Mr. Fox

I guess that is lucky for me that I don't really care much about single core performance. That is the main reason I gave 9950X a try. Otherwise the 285K would have been the better choice. It is a downgrade from everything else I own with not enough multi-thread mojo on top of being new with bugs that need to be ironed out first. I think I may have run single core Cinebench less than 10 times in my whole life, LoL.

Thank you. I've having to relearn everything that is for sure. I think I am getting way better results using fixed all core and manual voltage than PBO, which is actually exactly the same for me on Intel. eTVB and opportunistic core boost truly sucks. Here's my best so far... This is with the EK CR360 AIO. I am going to have to get the Heatkiller block on it until I buy a der8auer delid tool and Mycro bare die block. Brother @cylix it is the G.SKILL 8000 sticks that my Ryzen setup does not like. I put in the TeamGroup Xtreme 48GB kit and it boots and runs 8200 like a champ. Interesting. There must be something in the G.SKILL 8000 SPD default settings that causes it to fail to POST on Ryzen. What makes it ever more odd is the G.SKILL sticks run identical to the TeamGroup Xtreme in the Apex Encore. No differences in performance, settings or voltage. I just put the G.SKILL memory on the Apex Encore and *BOOM* there it is! The G.SKILL 8000 kits (both of them) have always worked fantastic. I only bought the TGX because it has no RGB and they're covered up with waterblocks so RGB may as well not exist.

Maybe this will help. Still not great considering what we are used to, but a MASSIVE improvement over where I started last night. That was pretty scary how bad it was. I've seen people that are ignorant about memory tuning with latency this bad on Intel. I'm pretty sure I can do better than this. Just now getting started in the process. Nothing works the same as Intel. Shake the Etch-A-Sketch and start over. Besides the latency, 8000 is comparatively very slow read/write/copy speeds on Ryzen.

Maybe Brother @chew can give us some pointers. My latency is lower than that but still very high. This seems a whole lot more convoluted that what I am used to. I think all the sliders and knobs have to be just right or it behaves badly, LOL. I getting closer to finding out where the sliders and knobs belong. Little by little. I think I am going to test my TeamGroup Xtreme 8200 48GB kit on this. No idea why the motherboard will not even try to boot with the G.SKILL 8000 48GB kit. It doesn't even try to boot. Just stops abruptly after about 1 second with Q-code 15 and CPU and RAM LEDs on.

I was pretty blunt in my response about how ugly it looks. https://community.hwbot.org/topic/239385-hwbot-new-gui-feedback-thread/?do=findComment&comment=680164

OK, I think maybe I am starting to get the hang of things a little better. I still need to catch up with Brother @Raiderman but I'm not as lost as when I gave up around 1:30 AM and went to bed, LOL. Ryzen is still as finnicky as it was with the 5950X, but this is an improvement from where it was for sure. Side note: Glad I don't give a hoot about WiFi. I guess WiFi 7 is a Winduhz 11-only piece of feces. No thanks. Stuff it, Micro$lop. @Papusan how do you like the icky new smartphone crap look of HWBOT? I submitted this and almost threw up in my mouth a little bit. Part of me wants to send a PM to Roman (der8aer) at oc.net and ask him why he allowed that. Everything is so big and bloated and ugly now. Things on the web page take up about 200% more browser space that necessary. Super idiotic. https://hwbot.org/submission/5685453_

Yes it it very strange. If I install the G.SKILL 8000 24x2 module and power it on it stops at Q-code 15 almost instantly with the CPU and RAM lights on. Maybe it is too early on the firmware. Or, maybe I need to RMA this motherboard and get a Taichi. I wish the Taichi Lite has the GPU release. It looks much better and I like the absence of RGB. What is interesting is none of the 8000 Expo memory seems to be available. The G.SKILL Expo 8000 doesn't even release to retail until 11/11/2024, so I am guessing the reviewers on YouTube received pre-release samples. OK, I dropped back to the prior BIOS with an older Agesa and I can get it to run using the MSI built-in 8000 profile. It still refuses to boot with the 8000 48GB kit. It does not even try to train or anything. Instantly shows code 15 and CPU and RAM lights on within like 1 second. I am guessing the 5600 base SPD clock and timings are not Ryzen-acceptable. @cylix is the 48GB kit you bought a default (non-XMP) of 4800 or 5600? I have not done any CPU tuning, just trying to figure out the memory. This seems a lot more convulted that what I am used to. I have a lot to relearn. For example, the BIOS shows the CPU clock is 4300 with everything on "Auto" but Windows shows like 5700 core clocks and it drops to like 5200 running Cinebench. Very weird and different. Without PBO enabled it runs a lot faster and gets a lot hotter.

