@@moldyshishkabob. You know that a lot of people have around 4 times the gb as cores for memory right? Many people have 6-8 cores with 32gb of tam. Same with those before who had 4 cores and 16GB…. So technically that’s is very common on the lowend.
I would have hoped that people were smart enough to open task manager while using discord/youtube and put 2 and 2 together but i quess people needed this video
@@killzonearmed who cares what HUB tests. Your set up is the one you should care about. So open up your epic CPU intensive multitasking and monitor the CPU usage, open up a game if you feel nothing a miss boom I just saved a youtuber hours of pointless work.
I watch videos all the time using Chrome while gaming on a 2600X, RX580. There are some scenarios where it causes a problem but the majority of games that I play don't seem to mind. The bigger problem is my brain can't multitask and I end up only half paying attention to each.
I use a 3600, 32gig ram 3200, GTX 1070 "hope to upgrade soon" Same as me. I watch vids on 2nd monitor, run chat apps, comp monitoring software etc while gaming on my 32inch primary monitor. Most of the time Multi-threading is normally not an issue but there are some times it is. Specifically in Space Engineers, or a couple single core heavy games. The single core games get affected when some other task decides it suddenly wants the prime cores. "Often some stupid windows service"
In my experience, what will definitely degrade gaming performance is playing a UA-cam video/Livestream on a second display, actively visible (not minimized). But that's on the gpu, not the cpu, and only if the gpu was already maxed out while playing
NVIDIA has addressed this saying the NVENC video decode part of the chip is completely separate to 3D/gaming part of the chip. (For GTX 1650 super and above) Video decoding does not affect 3D/gaming chip, unless you run multiple 4K youtube videos on multiple monitors and overpower the Video decode/encode chip. If you have performance issues, open up windows task manager, GPU section, and see if your Video decode chip is at 100% and what % the 3D/gaming chip is at
@@jcbc5950 But it still uses VRAM I/O so games which are more memory I/O bound than compute bound will see a bit of a reduction (but it's only a fairly small bit, even on low-ish end GPUs, as video-decode-and-copy-to-scanout-buffer bandwidth is fairly limited)
If you have mismatched refresh rates between monitors that is far more likely to be the cause of stuttering if you have a modern system, but even this has largely been addressed in the later versions of windows 10.
Which got me thinking... 1. would reducing the ram to 16gig from 32gig make a different since i'd imagine more people with R5 would be on 16 gig rams than 32 2. would changing to an Nvidia GPU make a difference since the known driver overhead issue of the Nvidia drivers
Really interesting benchmark! It would be great if you could add an i3 12100 or 12300 to the mix to see how a pretty powerful cpu that is a quad-core copes with the tests. Thanks for the videos and keep up the great work!
that would be more informative but that's not the point of this video I think, anyone who's buying 4 cores cpu is limiting himself in short term, so you're buying it knowing you're gonna be bottlenecked in this era, but the point of this video is to prove that the statement that many have that "6 cores cpus are not the sweet spot anymore since 8 cores will handle multitasking better" is wrong.
@@thepunic5261 I think it depends on the use cases. I would pair a 12100f with a RX 6600 without a second thought. It would be interesting to see if a system like that would suffer from multi tasking. Judging by how under stressed my 10100f is I'm willing to bet that it wouldn't be much of an issue.
@@ifrit35 Punic I agree with Mojo. With a lower end GPU anything below the rtx 2060 the i3 1200 is absolutely fine. I used my i3 6098p (2c4t) processor with my 1660 super for 2 years. Almost always in games my GPU was bottlenecked. Yes in some games my fps is higher and I got better frametimes with my 5600x but in general not that much of a big difference. The 12100f is a much better architecture and has 8 threads. It's a lot.
@@thepunic5261 I've yet to hear of anyone actually being unable to do reasonable multitasking because they only have four Alder Lake P-cores. There's a lot of theorizing, but the 12100F will be fine for almost anybody. Do most of us wildly overbuy anyway? Sure. But we should all admit that we could have been fine with much less.
I remember i was getting hitching with Twitch streams I’d leave running on my second monitor while playing games. I debated getting a 5800x to replace my 5600x but opted for upgrading to 32GB of RAM instead. I’m glad I did because I haven’t seen any hitches since.
Having as much ram as you can afford to and as much as the mb+cpu ( platform ) allows is a good idea for multi-tasking. After the amount of ram - timings ( speed ) is another overlooked aspect a lot. Now that AMD has introduced SAM ( Resizable Bar ) that will probably help multi-tasking out also.
@@Starscreamious Yeah the real amount of RAM you need is the what windows calls the "Committed Memory". The "In Use Memory" number it spits out is nonsense. People keep claiming that 16 GBs of RAM is good enough in modern gaming systems, which hasn't been true for nearly 6-8 years now. When you run out of RAM your systems starts using your SSD / HDD as virtual memory. Games like Insurgency Sandstorm / EA Battlefront 2 / Ghost Recon Wildlands will easily push your "Committed Memory" to 17-19 GBs of RAM while games like Total War games, Cities Skylines, and Anno 1800 easily use up to 22-28 GBs of RAM. Any gaming system that plays modern AAA games should definitely get 32 GBs of RAM now a days.
14:20 About streaming on a single PC, I typically will use my 5900X for streaming while gaming because most of the games that I play @1440p tend to have my RTX 2080 usage at 100% and whenever I use NVENC encoder it causes a lot of dropped frames on OBS (not in-game). So for me I find it better to get some use out of my CPU rather than have stream lag with NVENC.
@@hassosigbjoernson5738 Limiting FPS sometimes works, it's really dependent on the game. Though even sometimes when the GPU isn't at 90+% it still has some OBS frame drops.
I do exactly the same, as if I use my 6800 GPU I will have dropped frames due to encoder lag. So more cores on the CPU is definitely better in this scenario, which sadly this video doesn't address at all.
For me, I would leave the Intel GPU enabled, so that OBS could use QuickSync and leave the GPU alone. I'm specifically looking at these 2 CPU (5600 and 5700) wondering if the 2 cores would really make a different with CPU encoding. Since I'll be losing IQS. Right now I'm on Intel + Nvidia so I have QS and NVENC as options. I'm curious as of how much of a impact loosing IQS is if any. If the 5600 would be fine on its own or if the 2 extra cores will benefit on the encoding. Specially since some newer games are getting better at using multiple core/threads. But so far the video hasn't touch on that. -_-
One possible scenario is a user streaming UA-cam and discord with hardware encoding and acceleration disabled. Otherwise most of the workload is offloaded to the GPU.
i find that using hardware encoding can actually cause some video stutter (dropped frames) when loading things in a game. changing it to software encoding alleviates that for me.
@@Dracossaint I have found the same thing, but only on my 3080 and before that 2070s. When I use my 6800xt I have no such issues, so for me AMD has worked better in that regard.
@@Dracossaint really? never have issue with it. using RTX 3050, streaming 1080p 60fps is breeze. Sometimes, I even video conferencing with 3 different google meets and my GPU utilizations are still below 10%. Nvidia's NVENC sure is efficient, I believe the efficient encoder only available at GTX 1660s and all RTX series. Old GPU such as GTX 1060 6GB uses a lot of utilization to stream, my friend with better CPU but older GPU suffers from streaming issue.
@@Dracossaint Video encoding and decoding barely affects gaming with NVIDIA. NVIDIA has addressed this saying the NVENC video encode/video decode part of the chip is completely separate to CUDA/gaming part of the chip. While NVIDIA performance for 1080p/1440p gaming is abysmal compared to AMD, NVENC is very good so I have to use NVIDIA when streaming to Twitch.
This is exactly what I have wanted to see. I've stuck with 8C or higher since AM4 launched (I do a lot of encoding), but I still wanted to see how a Zen 3 6C would do long term against 8C consoles. I realize a 5600X is only slightly behind a 3700X in total performance, and the consoles are basically cut down, heavily down clocked 3700X (sort of). The other side programs is a good idea too. I regularly assign half of the threads to a batch while gaming, so I will stay with 8C or more, but my use case isn't normal. How were the .1% lows, similar behavior?
The Ryzen 3700X has a base clock of 3.6 Ghz. Thats exactly the clock speed of the Series X APU in SMT mode. In the 8c/8t non-smt mode, the Series X CPu has a clock speed of 3.8 Ghz. They reduced the cache for the console APU what also normally is the case for AMDs mobile and desktop APUs. The PS5 APU is clocked at 3.5 Ghz. So, not much of a difference. The only advantage clock wise is the boost clock the 3700X (4.4 Ghz) and normally every PC CPU has. But we all know the advantage of the PC isn't the potentially more powerful hardware, but the open character of the hardware platform.
@@OrkDiktator One of my PC's has a 3700x, and it never runs at 3.6Ghz when it's actually being put to work. Even all core 100% load. Same with Intel, when the board and cooling isn't garbage, and it has a decently power budget. I understand the spec side of it though. I agree for sure on what makes the PC advantageous. I also thoroughly enjoy the flexibility of modding games, fixing games when the devs refuse to add options or fix issues, and the ability to push settings higher years later, or even applying post processing to an old game. Although I would say the cache reduction is also a big disadvantage. Look at the 5800X3D vs the 5800X, and the latter has a higher clock speed. I know the SOC has advantages going for it as well.
