Mr. Fox

Awesome. I hope they get finanically raped by the litigation and left for dead. Then we can hope the same for lots of their evil accomplices in the tech space to suffer similar tragedies. We need to see some fortunes lost, empires ruined and some key decision-making people criminially prosecuted in order for a lot of this nonsense to get smashed for good.

That is EXACTLY what happened when I went from 1503 to the 9933 XOC BIOS that Safedisk provided last week. I did not try to use an old CMO file. I saved the text and set everything manually. It seemed stable at first but then started locking up in AIDA64 and TM5. I went back to 1503 and applied the stable profile and it was no longer stable like it was before 9933. I still have to do more messing with it to confirm, but I think the XMP Tweaked menu option is causing it now. It did not cause it before, which is weird. ASUS does some nice things with their BIOS, but they do some stupid things as well. When you flash firmware there are certain regions that are protected and never go back to what they were before when you flash the older firmware. That is why I generally think it is best to avoid firmware updates unless I am expecting something I need to happen. MSI and EVGA BIOS updates erase and re-write everything, in all regions. That is how it should be and ASUS needs to get a clue on this. If you pay close attention during an ASUS BIOS update and read the text at the bottom of the screen, sometimes it skips over certain regions and it never flashes some of them. They also need to stop insisting on Intel ME updates and bundling it with the BIOS. Intel ME updates are totally unnecessary. No Intel system requires the Intel ME to fully function. They need to provide the option to disable it. It serves no useful purpose for consumers and having it creates extra security risks that some people get wigged out about. No ME means no vulnerabilities associated with ME. Edit: I think I am going to try the BIOS Flashback recovery and then retest my saved profile with XMP Tweaked. BIOS flashback forces flashing of regions and allows some BIOS mods to be flashed that are blocked through a normal flash update.

Yeah, older BIOS is almost always better than newer. There was an exception as DDR5 technology was needing work, but we are way past that now. SA Bug: If your CPU has it, if you set System Agent voltage (VCCSA) above a certain value, usually a very low value, the system will start freezing. It would lock up in the BIOS or in Windows. This one 14900KF has it and if I left it set on Auto it would go too high and freeze with the CPU under load. I had to manually set it at 1.190V, which was fine up to 8200 CL38, but any higher memory clock than that was not stable because it needed more system agent voltage. I set 1.250V for 8400 CL38 and tight timings. That "no workey" with the SA bug. If you don't want to overclock the memory that high it is not a big deal, but it might lead one to believe something else is wrong if the freezing happens with the BIOS left on Auto and you don't know you have to manually decrease it from default. Using as little as you need is always good as long as you are not limited by the bug. Welcome back to bare die paradise. I thought you had lost your marbles there for a bit. I hate having an IHS on my CPUs. It sucks, LOL.

It may be another setting and not the SA bug. I went back to 1503 and applied my old profile and it was freezing in AIDA64 the same as 9933. Not sure why, so I applied XMP and manually set VCCSA to 1.350V and no freezing. So, something did change. I switched to XMP I (instead of XMP Tweaked) and applied my settings and now no more freezing. So, I may go back and test 9933 again with different settings to see for sure. It's weird that it was working fine and then suddenly XMP Tweaked would cause it to start freezing. The only reason I even considered updating the BIOS at all was to see if it fixed the SA bug on this one CPU. Otherwise, no need for BIOS updates. Waste of time and often brings less desirable performance, especially if "security" mitigations are part of the cancer payload. I was pleasantly surprised that it fixed that. Applying the microcode in Windows does NOT fix the SA bug. (I tried the Windows MC Prema shared and it did not correct the issue.) We are also only assuming it was the microcode update. Logical assumption, but it could be something else. I say logical because the SA bug was motherboard agnostic. Was the same on my MSI board as the ASUS and the BIOS update corrected it. If I had left this CPU in the MSI Z790I Edge ITX board it would have continued to be perfectly fine with the SA bug for the same reason you mention. Air cooled in a small chassis and poor circulation, 8200 is the edge of stability due to temperatures anyhow. Having low VCCSA because you can and because you want to is good. It's only an issue if you wanted to push it further and can't because of a flaw preventing it. Inconsequential otherwise. But, nobody like knowing there is a fly in the ointment. So, it is working without issue again. Will play with tightening up the timings again now that I am using the XMP I instead of XMP Tweaked option. Interesting change that makes no sense. The Intel ME was not updated, so I cannot blame it on that.

I spoke too soon. BIOS 9933 brought back the SA bug. Going back to 1503 and hoping it goes away again.

Just tested the new XOC BIOS for the Apex Encore and will apply it to the white Apex as well. It has the new 0x129 MC, but retains the ability to drop back to 11F. I tested it on the Encore first since the CPU with the SA bug that got fixed by 0x129 is installed in the Encore. Later today I am going to change to MC 11F and see if the SA bug rears its head again, or if that did something permanent to solve the bug. In other news, there is a ROG theme available in the latest HWiNFO64. It in typical ASUS fashion, it is the most tacky-looking theme one could ever imagines. It looks effing AWFUL, especially in dark mode. What is wrong with these nutjobs and their tacky and over-the-top garish aesthetic preferences? I am going to have to PM Martin over at HWiNFO forum and ask if the style of the theme was specified by ASUS or if he created it while taking drugs.

