A quick note. People are making the incorrect assumption that we didn't use the latest AMD chipset driver and 3D V-Cache Optimizer.... we did. AMD are aware of the performance related issues with the 7950X3D in some games and are working on a fix.
Also, Rainbow Six Siege and Rainbow Six Extraction aren't the same game. Gamers Nexus tested Siege, we tested Extraction and again AMD are aware of the issue with Extraction but couldn't provide a fix in time for the review. Gamers Nexus also test Shadow of the Tomb Raider using the built-in benchmark whereas we test in game, the village section which is far more CPU demanding than the canned benchmark.
I saw a few places that said Windows could not park cores unless the balanced power plan was used - not high performance. Can you confirm this is the case?
Not a very accurate simulation if they simply limited the applications or BIOS to half the 7950X3D. The 7950X3D has 128MB of L3 cache, which if you disable one CCD would leave you with 64MB of L3 cache. The 7800X3D has 96MB of L3 cache in a single CCD.
@@ACanadianDude you don't seem to realize that only one ccd on the 7950x3d has 3d vcache... I'm not sure how much vcache exactly it has but it's definitely more than 64mb
@@ACanadianDude You are wrong. one of the ccds in the 7950x3d has 96MB of L3 cache and the other one has 32mb of L3 cache, if they disable the non-3d cache ccd, it will give a pretty accurate 7800x3d simulation as they now both have 1 ccd with 8 cores and 96mb of L3 cache.
@@ACanadianDude What are you talking about? The 7950X has 64 MB of L3 cache (32 megs for each CCD). The 7950X3D comes with a 64MB additional vCache on one of those CCDs. Hence, if you disable the non-vCache CCD you are left with 32 MB + 64 MB = 96MB. I find it very disturbing that 4 people (so far) liked your comment, which goes to show how easily some people digest misleading information on the internet...
I think real one would better since 7800x3D simulation clock was 4800mhz but real one should be atleast 100mhz higher and even may hit 5ghz all cores since its a 8 core cpu that doesn't require special software like 7950x3D need ( i mean those new software still running background when simulating 7800x3D but real 7800x3D don't need that new software at all🤷 ) but maybe simulation is pretty accurate🤷 i hope it will be better than simulated result
Hard to say but it would come down to how the clocks are on CCD-0 of the 7950x3d vs the clocks of the 7800x3d. If the clocks are the same, then Gaming Performance will largely be the same.
There is a difference in cache sizes , 7800X3D is L2 at 8Mb, L3 at 96 Mb, 7950X3d is L2 at 16Mb, L3 at 128Mb. They will be close , maybe same, but I suspect the real 7800X3D will be 2-5% off what the simulated results show.
@@Elysiannthey disabled the extra ccd, it’s not using the cache in the other chipper because it’s a large latency hit across the fabric link to do so. It wouldn’t have any extra cache and if it did for some reason it would be slower due to the physics separation.
@@m8x425 pretty sure the CCX clocks better on the higher priced chip. Chiplet arch allows a way more intensive binning. It make sense that AMD uses that to make sure customers pay premium performances at premium prices.
You guys are the longest running subscription I ever had. ( patreon since 2017 iirc ) And you keep delivering, so I guess it wont end any time soon. Because if I had to choose one channel out of all the reviewers to keep in business, it would be hardware unboxed. Stay strong Steve 🙏( but also get some rest when "relevant" product cycle allows it)
Seems like the architecture and larger cache of Zen4 is “enough-ish” in more cases than with Zen3 making the extra cache less effective at getting that huge margin the 5800X3D was able to achieve. That said, the power consumption figures are impressive. Looks like the reduction in clock speeds brought the 7950X3D much deeper into its power/efficiency curve that the base parts clearly try to blow past like a 3090Ti does. HardwareCanucks even confirmed that air coolers are more than useable for this chip instead of massive liquid coolers. Great review, Steve.
Seems like 7800x3d is the way to go for the best gaming CPU on AM5, especially if you don't care about productivity. Otherwise the 5800x3d is still a very capable and impressive CPU in 2023
I'm gonna keep my 5800x3d for now. The jump is not bad from that to the projected performance of the 7800x3d, but i'd have to build a whole new rig and that would be way to expensive for the gains.
wait till HUB clarified whether they have configured the "3D V-Cache Optimizer" in the latest Chipset driver properly, the driver should be parking cores, and looking at HUB's review, it doesnt look like it is.
@@josejuanandrade4439 Only worth the upgrade, if you're still on Zen 2 or older. Zen 3 chips, especially the newer ones like the 5800X3D, are more than enough for gaming atm.
@@defouq he said he went on holiday at the end of this video so he didn't get to finish testing as this product release wasn't dated before he booked his holiday
The deficit of the full 7950X3D vs the simulated 7800X3D should have something to do with the windows game mode and the XBOX gamebar that recognizes an app as a game or not. Those ones that show the 1st slower than the 2nd one might be the ones not recognized properly and needed to do so manually. Kudos for the review!
Steve from GN explained that the power plan has to be set to Balanced so that the Chipset Driver can park the non-vcache cores when gaming. Personally I think it's pretty dumb coz it invalidates one of the oft-mentioned justifications for a 16-core desktop CPU: having a load of background tasks running while gaming. Intel is looking pretty smart now with how they solved this core affinity problem by implementing thread director with Alder Lake.
@@jihadjoe I know right? Who in their right mind wants to band-aid a $700 processor with several pieces of software and power use fixes? I'm not sure I even understand why AMD even put this out right now. How many average gamers are gonna know to take all of the steps (latest bios, latest windows and updates, extra software and power settings) to make this processor work properly? And then actually know if those band-aids actually are working for the games they play. No one is gonna want to bench test every game to see if everything is working as it should before they play it. I have a feeling that this is gonna get curb stomped by the 7800X3D when it finally comes out and this is why AMD put this thing out first. Bad move in my book.
@@rvman3359 Hopefully this will be solved through some windows update and "just work" in the future, that would be the best outcome. However, everyone currently on AM5 that buys one of these CPUs is going to have to update their BIOS regardless, otherwise the CPU won't work at all.
I wonder how this CPU will age. A few generations from now when 3D Vcache is on all cores, how likely is it that developers will optimize for this particular set of chips? That being said, the simulated 7800x3d results seem especially exciting! Thanks so much for testing that for us Steve!
Developers won't optimize for that. It is already a huge hassle to utilize multithreading and vectorized instructions sets such as SSE2 and AVX2. At best they can try to enforce thread residency so that critical threads won't be scheduled on e-cores or cores without e-caches. Other than that there is nothing to optimize for, compared to generalistic optimization strategies.
Developers aren’t the ones optimizing for these chips. Developers aren’t doing a thing. It’s purely AMD and Microsoft identifying if the app you are playing is a game or not and then disabling the second set if it’s a game. So developers don’t matter at all, just identifying whether the app is a game or not through windows.
