Sooo, you put a much higher-wattage CPU in a tiny case with no airflow, put an undersized cooler on top of the CPU, further gimped it with a weak GPU, and you regret it. Shocking turn of events.
The 9950X is $649.99 on Amazon and pulls about 200 Watts in Cinebench. This processor is both a terrible match for Bitwits cooler and is complete overkill for his old graphics card. You didn't even have to put this together to know that it would be a stupid mix of parts.
@@tanmaypanadi1414 He just knew that the RUclips algorithms would boost up any video with 9950x in the title even if there was no effort put into the internal logic of the build.
Its gonna be ass unless they fundamentally change the architecture. Dont expect a 9800X3D to be more than 4-5 percent faster than the 7800X3D. Honestly just buy a 7800X3D unless youre on AM4 or LGA1700 already then just be happy with what you got.
I see why this video is getting criticism and a decent amount is valid. The title is a little bit clickbaity --- the system definitely was not meant for such a power hungry chip but how do we know that until someone like Kyle takes his time to present us his experience. As an experiment, this video holds up, but everything could've been presented a little cleaner. As a community we shouldn't be harsh when someone tries to show their experience, even if it's not the most efficient decision. Keep up the good work
I miss your vlogs, Kyle. I don't care what you're doing. Fixing a hole in a wall, going to Home Depot, eating some weird green soup. You're a natural vlogger and entertainer. Please, consider bringing back some random stuff for us
@@BitwitI kinda just prefer the whatever you feel like doing when you feel like doing it. It’s a nice vibe, today I see sandwich vlog tomorrow I see 9959x sucks video. I’m here for it lol
Wrong. They marketed 9700x to match 7800x3D or even beat it. So, 9950x was implied to obliterate 7800x3D and we know how that went, based on countless reviews Check those slides from AMD at their presentations
@@OptimizingNetworkyep and falsified intel testing as well because they modified the profile to 135w. So I’m sure they modified zen4 as well to make their claims. Not impressed
9950X will beat 7000 X3D's, just wait for the new Agesa update. Also test these 9950X again on a newer X870 motherboard. Plus it comes with OC/PBO so you can overclock it and it will be getting very close to X3D performance if not beat it. While the 9000 X3D will be ahead of 9000 series and 7000 X3D's. But since the 9000 X3D are more power efficient the performance is slightly lower thus it won't match the 285K/9950X but instead will match the performance of 14900K/7950X in multi core workloads.
Im very glad you made this video. Im in a process of getting ready to build myself a powerful machine in a fractal terra case, and I was scared of 9950x not being able to cool at all, but seeing how you put it into 4l case with a very subpar cooler, im kinda impressed how well it perform still. Im planning on waiting until 870 mini-itx mobos become available to pull the trigger, terra fits my components very nicely (including 3090), but I was very afraid of cpu thermal limit before I saw this. Currently the cooler im looking at is Thermalright AXP90-X53 FULL copper, though it seems I gotta get it from aliexpress as its not available here. On paper it seems like an optimal choice(though I dont believe their claims that copper variant can dissipate 250watts lmao). Still, a bit of undervolt and underclock wont hurt much for my usecase.
When you said most people find the GPU pairing weird. I was like, "I do that too" lol. I have 3090's and bigger in other machines with lower CPU's, or vice versa. Just depends. I use my PC's for more work than play. If you're gaming, having lots of threads means you got work to do, not play. You'll save a lot of money and get equal performance with less cores. Thus when I hear the _950X series in any build. I instantly think of 2 different things: 1.) You do actual workstation level work. And need the high core count and fast single thread speed. 2.) You got it purely to brag or don't understand you could have spent 1/2 the money to get literally the same gaming performance.
Funny you say you wanted the 9950x for compute and work but then you say the 7950x3d wins when it obviously does not win for productivity and compute work like you said you wanted.
First off that is a sick micro build, I would love to try something like that. Also, nice thermal paste cactus, by far the best paste application method I've see yet. With that out of the way, I hate to say it buy the 9950x vs the 7950x3D is kind of a head scratcher. I have seen in some benchmarks that the 9950x does do a bit better here and there in some multicore workloads due to generational improvements. But it seems to not do nearly as good in gaming as the 7950x3D but a large amount. It really makes it feel like the best all around CPU for mixed workload is the 7950x3D and the 9950x would only be the better choice if you really do not intended to game on it what so ever.
*I cannot wait for the 9950X3D it should be close to* *14900K/7950X multi core performance while performing* *better in games then 7800X3D & 7950X3D which is CRAZY!!*
@@ilkercan84 9950X3D is better than all. In Geekbench the 9950X was slightly faster then 285K by low margin but 9950X3D will be much faster. Meaning it will not only beat 9950X in multicore but also beat 9800X3D/285K in single as it has higher single core performance. *9950X3D is a BEAST/KING of CPU'S!*
So the AMD boost algo will take into account: temps, average current limit (TDC), peak current limit (EDC), power limit (PPT) and the work load its running, to determine its boost speed. 6:17 CPU Boost clock overrride, set to Enable and set to 200, which is the max it will go. The CPU will now boost to 5950 (5750+200) when close to idle. Ideally set PPT to unlimited, and then find the sweet spot for EDC and TDC. The X3D chips do not need a voltage limit, because temp is their limiting factor due to the poor cooling performance. After figuring out EDC and TDC, you can set PPT really far and the 7950X3D uses whatever it needs. After EDC and TDC searching for the sweet spot, this largely determines its temperature properties. And after that you ll look for a PPT limit that is sensible. At some point, its going to brush against temp limits and diminishing returns on clock speed (faster = more leakage = hotter chip). You can then wonder if you think the next 20W are worth the speed boost. That said, temperature is usually not your limiting factor, because PPT already doesn't scale well. It would therefore be weird running the chip under what it is capable. Stock temp limits are just fine.
idea for the next video title: "Done regretting my upgrade to 9950X......" and in 10 min video install back the 7950X3D - will be very helpful, especially for people thinking to upgrade who have already watched the initial reviews and are currently gathering additional details and potential issues about the CPU before purchasing.
