when overclocked beyond a little bit past stock. Sometimes it runs smoothly and other times using the same MPT profile it is totally unstable. Very odd and frustrating, but this has been what I have always experienced with AMD products, so I am not complaining or surprised by it. What does make me surprised and want to complain is lack of Windows 7 functionality. If I attempt to boot Windows 7 with the 6900 XT it not only fails miserably, it wreaks all sorts of havoc with my Windows 10/11 installations. After Windows 7 fails to boot, I have a somewhat difficult time getting the other two OSes to boot again. Something with the ASROCK vBIOS is really messed up in this respect. It takes multiple chkdsk operations on each OS (individually when they are attempting to load) to get them to boot again. Out of curiosity, I pulled out the 6900 XT and dropped in the 3060 Ti and *BOOM* Windows 7 runs like a top again. No issues whatsoever. This really rubs me the wrong way. ASROCK did something wrong with the firmware, but I don't know what or why. If you have a dremmel tool grinding off a small portion of the "wing" on the left side of the IHS allows it to fit perfectly with the Thermalright CPU frame. The new copper IHS design is *almost* identical to stock. Had they made the left side wing exactly the same dimensions as stock it would not have any issues serving as a direct replacement with no modding needed to use the CPU frame. Since your stock IHS is probably going to get the left side dented if you use the Rockit delid tool, it can also be fixed and used with lapping. The damage can be corrected, but you'll have a "ruined" stock IHS when all is said and done. Not a deal breaker in the grand scheme of things, but inexcusable and distressing for anyone that is not technically savvy.

You last sentence seems to contradict the first paragraph. Do you mean the der8auer delidder that is NOT sold by EKWB is preferred? Are they different products (EK vs non-EK)? Or do you mean only for 12th Gen? The Rockit delidder has to be used with caution or the IHS will twist and tear off SMDs next to the IHS on the outside, not those under the IHS. And, so far every 12th and 13th Gen CPU I have delidded with the Rockit tool sustained damage to the IHS. It is poorly designed. You need the copper IHS because the original will be damaged in the process. It will require lapping to remove the damage caused by the delidder. The push block needs to be larger (longer) and not allow the IHS to rotate on top of the die. Warranty is becoming worthless at this point because there is always a lame excuse to shirk responsiblity and some fine print that nobody ever reads that tells them their warranty is worthless. I wonder if those same idiots know they will need to lie to Intel (or AMD) in the rare event they need the warranty? If they admit to overclocking the CPU or memory, Intel automatically denies the warranty now, just like AMD. Monkey see, monkey do. Honesty is no longer good policy. You'll end up screwed for doing the right thing because the people screwing you are dishonest. Geez, kids these days... I'd rather have a system that malfunctions with a warranty than one that works right with no warranty. Stupid is a nice way of putting it. Those same numbnuts love their BGA filth that overheats, so go figure. When stupidity is so common, the new normal always sucks. Good luck using the worthless warranty, silly kids.

Rockit delid tool works, but it's less than perfect. I don't have one to prove it yet, I believe the one from der8auer (available from EKWB) is superior without the risk of knocking off the SMDs or physically denting the left side of the IHS like a Rockit delid tool does with 12th and 13th Gen. The copper IHS will fit if you take time to grind away part of the "wing" on the left side of the IHS. It took me about 5 minutes of gentle trimming with my Dremmel tool to make it work. If you use the Rockit delid tool, do it in baby steps and check your progress. For some reason you have to watch and make sure the IHS does not start rotating clockwise while it is getting pushed off. If you just go in with guns blazing and push it off like a bull in a china cabinet you might end up with all of the SMDs on the lower right corner torn off. I need to buy one of the der8auer delid tools.

https://hwbot.org/submission/5230241_

Well, I am glad that worked out. DVI was popular when that card was released. Maybe the firmware mod removed support for HDMI and DP. Hard to know at this point if it was intentional or accidental. The XOC BIOS and vBIOS firmwares are not released by EVGA. They are aftermarket mods. Right now 3DMark is flaking out on me. This has happened to me dozens of times with 3060 Ti, 3090, 4090 and now 6900 XT. When you try to launch it the system freezes and eventually reboots if you wait long enough. It is their crappy SystemInfo scanner. Once it starts doing that the only thing that fixes it for me is uninstalling it and reinstalling it. 3DMark has become a real piece of trash. It seems to happen most often after they update their cancer to a newer version. Well, guess what they just did today... yup, an update.