Excellent point, it seems like at this point the 3D VCache is very immature. It's either going to completely take over and get way better than now, or it will be done and go away, with something else taking over.
great quality review! Not a single scent/trace of anything that could be called bias or "fanboyism". Thank you for being an amazing resource for all things PC!!
What I'd love to see is memory testing with the X3D. HUB has shown that Zen 4 is very sensitive to memory timing, but the extra cache could go a long way to eliminate this problem (it helps in this respect on the 5800X3D). If the X3D CPUs can use cheaper RAM just as effectively for gaming, that would be an interesting twist.
@@lukearbuckle4760 Amount of RAM doesn't make a difference in performance; if you have enough, you have enough. Memory speed makes the difference (when CPU-limited), so amount of memory is not the relevant part here.
They already had a video explaining why they don't. In 99% of the cases you will be hard limited by either CPU or GPU, and you can find out how your system will perform by taking the lower value of the two. CPU performance does not scale with resolution so you just need to take the 7800x3D performance and combine it with benchmarks for whatever GPU you have at 1440P to find out how it works. In many cases at 1440P, almost all at 4K you will be bound by GPU unless you have a 4090.
@@MegaTheJohny I never said you can’t want to, just that it’s objectively pointless and misleading (as so many reviewers keep trying to explain to their audience). But if you can’t take the minimum of two numbers that are both available online in most cases that’s completely your prerogative.
I think it would be interesting to see how the 7950X3D performs with process lasso where with setting where the game runs on the ccd with the 3d cache and all the other processes run on the other ccd.
You can set processor affinity for any process via powershell. It's easy to make a script that launches your game and ties it to the cores (hence CCD) you want.
@@andersjjensen Would be better if AMD releases their own custom launcher that does that though, instead of users having to build our own custom launch scripts for each game with psexec -a.
@@notenjoying666 Reminds me of when Alder Lake first came out. Using Process Lasso was very helpful before Windows optimized it's scheduling for the 12th gen parts.
@@jihadjoe Obviously, but the amount of people who need both a high core count consumer CPU AND want top dog gaming performance is going to be limited and probably mostly distributed among highly technical people.
@@andersjjensen there is nothing easy about PowerShell. It's like looking at a foreign language through two levels of google translate. However, if you are willing to post your scripts (maybe GitHub?) and are easy to modify to a given game... I'd very much be thankful!
It's very hard to do consistently is the problem. You cant control what others are doing around you. And having no benchmark is better than having wrong/inaccurate benchmarks for a game.
@@Jordan-ru8yf they can just do it in the marketplace of the MMO where most players gather or just make 3 system then play with tim and someone (pretty sure they have lots of parts for it). that way all three of them has the same experience. and yes its more effort but they only have to do it for 7950x3D chip vs 7950x chip vs 13900k.
intel is shit, even just while web browsing intel cpu is already going over 100W. It seem that intel just keeps overclocking their CPUs to crazy numbers instead of actually improving something. Bet next generation intel CPUs will be overclocked so high they will be reaching 500W power draw causing them to melt motherboards.
True, i also don't understand why total system cost includes MB and Memory but doesnt include additional 300 w for power supply and even more money for cooling solution.
@@c.f.m.8133 because they're using the same power supply and cooler...of course they could lower the price by buy a lower power supply or cheaper cooler. It's a first gen on the am5 mother board, so if you planning on getting new cpu after few years, that could save you a lot more.
Definitely. I think one of the great strengths of the 3D model is power efficiency in certain workloads. And they Frames per Watt charts were plain outright ignored again, which just confuses a sh*tload of viewers, telling the wrong message. I'd like to compare F/W between 7950X, its two ECO modes, and 79503D on the same chart with 13900k (because Intel deserves it big time). That would be a much different picture.
Great review as always 👍👍👍Your simulated data for the 7800x3D could explain why AMD decided to delay the release. If the benchmarks are true, the 7800x3D would be the much better option for gamers. I'm still going with the 7600x. In Germany the 7800x3D will probably cost 500€ (like the 5800x3D), which would be more than double than the 7600x right now. I don't see the point on spending so much money on a CPU, if you are GPU-bound anyway. Plus: With AM5 you have always the option to upgrade later.
Can you guys please do at least one or two game tests at 1440p and 4k so that we can see if it's at all worth it to upgrade from a 5000 series Ryzen. Most people with these CPUs are not gaming at 1080p. I understand you are doing the tests to show the differences between them properly but for a real world scenario it would be great to see higher resolution tests. Thanks again for your detailed analysis. My go to channel for reviews.
Really... What is stopping you from using a calculator and scale down. Firstly you lose "resolution" to the fact that you got lower numbers in FPS. Secondly you run into GPU limits and can't see the real difference. What does it help you too see one CPU does 10% more in 4k because it does 44 instead of 40 FPS, while seeing the real difference of, let's say 15% in 1080 because it does 115 FPS instead of 100?!? Total senseless request.
@@notenjoying666 Why? I game at 1440p like most people and would like to see the difference in performance with a RTX 4090. I'm very much aware that at higher resolutions the GPU becomes the bottleneck but with a card as powerful as the RTX 4090 I'm sure there will be some interesting results with these CPUs.
@denverbasshead2361 That's not why they don't benchmark it anymore. It's because the licensing when altering hardware locks them out after like 5 changes. It is a worthwhile benchmark, but you'll have to go elsewhere to find it. The 5800X3Ddoes phenomenal in MSFS2020 and I expect these new CPUs to do even better.
Interesting results! Thanks for testing this Steve! Seems it's not as big of a jump as it was with the 5800X3D, but I do like the innovation of the AMD team to take this road further.
I imagine it is a very good option for people who run virtual machines, as you can keep the 3dvcache cores for the host system to use in games and some other tasks while the guest system uses the other cores in different applications. Very niche use case, but if you're into it this might be the CPU you want to use.
although the assignment of the correct (non v-cache) cores / threads to the virtual machine could be difficult / the thread scheduler could have difficulties running the threads on the specific cores.
Bro how can you place the 7950x3d on the same level as a 13900ks? That Intel CPU burns literally money with that 500 w tdp. Also in your cost analysis you didn't mention you need a much bigger PSU if you want to run that abomination of a CPU. In Germany, just from that difference in TDP, you'd probably save over 200€ until that CPU hits end of life. Otherwise very good and concise review.
This actually is like a 2 in 1 processor - 7950 eco + 7800x3d in one. The ability to switch modes on the fly would make it the best cpu overall for creators & gamers with its lower power draw & solid performance...
Really great info on disabling in the bios, thanks for that❤️. Just bummed all of these reviewers never seem to test DCS, we really need to see these numbers especially in higher resolutions. 🤷
I find the varying settings from high/ultra to medium custom very annoying from one reviewer to the next. One will use High on one game medium on the next then the next channel you watch it is the opposite, then you find yourself watching and pausing two videos side by side to see WTF each did. Not very "clear" presentation at all.