@@thomasnostradamus it needs to be checked whether it's the same or not, I remember that somewhat recently Jayz had stability issue with AMD, but that's a completely different case than what's in this video. I get that clickbait titles work, but some just go a bit to far, especially when it doesn't live up to the expectation.
I have the 7950X3D and I overclocked the crap out of it, it does everything I need. (video encoding/editing/gaming/image editing/programming/3D modelling ..etc)
bro is testing on noctua l9a 16 cores 120w max 7950x3d and 9950x at 150w and saying that first was cooler… Bruh, this cooler was designed for 65w tdp, why r u setting ptt at 150w, def it will throttle to see where is the peak. Also default Tjmax on 3dvcache chips is 89 and non 3d is 95. Try to use processors in 65w profile and see whats better, your tests arent legit enough, U even didt show how much is average consumption, only peak. Bruh… Its cap.😂
Man, you just can’t beat the X3D. I’m loving my 7800x3D and it’s in a Fractal Terra case. I’m going to have to try undervolting it later for sure and see what happens. Thanks for the tips. 😊
There's something wrong with my 9950X then... You got 40k in Cinebench, in a SFF build. I got 36k with a full ATX case and really solid cooling! I've ruled out every possible issue, so the CPU must be under spec...
@@tyler6602 I started with stock settings + fresh Windows 11, installed chipset drivers, bios, checked RAM was running 6000 CL30 etc. When I got 36,500 score I tried using Curve Optimiser on -30. Score only reaches 39,000 but the system is unstable. And on stock it pulls the full 200W PPT and doesn't thermal throttle at all
My point of contention here is that you're missing a huge chunk of the puzzle when you skip Project Hydra 1.8E That's if you wish to be able to take full advantage of Ryzen. Consider the convenience when you need it. It's there to give you 3 performance choice at the touch of a button after HOC table creation. Plus disable things you don't want running, as a most pleasant reward for using it. Remember not to be a baddie by exceeding the TDP
Hey @Bitwit, I’m new to the channel, but I’ve had a crappy computer, and would like to ask your opinion on what parts just the pc (not the monitor). Parts that aren’t expensive or anything fancy, just enough for a good gameplay, no stutters, no packet loss, not pings or something, I don’t know anything about computers. And if you could or if it’s okay, would you be able to list the sites or places you’d go to in order to purchase them.😊
I undervolted my Ryzen 5 5600x because it was getting SUPER hot with the cooler it came with. Which I thought would have been adequate in my Arc Midi case with (6) 140mm fans. I lowered it to 1.1V and it's been stable and dropped 15C. Also going from a GTX 1080Ti to an RTX 4070 Ti Super introduced a ton more heat.
When people want to put top end processors into a mini cases (I have a good laugh at people who do this), you are going to have thermal problems. Just because you can, don't mean you should! if you do, expect the processor to perform way below optimal levels and it more than likely it will have throttling problems.
I really hope you're going through the same setup process as the other x3d models are required if you're not you're losing up to half of the CPU performance
hi nice video. If I were to build a new hypervisor server, what was your cpu recommendation for hypervisor use, which is usually idle. so idle power consumption has some significance. but should also have the ability to boot windows if I'm going to use it for gaming. I can't decide which of these Ryzen 9 9950x, 9900x, 9750x, 9750x3d or 7900x. The idea is that it should be forward-looking at least 5 years ahead but at the same time not lose a lot of unnecessary electricity use I think the choice is between 9950x or 9750x3d. but can I get the 9950x to use as little current as the 9750x3d by changing the tdp to 120 or 150 and increasing it to standard if I need a little extra cpu power
I have a similar setup, an ITX setup, but I used the SSUPD Meshroom, and have a 7950x and 7900xtx. I ALWAYS upgrade every generation, but I think for the first time in my life of building pcs (20+ years) I will not be "upgrading" and will stick with my 7950x.
This multi CCD situation drove me crazy. I figured get the 9900x then saw benchmarks that 7800x3d was winning So decided to go with 7900x3d because it has extra cores and only $30 more than the 7800x3d Then found out about the muti CCD layout and the 7800x3d beats the 7900x3d in gaming So I returned my 7900x3d and ordered a 7800x3d
So you never needed the extra cores. It's practically irrelevant that the 7800X3D is marginally faster at games, if you had good reasons to have more cores. The boost in multi-core is at a way higher margin than the loss in gaming is. Unless CS2/COD is life, you weren't going to actually miss those frames.
9950x is a weird CPU and it replaces the 7950x non-3d but for some reason in certain test the 7950x non-3d outperforms or the results are so close that it's not worth the upgrade. I'm glade I got the 7950x which is around $500 as of now. I think most of the issue with the 9950x is the new core parking feature which puts one of the CCD's to sleep as what JayztwoCents has found out.