You (we) have had to rely on luck, meaning not becoming a victim/loser in the silicon lottery, and knowledge. You have earned your spot. Those who receive cherry-picked parts for free have not. They still need to have knowledge, but their status is granted, not earned. When you earn something, it commands a higher level of respect from everyone, including yourself. That applies to other things in life as well. I find it difficult to respect status that is granted. I can still respect the person for their knowledge and skill, but I have to remind myself that I should.

Loved what you said here. It is not off topic when you consider the garbage we all have to put up with in a world where most technology has been reduced to trash... including that MXM adapter. Doing good things and striving for excellence requires more effort, but not as much effort as fixing the problems caused by people that do things half-assed. You can get stand-offs with longer threads and use nuts on the back side. Should not be an issue. What is an issue is the fact that is was produced and sold in a condition that is not turn-key and ready to use. That part is unacceptable. Instead of showing us how smart they are, the entity that did that showed us something else about them. It's not something good or nice, and it's definitely not something I would want anyone else to think about me if I were in their shoes.

I can't really speak to AMD's EPYC enterprise solutions, as I have no exposure to them and it is irrelevant in my personal space. If what I have experienced over the years with their consumer-oriented products is exemplary (probably not) their success in the enterprise space would be puzzling to me. My experience with the products geared toward consumers has generally seemed kind of half-baked and amateurish in terms of their execution. I would hope that is not tolerated in a commercial application. I also understand that they probably just assemble the parts, plug them in and put them to work, so heavy performance tuning and extracting better gaming and benchmarking results, and dealing with crummy/buggy firmware, software and driver support is probably a non-issue. It just needs to run and perform the task it was deployed to perform.

That is a very old image, so it does need to be updated. I need to add Intel, AMD, NVIDIA and Apple logos as well. They are all conspirators in the Techtarded Clown Mafia, and all of them are brokers of extreme stupidity.

Ruining laptops with BGA filth wasn't enough. When people with brain damage are allowed to make decisions that affect people that are not afflicted with the same kind of mental, ethical and moral defects, this is how things turn out. Welcome to 2023. This is what happens when stupidity is given permission to thrive.

Same things happens when you visit the Remond Retards' HQ... The sad reality...

Oh yeah... that, LOL. Doesn't do anything. Already tried that when they first introduce that scam. Worthless. ... There is a reason for this. Brother Tom MacDonald wrote a song about it. Here are the lyrics that describe the problem in great detail. The chorus is particularly on point... Stupid Evil, dishonest people create stupid products for a market dominated by stupid people that buy those stupid things.

How 'bout no, LOL... I wouldn't roll the dice on another Strix GPU... they're garbage. I think I would choose almost any other option over another Strix. If it is going to suck it should cost less, not more. Good to know this is normal for the crappy ASUS memory. The Strix 4090 should be priced like a PNY or Gigabutt Windfarts bottomfeeder 4090 because it performs like one of them, if not worse. I don't know what that is. HAGS? Let someone else find out. Better them than me. Strix stupid price should be a guaranteed golden ticket on binned GPU and memory, but it seems paying them extra for it is to only guarantee disappointment.

Here are some Linux 6900 XT benchmarks with overclocked core at 2800 and memory at 2200. Doesn't look like Indigo tests the 6900 XT.

OK... Here are some Linux benchmarks with 6900 XT overclocked core at 2800 and memory at 2200. Doesn't look like Indigo tests the 6900 XT.