@@crisnmaryfam7344 Its because you want to test a cpu bottleneck not a gpu bottleneck. And they tested those games with those settings so they have a baseline for the current cpu.
Great video as usual! Would love to see some benchmarks for RPCS3 and Switch emulation too, as Intel was leading quite a bit with their 13th gen stuff. Would also be great to see some temperature comparisons between this and Intel CPUs.
Maybe you should add Football Manager (FM) to the CPU testing. The CPU speed matters a lot in that game, it is a popular game (16. on steam), the community have made 4 benchmark save files with a methodology for how to test the game. And it is a completely different type of game compared to many of the games you are currently testing, probably most like Factorio, but it looks to matter much more what CPU you use in FM than in Factorio. I think it is a good candidate for a CPU gaming test. 1. Since the CPU matters a lot and the GPU looks to matter very little. 2. Is shows relatively big differences between CPU's. (it can use many cores) 3. A lot of people play the game and the game seems to change only a little from year to year, thus a CPU that was fast in the 22 version will also be fast in the 23 version. 4. The game might show differences between AMD's X and X3D versions as well as Intel's P and E cores.
X3d tech is coming along nicely. It's probably not the best investment (subjective opinion) when there is a 7600/7700 to start with on AM5 and both are very capable cpus for the next few years, but x3d definitely the part of choice when zen 6 come around and AM5 becomes a dead platform. Can then replace the 7700 with the latest 9800x3d(?).
Just FYI guys most of MSI motherboards are 6 layers only, which isn't great for ddr5. Ddr5 memory is a way more sensitive to amount of layers of PCB then ddr4.. So if you planning to overclock your ram or use 4 sticks, I would go with gigabyte or ASRock, their mid -high budget boards are all 8 layer PCB. It's a shame most of big tech TH-camrs don't say a word about it, when advertising MSI products.
@Hardwareunboxed can you guys please start testing PDX games in these CPU benchmarks? they are one of the few genuinely CPU-dependent games out there and a lot of people play them It's kinda weird that no one tests them for CPU benches
Gamers Nexus found that the cpu performs as expected with the newest driver. Not sure if you guys installed that for this test. But that may be why you had to simulate most of these results. Either way i think the 7800x3d is for the gamers and this is for the gamers who also do production work.
Hi Steve, much of the attractiveness of the 7950x was the ability to BIOS limit the CPU down to 65W(?) while still getting a slight performance bump over the 5950x running at stock power limits. With the 7950X3D losing to the 7950X in productivity workloads due to the lower TDP, how do the two compare when you manually limit the TDP? Personally I would be interested in a 65W comparison.
cost per frame is a good comparison. but if you look at frame per watt, i'd say that graph looks significantly in amds favor. i'm curious, did you try any of these tests with those custom memory timings? always enjoy the great content you guys produce :)
Cost per frame is useless. What matters is cost per additional frame and to know that I have to know what hardware you have and if you need a new motherboard, RAM etc. Nothing general can be said about that. E.g. the least cost per frame can be slower than what you already have if you have a 5800x3d. How useful is that?
Thank you as always. Also I know you guys already doing a lot of testing/benchmarking but would it be posssible to do a gaming benchmark on ultrawide monitors? I have a 1440p ultrawide and when I watch one of your bencmarking videos I always put "my results" somewhere between 1440p and 4K. It just would be nice to see some real life results:) Anyways keep up the good work.
Wow, that power draw is *very* impressive in all-core workloads. Also as expected, Windows scheduling is going to make or break these dual-CCD variants. It'll take quite a while for all the bugs and problems to be worked out... and yikes, that Factorio result. I expect to see a lot of similar issues for the first 3-6 months.
@@MizarcDev to be fair to HUB, the instructions from AMD didnt explicitly say anything about lowering power management settings to "balanced". GN had to find that out almost like a software bug. Where the Core Parking feature didnt work on high power settings.
Great video thanks for it. I know you mentioned it many times but I would love to see 4k results as well, as people like me who own old processor need that information to help us make better decisions when purchasing.
It won't help you, because you'll be limited by GPU. So look at the FPS you want from this review, and then look at the GPU reviews to see how much FPS you'll get at 4K. If it's less than what the CPU can deliver then you're good.
I’m gonna get the 7950x3d. It has great gaming performance, but also great multi core performance, even if it’s slightly slower than the regular model in the latter.
Even if the performance isn't drastically higher than the 13900ks it's important to remember that it's doing it using more than 210 watts less power to get there. That's pretty huge. The 7950x3d will be for people who don't want to compromise on productivity or have specific applications that benefit from the v-cache, the 7800x3d will be the one to get for people who just want to game and those simulated benchmarks are impressive, it will use even less power than the sepc we saw for the 7950x3d while still beating the 13900ks.
LOL don't worry man that's only with a 4090. Remember the 5800x3d is faster than the 12900k and 7950x because the 3090 ti isn't fast enough to saturate it completely. So it's only bottom of the barrel if you use GIGA expensive cards
Some impressive performance gains there! Nice to see the combination of more cores + extra v-cache instead of having to choose as with 5900X and 5800X3D at very similar pricing. Question is, can the simulated results be had using something like taskset to pin tasks to one CCD when launched rather than disabling in BIOS?
Man, a B450/B550 + 5800X3D is looking like it was a better deal every day. The 7800X will be interesting, but I doubt the cost-per-frame factoring in the mobo will beat the £65 Mortar Max + cpu combo
I'll never understand people that think it makes sense to buy a high-tier CPU, then shoot it in the kneecaps with shitty RAM and VRMs. If you're buying crap AM4 parts, just get a 5700X and be done with it.
@@rustler08 the B450 Mortar Max had some of the best VRMs of the generation (see buildzoids video), thanks to msi compensating for the bad press with overheating b350s by overengineering the follow up...
Great review! I'm in the crowd who is on a 3950x and both does productivity work and gaming on my machine, so it looks like this or a 13900KS will likely be the upgrade path for me. I do wonder though if AMD would be smart enough for situations like at 17:20 with CS:GO, where it's not a cache sensitive game if they would reverse disable the 3D V-Cache CCD to achieve higher clock speeds and thus a higher frame rate? I'm sure it's not something that most games would need, but for certain popular titles like CS:GO it could be enticing enough for AMD to attempt reverse core parking. That might also be something for you guys to look into!
Yep, love mine. I also don't just play games, I use modern emulators, and the 7700x is a beast in RPCS3, Xenia, Yuzu, Retroarch, etc.. I am really looking forward to see what the 7800x3d will do in games, although it will definitely be slower in emulation due to lower clocks.
In gaming, the non 3D Cores should actually be disabled. Perhaps you were not running the latest drivers? Clocks are higher on the 7950x3d than on the 7800x3d, so it will be slower but price performance wise, it will be in favor of the 7800x3d.
I feel like HU is head-to-head with Gamers Nexus in terms of editorial quality, information accuracy, and attention to detail. Thank you for all you do!!