Seeing Ryzen 9000 benchmarks lately and since you compared with the 7950X3D to the 9950X the results where kinda expected. The coloring of the graphs could be better like using different colors for the processors then a lighter/darker shade for the alternative results whether they be PBO, temp limited, undervolted, etc.
Me too. Just got a 7800X3D with an X670E-WiFi Plus and 64GB Dominator Titanium 6000MHz RAM yesterday and it's going in tomorrow. Not going to wait for the 9800X3D. It could be another 8 months away at this point.
The greys used on the graphs are pretty distinct, not sure why they would be an issue for the average person. It makes sense for people who are colorblind to have issues with it but I find it way more likely that there are many people with some crazy color balancing on their monitors lol
I'm not very tech-savvy but got a question, I'm benchmarking cyberpunk 2077 with ray tracing at its highest don't have path tracing on at 1080p, is it okay for my GPU to be using in the high '90s. I've heard mixed results where some people say why wouldn't you want to utilize your GPU for everything it's got, and others that say if you want your GPU to be long lasting you shouldn't have it at that high.
Not really an apples to apples comparisson, you should rerun the tests when they release the 9950X3D, the results were pretty much predictable which would perform better in games. in my SFF I have a really simple Ryzen 5 5600 and a 1660 Super, nothing high end but enough for what I want to do with it. My main rig weighs a ton so that small shoebox of a case is easy to transport.
You did not undervolt the 9950X, you gave it more headroom in the same power envelope of TDP170W/PPT230W (by curve optimization) and it performed better. IF you had lowered it's power envelope to TDP120W/PPT162W (the same as 7950X3D) it would have had the same power consumption as 7950X3D and better thermals compared to default 9950X (without the need to thermal limit it in bios) - hence it would have been a better comparision.
Why are people still undervolting the old way with these CPUs? You're losing a lot of efficiency and stability by doing this, use curve shaper instead.
I was going to wait for it. Then I said forget it. I just got a 7800X3D. I'm coming from a 3900X so I should see a lot of difference and I can either wait for Gen 6 or if the 9800X3D is that much better than the 7800X3D I can always get it later. My 3900X is bottlenecking my 4080 OC today, not in 6 months or whenever the 9800X3D is coming out.
@@HerrinSchadenfreude Oh that's a great upgrade that I can agree with since you have a 4080. I meant from his POV where he already had a 7950X3D. It was already shown in the reviews that the X3D chips for the 7000 series was better than the 9000 non x series.
How much would Cinebench score be if I limit 9950X to 70 C . Or wattage at that point. I have ordered 9950X for long term (15y) processor. I start with air cooling as saving money for some water cool solution. It is for programing and DAW use so it has those number crunch qualities.
Ideally you should've compared it to the 7950X, not the 7950X3d as its known that the 7950x3d runs alot cooler and draws way less power than the 7950X. Thus the 7950X would've been a much better comparison as the same test between the 7950X and 7950X3d would also have big differences in the results for heat and power consumption. You need the 9950X3d for this comparison to really mean anything.
Interesting build. But yeah: for productivity and gaming a big case is still very handy. I'm on the opposite site with an ancient huge Thermaltake level 10 gt build. Will probably downsize with my next build though. I have the also ancient 3950x and don't often need more performance. So yeah: I might go for a three times 8 core cluster this time. Would be fun!
Did you log in as Administrator on your computer or did you log in as the a user with Administrator access? Logging in with Administrator and re-running your tests likely will give you better results.
How does yalls CPU Temps average at 73? Mine never exceeds 65. Am I doing something wrong? It makes me feel like I may not be using my components to their potential.
The 9950 is a productivity forward cpu primarily for compute it’s not specifically a cpu for generalized all around computing- Dude , ya gotta know why you’re buying a specific CPU
Honestly I was expecting a lot worse since this is, well, a worst case scenario. Sure something like this is going to get handicapped either thermally or power draw wise w/ eco mode, but this experiment was never about practicality; its about that 1 niche “what if” case that we think about but know we’ll never get to see because it’s so niche
I don't like people gimping the Ryzen 9 7950X3D without showing it under a manual FCLK overclock. You can run a 7950X3D 8-core chip let using X3D cache at 5.450Ghz while the 7800X3D can only achieve 5.250Ghz. An overclocked 7950X3D CAN be faster in gaming than a 7800X3D, but it never will be unless a person spends the time and manually overclocks a locked processor. I just dislike people showing a handicapped 7950X3D because it makes other results look better while the 7950X3D is truly a beast of a processor while overclocked. 5.450Ghz on 8-X3D-Cores and 5.950Ghz on 8-Non-X3D-Cores. In addition, the dual-chiplet design of 7900X3D and 7950X3D allows for double the overall memory bandwidth between infinity fabric and memory, allowing for users to actually benefit from running DDR5 RAM above 6400MT/s, although running out-of-sync has diminishing returns and does not show much benefit until 7200MT/s WHILE also having tightened primary and subtimings. But, if you do the manual overclocking, it CAN be faster than 6400MT/s, it just takes immense effort to tighten your timings and tightened timings also increase CPU heat generated.