@@erroroflife187 In gaming fps yes. I was referring to multithread performance, if you were to make use of all of the CPU's processing power. I was mostly referring to what a 5600x would do compared to a Series X, if they were being fully utilized. I realize that most or all current titles benefit from the superior IPC of the 5600x.
@@TheCgOrion I fully agree with what you said. I'm still at a R 5 1600. Never checked when it runs on base clock. The Problem with implementing more cache is the space. There just isn't enough space for more cache on APUs. Maybe the 5800X3D will motivate the console manufacturers to implement more chance next generation. Both consoles are still great pieces of hardware for their pricepoints. I'd like to see similar APUs for Desktop PCs. They could completely replace dedicated low-end and maybe even lower mid-range GPUs and give AMD market leadership in that field. Btw.. my most played Game over the last 8 months was Skyrim on PC with over 270 mods (mostly new lands, quests and companions and some gameplay enhancements. I really like my Series X and Gamepass and all the stuff Xbox offers, but there's nothing that comes close to that nearly 11 years old game with no mod limitations.
You watch too much TechDeals, we don't need this at all. Anyone with half a brain knows tasks like this use hardly any resources. If you needed this video to know this you clearly couldn't tell your pc ran any slower.
I do livestreaming tests on a single PC system.. there are instances wherein extra cores do play a role even when using the GPU to encode (NVENC, H.264, Quicksync), of course this only applies to 'Warzone', the rest of the competitive games seem to be stable as what you have shown here. Great work as always, steve 👍
Great video, its good you acknowledge how peoples use cases differ and you mention modern processors. My 4770K struggles in games like Valorant and Tarkov, and there is a noticable improvement when not screen sharing and not watching UA-cam videos in the background. Just watching this video on UA-cam 1080p 60FPS on Firefox uses 10% with spikes to 20% CPU Usage.
Thank you for this! I wonder if this was an issue with immature OS's and weaker hardware. As cache, virtualized sandboxes, and scheduling improves, less is demand is put on having dedicated cores.
Yeah, the Vega 64 paired with a 2500K actually is NOT capable of playing back 4k60 by its own (so no playing a game even needed to knock down such a PC).
Now I want to test my old laptops and see if they can handle 4k60. Not that there's ever a reason to play high def videos while doing other things on an old machine, but just to see how old it has to be to struggle with it.
@@PainterVierax I think forcing youtube to provide h264 will also cause it to serve you 1080p60fps max, instead of 4k60fps, regardless of the browser used.
Excellent content, this is the type of workloads folks try to use their higher core CPUs for. Albeit the biggest test would be OBS and the various x264 CPU options with the visual advaned audio/visual effects processing options that might not be GPU accelerated, scaling, etc...
You should compare cpu utilization and talk about overall system smoothness when multitasking. Since 6 core cpus are already well into the hyperthreading most of the time in high demanding games.
Thanks for the testing. Good to have clarification on what has more of an impact. Looks like in going to be putting down a fat wad of cash for a crazy GPU and skipping out on getting a new CPU for now. I have 3 screens and the 5700XT is really feeling inadequate.
Great video! The real world practical benchmarking that we need. Would love to see more like this testing but with lots of open memory intensive applications and a ram intensive game to test how pagefiles are managed for background apps and how that affects gaming performance when low on memory.
Here's a question: Was the GPU decode/accelleration turned on? Was the limit to frame rate where it was close to equal down to GPU load not CPU? Edit: Testing one each of CPU and GPU limited games in the same circumstances (YT plus discord screen share) to see whether CPU and GPU rendering affects things more. As another commenter said, streaming a video in a format your GPU doesn't handle in hardware can seriously spike CPU usage. So this test may well benefit from tests on a more limited GPU.
You guys, Who work in this field are literally thee only ones who game and have exports running in the background. I find it amazing that you often cater to that small audience and it goes too show how hard you work!!!
If it's done on the gpu it won't (under normal circumstances) have any effect at all. It's a separate encoding chip on Nvidia cards. If you're doing cpu encoding however... then the result is obvious. Cpu encoding uses as much of the cpu as it can. If 99% is available it'll use that. So what exactly is the question? Would using 99% of your cpu affect your gaming fps? Yeah. You can't have both running at the same time unless you lower the quality in one of them. So you'll always be sacrificing something. Either fps or streaming/encoding quality. The only way around this is to use a very powerful cpu which can handle the encoding at 1080 or higher. But if you've got an expensive computer you've already paid to solve that problem.
Nice. This is the Type of Benchmark every Streamer and UA-camr needs. Playing a Game + several apps running in the Background like OBS, Afterburner, Restream Chat, Discord / TS3, etc...
@@edh615 a normal Gamer who doesn't do content, doesn't run OBS, nor Restream CHat and when he really wants to immerse himself into the Game, he turns off Discord/ TS too.
I think Steve is underestimating what some of us have running in the background... I almost never have just a single UA-cam video in the background. It's usually like 3 tabs of UA-cam, a couple of tabs of Twitch, Facebook, Instagram, Twitter, a bunch of tabs of things I was working on that I can't quite close yet (mostly text), a couple tabs of gmail on my personal account and business account, a few google spreadsheets... And that's when I'm not actively compiling or running code or using a VM... etc
Amazing benchmark, basically my experience with the 12400F, sure there's a FPS drop, but never relevant enough for me to even notice while playing. And if I notice a situation where this is the case, I can just well... close everything else but the game lol. No biggie, the 6-core is working for my needs pretty well.
I know this is not the same scenario. However, I upgraded my brother pc from ryzen 3 1200 to 5 1500 and what I realy did notice was not the gaming performance but the other applications. Windows gives all the resourses to the game, not the side applications running. I often hear him through the wall but not in discord. A half working solution to that problem was assinging a higher priority to discord in task bar. After ryzen 5, this problem was gone even with normal priority. So I think what you need is some extra resources for any random/running application for a smooth experience.
This is among a very few list of videos which provide me with the information I actually need, as opposed to the game FPS tests that make up for 99% of all youtube tests. I am moving from a rx570 to a APU 5700g because I no longer game. I am moving on with my life to only photoshop and blender, as well as moving onto a solar powered electrical source, so being able to undervolt with a low base TDP while keeping the capacity to process infrequently large files is going to be essential for me, all while keeping a GPU function close to that which have 2gb.
A nice start to the topic, and I like that you take the time to investigate people's assumptions even when you don't agree with them. As you said many of the games tested were lightly threaded games and so extra would do nothing. I would like to see more testing in games where all of the CPU cores are hit hard. The examples that come to mind are the end of Fortnite team rumbles matches where the settings are all low except for render scaling (100%), textures (high) and view distance (epic). Other examples are Shadow of the Tomb raider and Baldars Gate 3.
Thank you for this! I think tech deals channel is responsible for a lot of worry about this (for me at least) as he kept hammering that 6 core in general would be struggling in a few years (a few years ago :) ), although it's been a while since I've watched his content.
I think we are already living in a time if there will be jump in performance, it will be minuscule. The only way to improve performance today is either going full quantum, increase TDP and power consumption, change SoC, or everything combined. So, your current hardware, I think will still relevant in 5 - 7 years unless somehow Intel/AMD/Nvidia go nuts and move to new, more efficient architecture. We have seen x86, x64, and so on..., now ARM is a big contender. Let's see if Intel, AMD, Nvidia, and Microsoft can combine effort to make a new architecture. Where is Linux? They can improve themselves, they are impossible to kill and evolving everyday
Well, even games like RDR2 locked at 60 FPS 1440p HUB settings can use more than 6 threads. More is better, most of the time with PC resources. I only game though, no other applications are opened while I game.
@@takehirolol5962 not necessarily true. Just because games will use the cores doesn't mean more is better. A 5600X will destroy anything in the first 3 generations of Ryzen for gaming.
I've been gaming while watching UA-cam ever since I bought my second monitor. Even on my old i5 and GTX 1060 it was a fine experience. I mainly watched 1080p stuff. Nowadays with my i7-8700K 6-core CPU and a RTX 3080 Ti I play high-resolution games and watch a ton of UA-cam on the side, even 4K videos and even if I lose a frame or two I don't really care. It's wonderful to just turn off my mind and have some noise while grinding in game.
Why do you even waste your time on such games??? Jesus....If your game does not immerse you fully and you not only can, but want to have other backround stuff running, you are literally wasting your LIFE.
Amazing work! But I want to add that sometimes people use software decoding, which uses up quite a lot of CPU power, and I regularly use software decoding on my system to watch videos and streams. This is due to the fact that when I use HW decoding and my GPU is at a 100% load, video feed becomes extremely choppy or straight up locks up. This is probably due to an older GPU, as I use a 1080 Ti with a 5900x, or maybe due to my browser (firefox) I'd like if you were to explore this possibility. As a sidenote, I can certifiably say, that a 5900x is not enough to watch 8k60fps videos.
well.. no, if your GPU is at 100% load, it obviously has no time to do additional work. its going to drop frames if you are already at 100% usage and then ask it to do ANYTHING else. Also, Nvidia cards use a Software Scheduler that runs on the CPU, so if your CPU is fully utilized, your video card will start dropping frames wildly because the CPU doesnt have available time to manage the GPU. That being said, the 5900x shouldnt have any problem with the 1080ti. I've got my 1080ti driven by a ryzen 1700x, and it can fully max the GPU. The 1080Ti is a great card, but if any card is at 100%, expecting it to be smooth sailing when you introduce additional workloads to the device isnt going to go well. hehe.