So until a fix comes, if you want to get the best performance with games, you have to manually disable the non-3D CCD in most cases? I wonder if AMD will fix this or if it will be 'one of those things' that never properly get addressed.
no, just get the latest AMD chipset driver and config "3D V-Cache Optimizer" properly| watch gamers nexus review for a proper explainaton, I dont think the test rig is configured properly for this review given the CCD bahaviour.
@@quirkyitalianexploration and you have to config the power management to set it to "balanced" instead of "high performance' otherwise the 3D V-cache optimizer dont work properly.
I would love to see Valorant benchmarks alongside CS:GO to cover more CPU-limited titles. The 1% mins will be especially interesting as competitive FPS gamers usually run high refresh rate monitors.
If Factorio results are anything to go by, the 7800X3D maybe the most "future proof" CPU for gamers, ever. 10 years from now folks will be pairing them with their $10,000 GPUs and still getting decent frames
GN mentioned something about the balanced power profile being required to properly use the v cache. Did you use balanced? Just asking as you didn't mention it.
Something that would be nice is if for an example in Ryzen Master, we could say which program you want to run on the part of the chip with the stacked cache, and which part you prefer to run on the part with the higher clocks, or if you don't mind and might let it run however it might find it being the most convenient option.
I'll guess they will retest the 7900 XT and XTX when the other variants drop, since AMD had some driver issues/Performance issues when the cards first released
G'day Steve, In the intro regarding the 7900X3D "Probably because I don't think it's gonna make a whole lot of sense as it will just be 6cores with the 3D-V Cache" yeah I agree but I was wondering... 🤔Would it be possible to do a "7600X3D Simulated Result" video in gaming? while not being an actual CPU I think it would be VERY interesting.
A Japanese reviewer reported on 7950X3D on G.Skill DDR5-6000 and Jedec DDR5-4800 running the same in a few games (e.g FF15). Would like to see a deep dive on the difference.
Was considering returning my 13900k for the 7950x3d, especially with the AM5 socket being supported for the next few years but doesn’t seem worth the price. Add to the fact that AMD mobo’s cost more for comparable performance.. Will stick with what I have for now. Plus intel chips have higher resale value when I am ready to upgrade.
A quick note. People are making the incorrect assumption that we didn't use the latest AMD chipset driver and 3D V-Cache Optimizer.... we did. AMD are aware of the performance related issues with the 7950X3D in some games and are working on a fix.
Also, Rainbow Six Siege and Rainbow Six Extraction aren't the same game. Gamers Nexus tested Siege, we tested Extraction and again AMD are aware of the issue with Extraction but couldn't provide a fix in time for the review.
Gamers Nexus also test Shadow of the Tomb Raider using the built-in benchmark whereas we test in game, the village section which is far more CPU demanding than the canned benchmark.
Ok thanks guys for mentioning this. Glad AMD and the developers are aware and sorting this out, but not really a game changer for many titles as is.
what about video editing with it? am a freelancer thats why I'm wondering.
I saw a few places that said Windows could not park cores unless the balanced power plan was used - not high performance. Can you confirm this is the case?
@@LiamGand should be great. If you're only editing videos and not playing games, you might want the non x3d.
Man, I was hoping someone would simulate the 7800X3D, - and sure enough, you guys did it!!
Legends
Not a very accurate simulation if they simply limited the applications or BIOS to half the 7950X3D. The 7950X3D has 128MB of L3 cache, which if you disable one CCD would leave you with 64MB of L3 cache. The 7800X3D has 96MB of L3 cache in a single CCD.
@@ACanadianDude you don't seem to realize that only one ccd on the 7950x3d has 3d vcache... I'm not sure how much vcache exactly it has but it's definitely more than 64mb
@@ACanadianDude You are wrong. one of the ccds in the 7950x3d has 96MB of L3 cache and the other one has 32mb of L3 cache, if they disable the non-3d cache ccd, it will give a pretty accurate 7800x3d simulation as they now both have 1 ccd with 8 cores and 96mb of L3 cache.
@@ACanadianDude What are you talking about? The 7950X has 64 MB of L3 cache (32 megs for each CCD). The 7950X3D comes with a 64MB additional vCache on one of those CCDs. Hence, if you disable the non-vCache CCD you are left with 32 MB + 64 MB = 96MB.
I find it very disturbing that 4 people (so far) liked your comment, which goes to show how easily some people digest misleading information on the internet...
Now I REALLY want to see a comparison between the 7800X3D simulation vs the real one in april!! To know if it was accurate!
I think real one would better since 7800x3D simulation clock was 4800mhz but real one should be atleast 100mhz higher and even may hit 5ghz all cores since its a 8 core cpu that doesn't require special software like 7950x3D need ( i mean those new software still running background when simulating 7800x3D but real 7800x3D don't need that new software at all🤷 ) but maybe simulation is pretty accurate🤷 i hope it will be better than simulated result
Hard to say but it would come down to how the clocks are on CCD-0 of the 7950x3d vs the clocks of the 7800x3d. If the clocks are the same, then Gaming Performance will largely be the same.
There is a difference in cache sizes , 7800X3D is L2 at 8Mb, L3 at 96 Mb, 7950X3d is L2 at 16Mb, L3 at 128Mb. They will be close , maybe same, but I suspect the real 7800X3D will be 2-5% off what the simulated results show.
@@Elysiannthey disabled the extra ccd, it’s not using the cache in the other chipper because it’s a large latency hit across the fabric link to do so. It wouldn’t have any extra cache and if it did for some reason it would be slower due to the physics separation.
@@m8x425 pretty sure the CCX clocks better on the higher priced chip. Chiplet arch allows a way more intensive binning. It make sense that AMD uses that to make sure customers pay premium performances at premium prices.
You guys are the longest running subscription I ever had. ( patreon since 2017 iirc )
And you keep delivering, so I guess it wont end any time soon.
Because if I had to choose one channel out of all the reviewers to keep in business, it would be hardware unboxed.
Stay strong Steve 🙏( but also get some rest when "relevant" product cycle allows it)
Seems like the architecture and larger cache of Zen4 is “enough-ish” in more cases than with Zen3 making the extra cache less effective at getting that huge margin the 5800X3D was able to achieve.
That said, the power consumption figures are impressive. Looks like the reduction in clock speeds brought the 7950X3D much deeper into its power/efficiency curve that the base parts clearly try to blow past like a 3090Ti does.
HardwareCanucks even confirmed that air coolers are more than useable for this chip instead of massive liquid coolers. Great review, Steve.
Seems like 7800x3d is the way to go for the best gaming CPU on AM5, especially if you don't care about productivity. Otherwise the 5800x3d is still a very capable and impressive CPU in 2023
I'm gonna keep my 5800x3d for now. The jump is not bad from that to the projected performance of the 7800x3d, but i'd have to build a whole new rig and that would be way to expensive for the gains.