*MIXED feelings on this video...* If you don't watch the ENTIRE video it's easy to trash Bitwit. He did say he didn't need the extra multi-thread processing for his use case. So you paid MORE for a product that wasn't suited for your use case and then REGRET your decision? See, that's just strange. I'd rather you just did a straight-up benchmarking video and then talk about the pros and cons. Sort of the same thing, but it changes the tone. This is a little NEGATIVE when it's actually a great product... plus, there are performance issues Windows update may address but I guess you didn't know this 5 days ago... anyway, $125 more for 21% higher performance (and that may go closer to 30% once Windows is updated) is actually quite good... you have to think about this compared to the ENTIRE computer cost. If you spent $2125 vs $2000 then you spent 6.25% more money to get 21% higher performance. THAT is what you should be talking about in a more positive way. IMO... if it's mostly meant for gaming then you'd be looking at an R7-7800X3D or maybe waiting for the R7-9800X3D at this budget... So BITWIT is like "it doesn't really benefit GAMING and I personally don't QUITE need all the processing power for multi-threading performance." Ya, this product isn't meant for YOU then.
Why are there no renewed reviews with the new update that fixes the performance issues? It a pretty big deal with these and no one is re benchmarking after the update.
Almost impossible to visually see the color differences on the graphs in a room with natural light. Would be nice to use colors that are not very close to each other, especially for those with color blindness.
Which set-up is better.? 14700k (or 9700x) + ARC a770 (16gb) Vs 14700k (or 9700x)+ rtx 3060 (12gb) Vs 14700k (or 9700x) + AMD 7800xt (16GB) For Davinci Resolve My Canon R6 m2 shoots 4k in h.264 n h.265
I updated my Gigabyte b650m Elite Ax to the latest BIOS and upgraded from a 7700X to a 9900X and now it won't boot. It just hangs with the DRAM light. I dropped to a single stick and waited forever, still can't get in. I cleared CMOS several times and even reverted back to the 7700X but my board seems to be completely hosed. Freakin blows
I have an old 5800X and I also undervolted using PBO II. My cores will take a negative 25 as well. At -30 it's not stable and I have neither the time nor the patience to do a by core optimisation. P.S. Kyle, get a haircut man! 😄
some of you mentions why put a power hungry cpu into a tiny case with that low pro cooler. this is actually interesting test, more real world scenario especially those need powerful cpu in a sff for their always on the go job, rather than open test bench in a controlled environment.
Sooo, you put a much higher-wattage CPU in a tiny case with no airflow, put an undersized cooler on top of the CPU, further gimped it with a weak GPU, and you regret it.
Shocking turn of events.
imaging a tech tubing doing this so a consumer doesn't make the mistake. shocker
@@mindurbusiness-b3u Imagine being and cosumer and seeeing, this type of video for the 100th time...
🤣
I actually like seeing it run this way, it's an alternative to the big open bench tests
lolllllllllllllllll
Am I the only person colorblind just right to not be able to read any of those charts. I can't make out the 2 shades of gray.
Not colorblind but yeah those grays (2:34) aren't exactly the greatest choice. :-/
Still, the length of the bar and the numbers at the end are more than enough to understand the data conveyed.
Could just be a monitor calibration problem, check your contrast levels.
@@mark.082 Extrapolate!
my eye sight might be an issue. but the contrast could be better for some of the grey charts @@Skyflairl2p
The 9950X is $649.99 on Amazon and pulls about 200 Watts in Cinebench. This processor is both a terrible match for Bitwits cooler and is complete overkill for his old graphics card. You didn't even have to put this together to know that it would be a stupid mix of parts.
He probably wanted to just nerd out and build something with the new cpu.
@@tanmaypanadi1414 He just knew that the RUclips algorithms would boost up any video with 9950x in the title even if there was no effort put into the internal logic of the build.
I don’t know if I would call that an old gpu, it’s pretty recent. Underpowered for that cpu sure
It's actually impressive that the performance is still as good as it is with that cooler. The Intel version is subpar for even a 12600K.
$1200AUD here on Amazon
I’m just waiting for the inevitable X3D at this point. No reason to jump on this one from my 5950x
Isn't happening this time around
X3D is sweet. Mo cache mean mo betta
@@skunk2spec489 It literally is. Probably before the end of the year.
Just boycott AMD, they should've release X3D first, but no, they pulled an Nvidia Super stunt.
Its gonna be ass unless they fundamentally change the architecture. Dont expect a 9800X3D to be more than 4-5 percent faster than the 7800X3D. Honestly just buy a 7800X3D unless youre on AM4 or LGA1700 already then just be happy with what you got.
I'm not colorblind, but these graphs using different shades of grey makes them very hard to understand!
seriously lol
an apt metaphor for this particular CPU tho
I see why this video is getting criticism and a decent amount is valid. The title is a little bit clickbaity --- the system definitely was not meant for such a power hungry chip but how do we know that until someone like Kyle takes his time to present us his experience. As an experiment, this video holds up, but everything could've been presented a little cleaner. As a community we shouldn't be harsh when someone tries to show their experience, even if it's not the most efficient decision. Keep up the good work
I miss your vlogs, Kyle. I don't care what you're doing. Fixing a hole in a wall, going to Home Depot, eating some weird green soup. You're a natural vlogger and entertainer. Please, consider bringing back some random stuff for us
Are you suggesting I answer my true calling and change this to a vlog channel? 🤔
@@Bitwit YES! Or daily vlog on WRKHRSE
@@Bitwitor both, don't forget to include Lyle 😅
@@BitwitI kinda just prefer the whatever you feel like doing when you feel like doing it. It’s a nice vibe, today I see sandwich vlog tomorrow I see 9959x sucks video. I’m here for it lol
This is a "How does Noctua NH-L9A handle ...." video. 😁
iye boss
Hawk Tuah is a trending.... cooler.
are u kidding? 9950x is never supposed to win in Games vs a X3d model. AMD created internal competition.
amd created bunch of lies at ces
Wrong. They marketed 9700x to match 7800x3D or even beat it.