I am trying to get the memory overclock figured out. The motherboard will not boot with 24GB sticks and doing more research it seems there are very few 48GB memory kits that are compatible with Ryzen. The MSI QVC shows modules tested as compatible that I cannot find for sale. So, right now I am using an old G.SKILL XMP 7200 32GB kit with no Expo profile. The XMP will not boot, but using the MSI 7200 profile it does. This is going to take me a while to figure out. I may end up having to purchase a newer/compatible memory kit later on. It looks like it is easily doing 45K in Cinebench R23 with no CPU overclock, just PBO enabled. FYI - on the X870E Carbon I can use the PCIe 5.0 slot above the CPU and the GPU works at x16, but if I populate that and the first (PCIe 5.0) M.2 slot under the GPU with any kind of NVMe the GPU drops to x8. So, same as the Encore, I have one M.2 slot that is unusable unless I want to castrate the GPU.

The artist responsible for DOOM and Wolfenstein music. 🤘

I agree with you. The trouble is that if you do not want a CPU made with the glued together tiled mess from Intel you are left with the same kind of undesirable contraption from AMD. So, there is no alternative and no solution to that issue unless you stick with older tech built on the superior monolithic design. 😑 Thus, I went with Ryzen 9, not Core Ultra 9. Neither one ideal, but the Intel part is the weaker and immature (needing more work) product now. Had they released a 32 thread 285K the weaker part would not be true, but there would still be the guinea pig experience to contend with. Being an early adopter of anything is usually not a smart move. You must choose if you prefer vomiting or diarrhea. At least you get to choose the form of malaise. 😑

Newer is always better newer

I know a lot of people think brother @Papusan and @Mr. Fox are bat-poop crazy for hating on Winduhz 11 so hard, especially newer versions, (also including the newer Windoze 10 cancer,) but we actually do our homework. If performance was better we would actually be using it as our preferred OS. The ugly part is subjective and fixable. Try as we may, people are still going to think we are nuts and form their own misinformed opinions and continue to believe the lies that it works better. Well, bless their hearts. The only way to know is to test. All of my systems boot multiple operating systems and it immediately becomes as plain as the nose on your face which one sucks when you can reboot and run the same tests back-to-back among OS versions and watch the newer ones reveal their suckiness. It leaves no wiggle room for speculation on which one deserves the most hate. CPU and memory performance have declined and degraded with newer OS versions. Graphics performance sometimes improves and sometimes regresses. That part is inconsistent. The Redmond Retards add silly gimmick features to entice gamers into running their crappy cancer OS and pray nobody notices what a steaming deep dish colon loaf it is. And, guess what? Most of the gamerboys don't notice. No, my surprised look is not on my face. Us old farts know what we're talking about most of the time. Windows 11 is a pile of crap. Even Jufus the potty mouth has documented that the claims that newer versions of Winduhz run better is a load of crap. Even as recently as the 9900/9950X release he debunked the excuses that AMD fanboys were making that Windoze 11 needed an update for the new CPU to work right. I generally view the opinion that newer OSes work better as a sign of inexperience and ignorance. Some smart people believe that and it is probably because they make assumptions rather than exercising due dilligence and testing, or they are not doing the right tests. Even if the latest cancer OS performed exactly the same as every version before it, who wants to own a zombie computer and have the Redmond Reprobates stealing data and making all of the decisions about what happens with the OS? Hard to believe there are many otherwise intelligent people that drink that Kool-Aid. Letting Micro$oft decide what happens is essentially a plan for failure if you like the idea of having a system perform at its best. Believing the next update or version release of Winduhz will improve performance is utterly insane, and contrary to historical evidence that suggests the complete opposite should be expected. It will be interesting to play with the 9950X and discover what my opinion will be in the end. Parts arrived yesterday so I will install them today or tomorrow. I'm not sold on 3D v-cache being all that, but probably because I am not a gamer and don't value that as much as gamers do. It is good that 9800X3D is not as terrible at productivity as its predecessor. The idea of buying an 8-core CPU doesn't really do anything for me other than invoking a feeling of disgust. That seems very lame to me by current standards, kind of like saying 8GB of RAM is enough. It might work, but it's barely adequate for much more than email and web surfing. For the same reason, I was offended that the Core Ultra 9 flagship was downgraded to 24 threads from its predecessor's 32-thread i9 flagship and I didn't like the 285K before it released for that reason alone. I was surprised that it didn't perform worse than it does, but losing 8 threads and having lower boost clocks with less P-core overclock headroom is a downgrade to me no matter what angle I choose to look at it. It would be hard for me to find something to be happy about because of both of those things. I never like the concept of doing more with less. That always rubs me the wrong way no matter where the concept is applied. Looks like I lucked out on the 9950X purchase. It is now sold out at NewEgg and I purchased it for $110 less than launch price. Maybe Core Ultra 9 adopters will get lucky with a similar price drop in a couple of months.

I stopped caring about what they want, or what is best for them, shortly after W8 dropped from their rectum and now I want only what is the worst of everything for them. Since the day they spewed that filth they've done nothing but one screwup after another.