@@josejuanandrade4439 Enjoy your CPU mate!
@@josejuanandrade4439 I'm keeping my 12400🤣
wait till HUB clarified whether they have configured the "3D V-Cache Optimizer" in the latest Chipset driver properly, the driver should be parking cores, and looking at HUB's review, it doesnt look like it is.
@@josejuanandrade4439 Only worth the upgrade, if you're still on Zen 2 or older. Zen 3 chips, especially the newer ones like the 5800X3D, are more than enough for gaming atm.
LG 27GR95QE on LG.com: bit.ly/3xLXdAN
Cinebench R23 doesn't use AVX-512.
WHERE IS THE DOTA 2 AND VALORANT BENCHMARK??
@@defouq he said he went on holiday at the end of this video so he didn't get to finish testing as this product release wasn't dated before he booked his holiday
The deficit of the full 7950X3D vs the simulated 7800X3D should have something to do with the windows game mode and the XBOX gamebar that recognizes an app as a game or not. Those ones that show the 1st slower than the 2nd one might be the ones not recognized properly and needed to do so manually. Kudos for the review!
Let's let others be beta testers! Do not buy this crap
Steve from GN explained that the power plan has to be set to Balanced so that the Chipset Driver can park the non-vcache cores when gaming. Personally I think it's pretty dumb coz it invalidates one of the oft-mentioned justifications for a 16-core desktop CPU: having a load of background tasks running while gaming. Intel is looking pretty smart now with how they solved this core affinity problem by implementing thread director with Alder Lake.
@@jihadjoe I know right? Who in their right mind wants to band-aid a $700 processor with several pieces of software and power use fixes? I'm not sure I even understand why AMD even put this out right now. How many average gamers are gonna know to take all of the steps (latest bios, latest windows and updates, extra software and power settings) to make this processor work properly? And then actually know if those band-aids actually are working for the games they play. No one is gonna want to bench test every game to see if everything is working as it should before they play it. I have a feeling that this is gonna get curb stomped by the 7800X3D when it finally comes out and this is why AMD put this thing out first. Bad move in my book.
@@rvman3359 Your average gamers aren't gonna buy this CPU.
@@rvman3359 Hopefully this will be solved through some windows update and "just work" in the future, that would be the best outcome. However, everyone currently on AM5 that buys one of these CPUs is going to have to update their BIOS regardless, otherwise the CPU won't work at all.
I wonder how this CPU will age. A few generations from now when 3D Vcache is on all cores, how likely is it that developers will optimize for this particular set of chips?
That being said, the simulated 7800x3d results seem especially exciting! Thanks so much for testing that for us Steve!
Bold of you to assume that the next generations of cpus will have 3d vcache on all cores.
Very unlikely. They don't even optimize for multithreading, specially games
Developers won't optimize for that. It is already a huge hassle to utilize multithreading and vectorized instructions sets such as SSE2 and AVX2. At best they can try to enforce thread residency so that critical threads won't be scheduled on e-cores or cores without e-caches. Other than that there is nothing to optimize for, compared to generalistic optimization strategies.
Developers aren’t the ones optimizing for these chips. Developers aren’t doing a thing. It’s purely AMD and Microsoft identifying if the app you are playing is a game or not and then disabling the second set if it’s a game. So developers don’t matter at all, just identifying whether the app is a game or not through windows.
Excellent point, it seems like at this point the 3D VCache is very immature. It's either going to completely take over and get way better than now, or it will be done and go away, with something else taking over.
great quality review! Not a single scent/trace of anything that could be called bias or "fanboyism". Thank you for being an amazing resource for all things PC!!
Love the videos guys!
What I'd love to see is memory testing with the X3D. HUB has shown that Zen 4 is very sensitive to memory timing, but the extra cache could go a long way to eliminate this problem (it helps in this respect on the 5800X3D). If the X3D CPUs can use cheaper RAM just as effectively for gaming, that would be an interesting twist.
I want to see the memory scaling on the X3D too, as it could make a big difference to the cost of a build, or the amount of ram a build gets!
@@lukearbuckle4760 Amount of RAM doesn't make a difference in performance; if you have enough, you have enough. Memory speed makes the difference (when CPU-limited), so amount of memory is not the relevant part here.
Wish I could thumbs up more than once. Pause.
@@dennisjungbauer4467 I think that Luke's point was that if you buy slower RAM then you can afford more RAM.
Great job! I'd love to see 5800x3D vs 7800x3D at 1440p in the top popular games (real world test) once the chip releases!
They already had a video explaining why they don't. In 99% of the cases you will be hard limited by either CPU or GPU, and you can find out how your system will perform by taking the lower value of the two. CPU performance does not scale with resolution so you just need to take the 7800x3D performance and combine it with benchmarks for whatever GPU you have at 1440P to find out how it works. In many cases at 1440P, almost all at 4K you will be bound by GPU unless you have a 4090.
Well I do have 4090 and I would like to see 1440p benchmarks
@@MegaTheJohny I never said you can’t want to, just that it’s objectively pointless and misleading (as so many reviewers keep trying to explain to their audience). But if you can’t take the minimum of two numbers that are both available online in most cases that’s completely your prerogative.
@@Jordan-ru8yf 100% not pointless. Also they only didn't do it because he didn't have time
@@MegaTheJohny Long story short: platform upgrade not worth it.
I think it would be interesting to see how the 7950X3D performs with process lasso where with setting where the game runs on the ccd with the 3d cache and all the other processes run on the other ccd.
You can set processor affinity for any process via powershell. It's easy to make a script that launches your game and ties it to the cores (hence CCD) you want.
@@andersjjensen Would be better if AMD releases their own custom launcher that does that though, instead of users having to build our own custom launch scripts for each game with psexec -a.
@@notenjoying666 Reminds me of when Alder Lake first came out. Using Process Lasso was very helpful before Windows optimized it's scheduling for the 12th gen parts.
@@jihadjoe Obviously, but the amount of people who need both a high core count consumer CPU AND want top dog gaming performance is going to be limited and probably mostly distributed among highly technical people.
@@andersjjensen there is nothing easy about PowerShell. It's like looking at a foreign language through two levels of google translate.
However, if you are willing to post your scripts (maybe GitHub?) and are easy to modify to a given game... I'd very much be thankful!
I’m waiting for the 7800X3D for sure, thanks for this!
You guys need to put some MMO's into your Benchmark pool. Thats where those 3D parts really beginn to shine!
They don't even do Warzone LOOL
It's very hard to do consistently is the problem. You cant control what others are doing around you. And having no benchmark is better than having wrong/inaccurate benchmarks for a game.
@@Jordan-ru8yf they can just do it in the marketplace of the MMO where most players gather or just make 3 system then play with tim and someone (pretty sure they have lots of parts for it). that way all three of them has the same experience. and yes its more effort but they only have to do it for 7950x3D chip vs 7950x chip vs 13900k.
But do you really need 500 FPS on an MMO...?