So, 9950x was implied to obliterate 7800x3D and we know how that went, based on countless reviews
Check those slides from AMD at their presentations
@@OptimizingNetworkyep and falsified intel testing as well because they modified the profile to 135w. So I’m sure they modified zen4 as well to make their claims. Not impressed
9950X will beat 7000 X3D's, just wait for the new Agesa update.
Also test these 9950X again on a newer X870 motherboard.
Plus it comes with OC/PBO so you can overclock it and
it will be getting very close to X3D performance if not beat it.
While the 9000 X3D will be ahead of 9000 series and 7000 X3D's.
But since the 9000 X3D are more power efficient the performance
is slightly lower thus it won't match the 285K/9950X but instead
will match the performance of 14900K/7950X in multi core workloads.
@@kealz318 just dream more!!same bullshits AMD talks when Bulldozer came out
I'm not sure what you were expecting comparing an x3d chip to it
you're actually stoked on it. i can tell that you can't believe how well it did here and that you love it so so much.
Have a shot every time Kyle spins the cooler's fan.
Im very glad you made this video.
Im in a process of getting ready to build myself a powerful machine in a fractal terra case, and I was scared of 9950x not being able to cool at all, but seeing how you put it into 4l case with a very subpar cooler, im kinda impressed how well it perform still.
Im planning on waiting until 870 mini-itx mobos become available to pull the trigger, terra fits my components very nicely (including 3090), but I was very afraid of cpu thermal limit before I saw this.
Currently the cooler im looking at is Thermalright AXP90-X53 FULL copper, though it seems I gotta get it from aliexpress as its not available here. On paper it seems like an optimal choice(though I dont believe their claims that copper variant can dissipate 250watts lmao). Still, a bit of undervolt and underclock wont hurt much for my usecase.
The fact that there are people that would put a 4090 in a case this small is the first sign we are doomed
When you said most people find the GPU pairing weird. I was like, "I do that too" lol. I have 3090's and bigger in other machines with lower CPU's, or vice versa. Just depends. I use my PC's for more work than play. If you're gaming, having lots of threads means you got work to do, not play. You'll save a lot of money and get equal performance with less cores. Thus when I hear the _950X series in any build. I instantly think of 2 different things:
1.) You do actual workstation level work. And need the high core count and fast single thread speed.
2.) You got it purely to brag or don't understand you could have spent 1/2 the money to get literally the same gaming performance.
nice to see you! we miss you!
Next video will be with installing v8 engine in B-class car?
7950X3D has Tjmax 89c and the 9950X has Tjmax 95c. So both chips run to their limit that they were designed. I do not see an issue...
Holy bawls you back doing tech and I missed it? Shweet!
I thought the same thing how did I miss that?!
Funny you say you wanted the 9950x for compute and work but then you say the 7950x3d wins when it obviously does not win for productivity and compute work like you said you wanted.
First off that is a sick micro build, I would love to try something like that. Also, nice thermal paste cactus, by far the best paste application method I've see yet. With that out of the way, I hate to say it buy the 9950x vs the 7950x3D is kind of a head scratcher. I have seen in some benchmarks that the 9950x does do a bit better here and there in some multicore workloads due to generational improvements. But it seems to not do nearly as good in gaming as the 7950x3D but a large amount. It really makes it feel like the best all around CPU for mixed workload is the 7950x3D and the 9950x would only be the better choice if you really do not intended to game on it what so ever.
*I cannot wait for the 9950X3D it should be close to*
*14900K/7950X multi core performance while performing*
*better in games then 7800X3D & 7950X3D which is CRAZY!!*
7950x3d is better than all of those
@@ilkercan84 9950X3D is better than all.
In Geekbench the 9950X was slightly faster then 285K by low margin but 9950X3D will be much faster. Meaning it will not only beat 9950X in multicore but also beat 9800X3D/285K in single as it has higher single core performance.
*9950X3D is a BEAST/KING of CPU'S!*
So the AMD boost algo will take into account: temps, average current limit (TDC), peak current limit (EDC), power limit (PPT) and the work load its running, to determine its boost speed.
6:17 CPU Boost clock overrride, set to Enable and set to 200, which is the max it will go. The CPU will now boost to 5950 (5750+200) when close to idle.
Ideally set PPT to unlimited, and then find the sweet spot for EDC and TDC. The X3D chips do not need a voltage limit, because temp is their limiting factor due to the poor cooling performance. After figuring out EDC and TDC, you can set PPT really far and the 7950X3D uses whatever it needs.
After EDC and TDC searching for the sweet spot, this largely determines its temperature properties. And after that you ll look for a PPT limit that is sensible. At some point, its going to brush against temp limits and diminishing returns on clock speed (faster = more leakage = hotter chip). You can then wonder if you think the next 20W are worth the speed boost.
That said, temperature is usually not your limiting factor, because PPT already doesn't scale well. It would therefore be weird running the chip under what it is capable. Stock temp limits are just fine.