I find it very hard to see any visual difference in the benchmark no matter what settings are used. The issue using DLSS on 2080 Ti and 3090 Ti is there is no frame generation, and the benchmark does not have options for performance vs quality. It was interesting seeing how the different settings affected the different GPUs. I agree. I don't like FSR and generally don't use it. On a GPU with no DLSS+frame generation, at least in this benchmark, it provides a better framerate with the FSR frame generation. Unless something unusual changes my mind, I have no intention of buying the game though (I seldom ever enjoy playing RPG) so it's all just for benching fun and won't matter in the grand scheme of things. Howdy stranger. Nice seeing you. Hope you can stay for a spell. I think Steel Nomad is a joke. 3DMark Suite in general has become a bloated piece of garbage benchmark. The more they add to it, the trashier it seems to get. Things started circling the drain with Port Royal and have gotten progressively worse. The best benchmarks in the suite are those with CPU and combined scores, and the very best (Sky Diver) is "no longer supported" because UL is run by idiots. Of course, these are my personal biases, so I know not everyone will agree with them.

Things start getting sketchy and the 2080 Ti starts showing its age with Super Resolution at 100%, which is tough even for the 4090. Basically nothing can produce a smooth experience cranks up like that. Even so, with everything maxed out it still performs stronger than I expected it to. It would not be a good gaming experience (slideshow)... but still... just saying... One thing I did not expect is that with RT turned off the GPU power increased by about 50W. Because there was not a pronounced difference in the image quality, I am assuming this is because 100% of the workload shifts to the rendering cores and the tensor cores take a nap. If that is the correct explanation it makes sense, but I would have assumed that ray tracing enabled would increase the power utilization.. It seems to balance it out and relieve some stress on the GPU core.

I tested Wukong on the 2080 Ti and it is interesting. It works better than I expected it to and it appears the old Xeon has plenty of juice to not be a bottleneck to the GPU. The CPU was generally 25~35% utilization and the GPU at or close to 100%. And, the graphics look good no matter what level of detail or quality preset. Hard for me to see any difference except with ray tracing disabled. Then it does lose some quality, but still looks very good. Hardly any difference between FSR and TSR that I can see.

I view gaming as an extended form of benching. I pay as much attention to my system as I do the game. I own games that I do not like and having no interest in playing simply for the included benchmark. I love it when a game dev releases a free standalone benchmark tool like the one we are playing with because then I have no reason to purchase the game. It looks beautiful, but I can say with a fair level of confidence that I would not like playing the actual game. This applies also to the Final Fantasy benchmarks. You could not even pay me to play those games. On top of being subscription-based I think they are icky, (based on genre and gameplay style,) but the free standalone benchmarks are fun to run.

You should put something like this on it to allow air flow but direct it away from the system, either toward the ceiling or forward (into the room) and away from the bench. You could attach it with double-sided rubber tape or even use adhesive magnets if the backplate is not aluminum. It would look nice and probably be better for cooling. You could also attach it with a bead of T-7000 adhesive, which would be strong enough to hold it forever, but not so permanent that you could not remove it later if you wanted to. https://www.amazon.com/ORLANG-Marine-Louvered-Mounting-Suitable/dp/B0CS61SZ9K?th=1

Both of us have the updated PCB design without that weakened hook area.

I gritted my teeth and flashed the new BIOS again and found that DTPF setting present where expected. No idea why it is not visible in the January BIOS. So, I was able to test this benchmark with the Intel Optimizer and found it provided no benefit. FPS were the same as without it and it made it feel like the FPS was lower due to some stuttering. I hope I don't have the same issues with the M.2 port dropping out and the BIOS loading menus incorrectly. If that happens even once then it is back to the January BIOS. Crossing my fingers. So far, so good. I do like that MSI does not interfere with flashing older firmware and does not bundle the ME firmware cancer with their BIOS updates. I love that it is my decision and not theirs and I hate that ASUS doesn't leave that decision entirely up to my own discretion. So, each brand has its pros and cons. ASUS gives a lot of options, but neglects a couple of things I consider super important. EVGA was the best. If I were the king of the world and could do things my way, I would pick and choose the best each brand had to offer and reject their crap. There would be no restrictions on the choice of firmware versions that could be flashed, and no firmware signature crap. The BIOS would also include the ability to control the RGB/ARGB within the UEFI and eliminate the cancer software trash installed in the OS. Boards like the Apex wouldn't even have any RGB crap on them. I'd also shut down Micro$lop Store and ban the development of UWP smartphone app for PC filth.