@@mintymus you obviously don't know what's 200-500 players at one place!
The discrepancy in total system consumption between Intel and AMD is mind boggling to me
intel is shit, even just while web browsing intel cpu is already going over 100W. It seem that intel just keeps overclocking their CPUs to crazy numbers instead of actually improving something. Bet next generation intel CPUs will be overclocked so high they will be reaching 500W power draw causing them to melt motherboards.
True, i also don't understand why total system cost includes MB and Memory but doesnt include additional 300 w for power supply and even more money for cooling solution.
@@c.f.m.8133 because they're using the same power supply and cooler...of course they could lower the price by buy a lower power supply or cheaper cooler.
It's a first gen on the am5 mother board, so if you planning on getting new cpu after few years, that could save you a lot more.
If you are a gamer it must not bother you at all, because average gaming power consumption is the same between intel and AMD.
You should google average and psu tripping...
Great review! I would love to see "Frames per Watt" and/or "Total System Power Consumption While Gaming" become standard metrics.
Meaningless, because it's highly dependent to a GPU and GPU drivers.
@@dat_21 FPS also highly depends on a good driver, so with the same logic, FPS is also meaningless. Also thermal throttling, etc.
Definitely. I think one of the great strengths of the 3D model is power efficiency in certain workloads. And they Frames per Watt charts were plain outright ignored again, which just confuses a sh*tload of viewers, telling the wrong message. I'd like to compare F/W between 7950X, its two ECO modes, and 79503D on the same chart with 13900k (because Intel deserves it big time). That would be a much different picture.
I am so excited for this. Thank you so much Steve. I love all the work you guys put in. You guys provide the best benchmarks and advice by far!
I'm not
@@wallacesousuke1433 I was referring the video and benchmarks. not the CPU necessarily.
The simulated benchmarks were very close. Top job!
Perfect review. Hope we will see some 1440p and 2160p ultra benchmarks to see if 3d cash heps there.
Great review as always 👍👍👍Your simulated data for the 7800x3D could explain why AMD decided to delay the release. If the benchmarks are true, the 7800x3D would be the much better option for gamers.
I'm still going with the 7600x. In Germany the 7800x3D will probably cost 500€ (like the 5800x3D), which would be more than double than the 7600x right now. I don't see the point on spending so much money on a CPU, if you are GPU-bound anyway. Plus: With AM5 you have always the option to upgrade later.
please dont go amd if youre not going for an x3d cpu the motherboards and ram are so expensive on am5
7800x3d was always going to be the best gaming part. 1CCD vs 2CCD.
@@tnt_gaming21 Don’t be so poor.
7600x is already monstrous enough, you dont need the best of the best in the gaming world unless you dont go for 7900XTX+
i'd personally go with a 5800x3d instead. Then use the money you save for a better upgrade in the future.
Can you guys please do at least one or two game tests at 1440p and 4k so that we can see if it's at all worth it to upgrade from a 5000 series Ryzen. Most people with these CPUs are not gaming at 1080p. I understand you are doing the tests to show the differences between them properly but for a real world scenario it would be great to see higher resolution tests. Thanks again for your detailed analysis. My go to channel for reviews.
Really... What is stopping you from using a calculator and scale down.
Firstly you lose "resolution" to the fact that you got lower numbers in FPS.
Secondly you run into GPU limits and can't see the real difference.
What does it help you too see one CPU does 10% more in 4k because it does 44 instead of 40 FPS, while seeing the real difference of, let's say 15% in 1080 because it does 115 FPS instead of 100?!?
Total senseless request.
@@notenjoying666 Why? I game at 1440p like most people and would like to see the difference in performance with a RTX 4090. I'm very much aware that at higher resolutions the GPU becomes the bottleneck but with a card as powerful as the RTX 4090 I'm sure there will be some interesting results with these CPUs.
th-cam.com/video/Zy3w-VZyoiM/w-d-xo.html
MSFS 2020 is a popular flight sim that would be interesting to see benchmarks on …
It's GPU bound, it used to be in their videos when it first launched. No scaling so it's useless
@denverbasshead2361 That's not why they don't benchmark it anymore. It's because the licensing when altering hardware locks them out after like 5 changes. It is a worthwhile benchmark, but you'll have to go elsewhere to find it. The 5800X3Ddoes phenomenal in MSFS2020 and I expect these new CPUs to do even better.
@@denverbassheadNot true. The more complex aircraft are heavily bound by single core performance on the CPU.
@@denverbasshead Tom's Hardware website, page 6 of x3d review, lists 50% performance gain in MSFS on a 4090 at 1080p and 1400 resolutions. Impressive.
@@denverbasshead Way to spread clearly verifiable misinformation. Yikes.
The 7950x3d is a complete Badass for the money obviously we'll have to wait for the 7800x3d on April 6th
Interesting results! Thanks for testing this Steve! Seems it's not as big of a jump as it was with the 5800X3D, but I do like the innovation of the AMD team to take this road further.
AMD keeps on pushing the technology forward. Huge props for them
Quite impressive power consumption, great review as always. Enjoy your well-earned break, Steve. :)
Thx 4 the simulation of the7800X3D. 👏🏻
This is really interesting. Especially the 7800x3d simulated. This is a good reason to wait until april.
I imagine it is a very good option for people who run virtual machines, as you can keep the 3dvcache cores for the host system to use in games and some other tasks while the guest system uses the other cores in different applications. Very niche use case, but if you're into it this might be the CPU you want to use.
although the assignment of the correct (non v-cache) cores / threads to the virtual machine could be difficult / the thread scheduler could have difficulties running the threads on the specific cores.
Thank you for the Simulated Results !
Bro how can you place the 7950x3d on the same level as a 13900ks? That Intel CPU burns literally money with that 500 w tdp. Also in your cost analysis you didn't mention you need a much bigger PSU if you want to run that abomination of a CPU. In Germany, just from that difference in TDP, you'd probably save over 200€ until that CPU hits end of life.
Otherwise very good and concise review.
even if I had 3000USD in my wallet for a CPU upgrade last thing I would get is that KS i9. i3 is better that that hot spreeder
Incredible job as always Steve!!!
This actually is like a 2 in 1 processor - 7950 eco + 7800x3d in one. The ability to switch modes on the fly would make it the best cpu overall for creators & gamers with its lower power draw & solid performance...
Really great info on disabling in the bios, thanks for that❤️. Just bummed all of these reviewers never seem to test DCS, we really need to see these numbers especially in higher resolutions. 🤷
Bro, the level of analytical ability, attention to detail and clear presentation in Hardware Unboxed is totally insane.
I find the varying settings from high/ultra to medium custom very annoying from one reviewer to the next. One will use High on one game medium on the next then the next channel you watch it is the opposite, then you find yourself watching and pausing two videos side by side to see WTF each did. Not very "clear" presentation at all.
@@crisnmaryfam7344 Its because you want to test a cpu bottleneck not a gpu bottleneck. And they tested those games with those settings so they have a baseline for the current cpu.