Hello Bruv, Long Time No See. Nice To See You Once Again
idea for the next video title: "Done regretting my upgrade to 9950X......" and in 10 min video install back the 7950X3D - will be very helpful, especially for people thinking to upgrade who have already watched the initial reviews and are currently gathering additional details and potential issues about the CPU before purchasing.
JayzTwoCents does something similar with his AMD summer upgrade, then a “Why I’m switching back to Intel” video in the fall.
@@thomasnostradamus it needs to be checked whether it's the same or not, I remember that somewhat recently Jayz had stability issue with AMD, but that's a completely different case than what's in this video. I get that clickbait titles work, but some just go a bit to far, especially when it doesn't live up to the expectation.
I have the 7950X3D and I overclocked the crap out of it, it does everything I need. (video encoding/editing/gaming/image editing/programming/3D modelling ..etc)
Came for the Lyle, stayed for the Kyle
bro is testing on noctua l9a 16 cores 120w max 7950x3d and 9950x at 150w and saying that first was cooler… Bruh, this cooler was designed for 65w tdp, why r u setting ptt at 150w, def it will throttle to see where is the peak.
Also default Tjmax on 3dvcache chips is 89 and non 3d is 95. Try to use processors in 65w profile and see whats better, your tests arent legit enough, U even didt show how much is average consumption, only peak. Bruh… Its cap.😂
Man, you just can’t beat the X3D. I’m loving my 7800x3D and it’s in a Fractal Terra case. I’m going to have to try undervolting it later for sure and see what happens. Thanks for the tips. 😊
I never regret watching a Bitwit video
There's something wrong with my 9950X then... You got 40k in Cinebench, in a SFF build. I got 36k with a full ATX case and really solid cooling! I've ruled out every possible issue, so the CPU must be under spec...
Are you using pbo?
@@tyler6602 I started with stock settings + fresh Windows 11, installed chipset drivers, bios, checked RAM was running 6000 CL30 etc. When I got 36,500 score I tried using Curve Optimiser on -30. Score only reaches 39,000 but the system is unstable.
And on stock it pulls the full 200W PPT and doesn't thermal throttle at all
My point of contention here is that you're missing a huge chunk of the puzzle when you skip Project Hydra 1.8E
That's if you wish to be able to take full advantage of Ryzen.
Consider the convenience when you need it. It's there to give you 3 performance choice at the touch of a button after HOC table creation.
Plus disable things you don't want running, as a most pleasant reward for using it. Remember not to be a baddie by exceeding the TDP
Hey @Bitwit, I’m new to the channel, but I’ve had a crappy computer, and would like to ask your opinion on what parts just the pc (not the monitor). Parts that aren’t expensive or anything fancy, just enough for a good gameplay, no stutters, no packet loss, not pings or something, I don’t know anything about computers. And if you could or if it’s okay, would you be able to list the sites or places you’d go to in order to purchase them.😊
Great! I absolutely love small factor solutions.
I undervolted my Ryzen 5 5600x because it was getting SUPER hot with the cooler it came with. Which I thought would have been adequate in my Arc Midi case with (6) 140mm fans. I lowered it to 1.1V and it's been stable and dropped 15C. Also going from a GTX 1080Ti to an RTX 4070 Ti Super introduced a ton more heat.
I guess this is a very good result for games man
When people want to put top end processors into a mini cases (I have a good laugh at people who do this), you are going to have thermal problems. Just because you can, don't mean you should! if you do, expect the processor to perform way below optimal levels and it more than likely it will have throttling problems.
I really hope you're going through the same setup process as the other x3d models are required if you're not you're losing up to half of the CPU performance
hi nice video. If I were to build a new hypervisor server, what was your cpu recommendation for hypervisor use, which is usually idle. so idle power consumption has some significance.
but should also have the ability to boot windows if I'm going to use it for gaming.
I can't decide which of these Ryzen 9 9950x, 9900x, 9750x, 9750x3d or 7900x.
The idea is that it should be forward-looking at least 5 years ahead but at the same time not lose a lot of unnecessary electricity use
I think the choice is between 9950x or 9750x3d. but can I get the 9950x to use as little current as the 9750x3d by changing the tdp to 120 or 150 and increasing it to standard if I need a little extra cpu power
I have a similar setup, an ITX setup, but I used the SSUPD Meshroom, and have a 7950x and 7900xtx. I ALWAYS upgrade every generation, but I think for the first time in my life of building pcs (20+ years) I will not be "upgrading" and will stick with my 7950x.
I'm curious with the x870 boards and the faster ram would it make the choice for the 9950x more viable.
This multi CCD situation drove me crazy. I figured get the 9900x then saw benchmarks that 7800x3d was winning
So decided to go with 7900x3d because it has extra cores and only $30 more than the 7800x3d
Then found out about the muti CCD layout and the 7800x3d beats the 7900x3d in gaming
So I returned my 7900x3d and ordered a 7800x3d
So you never needed the extra cores. It's practically irrelevant that the 7800X3D is marginally faster at games, if you had good reasons to have more cores. The boost in multi-core is at a way higher margin than the loss in gaming is. Unless CS2/COD is life, you weren't going to actually miss those frames.
What a cool little box, though!
If you undervote the CPU in the processors, how does effect on high-end gaming? And rendering like Adobe premiere and other stuff exporting
9950x is a weird CPU and it replaces the 7950x non-3d but for some reason in certain test the 7950x non-3d outperforms or the results are so close that it's not worth the upgrade. I'm glade I got the 7950x which is around $500 as of now. I think most of the issue with the 9950x is the new core parking feature which puts one of the CCD's to sleep as what JayztwoCents has found out.