That would be nice. Although, I am pretty sure that is never going to happen. You have to be part of the right clique and none of us are. Our results don't matter except to the extent that they become a threat to the Chosen Ones. I am pretty sure that CENS pursued the golden sample CPU that @tps3443 had because he wasn't "supposed to have it" and not because CENS actually wanted or needed it. Someone at Intel must have let it make its way into the retail channel by mistake. And, when he bought it from our Brother here it would not surprise me if it was with someone else's money. Well, this benchmark runs nice enough on 3090 Ti FTW3 if you don't go too crazy on the settings. There is no NVIDIA frame generation or DLSS 3.0 on pre-40 series RTX, so cranking the settings will crush it just as it does the 4090 with frame generation disabled. UE5 has its own though, and works fairly well. This system is connected to two 1440p displays for work.

I would not even be able to test the Intel Application Optimizer on the Z790I Edge if I wanted to. I was going to see if it made any difference with the 3090 Ti but the menu option for Intel DTPF is missing from where it is supposed to be according to MSI documentaion (right below the setting for CFG Lock). I wish I knew what the key combo is for MSI desktop firmware developer mode to access all menus. The old key combo that worked on laptops does not work on the desktop motherboard. It is stupid that MSI also hides the IA VR Limit menu. So, you cannot set a cap on the VRM voltage output (and therefore can't limit VCore properly) in their firmware. Super dumb.

The nice thing about 4 FPS is if you blink while watching you only miss one or two frames.

OK this benchmark draws insane amounts of power for some reason if run on Windows 11. And, it is not only during shader compiling. It's from start to finish. Just for giggles I wanted to see if the FPS got better or worse, so I booted into Windows 11. I tripped my circuit breaker twice and had to leave my work computer turned off when testing on Windows 11 for some reason. I almost never use Windows 11 and now I find myself wondering if that (Windows 11) has something to do with the Intel CPU degradation being complained about. This did not happen on Windows 10 (circuit breaker tripping thing). It seems like the power utilization spirals out of control on Windows 11, but does not when I am running the benchmark on Windows 10. On Windows 11, my framerates are higher by about 5 FPS, but the power draw increases by more than the minor increase in FPS. The WireView meter on the 4090 Suprim shows up to 490W (almost 100W more than Windows 10 does) and almost 100W more on the CPU on Windows 11. My Kill-A-Watt showed up to 790W being pulled from the wall, almost the entire duration of the benchmark the GPU is above 400W and over 700W from the wall the entire time while the benchmark is running. That is a bug in the game, not unique to you. I experienced exactly the same thing. To get the scale to 66 in the benchmark I have to set it to 67 or the benchmark runs at 65%. But the scores are identical with 65 or 66. No measurable difference. This benchmark is also very consistent. Back-to-back runs stay the same in terms of FPS and vary by zero to 1 FPS between runs. 65 (settings show 66) 66 (settings show 67)

So, I ran this on the Apex and it fell in the middle between your results and the Encore results previously submitted. I think this is all within a margin of error. When framerates are too low to be playable they no longer matter if it is 1 or 2 FPS difference. OK, I have to lead an operations meeting in a few minutes, but I will try that later.

@electrosoft just FYI. I went back to the old (January) BIOS on the Z790I Edge motherboard. The new BIOS with 0x129 MC was causing one of my M.2 slots to randomly stop working. I noticed it because one of my SN770 SSDs that I use for data/file storage at work was randomly disappearing. I could tell when it was going to not work because it would go through a couple of boot cycles like memory training. Sometimes it would come back after a reboot, but I also noticed when the M.2 port would go AWOL in the BIOS was showing me menus for MSI Developer mode. So something buggy with their BIOS. After going back to the old BIOS I have had no more issues with that M.2 port vanishing. I doubt it is the MC change. Probably something else MSI buggered up with the firmware modifications to appease those that want Intel defaults.

It was enabled when I ran it, so that base was covered. I disabled it and disabled DTPF in the BIOS and the lost FPS was recovered (identical avg/min/max FPS) and it no longer is extra stuttery feeling like it is with DTPF enabled with the Optimizer turned on.

Well the Intel Optimizer (with DTPF enabled in the BIOS) actually hindered performance. Decreased my results by 1 FPS but also looks more stuttery than it does with it disabled. It looks like the App doesn't have any support for this game. Metro Exodus is the only one listed that I play. Most of the supported games I either do not own or have zero interest in owning/playing them. I am going to check some of CPU benchmarks to see if they were diminished and probably disable that in the BIOS. I have always hated DTPF and Intel ME feces. It doesn't look like I have any legitimate need for it or the Optimizer app.

Does this require DTPF be enabled in the BIOS? I have that disabled. I don't like it. The Optimizer says "Failed to Connect" probably because of that.

I've never installed that Intel Optimizer thing. I may tr it. My boost is only holding 2775. So, you're 200MHz higher on core, which would explain that 2 FPS difference. I will try the Intel Optimizer first and see if that makes any difference, and then lock my boost to match yours and see what it does.

Oh, that's right. I forgot. Your ambients are probably not as high as mine are. I would imagine it would be getting even hotter than that if they were. Even if it is not the best silicon quality and won't overclock as high as some that are better, it should do everything you do in terms of gaming extremely well. More likely that it is "average" rather than "poor" quality. I ran all these on the Gigabyte 4090. I will try the same on the 4090 Suprim to see if there is a difference. That GPU is better bin than the Gigabyte GPU.