Great video as usual! Would love to see some benchmarks for RPCS3 and Switch emulation too, as Intel was leading quite a bit with their 13th gen stuff. Would also be great to see some temperature comparisons between this and Intel CPUs.
Big thanks for the review, I've waited for over half a year to upgrade, might as well wait a little bit more for the 7800x3D.
Maybe you should add Football Manager (FM) to the CPU testing.
The CPU speed matters a lot in that game, it is a popular game (16. on steam), the community have made 4 benchmark save files with a methodology for how to test the game. And it is a completely different type of game compared to many of the games you are currently testing, probably most like Factorio, but it looks to matter much more what CPU you use in FM than in Factorio.
I think it is a good candidate for a CPU gaming test.
1. Since the CPU matters a lot and the GPU looks to matter very little.
2. Is shows relatively big differences between CPU's. (it can use many cores)
3. A lot of people play the game and the game seems to change only a little from year to year, thus a CPU that was fast in the 22 version will also be fast in the 23 version.
4. The game might show differences between AMD's X and X3D versions as well as Intel's P and E cores.
X3d tech is coming along nicely. It's probably not the best investment (subjective opinion) when there is a 7600/7700 to start with on AM5 and both are very capable cpus for the next few years, but x3d definitely the part of choice when zen 6 come around and AM5 becomes a dead platform. Can then replace the 7700 with the latest 9800x3d(?).
Probably. Especially if your not gaming at 1080p
Just FYI guys most of MSI motherboards are 6 layers only, which isn't great for ddr5. Ddr5 memory is a way more sensitive to amount of layers of PCB then ddr4.. So if you planning to overclock your ram or use 4 sticks, I would go with gigabyte or ASRock, their mid -high budget boards are all 8 layer PCB. It's a shame most of big tech TH-camrs don't say a word about it, when advertising MSI products.
Useful info thanks.
@Hardwareunboxed can you guys please start testing PDX games in these CPU benchmarks? they are one of the few genuinely CPU-dependent games out there and a lot of people play them
It's kinda weird that no one tests them for CPU benches
Gamers Nexus found that the cpu performs as expected with the newest driver. Not sure if you guys installed that for this test. But that may be why you had to simulate most of these results. Either way i think the 7800x3d is for the gamers and this is for the gamers who also do production work.
What an idiotic statement lol the non-x3d versions are excellent for gaming and even better for productivity
Hi Steve, much of the attractiveness of the 7950x was the ability to BIOS limit the CPU down to 65W(?) while still getting a slight performance bump over the 5950x running at stock power limits. With the 7950X3D losing to the 7950X in productivity workloads due to the lower TDP, how do the two compare when you manually limit the TDP? Personally I would be interested in a 65W comparison.
cost per frame is a good comparison. but if you look at frame per watt, i'd say that graph looks significantly in amds favor. i'm curious, did you try any of these tests with those custom memory timings? always enjoy the great content you guys produce :)
Cost per frame is useless. What matters is cost per additional frame and to know that I have to know what hardware you have and if you need a new motherboard, RAM etc. Nothing general can be said about that. E.g. the least cost per frame can be slower than what you already have if you have a 5800x3d. How useful is that?
You can easily and quickly do your own cost per frame calculation yourself using their benchmarks. And you can adjust it for sales and price drops.
And your computer is idle way more than not. At idle, Intel wins easily.
@@mintymus my computer is rarely at idle. if it is left inactive for more then 10 minutes, it goes to sleep.
@@jayjonah2762 So your computer is always at 100%?
Glad there are multiple tech reviewers to show the different date
Thank you as always. Also I know you guys already doing a lot of testing/benchmarking but would it be posssible to do a gaming benchmark on ultrawide monitors? I have a 1440p ultrawide and when I watch one of your bencmarking videos I always put "my results" somewhere between 1440p and 4K. It just would be nice to see some real life results:) Anyways keep up the good work.
lol this wouldn't make any sense for a cpu benchmark, increasing the resolution would mostly hit the gpu more
You should get about 74% of the performance of 1440p results.
@@rustler08 That's about how I figure out my results:) It just would be nice to see it in action;)
@@General_Li_Shin I know I posted it under a cpu benchmark video but I meant for testing gpu on ultrawide. Sorry
Steve you the only one with the simulated results, bravo 👏
Wow, that power draw is *very* impressive in all-core workloads.
Also as expected, Windows scheduling is going to make or break these dual-CCD variants. It'll take quite a while for all the bugs and problems to be worked out... and yikes, that Factorio result. I expect to see a lot of similar issues for the first 3-6 months.
That 7800X3D looks so good 😁😁😁
I was worried about the hybrid CCD issues causing certain tasks to be performed on the wrong chip, and damn that Factorio result really shows it :/
Watch GN's review, i have a feeling this test rig didnt configure "3D V-Cache Optimizer" properly in the chipset driver.
@@MrFredscrap You appear to be right. There for sure seems to be methods in place to prioritize or outright shut off the second CCD while gaming.
@@MizarcDev to be fair to HUB, the instructions from AMD didnt explicitly say anything about lowering power management settings to "balanced".
GN had to find that out almost like a software bug. Where the Core Parking feature didnt work on high power settings.
By far the best review, very interesting with the second CCD disabled too. Thanks.
Great video thanks for it. I know you mentioned it many times but I would love to see 4k results as well, as people like me who own old processor need that information to help us make better decisions when purchasing.
It won't help you, because you'll be limited by GPU. So look at the FPS you want from this review, and then look at the GPU reviews to see how much FPS you'll get at 4K. If it's less than what the CPU can deliver then you're good.
@@ojassarup258 i already have 4080 I am not limited by gpu
I’m gonna get the 7950x3d. It has great gaming performance, but also great multi core performance, even if it’s slightly slower than the regular model in the latter.
Even if the performance isn't drastically higher than the 13900ks it's important to remember that it's doing it using more than 210 watts less power to get there. That's pretty huge. The 7950x3d will be for people who don't want to compromise on productivity or have specific applications that benefit from the v-cache, the 7800x3d will be the one to get for people who just want to game and those simulated benchmarks are impressive, it will use even less power than the sepc we saw for the 7950x3d while still beating the 13900ks.
Impressive how the top of the line 5800x3D is now bottom of the barrel currently, I'm still keeping mine for many years
LOL don't worry man that's only with a 4090. Remember the 5800x3d is faster than the 12900k and 7950x because the 3090 ti isn't fast enough to saturate it completely. So it's only bottom of the barrel if you use GIGA expensive cards
Some impressive performance gains there! Nice to see the combination of more cores + extra v-cache instead of having to choose as with 5900X and 5800X3D at very similar pricing. Question is, can the simulated results be had using something like taskset to pin tasks to one CCD when launched rather than disabling in BIOS?
damn, that factorio looks good
Small correction there, Tjmaxx is actually 89 C as you can see on AMD's spec sheet on their website.