"I am not ready to be without you"
Seeing Ryzen 9000 benchmarks lately and since you compared with the 7950X3D to the 9950X the results where kinda expected.
The coloring of the graphs could be better like using different colors for the processors then a lighter/darker shade for the alternative results whether they be PBO, temp limited, undervolted, etc.
Still sitting on a 3900X, still on the fence about upgrading. May wait for Zen 6 "Medusa", or see how Qualcomm desktop chips perform.
Me too. Just got a 7800X3D with an X670E-WiFi Plus and 64GB Dominator Titanium 6000MHz RAM yesterday and it's going in tomorrow. Not going to wait for the 9800X3D. It could be another 8 months away at this point.
@@HerrinSchadenfreude 8 months wait for 5% speed boost. Low hopes after seeing 9000 lineup. 7800x3D will be fine for the next 5 years at least.
The greys used on the graphs are pretty distinct, not sure why they would be an issue for the average person. It makes sense for people who are colorblind to have issues with it but I find it way more likely that there are many people with some crazy color balancing on their monitors lol
I could see the difference and still thought it was an odd choice.
you trying to save ink with the black and white slides?
I'm not very tech-savvy but got a question, I'm benchmarking cyberpunk 2077 with ray tracing at its highest don't have path tracing on at 1080p, is it okay for my GPU to be using in the high '90s. I've heard mixed results where some people say why wouldn't you want to utilize your GPU for everything it's got, and others that say if you want your GPU to be long lasting you shouldn't have it at that high.
Wait for BIOS/software/driver updates, the performance gains show in Linux, so the problem lies with Windows.
Awesome Kyle!
Not really an apples to apples comparisson, you should rerun the tests when they release the 9950X3D, the results were pretty much predictable which would perform better in games.
in my SFF I have a really simple Ryzen 5 5600 and a 1660 Super, nothing high end but enough for what I want to do with it. My main rig weighs a ton so that small shoebox of a case is easy to transport.
The white PC case on your desk what's the name of it?
After watching the last video, I built this but opted for a 7800X3D and hot no heat or performance issues at stock.
You did not undervolt the 9950X, you gave it more headroom in the same power envelope of TDP170W/PPT230W (by curve optimization) and it performed better. IF you had lowered it's power envelope to TDP120W/PPT162W (the same as 7950X3D) it would have had the same power consumption as 7950X3D and better thermals compared to default 9950X (without the need to thermal limit it in bios) - hence it would have been a better comparision.
Why are people still undervolting the old way with these CPUs? You're losing a lot of efficiency and stability by doing this, use curve shaper instead.
Is not curveshaper limited to motherboards not yet released. The x870 models. It is not on current/old MBs
Is not curveshaper limited to motherboards not yet released. The x870 models. It is not on current/old MBs
@@djchotus1 No, its out and it works on current boards already.
Why not run it at 90c if 95c is safe ? Or even 94c just change fan curve .
good built I also need a higher core CPU an a not so powerful GPU, do you have the link of the GPU that is in the video?
Why didn't you wait for the X3D 9000 series
I was going to wait for it. Then I said forget it. I just got a 7800X3D. I'm coming from a 3900X so I should see a lot of difference and I can either wait for Gen 6 or if the 9800X3D is that much better than the 7800X3D I can always get it later. My 3900X is bottlenecking my 4080 OC today, not in 6 months or whenever the 9800X3D is coming out.
@@HerrinSchadenfreude Oh that's a great upgrade that I can agree with since you have a 4080. I meant from his POV where he already had a 7950X3D. It was already shown in the reviews that the X3D chips for the 7000 series was better than the 9000 non x series.
You guys know he gets the 9950x for free right
How much would Cinebench score be if I limit 9950X to 70 C . Or wattage at that point. I have ordered 9950X for long term (15y) processor. I start with air cooling as saving money for some water cool solution. It is for programing and DAW use so it has those number crunch qualities.
If u go from a X3D to a X and or comparing the two and complain u should not be around computer hardware.. whether AM4 or AM5 X3D out preforms
Ideally you should've compared it to the 7950X, not the 7950X3d as its known that the 7950x3d runs alot cooler and draws way less power than the 7950X.
Thus the 7950X would've been a much better comparison as the same test between the 7950X and 7950X3d would also have big differences in the results for heat and power consumption.
You need the 9950X3d for this comparison to really mean anything.
Interesting build. But yeah: for productivity and gaming a big case is still very handy. I'm on the opposite site with an ancient huge Thermaltake level 10 gt build. Will probably downsize with my next build though. I have the also ancient 3950x and don't often need more performance. So yeah: I might go for a three times 8 core cluster this time. Would be fun!
Hope life is going good my man.
Did you log in as Administrator on your computer or did you log in as the a user with Administrator access? Logging in with Administrator and re-running your tests likely will give you better results.
I had trouble with a 65w chip in my itx build I can’t imagine running a 170w chip in there.?
How does yalls CPU Temps average at 73? Mine never exceeds 65. Am I doing something wrong? It makes me feel like I may not be using my components to their potential.
I'll never understand the stupidly small pc cases.
Saves space or something. I admire the tiny ITX builds, but you are right though, sometimes they are almost "too" small.
I'll never understand stupidly large ones. Huge unused space is dumb and pointless.
4:58 is that GPU sag or that just a camera thing?
