Jolyon Chen

Me too, I only bought these back up furnace CPUs to play this game maxed out with 300 leagues after checking this topic and this topic alone for over... 4 years. It's still the best on the net without breaking into same data center Admins should pin this page in all honesty.

You shouldn't decipher that from this chart. You'd need to know the vrm power phases as they influence how precisely the voltage can be "send" to the CPU (roughly said), and then, we need to know which cpu cooler everyone used too, as that then shows if the user can max out the CPU before thermal throttling, edp limit throttling, ring throttling etc etc. I mean, I'm running a 13900ks on a slightly underpowered motherboard, and when I tested that chip in a maxed out Z690 board of a friend I hit 09:30 (only requiring 288W, wth Intel) on average over 3 runs in Benchmark 4. But that friend runs a noctua and I run a 360mm AIO, so that dilutes the accuracy of the result again, and then you have people turning on ECO mode, not turning on the right energy mode and companies turning on Multicore Enhancement and posting in here, bla... bla...bla, everyone should test on stock settings, with the same systems and then we could tell. Bottom line: The fact Intel and Windows had the scheme so hard running a windows scheduler and letting AMD rely on third party implementation on per game adjustment with the Xbox Game Bar or project Lasso, should you tell that AMD is prob faster on pure CPU power, although if it would make you happy I'd happily do that maxed out mobo run and to go topping all the charts here, just to say INTEL IS KING!

Thx for this. This is great data. Windows actively tanks AMD hardware with their scheduler, so you'll be even better when they fix that... only problem could be that if there is some kind of cartel agreement, that fix to assign all the cores seamlessly for AMD won't come until Windows 12 around Christmas 2024. But I'd be happy with this CPU. I'd would not even upgrade that CPU until I tested it with W12 tbh. I do advise you to check out Project Lasso to handle that CPU better. Btw, @Brother Ben , just a quick heads up. I was running 132GB of ram in my bench. That is supported through hardware updates since 2022. I even set some aside for virtual machine, SSD and pagefile (on that VM) as well, hence the strange nr. It's not 32Gb like you listed in the spreadsheet.

FM is one of 1% of games that uses all cores past 8 in that specific scenario like Bench 4, even though the rest of the game is single threaded. The Intel and Windows thread schedulers and their poorer performance with Ryzen helps Intel in all most all other test scenarios. Whether it's intentional, I'll leave to you. Simulator games that are well coded is a very niche market (even most FM players don't ever play like benchmark 4), and everyone is way too busy nowadays, most people just go on Reddit or ask that one tech-savvy friend for a CPU, almost nobody tests things themselves for free for their specific scenario like we here do. Benchmarkers are 99% clickbait, scripted or don't benchmarks simulation games with very specific scenarios. Intel is running behind again in development and removed that voltage regulator into the bios that could make 13000s more efficient, so they couldn't push the chip further in our specific bench 4 load. Most people won't notice or need this. These reasons make Ryzen 9 better for well coded multi-core simulation games a quite unknown secret. Sources in spoilers;

@Brother Ben PS: I fixed the problem I tagged you about last week. Anyway, hope you hit 100 this year. Type: Desktop Model: Custom CPU Model: 13900KS CPU Base Frequency: 3.20 GHz CPU Turbo Frequency: 6.0 GHz RAM: 132GB RAM Clockspeed: 4000Mhz GPU: AMD Radeon 6900 XT Graphics Level in 3D: Very High Benchmark 1: 36 sec Benchmark 2: 1 min 09 sec Benchmark 3: 5 min 23 sec Benchmark 4: 10 min 30 Sec Stock bios settings upped the voltage like a lunatic in a new special mode called EPD for the 13900k, so disable that if you play like benchmark 4.

Benchmark 2 and 4 also crashed for me. Ran Cinebench just fine, ram also stock, updates my game drives & drivers, verified the game cache, etc etc etc etc.@Brother Ben Would you help us please?

So sad that Intel has - according to bios manufacturers - cancelled Intel Digital Linear Voltage Regulator. 20% better power management just went down the bin, and was cancelled very late into production. Now that Meteor Lake looks for desktop looks delayed to 2024 or even cancelled, a AM5 mobo would for the 3D 7000s at least looks better for FM as the temp issues will remain. I'm still buying the 13900ks as I have a z690 already, but I wouldn't advise it purely for this game anymore. If anyone wants me to benchmark the difference between a 12900ks and 12900k/13900k here, or wants me to benchmark everything before getting the 13900ks, let me know.

Multiple people already gave you the answer. 1. You'd gain 30 sec per week in FM, at most. You don't use the pc really enough to notice the PCI, nvme or usb upgrade. 2. You don't have to do it for better ram speeds in this game, as this game doesn't care about that. 3. Davinchi resolve is gpu-intensive. It doesn't care that much about your cpu. So. If you earn your money with Davinchi, I'd just upgrade the MB, CPU and actually, most importantly, a much better GPU if you do earn money with editing. Otherwise, If you wanna gain as much with just a cpu upgrade, go to the cpu support page of your bios manufacturer and check the best supported bios (I guess the 5950x). The data would advise you btw to update to a 5800x3d (if you play other games) or 5900. Conclusion, I wouldn't pay $400 for a cpu upgrade to gain 30 sec in a game if I don't play other games or need it for work.