Congrats, bro. That's awesome. It is good that your first crack at it yielded results because it gives incentive to continuing pecking at it more. (Some people have a rough initial start, get frustrated and give it up to move onto something easier.) Memory overclocking can be fun when the hardware cooperates. Not so much fun when finding stability is difficult. Memory overclocking makes a small difference in normal performance overall, but can be very beneficial with some games and benchmarks. 1.520V should be no problem at all unless it gets hot enough to start experiencing errors. That is not dangerously hot, but memory starts to error out when you push beyond 45°C. Anything above 50°C will make finding stability very difficult. It will be interesting to see what has changed with AMD memory overclocking since owning the X570 mess 3 years ago. I hope it is easier and less frustrating than it was on the DDR4 platform with buggy Agesa firmware.

That sounds a bit like the stupid stunt Intel pulled with Core Ultra voltage capping crap. Maybe they are scared of their CPUs catching fire like what happened a year or two ago. But, it's made for gamers, not overclockers.

And, you would be rightfully so. If they are going to do that it needs to be the first model launched and make everyone else wait for the non-Ti 5090 to follow later. Pulling that stunt on people that already purchase the flagship X090 is just playing dirty pool. Despicable and scummy even.

No there is not. Those work great, even with multiple NVMe on the card. I have 8 NVMe SSDs on the Apex Encore and 9 on the Apex, including add-in PCIe cards. On the Velocita I have a single NVMe card in the bottom PCIe slot and all work marvelously. This one is my favorite, but not because I wrote the review: https://reviews.extremehw.net/all-reviews/sabrent-4-drive-nvme-m-2-ssd-to-pcie-3-0-x4-adapter-card-review/

Sadly, it is only 8 cores. Makes it a gamer-only solution. Good if gaming trumps everything, but not good otherwise. On the topic of SAD, why would anyone flush money down the toilet on this rubbish memory? Crucial DDR5 Pro. But, the question is "Pro" what? I got this free with a Gen5 M.2 review for Crucial a long time ago, so free is good and I cannot complain on that basis. But, I cannot imagine anyone buying this on purpose. I put this in the Velocita since I will be using the 8000 kit in the Carbon. Glad I have something to use, but still... LOL. I bumped it from 5600 to 6000 and tightened the timings as far as they can go without failing to boot and it still performs horribly. Truly pathetic and unworthy of consideration for any purpose.

Thank you, Jesus! It is interesting that the X870E Carbon manual makes no mention of this that I can find, but I remember in one of the videos I watched (don't recall if it was KitGuru or Wendell at L1) it was mentioned that if both Gen5 M.2 slots are populated it will do that. I also do not remember if it was a specific board or all of them. The one under the GPU PCIe x16 slot is shared with the slot if I remember correctly. It would have been better to have only one Gen5 slot for sure. I love that the Apex has none. It makes it better than the Encore for that reason alone. If you want Gen5 you can use the included add-in card and cripple your GPU. It is unfortunate as well that many new Intel and AMD motherboard have only 4 SATA ports. In some cases one of the SATA ports is shared with one M.2 and both cannot be used. (I think only the lower tier chipsets, not Z or X have this handicap.) That really sucks. X870E Carbon User Manual

Makes a lot of sense, doesn't it? Build something for gamers and overclocking enthusiasts and give it "special" features that no gamer or overclocking enthusiast would want. That's the new way everyone rolls in tech world now. Pat on the back and a knife to the gut.

Welcome to more Gen5. The tech nobody needs. If my understanding is correct (and it may not be) if you have four M.2 slots that happens if you use both of the Gen5, but not if you use only the one above the GPU. So you could only use two Gen4 and one Gen5. A$U$ made the top M.2 Gen5 on the Apex Encore and it happens on it, too. The Apex is all Gen4 M.2 and you can fill them all with no impact.

Yes, 9950X. Just because I honestly do not care about having a gaming-focused CPU. That is not important to me. I want what performs best in CPU centric overclocked computational workloads and 3D v-cache isn't that. We know nothing really about 9950X3D other than rumors. It is supposedly going to be unlocked, but that doesn't mean it will be good at overclocking.

There are a couple more features on the newer board. I could have saved about $100 on the older Carbon, but it did not have some of the things I always look for like on-board buttons for things like power and reset. But, the way AMD does their "chipset" number is sort of misleading (as noted in the KitGuru video) and the only difference between X670E and X870E is USB 4.0 is mandatory on the later, optional on the former (update to board partner to choose). But, in another way is it not misleading because the motherboards have differences in design and aesthetics even if the actual chipset didn't change. It would get confusing if they had multiple motherboards that had the same chipset name and the same motherboard model name with different features and design changes. So in that respect it is more necessary than misleading. There could also be differences (hopefully improvements) in PCB and SMD components. This would be less confusing than having board version/revision numbers like Gigabyte sometimes does.