In price comparison you forgot 200€ for a cooler that can cool that 13900ks 😅
Awesome B-Roll!!
Man, a B450/B550 + 5800X3D is looking like it was a better deal every day. The 7800X will be interesting, but I doubt the cost-per-frame factoring in the mobo will beat the £65 Mortar Max + cpu combo
I'll never understand people that think it makes sense to buy a high-tier CPU, then shoot it in the kneecaps with shitty RAM and VRMs.
If you're buying crap AM4 parts, just get a 5700X and be done with it.
@@rustler08 the B450 Mortar Max had some of the best VRMs of the generation (see buildzoids video), thanks to msi compensating for the bad press with overheating b350s by overengineering the follow up...
Holy crap, a review where I can’t spot any obvious skewing or bias.
This is the best HUB review so far.
I wonder if the actual 7800X3D will have the same amount of v-cache as your simulated setting.
Specs are announced, it will.
Most likely yes
Both do have 96 MB, yes.
Great review, one more thing i would like to see is pbo and curve optimizer on x3d impact on results
Great review! I'm in the crowd who is on a 3950x and both does productivity work and gaming on my machine, so it looks like this or a 13900KS will likely be the upgrade path for me. I do wonder though if AMD would be smart enough for situations like at 17:20 with CS:GO, where it's not a cache sensitive game if they would reverse disable the 3D V-Cache CCD to achieve higher clock speeds and thus a higher frame rate? I'm sure it's not something that most games would need, but for certain popular titles like CS:GO it could be enticing enough for AMD to attempt reverse core parking. That might also be something for you guys to look into!
Very high quality. Thank you for doing the tests twice.
Will be good to see more detailed comparison between power consumption of 13900K(S) vs 7950X3D.
Still very happy with my 7700X!
Yep, love mine. I also don't just play games, I use modern emulators, and the 7700x is a beast in RPCS3, Xenia, Yuzu, Retroarch, etc..
I am really looking forward to see what the 7800x3d will do in games, although it will definitely be slower in emulation due to lower clocks.
In gaming, the non 3D Cores should actually be disabled. Perhaps you were not running the latest drivers? Clocks are higher on the 7950x3d than on the 7800x3d, so it will be slower but price performance wise, it will be in favor of the 7800x3d.
No PBO tests? I hope that's the next video. Apparently the overclocking isn't that bad this time.
I feel like HU is head-to-head with Gamers Nexus in terms of editorial quality, information accuracy, and attention to detail. Thank you for all you do!!
POV: These guys are better than GN, imho.
For pure hardware testing they are the best in the game. GN excels at investigative journalism.
@@FroschYankee Not even close to the same level.
@@FroschYankee these guys are like GN Lite
@@FroschYankee GN included the "3D cache optimiser". Did HUB mention it?
Best review I have seen so far. Thank you for the 7800x3d (simulated), great work.
So until a fix comes, if you want to get the best performance with games, you have to manually disable the non-3D CCD in most cases? I wonder if AMD will fix this or if it will be 'one of those things' that never properly get addressed.
no, just get the latest AMD chipset driver and config "3D V-Cache Optimizer" properly|
watch gamers nexus review for a proper explainaton, I dont think the test rig is configured properly for this review given the CCD bahaviour.
@@MrFredscrap Good point. I just watched Gamers Nexus and see that they updated their drivers to make it work properly.
@@quirkyitalianexploration and you have to config the power management to set it to "balanced" instead of "high performance' otherwise the 3D V-cache optimizer dont work properly.
FYI we're using the latest AMD chipset driver and 3D V-Cache Optimizer.
As always, great review & thank you for including titles like ACC
It seems like these are worth a revisit in a few months once the drivers and Windows updates get more work.
Amazing. Will be going for 7800x3d.
I would love to see Valorant benchmarks alongside CS:GO to cover more CPU-limited titles. The 1% mins will be especially interesting as competitive FPS gamers usually run high refresh rate monitors.
The impromptu motherboard review was a bonus on this excellent review. Thanks, Steve. ;P
Interesting for 5800x3d owners. great video as always guys, you always have the best data on new components
If Factorio results are anything to go by, the 7800X3D maybe the most "future proof" CPU for gamers, ever. 10 years from now folks will be pairing them with their $10,000 GPUs and still getting decent frames
Maybe so but so should the 7950X3D be the same. It isnt in this Chart however so there is clearly something VERY wrong.
0:22 - While it may not make sense, I believe it'll end up with more L3 cache per core, which may be interesting to investigate...
I'm calling it now... please when reviewing the 7800X3D, include the 7950X3D 'simulated 7800X3D' results. IMO could be interesting...
GN mentioned something about the balanced power profile being required to properly use the v cache. Did you use balanced? Just asking as you didn't mention it.
Something that would be nice is if for an example in Ryzen Master, we could say which program you want to run on the part of the chip with the stacked cache, and which part you prefer to run on the part with the higher clocks, or if you don't mind and might let it run however it might find it being the most convenient option.
You don't need to disable 8 cores completely.
There is a setting to which type of CCD should be prioritized.
The fact it can perform so well, at under half the wattage of the Intel KS.... Is pretty insane.
7800X3D WITH 7900XTX NITRO+ PLEASE!!
I'll guess they will retest the 7900 XT and XTX when the other variants drop, since AMD had some driver issues/Performance issues when the cards first released
Always the best Steve, cheers for all the hard work, enjoy the holiday!
Fuck yes! Starting right off the bat with Factorio! Thanks so much!
G'day Steve,
In the intro regarding the 7900X3D "Probably because I don't think it's gonna make a whole lot of sense as it will just be 6cores with the 3D-V Cache" yeah I agree but I was wondering...
🤔Would it be possible to do a "7600X3D Simulated Result" video in gaming? while not being an actual CPU I think it would be VERY interesting.
A Japanese reviewer reported on 7950X3D on G.Skill DDR5-6000 and Jedec DDR5-4800 running the same in a few games (e.g FF15).
Would like to see a deep dive on the difference.
i doubt theres difference in ff, not many i recall f1, battlefield, assassin creed, and prob all cpu heavy titles
You guys should have more subs!
Even with the 13900K being better value I think it makes more sense to get on the AM5 platform for future upgrade options.
Not the title I wanted to see but I'm glad you're honest about it.
Was considering returning my 13900k for the 7950x3d, especially with the AM5 socket being supported for the next few years but doesn’t seem worth the price. Add to the fact that AMD mobo’s cost more for comparable performance.. Will stick with what I have for now. Plus intel chips have higher resale value when I am ready to upgrade.
20% is too much for a maybe 5 year upgrade path?
Thx for testing Factorio!
at 3:13 you refer to, I believe, the 7950X being reviewed here, as "59X3D" which sounds a bit confusing.
This does look promising. I hope renewing my windows 10 with motherboard change from Am4 to Am5 wont suck big time :D