Thanks Steve for making this detailed video. Would it be exciting to cover a 9900k revisit? Now that we are on heels on Raptor Lake and Zen 4 and many are looking to upgrade.
it’d be interesting to see if this is the same behaviour on Alder Lake E-cores as those have been sold as being good for background tasks. You could test the i5-12600K vs the i5-12600 (or ideally the i5-12490F as it has the larger L3 cache).
You see the point is, even with P cores alone can handle ALL light-threaded background tasks, which negates the point of E cores taking care of background while P cores concentrate on 'important' tasks. Hell even many P cores sit in idle past certain point as the video suggests. The question is then, would E cores save more power than just using P cores for all the tasks? Unfortunately no. As much as marketing guys in Intel want you to believe E cores are 'efficient', E cores are just as power hungry as P cores. So what the hell E cores are for? For all core performance such as rendering where latency isn't the issue. Other than that E cores aren't useful for saving energy contrary to Intel's marketing.
Active applications trigger i/o and access memory. Steve explained spare cores idling DO NOT help, when you have lightly loaded cores already. This obsession with core count is simply wrong.
An excellent video, thanks! What I'd like to see is how much Unreal Engine 5 + DirectStorage in games (when they arrive) changes the calculus here. How many cores will those games like to see?
Unreal 5 only uses 8 cores. So not much change there. Unless they actually decide to use more threads and go into Async style of cpu utilization. Quite disappointing really. But considering PS5 and XboX X devs have the ability to limit the consoles to use only 8 cores instead of 16. I think we will see 8 cores be the max for a while still.
Not sure if my question is similar, but I'm building a PC for game development (including UE5 as one of the engines I consider using) and I wonder if the 5700X is worth the cost compared to the 5600.
@@MrGraywolf09 Going forward you will need a 8gb vram card minimum to run at 30 fps. So a 5700xt and a 1080 will run at that on the example/sample projects. You can get away with a 6600xt/6650xt and a 3060/ti but running the project at medium settings will be alright. A basic project takes 2gb of ram with barely anything in it. If you can save up and get a 6700xt or 6800xt you should be fine for say 6 years if it lasts that long. You could pick the equivalent level card from nvidia. 64gb of ram or more with 16 cores cpu. 8 cores 16 threads. nvme recommended going forward something like 256gb os and 2tb. You can get away with hard drive but it will take minutes to load instead of seconds. UE5 is next gen so it has a heavy cost.
Great video. Your comment about overall performance rather than a strict core count is so true and absent from so many discussions. I think the issue for using the 5600 is that it's such a powerful PC. In all the games tested there isn't one that is both CPU limited and using most of the cores. You'd have to pick a recent triple A game but also pair it with a top end GPU playing at 1080p low graphical settings, my only suggestion is Cyperpunk but if Watch Dogs Legion didn't have an issue I doubt Cyperpunk will either. I think this will only be an issue for older CPUs. For example a few years ago I choose a 2nd hand i7-3960x over the i7-4790k. Both great CPUs, but I thought I'd prefer the 6 cores of the i7-3960x over the slight faster IPC speed of the i7-4790k. In a modern game the i7-4790k could see all cores heavily utilised where extra background tasks could hit the 1% low more than the i7-3960x, however we are talking about these CPUs playing games released 8-10 years after they came out. Try it again with older CPUs which have 60-70% usage in a game. Also for a background task can you add recording the gameplay with Geforce Experience or the AMD equivilant.
I have a 5800X with a TUF RX6800XT, that I use for both engineering work.. and gaming.. I still use my MAC Book Pro 2017, for UA-cam videos while I game or do any simulation/solidworks😂
Awesome benchmark! Would be also cool to see a similiar one with lower end hardware, like CPU with 4 cores, with/without HT/SMT and some low end GPU as well! Thanks for the video!
This is strange. While watching this video at 1440p and playing at 1.5x, my GPU kicks up from 0%-1% to low 30s. I'm playing in an Edge browser with hardware acceleration turned off. This is on a 3440x1440 monitor. _NOTE:_ With Edge + no hardware acceleration, the GPU used Client Server Runtime Process. When I switched to Brave _(hardware acceleration default on),_ the GPU utilization is in the low 40s. Same result with FireFox. The two top GPU usage apps were the browser + Desktop Window Manager, with the browser taking up most of the GPU usage. When I used Edge + no hardware acceleration, my CPU usage was between 12% and 18%. With FireFox and Brave, my CPU usage was between 3% - 6%. When playing games, I use Edge because with hardware acceleration turned on, UA-cam would kick my quality down from 4k or 2k down to 360p every single time. Now it doesn't.
More of this real world scenarios testing please, one thing you can try is perhaps just adding more and more background task and see when does it impact performance ? Maybe running adobe suite or video editing program and other workloads.
For Stream dedicated PCs, in my experience I would recommend getting a platform that can have a decent PCIe capacity, for example I decided to keep my 5820k from 6 years ago since it had a decent capacity, besides lots of PCIe connectors, which those came really handy when adding devices like for example, the Elgato 4K mk2 capture card. I currently run two 4K capture cards with a 1650 Super, so that i can encode with the NVENC part, as well as record in higher bitrate, using X264 to the streaming pc's hard drive. Now since X99 is a bit of a rare platform unless you do a build from like aliexpress, I suggest that if you are looking more for a usual consumer CPU, something like the Ryzen 1600 will be plenty for that matter, and these can be found relatively cheap nowadays, as for integrated graphics CPUs, a Ryzen 3400G and 16gb of ram since you need a bit more ram for the iGPU, should do well as long as the encoding is set to X264 (Cpu based on OBS), with that is it possible to "get away" on not having to buy a dedicated GPU for a streaming only PC. As for motherboards, besides having enough VRM for your chosen CPU, if you want to future-proof it to later have a dedicated GPU + a PCIe Capture card like one of the mentioned above, a good way to find one easily, is that it is "Crossfire compatible", that way it will be easier to search for motherboards that have more than a x16 slot conector, which not all budget options tend to have. Hope this helps the ones interested on Streaming dedicated PCs.
Nah, it'd be cheaper to add 2 to 4 cores to the existing platform or just use hardware encoding. An extra PC will add 500€ or so, which would basically upgrade you to a 5950X that will handle it all and you don't have to switch PCs all the time
@@cromefire_ as someone who has a 5900X system, i still kept my X99 because of the reasons mentioned above, besides your GPU still gets a performance hit as well since it needs to render the scenes for the other applications that you use too, it is not "CPU upgrade bound" only. Besides that, you will also be limited on PCIe bandwith as well, using M.2 drives and your GPU, is very likely that your motherboard wont have enough PCIe lanes alone for adding a 4K capture card, and if you do, then you would limiting the PCIe bandwith on your GPU already by half. The main purpose of using an external system is to have next to no performance impacts on your daily drive/gaming pc. Also the upgrade that you suggest alone can mean as much as a simple, used but cost effective system enough to be used for a stream dedicated system, even more if you skip the capture card and use a LAN bandwith system like NDI for example.
@@XxDeViLBrInGeRxX Well you'd probably use the motherboard PCI-e for the capture card, my rough estimate would be that 4x should probably be enough. And even if not PCI-e x8 is still enough with a 3090 to only get like a very few percent. Regarding HW encoding: no HW encoding is done on dedicated silicon, so it's fine and with a 5950x it's negligible performance anyways. I mean after all you only need 1080p60 (some even do 800somthingP) for streaming anyway, at least on twitch, for proper 4K you have to go to UA-cam I think (which few do). After all I can play games fine, even when doing a software AV1, 16 thread encode, so even a 5700 or 5900X will probably be fine. Just be sure to pin the threads currently so they don't interfere and everything is fine. And I mean besides price you just loose feature wise too because things like window capture obviously don't work and you need the capturing card. On a single system a capture card only even starts to interfere if you both game on PC and console anyway as you don't need the the GPU with console gaming and you don't need the capture card with PC gaming. Add the second keyboard and amonitor that's for the second PC and it doesn't make much sense for most. If I'd built a PC for gaming nowadays, I'd just get a i5-12600K (non-F) probably and just use QuickSync. Even with a second PCI-e slot used you will still have PCI-e 4.0x8 for current gen and 5.0x8 for next gen, so plenty of bandwidth and if you want to do CPU encoding just put OBS on the E-Core, video encoding is something they do great at (in fact E-Core ≈ 1 P-Core Hypethread), so think of it as a dedicated older quad-core for encoding only. But of course you do you if a dedicated machine is what you desire nobody can stop you, it's just probably ~50% more expensive and I find it more annoying than the easy solution (which is a 12600K which is like 100-200€ more I think than a 12400F) and what you gain is ~5% performance in the best case, of which the stream won't see anything because they neither see 4K nor > 60Hz.