They did the same thing with Fire Strike. That started with Windows 8. You get a fake lower score with Windows 7 because they deliberately built the benchmark to not work as well. They call it "optimization" but it is actually manipulation and misrepresentation. Some people will call it tin foil hat, but I honestly believe the Redmond Reprobates collude with "partners" in the tech industry to manipulate everything they can to deceive people into embracing trash. Lies and deception are their primary vehicles to accomplishing that. Intel Speedshift, DX12, resizable bar, ray tracing and core scheduling are additional examples of misrepresented need for a newer OS downgrade to get "compatibility" with features that only require the new OS due to deliberate fabrication of fake incompatibilities. Their fraud and deception is supported through a complex and highly abusive use of cancer drivers and malignant firmware. All of these technologies do, in fact, function exactly the same with Windows 7. They do not require Windows 10 or 11 to function. I would love to know how much money the Redmond Reprobates have paid to organize the collusive software and firmware engineering and coordination of the marketing propaganda. They haven't bothered to extend the same lies to Linux because the tiny market share it holds is not a serious threat like the one their better and now obsolete operating system poses.

Unless you are foolish about where you purchase parts, you can buy a 13900KS and a high end mobo like an Apex or Dark, and a decent 360mm AIO and still have at least a $100 USD change left over compared to the price of a better model of RTX 4090. If you're a savvy tech buyer, enough to cover 32GB of Hynix A-die. Another thing I have noticed a few times with the 6900 XT is that upon entering the BIOS on the first screen there is corruption and bizarre colors until the mouse cursor is moved or I use the arrow keys on the keyboard on the first BIOS screen to begin navigation, then it instantly returns to normal. It does not do it every time and no error is produced when it occurs. Just an observation of something weird that happens randomly and has not caused a problem so far.

Maybe one weekend my curiosity or boredom will be great enough to find out. Having to mess with water and tubing beyond using a QDC fitting often causes me to procrastinate.

It is a meaningless and patronizing gesture. It is certainly not evidence that they are interested in offering value. I find the €40 price drop to be insulting, at best. At worst, it reinforces the thought that the end users that purchase their overpriced products are as dumb as a box of rocks. Many of them are. Some are smart enough to know better, but want what they want and they are willing to get screwed to have what they want. A lot of us fit into that bucket. The Green Goblin knows that both types of end user exist. The price is the same for the stupid as it is the foolish.

Yup. It is only a matter of time. The digital genocide will come. It is everyone's destiny to get screwed if they are happy to keep using the GPU. The Green Goblin will create a way to make them unhappy, and the Red Robinhood will make sure the drivers malfunction. First will come the fake performance degradation, then comes the EOL for driver support and removal of previously released drivers for download. Remember... we are not customers. We are end users. They don't love us and don't care if we love them... as long as the money keeps flowing.

Maybe. I cannot boot Windows 7 in pure UEFI mode on the ASUS mobo. I have tried and it fails. I can with the EVGA mobo. Even resizable bar support works in pure UEFI mode on the EVGA mobo. In both cases that is with NVIDIA GPUs. This is the only AMD GPU tested. The problem may be in reverse. I probably won't know the answer to that question unless I decide to buy the 6900 XT because it will go into the ASUS-powered system when I am done benching it. If I decide I do not want to buy it, then I may not want to invest the effort into finding out. Had I known of this issue with Windows 7 before putting a waterblock on the 6900 XT I would have tested in both desktops to see if the problem is different on one machine versus the other, or the same on both. I am only guessing until I jump through the hoops, but the only thing that has changed is going from an NVIDIA GPU to the 6900 XT. That is when Windows 7 gets broken on the EVGA mobo. I am not sure why they would need to make one. Given the fact that you can buy a cable from Cablemods for $20 USD they probably have no financial incentive to make one. If they did and had to charge more to be profitable, nobody (or many people) would not be willing to pay extra for the cable from EVGA. I know that I would not.