Your CPU in HWinfo64 shows it slows itself down for a technical reason briefly (IA limit reason means very simplified: it defends its longevity by slowing down the core speed), that is normal behavior. It's a design choice. For the overwhelming majority of times, It's normally at 3,2, with that energy profile it now appears good. It's not stuck, as it boosted to 3,6 for one core and higher than 3,2 for multiple cores for a less than a few minutes. Powermonger's answer is correct. Your motherboard doesn't support XMP, due to costs, time, it's an old thing, made by lenovo for their own hardware only, etc. So no, your ram will stay stuck on your current speed. It's still the normal recommended speed for your cpu though, so it's not like you're losing performance. For the 1% extra performance in case a new bios profile unlocks XMP by installing a new bios, I wouldn't advise upgrading to new bios anyway. Because new motherboards have a reserve bios, so in case your power goes out or smth during your bios update, you can still start it up. With your rig, if something happens out of the blue, you're out of luck. If you don't care about losing and killing the rig, try to find a new bios by going to the Lenovo site, click on support top left and “detect product”. The bios download that may show in a long list has a manual explaining the process further. I btw can't find your model and bios file for you because the motherboard name is partly hidden in hwinfo, and I don't know all Lenovo home-made mobo's from the top of my head. And tbh, it's such an old model, not easily found anymore, if I give you the wrong bios file, I'd feel so sorry I'd send you a new pc.

@Earnie is God! Hi, So, to clear some things up: 1. XMP doesn't care about overclocked or “non-overclockable” CPUs, that should just work. 2. Your RAM profile without xmp should just dial back to an old slower version when it doesn't work, but that is not the same number as CPU speed. 2. 6500 and a 4770 processor both officially can't be overclocked (only with very old bios' firmware, long story, doesn't seem still enabled with you here). 3. Question: how did you see your processor speed only hit 2133Mhz. As I think you've confused your RAM profile with your CPU speed here? If not, follow step 2; Step 2: turn on “high performance” in your “energy management” app in windows like Ben always tells us. If there is another plan there like “full performance” “max power”, "Lenovo/MSI/Evil Inc Gamer 1000%" don't use it. Can you download the program called hwinfo64 and run benchmark “A” again? Take a screenshot when you're done of that program. Like this:

Didn't even ask your tech support guy. 7950x looks underwhelming. 12900k still outperforms it the majority of games tested on nearly all resolutions. Only advantage is work applications. Most reviewers are not positive. AMD probably has gone all in on 3D chips. Really hope it's just an early adaptor problem. I had FX bulldozer, this feels the same.

Great news Ben! Regarding ideas, saw some suggestions in this thread but; benchmarks have shown Windows version doesn't matter (only turning on Virtualization-based Security matters), Ram speed or latency is for 95% of test cases within the margin of error. Clock speed only matters for heavy modders and even apple is maxing out clock speed. Only thing I'd add is a TLDR section for people with short attention spans and those not liking charts; “I usually play with (amount) of leagues, with (amount) in full detail. What CPU do I need?“.

After a mobo switch, a cooling switch (vrm, cpu loop to aio) and a ram oc to (4400mhz, 128 GB, 32cl), I figured out that the non-convex coldplate of the water cooler messes up the alder lake IHS bending problem even more. So, a 3rd party contact frame or an air cooler with convex cold plate will cool the 12900k(s) way better, causing the TVB to boost the first two benchmarks way better under 50-70 degrees Celsius. So, if you're gonna play FM22 or FM23 not modded and with less than 25 leagues and less than 200k players with a I9: You'll have worse performance if you don't use a cooler with a convex coldplate, don't use a contact frame (or thick washers under your IHS) or don't undervolt your cpu. Sure, Raptor lake will have per core and package oc TVB support, but that just looks like a compensation for running 8 more E-cores as it's the same design and a stopgap according to Intel = same cooling problems expected. Not to deny, FM has excellent multicore support, this is more on Intel trying to rule 1080p benchmarks by squeezing out lunatic thermals. So as I'm stuck trying to finish my career with 200 countries hoping I'll finish it before Raptor lake launches so that I can play less CPU-intensive games and wait for the 13900ks around March/May '23. I'm predicting, I could redo the benchmarks, but after testing in Cinebench, I think I'll prob be just 5 to 10 seconds faster in the first two benchmarks, so maybe the trend of only better perf in heavier loads is a better warning for users visiting the first page. I'm thinking this is way too complex tech talk for most so; Simple TLDR: If you buy Rocket Lake or Alder Lake mostly for FM23 and don't mod or load all leagues like in benchmarks C or D, just buy an I7.

Type: Desktop Model: Custom CPU Model: 12900KS CPU Base Frequency: 3,4 Ghz (P-Cores), 2,5 GHz (E-Cores). CPU Turbo Frequency: 5,7 GHz (P-cores) - 5,2 All-core - 4,1 GHz (E-cores) RAM: 128GB RAM Clockspeed: 3600Mhz GPU: AMD Radeon 6900 XT Graphics Level in 3D: Very High Storage Type: SSD (Samsung 980 Evo Pro 2TB) Benchmark A: 41 sec Benchmark B: 3 min 33 Sec Benchmark C: 5 min 55 Sec Benchmark D: 15 min 29 Sec Hey @Brother Ben, how are you doing? Hopefully this benchmark can still be valuable. I'm hitting the E-TVB limit in Benchmark 2. So it doesn't boost the one core running the benchmark at 90% because it appears to load very small workloads on other cores. I've tried raising the temp limits for E-TVB, but raising it by 10c-20 doesn't make it difference, and raising it by 30 or disabling or playing with TVB optimizations or E-TVB and OCTVB caused a terror of BSODs, So... yeah I'm giving up on that one.