From personal experience with r5-3600 + rx480, i do feel performance loss in games like Division 2 or AC Odyssey if i have twitch or youtube open and not minimized. Somehow youtube and twitch does impact my games when games are gpu bottle-necked.
Discord had a bug where their active noise cancellation software would turn itself off due to "high system load" on Ryzen CPUs, but it has been fixed. Something else that I have noticed is that video playback will drop a lot of frames while I am in a game that is maxing out my VRAM (I still have a 580 4gb) It may be nice to test the performance loss when approaching 100% utilization of RAM or VRAM.
Any time you go near your max ram or vram things stop working well. That's to be expected. If you can juggle 5 balls at a time and someone says "can you please do 6?" what do you think will happen? You're going to fumble and drop some.
@@flameshana9 Obviously. The point would be to measure how much of a performance impact there is, and how much ram/VRAM is enough for modern use cases. This can be an issue with new hardware as well. For instance the 3080 10gb can run into performance issues at 4k in some games but not others. Potentially adding in a little bit of multitasking on a second monitor could push the card to it's VRAM limit
Since the "invent" of tabs in browsers, it's not unusual for me to have 30+ tabs open. And because 16GB* was my minimum in a PC from 2012 or so (when 7970 was released), and now 32GB*, I never close the programs, I almost never reboot the PC (just put to sleep). I never experienced the drop in performance with these kind of background tasks. But when I have system uptime 10 days+, it sometimes happens that the whole system just start to stutter even if I close the browser with said 30+tabs and opened programs. That's when I have to reboot, or if Windows Update forces me first. But I was curious in hard numbers how much these background tasks affect performance. My guess was 0.1 and 1% would be much more affected then what was shown here. Nice to have a video to link when someone brings up this topic. *Why I had a need for 16GB and now 32GB RAM? Because I'm a software developer, and need the RAM. And because I'm lazy and leave all the applications open, until Windows forces me to reboot.
@@MrNota500 I know. But it's easier to buy more RAM and stay lazy. I have 2 monitor setup for software development. I have so many applications opened all time. Visual Studio, our company collaboration tools, Opera with 30+tabs,...Why would I close applications? It doesn't hurt performance. Except when Windows fails and have to restart.
Yup having 50+ tabs here i defintely notice an issue with browser memory leaks, after 6-8hrs ro sometimes 28-32hrs that ram usage will creep up over time despite having the same amount of tabs or going a bit less.
@@scarletspidernz I have a different problem. Even if I close all the apps, and have the RAM freed from everything, sometimes Display Drivers start misbehaving. This can result in poor gaming performance, like stuttering, low FPS or outright fail to open a game. Or if I open a browser with only 1 tab, and go to youtube, playback will start pausing (not buffering). Then youtube downgrades a resolution, even to the lowest 144p, video doesn't want to play normally. This happens when the System Uptime is very long (weeks), when PC went through a couple of weeks of daily Sleep, working, Sleep,... Only restart helps. This happens with some drivers more often. I must say that I haven't experienced this for a couple of months now. As I understand, AMD claims they have rewritten their graphics driver almost from the ground up, starting with May Radeon drivers. Maybe this had something to do with it.
Amazing video as always!!! Can't wait to see Hardware Unboxed reviews of the new CPU and GPU to come.... (not talking about intel GPU of course, those are a complete mistery, even for intel).
I feel like having videos open while playing a game causes more input lag and latency issues as the GPU is doing more things at once. Could be an interesting thing to test out.
Wow, I have never seen another channel benchmark my two most played games: Cities Skylines and StarCraft 2!!! I am going to build my first PC later this year and it is really helpful to hear and see how those games utilize the CPU from an expert. Thank you so much!!!
Question Steve: Were you using all these applications with Hardware Acceleration off? Hardware Acceleration uses the GPU instead of the CPU to handle most of the tasks. That's more likely what the comments meant because even with a 5900x I can use up to 20% of my CPU with just Firefox running with multiple tabs and a 4k video running with HA off. For YT or Twitch watching, it means if the GPU is maxed out in game, it'll most likely lag or freeze the video. Whereas you'll never typically hit 100% CPU usage in game so that's why I opt to not use hardware acceleration. PS: RGB software would be interesting to test in the background as well. From the Gamers Nexus video we've seen some can be very resource hogs even in the background.
"For YT or Twitch watching, it means if the GPU is maxed out in game, it'll most likely lag or freeze the video." Actually, this is extremely unlikely. The video encoder/decoder(s) (a.k.a. "media engine") is a separate sub-unit/processor. Not specific to Nvidia but in their words, "NVIDIA GPUs contain one or more hardware-based decoder and encoder(s) (separate from the CUDA cores) which provides fully-accelerated hardware-based video decoding and encoding for several popular codecs. With decoding/encoding offloaded, the graphics engine and the CPU are free for other operations." Additionally, if you have a non-F Intel CPU, you could enable the IGP and perhaps move the decoding to that "engine" or split the workload up further (assuming an app allows choosing which hardware decoder to utilize).
As a Vtuber I actually run dual PC's because how intensive the multi-tasking is when streaming... But that's because a lot of things are running: OBS (10-15%), Vtube Studio (20-30%), audio player (2-5%), Web Browser (5-10%), even with a second PC that is doing this so the other system can game you sometimes add other things as well like a video capture program (used for lighting the model) or the Twitch Integrated Throwing System. I have run this all on one PC, but the experience was horrible.
Why don't you just show yourself instead of some 3D model? Never got this "Vtuber" thing. I suppose I get it if you're too shy to do so, but you'd be great, I promise :3
Web Browser at 5-10%, i call BS, I have 30-60 tabs open at any giving time and I only see at most 5%, and this is with a laptop processor, i9 10980hk and 64gb of ram. I do movie editing for part time work.
@@crylune Because that's not an option if I value my security. Showing my face means any crazy (including all to many of my neighbors) can find out way to much information about me with enough time. Don't like Vtubers? Don't watch. That's on you, but I'll take being able to be myself online and not needing to worry about getting SWATed or harassed.
Amazing work Steve. Although I expect nothing less of you I didn't expect you are this fast doing benchmarks. In the meantime I'm still waiting for your RX 6800 vs RTX 3070 Ti 50 games in 2022 benchmark blockbuster video!
Thank you for this. I feel that for streaming having an intel with an igpu helps. As the igpu handles the encoding the gpu can handle gaming or other tasks. That would be an interesting test to see.
I have done cpu based recording/streaming in obs (I think I did 720p 60fps fast preset) while gaming on my old 4770 and it used about 20%. If your just doing discord and youtube I doubt it would make a noticeable difference unless you are already on the edge of your fps target. My guess would be you would have to go back to a 4 thread cpu like the old i5s to see a big difference from discord and youtube (but those barely run modern games anyway so does it really matter?)
As long AMD don't fix their GPU encoder, anyone with a quest 2 and playing with oculus link or virtual desktop will need Nvenc or more CPU cores. Niche case? Yes. Just gaming? Also yes.
How about.. what difference does it make to use NVENC to record, stream on discord or on twitch, compared to not using it with various cards? Since it’s the GPU that does the work here, it would be really interesting to see how it scales across all cards. Of course AMD uses something different but it would be good to see that too!
To be fair, UA-cam videos are pretty intensive, better go 64 cores.
Your content is awesome and you’re a phenomenal human being 🤘💪
Don't forget that it's best to have at least four times as much GB as number of cores. 256GB should be a good start.
Also requires RGB on as much components as possible to allow for lit gaming performance.
Oh please don't become the "Tech deals" dude lmao.
@@moldyshishkabob. You know that a lot of people have around 4 times the gb as cores for memory right? Many people have 6-8 cores with 32gb of tam. Same with those before who had 4 cores and 16GB…. So technically that’s is very common on the lowend.
This is the type of real world content that i have been looking for! Cheers boys
I would have hoped that people were smart enough to open task manager while using discord/youtube and put 2 and 2 together but i quess people needed this video
@@Takashita_Sukakoki yea because I definitely have the components to buy and test it
@@killzonearmed who cares what HUB tests. Your set up is the one you should care about. So open up your epic CPU intensive multitasking and monitor the CPU usage, open up a game if you feel nothing a miss boom I just saved a youtuber hours of pointless work.
@@Takashita_Sukakoki I'm going to upgrade my pc, that's why it's useful to have these videos, I think you don't understand that
as HUB keeps saying there is no "future proofing" so buy a capable 6 core alder lake/zen3 or next gen cpus and no need to overthink it.
I watch videos all the time using Chrome while gaming on a 2600X, RX580. There are some scenarios where it causes a problem but the majority of games that I play don't seem to mind. The bigger problem is my brain can't multitask and I end up only half paying attention to each.
so relatable!