The 9950 is a productivity forward cpu primarily for compute it’s not specifically a cpu for generalized all around computing- Dude , ya gotta know why you’re buying a specific CPU
Honestly I was expecting a lot worse since this is, well, a worst case scenario. Sure something like this is going to get handicapped either thermally or power draw wise w/ eco mode, but this experiment was never about practicality; its about that 1 niche “what if” case that we think about but know we’ll never get to see because it’s so niche
really like this build
Nice seeing you again dude! Long time sub here
I don't like people gimping the Ryzen 9 7950X3D without showing it under a manual FCLK overclock.
You can run a 7950X3D 8-core chip let using X3D cache at 5.450Ghz while the 7800X3D can only achieve 5.250Ghz.
An overclocked 7950X3D CAN be faster in gaming than a 7800X3D, but it never will be unless a person spends the time and manually overclocks a locked processor.
I just dislike people showing a handicapped 7950X3D because it makes other results look better while the 7950X3D is truly a beast of a processor while overclocked.
5.450Ghz on 8-X3D-Cores and 5.950Ghz on 8-Non-X3D-Cores.
In addition, the dual-chiplet design of 7900X3D and 7950X3D allows for double the overall memory bandwidth between infinity fabric and memory, allowing for users to actually benefit from running DDR5 RAM above 6400MT/s, although running out-of-sync has diminishing returns and does not show much benefit until 7200MT/s WHILE also having tightened primary and subtimings.
But, if you do the manual overclocking, it CAN be faster than 6400MT/s, it just takes immense effort to tighten your timings and tightened timings also increase CPU heat generated.
For the form factor of the build that cooler is too small. i wiuld have used a 7700 and moved on with my life.
why didn't you wait until the 900x3d cpu's are released ?
I feel like color blinded looking at the charts
The L9a has a rating for 65W TDP, the 9950X has a TDP nearly triple that.
Does he have a part list for this build?
*MIXED feelings on this video...*
If you don't watch the ENTIRE video it's easy to trash Bitwit. He did say he didn't need the extra multi-thread processing for his use case. So you paid MORE for a product that wasn't suited for your use case and then REGRET your decision? See, that's just strange. I'd rather you just did a straight-up benchmarking video and then talk about the pros and cons. Sort of the same thing, but it changes the tone. This is a little NEGATIVE when it's actually a great product... plus, there are performance issues Windows update may address but I guess you didn't know this 5 days ago... anyway, $125 more for 21% higher performance (and that may go closer to 30% once Windows is updated) is actually quite good... you have to think about this compared to the ENTIRE computer cost. If you spent $2125 vs $2000 then you spent 6.25% more money to get 21% higher performance. THAT is what you should be talking about in a more positive way. IMO... if it's mostly meant for gaming then you'd be looking at an R7-7800X3D or maybe waiting for the R7-9800X3D at this budget...
So BITWIT is like "it doesn't really benefit GAMING and I personally don't QUITE need all the processing power for multi-threading performance." Ya, this product isn't meant for YOU then.
man..... the charts.... was soooooooooo dificult to read. I dont know it was inttencional! but... was hardddddd..
when are you going to go back to the actual vlogs ? and not just slide shows! thats where i got hooked on the channel!
Trying to buy one from Best buy but they have 0 ryzen 9 9950x selling my ryzen 9 7900x for $150 😮❤
Why are there no renewed reviews with the new update that fixes the performance issues? It a pretty big deal with these and no one is re benchmarking after the update.
Still rocking my 5900x because I don’t listen to the tech hype garbage. Everything runs just fine and snappy.
The 3d cache varients of the cpus have not convinced me to move over to them.
Almost impossible to visually see the color differences on the graphs in a room with natural light. Would be nice to use colors that are not very close to each other, especially for those with color blindness.
Why black&white graphs?
I still think this was a fun video. Chin up brother
you should get jacked again kyle. unless you have a sleeper build you look less jacked... love you still pookie! :)
Which set-up is better.?
14700k (or 9700x) + ARC a770 (16gb)
Vs
14700k (or 9700x)+ rtx 3060 (12gb)
Vs
14700k (or 9700x) + AMD 7800xt (16GB)
For Davinci Resolve
My Canon R6 m2 shoots 4k in h.264 n h.265
Just released, yet has a 2023 date stamped on it? How long ago were they actually made?
Not sure why you're using white and grey for your graphs. It's sometimes hard to decipher
I updated my Gigabyte b650m Elite Ax to the latest BIOS and upgraded from a 7700X to a 9900X and now it won't boot. It just hangs with the DRAM light. I dropped to a single stick and waited forever, still can't get in. I cleared CMOS several times and even reverted back to the 7700X but my board seems to be completely hosed. Freakin blows
Damn thats impressive. 43500 in cinebench with that tiny cooler. Try getting 40k on any other cpu on the planet with that cooler.
After the win11 update?
I have an old 5800X and I also undervolted using PBO II. My cores will take a negative 25 as well. At -30 it's not stable and I have neither the time nor the patience to do a by core optimisation. P.S. Kyle, get a haircut man! 😄
My 3950x overclocked to 6.9Ghz finally booted before it crashed make sure you set the voltage to 2.5 😮....
some of you mentions why put a power hungry cpu into a tiny case with that low pro cooler. this is actually interesting test, more real world scenario especially those need powerful cpu in a sff for their always on the go job, rather than open test bench in a controlled environment.