It also could be a lot smoother
I use a 3600, 32gig ram 3200, GTX 1070 "hope to upgrade soon" Same as me. I watch vids on 2nd monitor, run chat apps, comp monitoring software etc while gaming on my 32inch primary monitor. Most of the time Multi-threading is normally not an issue but there are some times it is. Specifically in Space Engineers, or a couple single core heavy games. The single core games get affected when some other task decides it suddenly wants the prime cores. "Often some stupid windows service"
brain needs more cores, confirmed
No one can actually multitask. Humans aren’t capable of multitasking. “Good multitaskers” really are just good at switching rapidly between tasks.
In my experience, what will definitely degrade gaming performance is playing a UA-cam video/Livestream on a second display, actively visible (not minimized). But that's on the gpu, not the cpu, and only if the gpu was already maxed out while playing
Which is because if a video is visible it's using hardware decoding (part of the gpu). If it's minimized that decoding is heavily reduced.
NVIDIA has addressed this saying the NVENC video decode part of the chip is completely separate to 3D/gaming part of the chip. (For GTX 1650 super and above) Video decoding does not affect 3D/gaming chip, unless you run multiple 4K youtube videos on multiple monitors and overpower the Video decode/encode chip. If you have performance issues, open up windows task manager, GPU section, and see if your Video decode chip is at 100% and what % the 3D/gaming chip is at
@@jcbc5950 But it still uses VRAM I/O so games which are more memory I/O bound than compute bound will see a bit of a reduction (but it's only a fairly small bit, even on low-ish end GPUs, as video-decode-and-copy-to-scanout-buffer bandwidth is fairly limited)
If you have mismatched refresh rates between monitors that is far more likely to be the cause of stuttering if you have a modern system, but even this has largely been addressed in the later versions of windows 10.
Did they run the tests with YT on second monitor or minimized? I missed that part.
looking at the R6 Siege results it looks more like the background tasks are bogging down the memory controller rather than the CPU cores.
Which got me thinking...
1. would reducing the ram to 16gig from 32gig make a different since i'd imagine more people with R5 would be on 16 gig rams than 32
2. would changing to an Nvidia GPU make a difference since the known driver overhead issue of the Nvidia drivers
Happy to see you deliver what's asked for, and that your insights are appreciated and as usual, on the money.
was about a time to shut up all those tech idiots with bloatware and malware on their PCs 🤡😂
Great video. You could try doing it on older CPU's too to see the long term implications of having less cores.
Really interesting benchmark! It would be great if you could add an i3 12100 or 12300 to the mix to see how a pretty powerful cpu that is a quad-core copes with the tests. Thanks for the videos and keep up the great work!
Yes that will be perfect bcs i don’t know if i should buy the 12100f or spend 100 euro more on the 12400f
that would be more informative but that's not the point of this video I think, anyone who's buying 4 cores cpu is limiting himself in short term, so you're buying it knowing you're gonna be bottlenecked in this era, but the point of this video is to prove that the statement that many have that "6 cores cpus are not the sweet spot anymore since 8 cores will handle multitasking better" is wrong.
@@thepunic5261 I think it depends on the use cases. I would pair a 12100f with a RX 6600 without a second thought.
It would be interesting to see if a system like that would suffer from multi tasking. Judging by how under stressed my 10100f is I'm willing to bet that it wouldn't be much of an issue.
@@ifrit35 Punic I agree with Mojo. With a lower end GPU anything below the rtx 2060 the i3 1200 is absolutely fine. I used my i3 6098p (2c4t) processor with my 1660 super for 2 years. Almost always in games my GPU was bottlenecked. Yes in some games my fps is higher and I got better frametimes with my 5600x but in general not that much of a big difference.
The 12100f is a much better architecture and has 8 threads. It's a lot.
@@thepunic5261 I've yet to hear of anyone actually being unable to do reasonable multitasking because they only have four Alder Lake P-cores. There's a lot of theorizing, but the 12100F will be fine for almost anybody. Do most of us wildly overbuy anyway? Sure. But we should all admit that we could have been fine with much less.
I remember i was getting hitching with Twitch streams I’d leave running on my second monitor while playing games. I debated getting a 5800x to replace my 5600x but opted for upgrading to 32GB of RAM instead. I’m glad I did because I haven’t seen any hitches since.
You can easily see the RAM usage in the Task Manager. If the RAM does not run full, it isn't. But dual channel and timings can make a difference.
@@hassosigbjoernson5738 Windows lies about system RAM usage
Having as much ram as you can afford to and as much as the mb+cpu ( platform ) allows is a good idea for multi-tasking. After the amount of ram - timings ( speed ) is another overlooked aspect a lot. Now that AMD has introduced SAM ( Resizable Bar ) that will probably help multi-tasking out also.
@@Starscreamious Yeah the real amount of RAM you need is the what windows calls the "Committed Memory". The "In Use Memory" number it spits out is nonsense. People keep claiming that 16 GBs of RAM is good enough in modern gaming systems, which hasn't been true for nearly 6-8 years now. When you run out of RAM your systems starts using your SSD / HDD as virtual memory. Games like Insurgency Sandstorm / EA Battlefront 2 / Ghost Recon Wildlands will easily push your "Committed Memory" to 17-19 GBs of RAM while games like Total War games, Cities Skylines, and Anno 1800 easily use up to 22-28 GBs of RAM.
Any gaming system that plays modern AAA games should definitely get 32 GBs of RAM now a days.
@@armybear2 committed doesn't seem relevant to me. I have 32 Gb, 8 are on standby and 4 are literally free, but the committed RAM is 48Gb
14:20 About streaming on a single PC, I typically will use my 5900X for streaming while gaming because most of the games that I play @1440p tend to have my RTX 2080 usage at 100% and whenever I use NVENC encoder it causes a lot of dropped frames on OBS (not in-game). So for me I find it better to get some use out of my CPU rather than have stream lag with NVENC.
Do you limit your fps to 60 when streaming, because you only can stream like 1080 60 Hz? That could take of some load from the GPU.
better to have an Intel iGPU to do the encoding.
@@hassosigbjoernson5738 Limiting FPS sometimes works, it's really dependent on the game. Though even sometimes when the GPU isn't at 90+% it still has some OBS frame drops.
I do exactly the same, as if I use my 6800 GPU I will have dropped frames due to encoder lag. So more cores on the CPU is definitely better in this scenario, which sadly this video doesn't address at all.
For me, I would leave the Intel GPU enabled, so that OBS could use QuickSync and leave the GPU alone. I'm specifically looking at these 2 CPU (5600 and 5700) wondering if the 2 cores would really make a different with CPU encoding. Since I'll be losing IQS. Right now I'm on Intel + Nvidia so I have QS and NVENC as options. I'm curious as of how much of a impact loosing IQS is if any. If the 5600 would be fine on its own or if the 2 extra cores will benefit on the encoding. Specially since some newer games are getting better at using multiple core/threads. But so far the video hasn't touch on that. -_-
One possible scenario is a user streaming UA-cam and discord with hardware encoding and acceleration disabled. Otherwise most of the workload is offloaded to the GPU.
True some people do use a 6500xt and also it would be interesting if you could get more in game performance by shifting load of the GPU
i find that using hardware encoding can actually cause some video stutter (dropped frames) when loading things in a game. changing it to software encoding alleviates that for me.
@@Dracossaint I have found the same thing, but only on my 3080 and before that 2070s. When I use my 6800xt I have no such issues, so for me AMD has worked better in that regard.
@@Dracossaint really? never have issue with it. using RTX 3050, streaming 1080p 60fps is breeze.
Sometimes, I even video conferencing with 3 different google meets and my GPU utilizations are still below 10%.
Nvidia's NVENC sure is efficient, I believe the efficient encoder only available at GTX 1660s and all RTX series.
Old GPU such as GTX 1060 6GB uses a lot of utilization to stream, my friend with better CPU but older GPU suffers from streaming issue.
@@Dracossaint Video encoding and decoding barely affects gaming with NVIDIA. NVIDIA has addressed this saying the NVENC video encode/video decode part of the chip is completely separate to CUDA/gaming part of the chip. While NVIDIA performance for 1080p/1440p gaming is abysmal compared to AMD, NVENC is very good so I have to use NVIDIA when streaming to Twitch.
This is exactly what I have wanted to see. I've stuck with 8C or higher since AM4 launched (I do a lot of encoding), but I still wanted to see how a Zen 3 6C would do long term against 8C consoles. I realize a 5600X is only slightly behind a 3700X in total performance, and the consoles are basically cut down, heavily down clocked 3700X (sort of). The other side programs is a good idea too. I regularly assign half of the threads to a batch while gaming, so I will stay with 8C or more, but my use case isn't normal. How were the .1% lows, similar behavior?
The Ryzen 3700X has a base clock of 3.6 Ghz. Thats exactly the clock speed of the Series X APU in SMT mode. In the 8c/8t non-smt mode, the Series X CPu has a clock speed of 3.8 Ghz. They reduced the cache for the console APU what also normally is the case for AMDs mobile and desktop APUs. The PS5 APU is clocked at 3.5 Ghz. So, not much of a difference. The only advantage clock wise is the boost clock the 3700X (4.4 Ghz) and normally every PC CPU has. But we all know the advantage of the PC isn't the potentially more powerful hardware, but the open character of the hardware platform.
@@OrkDiktator One of my PC's has a 3700x, and it never runs at 3.6Ghz when it's actually being put to work. Even all core 100% load. Same with Intel, when the board and cooling isn't garbage, and it has a decently power budget. I understand the spec side of it though. I agree for sure on what makes the PC advantageous. I also thoroughly enjoy the flexibility of modding games, fixing games when the devs refuse to add options or fix issues, and the ability to push settings higher years later, or even applying post processing to an old game. Although I would say the cache reduction is also a big disadvantage. Look at the 5800X3D vs the 5800X, and the latter has a higher clock speed. I know the SOC has advantages going for it as well.
5600 is actually quite a bit better than a 3700x
@@erroroflife187 In gaming fps yes. I was referring to multithread performance, if you were to make use of all of the CPU's processing power. I was mostly referring to what a 5600x would do compared to a Series X, if they were being fully utilized. I realize that most or all current titles benefit from the superior IPC of the 5600x.
@@TheCgOrion I fully agree with what you said. I'm still at a R 5 1600. Never checked when it runs on base clock. The Problem with implementing more cache is the space. There just isn't enough space for more cache on APUs. Maybe the 5800X3D will motivate the console manufacturers to implement more chance next generation. Both consoles are still great pieces of hardware for their pricepoints. I'd like to see similar APUs for Desktop PCs. They could completely replace dedicated low-end and maybe even lower mid-range GPUs and give AMD market leadership in that field. Btw.. my most played Game over the last 8 months was Skyrim on PC with over 270 mods (mostly new lands, quests and companions and some gameplay enhancements. I really like my Series X and Gamepass and all the stuff Xbox offers, but there's nothing that comes close to that nearly 11 years old game with no mod limitations.
Thank you for your continued high quality and informative videos!
I love how this video was essentially a direct result of responding to the community. Hope you guys reach 1M subs soon
Thank you! Very good content for those who are looking into this theme. The modern 6 cores CPUs can take the "load" of this simple tasks.
The kind of benchmarking people actually need 😳
*think they need :P
Ya bro
You watch too much TechDeals, we don't need this at all. Anyone with half a brain knows tasks like this use hardly any resources. If you needed this video to know this you clearly couldn't tell your pc ran any slower.
@@registeredblindgamer4350 true as well, but I'm not upset with the upload of this video
Keep up the good work guys! Love the content! Really useful information.
I do livestreaming tests on a single PC system.. there are instances wherein extra cores do play a role even when using the GPU to encode (NVENC, H.264, Quicksync), of course this only applies to 'Warzone', the rest of the competitive games seem to be stable as what you have shown here. Great work as always, steve 👍
Great video, its good you acknowledge how peoples use cases differ and you mention modern processors. My 4770K struggles in games like Valorant and Tarkov, and there is a noticable improvement when not screen sharing and not watching UA-cam videos in the background. Just watching this video on UA-cam 1080p 60FPS on Firefox uses 10% with spikes to 20% CPU Usage.
Love the videos lately. Y'all have become my current favorite tech channel! LOVE the monitor dedicated content as well
Thank you for this! I wonder if this was an issue with immature OS's and weaker hardware. As cache, virtualized sandboxes, and scheduling improves, less is demand is put on having dedicated cores.
So glad you did this, its a much more relevant set of tests for real use. Great info!
Upvoted even before watching, this is testing that we really want! Thank you HWunboxed team!
Excellent and interesting content as always. :D What I noticed if you have no VP9 Decoder for UA-cam (like my vega 56), the CPU usage skyrockets.
Yeah, the Vega 64 paired with a 2500K actually is NOT capable of playing back 4k60 by its own (so no playing a game even needed to knock down such a PC).
@@twinjuke even a 4670K stuttered for me on 4k60.
Now I want to test my old laptops and see if they can handle 4k60. Not that there's ever a reason to play high def videos while doing other things on an old machine, but just to see how old it has to be to struggle with it.
I don't know for Chrome but Firefox have some extensions to force YT providing H.264 content instead.
@@PainterVierax I think forcing youtube to provide h264 will also cause it to serve you 1080p60fps max, instead of 4k60fps, regardless of the browser used.
Excellent content, this is the type of workloads folks try to use their higher core CPUs for. Albeit the biggest test would be OBS and the various x264 CPU options with the visual advaned audio/visual effects processing options that might not be GPU accelerated, scaling, etc...
thank you so much for testing this, although i would happily watch more content from this subject and stuffs similar to it.
Thank you so much for this! this is what i needed for my planned build!
You should compare cpu utilization and talk about overall system smoothness when multitasking. Since 6 core cpus are already well into the hyperthreading most of the time in high demanding games.
Thanks for the testing. Good to have clarification on what has more of an impact.
Looks like in going to be putting down a fat wad of cash for a crazy GPU and skipping out on getting a new CPU for now. I have 3 screens and the 5700XT is really feeling inadequate.
It’s nice to get clarity on this, thanks :)
excellent testing Steve. Thank you.
Great video! The real world practical benchmarking that we need. Would love to see more like this testing but with lots of open memory intensive applications and a ram intensive game to test how pagefiles are managed for background apps and how that affects gaming performance when low on memory.
Here's a question: Was the GPU decode/accelleration turned on? Was the limit to frame rate where it was close to equal down to GPU load not CPU?
Edit: Testing one each of CPU and GPU limited games in the same circumstances (YT plus discord screen share) to see whether CPU and GPU rendering affects things more. As another commenter said, streaming a video in a format your GPU doesn't handle in hardware can seriously spike CPU usage. So this test may well benefit from tests on a more limited GPU.
Superb work, this answers alot of questions for me.
You guys,
Who work in this field are literally thee only ones who game and have exports running in the background.
I find it amazing that you often cater to that small audience and it goes too show how hard you work!!!
I'd like to see 1080p recording/live streaming with OBS or similar added to this kind of benchmark.
If it's done on the gpu it won't (under normal circumstances) have any effect at all. It's a separate encoding chip on Nvidia cards.
If you're doing cpu encoding however... then the result is obvious. Cpu encoding uses as much of the cpu as it can. If 99% is available it'll use that. So what exactly is the question? Would using 99% of your cpu affect your gaming fps? Yeah. You can't have both running at the same time unless you lower the quality in one of them. So you'll always be sacrificing something. Either fps or streaming/encoding quality.
The only way around this is to use a very powerful cpu which can handle the encoding at 1080 or higher. But if you've got an expensive computer you've already paid to solve that problem.
Nice. This is the Type of Benchmark every Streamer and UA-camr needs. Playing a Game + several apps running in the Background like OBS, Afterburner, Restream Chat, Discord / TS3, etc...
that's just a normal gaming user.
@@edh615
a normal Gamer who doesn't do content, doesn't run OBS, nor Restream CHat and when he really wants to immerse himself into the Game, he turns off Discord/ TS too.
Been waiting for something like this. Thank you!
Thank you Steve, been looking forward to a video like this.
I think Steve is underestimating what some of us have running in the background... I almost never have just a single UA-cam video in the background. It's usually like 3 tabs of UA-cam, a couple of tabs of Twitch, Facebook, Instagram, Twitter, a bunch of tabs of things I was working on that I can't quite close yet (mostly text), a couple tabs of gmail on my personal account and business account, a few google spreadsheets...
And that's when I'm not actively compiling or running code or using a VM... etc
Amazing benchmark, basically my experience with the 12400F, sure there's a FPS drop, but never relevant enough for me to even notice while playing. And if I notice a situation where this is the case, I can just well... close everything else but the game lol. No biggie, the 6-core is working for my needs pretty well.
I know this is not the same scenario. However, I upgraded my brother pc from ryzen 3 1200 to 5 1500 and what I realy did notice was not the gaming performance but the other applications. Windows gives all the resourses to the game, not the side applications running. I often hear him through the wall but not in discord. A half working solution to that problem was assinging a higher priority to discord in task bar. After ryzen 5, this problem was gone even with normal priority. So I think what you need is some extra resources for any random/running application for a smooth experience.
Thanks, this is really interesting data even though it shows minimal difference. It's nice to see this made in response to the community question.
epic video thanks. i was lookin for this exact test.
Great video! I would love to see some follow-up with some 2000 or 3000 series parts in the future to see how they hold up
Ha! yeahhh, expecting these guys to develop a multitasking benchmark or follow up on anything will just leave us disappointed.
This is among a very few list of videos which provide me with the information I actually need, as opposed to the game FPS tests that make up for 99% of all youtube tests. I am moving from a rx570 to a APU 5700g because I no longer game. I am moving on with my life to only photoshop and blender, as well as moving onto a solar powered electrical source, so being able to undervolt with a low base TDP while keeping the capacity to process infrequently large files is going to be essential for me, all while keeping a GPU function close to that which have 2gb.
or you can just buy a laptop for that
@@gabriellakotainegabor1902 no
A nice start to the topic, and I like that you take the time to investigate people's assumptions even when you don't agree with them.
As you said many of the games tested were lightly threaded games and so extra would do nothing. I would like to see more testing in games where all of the CPU cores are hit hard. The examples that come to mind are the end of Fortnite team rumbles matches where the settings are all low except for render scaling (100%), textures (high) and view distance (epic). Other examples are Shadow of the Tomb raider and Baldars Gate 3.
This is why I subscribed to this channel. Great info, thank you!
Thank you for this! I think tech deals channel is responsible for a lot of worry about this (for me at least) as he kept hammering that 6 core in general would be struggling in a few years (a few years ago :) ), although it's been a while since I've watched his content.
I think we are already living in a time if there will be jump in performance, it will be minuscule.
The only way to improve performance today is either going full quantum, increase TDP and power consumption, change SoC, or everything combined.
So, your current hardware, I think will still relevant in 5 - 7 years unless somehow Intel/AMD/Nvidia go nuts and move to new, more efficient architecture.
We have seen x86, x64, and so on..., now ARM is a big contender.
Let's see if Intel, AMD, Nvidia, and Microsoft can combine effort to make a new architecture.
Where is Linux? They can improve themselves, they are impossible to kill and evolving everyday
Yeah. He trolled me for using a perfectly working dual core laptop for office work. He thinks everyone needs atleast 8 cores and 64 gb ram.
Well, even games like RDR2 locked at 60 FPS 1440p HUB settings can use more than 6 threads.
More is better, most of the time with PC resources. I only game though, no other applications are opened while I game.
Never really liked Tech Deals, dunno, his opinions don't sync with mine I guess
@@takehirolol5962 not necessarily true. Just because games will use the cores doesn't mean more is better. A 5600X will destroy anything in the first 3 generations of Ryzen for gaming.
I've been gaming while watching UA-cam ever since I bought my second monitor. Even on my old i5 and GTX 1060 it was a fine experience. I mainly watched 1080p stuff.
Nowadays with my i7-8700K 6-core CPU and a RTX 3080 Ti I play high-resolution games and watch a ton of UA-cam on the side, even 4K videos and even if I lose a frame or two I don't really care. It's wonderful to just turn off my mind and have some noise while grinding in game.
No one asked
@@usonlyus7148 jealous
Why do you even waste your time on such games??? Jesus....If your game does not immerse you fully and you not only can, but want to have other backround stuff running, you are literally wasting your LIFE.
@gabriellakotainegabor1902 how is he wasting his life if he's doing what he wants to do?
Excellent quick peek into a very simple but very testable concept!
So glad you did this!
Amazing work! But I want to add that sometimes people use software decoding, which uses up quite a lot of CPU power, and I regularly use software decoding on my system to watch videos and streams. This is due to the fact that when I use HW decoding and my GPU is at a 100% load, video feed becomes extremely choppy or straight up locks up. This is probably due to an older GPU, as I use a 1080 Ti with a 5900x, or maybe due to my browser (firefox)
I'd like if you were to explore this possibility. As a sidenote, I can certifiably say, that a 5900x is not enough to watch 8k60fps videos.
Firefox doesn't behave well with long yt videos or streams. It eats memory till it becomes unusable. At least it does on my system.
@@batbiebel I've noticed that too
well.. no, if your GPU is at 100% load, it obviously has no time to do additional work. its going to drop frames if you are already at 100% usage and then ask it to do ANYTHING else. Also, Nvidia cards use a Software Scheduler that runs on the CPU, so if your CPU is fully utilized, your video card will start dropping frames wildly because the CPU doesnt have available time to manage the GPU. That being said, the 5900x shouldnt have any problem with the 1080ti. I've got my 1080ti driven by a ryzen 1700x, and it can fully max the GPU. The 1080Ti is a great card, but if any card is at 100%, expecting it to be smooth sailing when you introduce additional workloads to the device isnt going to go well. hehe.
Thanks Steve for making this detailed video. Would it be exciting to cover a 9900k revisit? Now that we are on heels on Raptor Lake and Zen 4 and many are looking to upgrade.
Nice informative series !!
i love the fact that you test ACC, you're the only channel i see testing, and i love it because i race there so much :D
it’d be interesting to see if this is the same behaviour on Alder Lake E-cores as those have been sold as being good for background tasks. You could test the i5-12600K vs the i5-12600 (or ideally the i5-12490F as it has the larger L3 cache).
You see the point is, even with P cores alone can handle ALL light-threaded background tasks, which negates the point of E cores taking care of background while P cores concentrate on 'important' tasks. Hell even many P cores sit in idle past certain point as the video suggests.
The question is then, would E cores save more power than just using P cores for all the tasks? Unfortunately no. As much as marketing guys in Intel want you to believe E cores are 'efficient', E cores are just as power hungry as P cores.
So what the hell E cores are for? For all core performance such as rendering where latency isn't the issue. Other than that E cores aren't useful for saving energy contrary to Intel's marketing.
Active applications trigger i/o and access memory.
Steve explained spare cores idling DO NOT help, when you have lightly loaded cores already.
This obsession with core count is simply wrong.
An excellent video, thanks! What I'd like to see is how much Unreal Engine 5 + DirectStorage in games (when they arrive) changes the calculus here. How many cores will those games like to see?
Unreal 5 only uses 8 cores. So not much change there. Unless they actually decide to use more threads and go into Async style of cpu utilization. Quite disappointing really. But considering PS5 and XboX X devs have the ability to limit the consoles to use only 8 cores instead of 16. I think we will see 8 cores be the max for a while still.
Not sure if my question is similar, but I'm building a PC for game development (including UE5 as one of the engines I consider using) and I wonder if the 5700X is worth the cost compared to the 5600.
@@MrGraywolf09 Going forward you will need a 8gb vram card minimum to run at 30 fps. So a 5700xt and a 1080 will run at that on the example/sample projects. You can get away with a 6600xt/6650xt and a 3060/ti but running the project at medium settings will be alright. A basic project takes 2gb of ram with barely anything in it. If you can save up and get a 6700xt or 6800xt you should be fine for say 6 years if it lasts that long. You could pick the equivalent level card from nvidia. 64gb of ram or more with 16 cores cpu. 8 cores 16 threads. nvme recommended going forward something like 256gb os and 2tb. You can get away with hard drive but it will take minutes to load instead of seconds. UE5 is next gen so it has a heavy cost.
Thanks for making this!!
Great video. Your comment about overall performance rather than a strict core count is so true and absent from so many discussions.
I think the issue for using the 5600 is that it's such a powerful PC. In all the games tested there isn't one that is both CPU limited and using most of the cores. You'd have to pick a recent triple A game but also pair it with a top end GPU playing at 1080p low graphical settings, my only suggestion is Cyperpunk but if Watch Dogs Legion didn't have an issue I doubt Cyperpunk will either.
I think this will only be an issue for older CPUs. For example a few years ago I choose a 2nd hand i7-3960x over the i7-4790k. Both great CPUs, but I thought I'd prefer the 6 cores of the i7-3960x over the slight faster IPC speed of the i7-4790k. In a modern game the i7-4790k could see all cores heavily utilised where extra background tasks could hit the 1% low more than the i7-3960x, however we are talking about these CPUs playing games released 8-10 years after they came out.
Try it again with older CPUs which have 60-70% usage in a game. Also for a background task can you add recording the gameplay with Geforce Experience or the AMD equivilant.
You guys rock for doing this 👍🏻
I have a 5800X with a TUF RX6800XT, that I use for both engineering work.. and gaming.. I still use my MAC Book Pro 2017, for UA-cam videos while I game or do any simulation/solidworks😂
Awesome benchmark!
Would be also cool to see a similiar one with lower end hardware, like CPU with 4 cores, with/without HT/SMT and some low end GPU as well!
Thanks for the video!
The outro bugged out. Great video as always:)
One of the best ideas for benchmarking I have seen in ages. Awesome as always. Thanks guys
This is strange. While watching this video at 1440p and playing at 1.5x, my GPU kicks up from 0%-1% to low 30s. I'm playing in an Edge browser with hardware acceleration turned off. This is on a 3440x1440 monitor. _NOTE:_ With Edge + no hardware acceleration, the GPU used Client Server Runtime Process.
When I switched to Brave _(hardware acceleration default on),_ the GPU utilization is in the low 40s. Same result with FireFox. The two top GPU usage apps were the browser + Desktop Window Manager, with the browser taking up most of the GPU usage.
When I used Edge + no hardware acceleration, my CPU usage was between 12% and 18%. With FireFox and Brave, my CPU usage was between 3% - 6%. When playing games, I use Edge because with hardware acceleration turned on, UA-cam would kick my quality down from 4k or 2k down to 360p every single time. Now it doesn't.
Props for having the patience to test this
More of this real world scenarios testing please, one thing you can try is perhaps just adding more and more background task and see when does it impact performance ? Maybe running adobe suite or video editing program and other workloads.
For Stream dedicated PCs, in my experience I would recommend getting a platform that can have a decent PCIe capacity, for example I decided to keep my 5820k from 6 years ago since it had a decent capacity, besides lots of PCIe connectors, which those came really handy when adding devices like for example, the Elgato 4K mk2 capture card. I currently run two 4K capture cards with a 1650 Super, so that i can encode with the NVENC part, as well as record in higher bitrate, using X264 to the streaming pc's hard drive.
Now since X99 is a bit of a rare platform unless you do a build from like aliexpress, I suggest that if you are looking more for a usual consumer CPU, something like the Ryzen 1600 will be plenty for that matter, and these can be found relatively cheap nowadays, as for integrated graphics CPUs, a Ryzen 3400G and 16gb of ram since you need a bit more ram for the iGPU, should do well as long as the encoding is set to X264 (Cpu based on OBS), with that is it possible to "get away" on not having to buy a dedicated GPU for a streaming only PC. As for motherboards, besides having enough VRM for your chosen CPU, if you want to future-proof it to later have a dedicated GPU + a PCIe Capture card like one of the mentioned above, a good way to find one easily, is that it is "Crossfire compatible", that way it will be easier to search for motherboards that have more than a x16 slot conector, which not all budget options tend to have.
Hope this helps the ones interested on Streaming dedicated PCs.
Nah, it'd be cheaper to add 2 to 4 cores to the existing platform or just use hardware encoding. An extra PC will add 500€ or so, which would basically upgrade you to a 5950X that will handle it all and you don't have to switch PCs all the time
@@cromefire_ as someone who has a 5900X system, i still kept my X99 because of the reasons mentioned above, besides your GPU still gets a performance hit as well since it needs to render the scenes for the other applications that you use too, it is not "CPU upgrade bound" only. Besides that, you will also be limited on PCIe bandwith as well, using M.2 drives and your GPU, is very likely that your motherboard wont have enough PCIe lanes alone for adding a 4K capture card, and if you do, then you would limiting the PCIe bandwith on your GPU already by half. The main purpose of using an external system is to have next to no performance impacts on your daily drive/gaming pc.
Also the upgrade that you suggest alone can mean as much as a simple, used but cost effective system enough to be used for a stream dedicated system, even more if you skip the capture card and use a LAN bandwith system like NDI for example.
@@XxDeViLBrInGeRxX Well you'd probably use the motherboard PCI-e for the capture card, my rough estimate would be that 4x should probably be enough. And even if not PCI-e x8 is still enough with a 3090 to only get like a very few percent. Regarding HW encoding: no HW encoding is done on dedicated silicon, so it's fine and with a 5950x it's negligible performance anyways. I mean after all you only need 1080p60 (some even do 800somthingP) for streaming anyway, at least on twitch, for proper 4K you have to go to UA-cam I think (which few do). After all I can play games fine, even when doing a software AV1, 16 thread encode, so even a 5700 or 5900X will probably be fine. Just be sure to pin the threads currently so they don't interfere and everything is fine. And I mean besides price you just loose feature wise too because things like window capture obviously don't work and you need the capturing card. On a single system a capture card only even starts to interfere if you both game on PC and console anyway as you don't need the the GPU with console gaming and you don't need the capture card with PC gaming. Add the second keyboard and amonitor that's for the second PC and it doesn't make much sense for most.
If I'd built a PC for gaming nowadays, I'd just get a i5-12600K (non-F) probably and just use QuickSync. Even with a second PCI-e slot used you will still have PCI-e 4.0x8 for current gen and 5.0x8 for next gen, so plenty of bandwidth and if you want to do CPU encoding just put OBS on the E-Core, video encoding is something they do great at (in fact E-Core ≈ 1 P-Core Hypethread), so think of it as a dedicated older quad-core for encoding only.
But of course you do you if a dedicated machine is what you desire nobody can stop you, it's just probably ~50% more expensive and I find it more annoying than the easy solution (which is a 12600K which is like 100-200€ more I think than a 12400F) and what you gain is ~5% performance in the best case, of which the stream won't see anything because they neither see 4K nor > 60Hz.
From personal experience with r5-3600 + rx480, i do feel performance loss in games like Division 2 or AC Odyssey if i have twitch or youtube open and not minimized. Somehow youtube and twitch does impact my games when games are gpu bottle-necked.
Love this channel man
Love the quality and flavor of your B-roll material!
What about 4c 8thr cpu like 12100 or 4100?
Discord had a bug where their active noise cancellation software would turn itself off due to "high system load" on Ryzen CPUs, but it has been fixed.
Something else that I have noticed is that video playback will drop a lot of frames while I am in a game that is maxing out my VRAM (I still have a 580 4gb)
It may be nice to test the performance loss when approaching 100% utilization of RAM or VRAM.
Any time you go near your max ram or vram things stop working well. That's to be expected. If you can juggle 5 balls at a time and someone says "can you please do 6?" what do you think will happen? You're going to fumble and drop some.
@@flameshana9 Obviously.
The point would be to measure how much of a performance impact there is, and how much ram/VRAM is enough for modern use cases.
This can be an issue with new hardware as well. For instance the 3080 10gb can run into performance issues at 4k in some games but not others. Potentially adding in a little bit of multitasking on a second monitor could push the card to it's VRAM limit
@@flameshana9 how many do you think your mom can handle?
Thanks for this. I always wondered this.
Great video. I have always been a Multi-tasker. I try to build my computers to handle Multi-Tasking. Thanks for making this video.
I want to see the performance hit on Discord using Discords GP acceleration feature.
Since the "invent" of tabs in browsers, it's not unusual for me to have 30+ tabs open. And because 16GB* was my minimum in a PC from 2012 or so (when 7970 was released), and now 32GB*, I never close the programs, I almost never reboot the PC (just put to sleep). I never experienced the drop in performance with these kind of background tasks. But when I have system uptime 10 days+, it sometimes happens that the whole system just start to stutter even if I close the browser with said 30+tabs and opened programs. That's when I have to reboot, or if Windows Update forces me first. But I was curious in hard numbers how much these background tasks affect performance. My guess was 0.1 and 1% would be much more affected then what was shown here. Nice to have a video to link when someone brings up this topic.
*Why I had a need for 16GB and now 32GB RAM? Because I'm a software developer, and need the RAM. And because I'm lazy and leave all the applications open, until Windows forces me to reboot.
nice clown, I bet it's so practical having bloat tabs mixed among those few you use 🤡🤡🤡🤡🤡🤡 30 tabs 🤣🤣
I think you have a tab problem, might want to seek some help.
@@MrNota500 I know. But it's easier to buy more RAM and stay lazy. I have 2 monitor setup for software development. I have so many applications opened all time. Visual Studio, our company collaboration tools, Opera with 30+tabs,...Why would I close applications? It doesn't hurt performance. Except when Windows fails and have to restart.
Yup having 50+ tabs here i defintely notice an issue with browser memory leaks, after 6-8hrs ro sometimes 28-32hrs that ram usage will creep up over time despite having the same amount of tabs or going a bit less.
@@scarletspidernz I have a different problem. Even if I close all the apps, and have the RAM freed from everything, sometimes Display Drivers start misbehaving. This can result in poor gaming performance, like stuttering, low FPS or outright fail to open a game. Or if I open a browser with only 1 tab, and go to youtube, playback will start pausing (not buffering). Then youtube downgrades a resolution, even to the lowest 144p, video doesn't want to play normally. This happens when the System Uptime is very long (weeks), when PC went through a couple of weeks of daily Sleep, working, Sleep,... Only restart helps. This happens with some drivers more often. I must say that I haven't experienced this for a couple of months now. As I understand, AMD claims they have rewritten their graphics driver almost from the ground up, starting with May Radeon drivers. Maybe this had something to do with it.
Amazing video as always!!! Can't wait to see Hardware Unboxed reviews of the new CPU and GPU to come.... (not talking about intel GPU of course, those are a complete mistery, even for intel).
Very interesting analysis!
I feel like having videos open while playing a game causes more input lag and latency issues as the GPU is doing more things at once. Could be an interesting thing to test out.
Having just bought a 5700x to upgrade from a 1600, I'm glad it was only $25 more than the 5600.
Wow, I have never seen another channel benchmark my two most played games: Cities Skylines and StarCraft 2!!! I am going to build my first PC later this year and it is really helpful to hear and see how those games utilize the CPU from an expert. Thank you so much!!!
Great content, keep going!
Question Steve:
Were you using all these applications with Hardware Acceleration off? Hardware Acceleration uses the GPU instead of the CPU to handle most of the tasks. That's more likely what the comments meant because even with a 5900x I can use up to 20% of my CPU with just Firefox running with multiple tabs and a 4k video running with HA off.
For YT or Twitch watching, it means if the GPU is maxed out in game, it'll most likely lag or freeze the video. Whereas you'll never typically hit 100% CPU usage in game so that's why I opt to not use hardware acceleration.
PS: RGB software would be interesting to test in the background as well. From the Gamers Nexus video we've seen some can be very resource hogs even in the background.
"For YT or Twitch watching, it means if the GPU is maxed out in game, it'll most likely lag or freeze the video." Actually, this is extremely unlikely. The video encoder/decoder(s) (a.k.a. "media engine") is a separate sub-unit/processor. Not specific to Nvidia but in their words, "NVIDIA GPUs contain one or more hardware-based decoder and encoder(s) (separate from the CUDA cores) which provides fully-accelerated hardware-based video decoding and encoding for several popular codecs. With decoding/encoding offloaded, the graphics engine and the CPU are free for other operations." Additionally, if you have a non-F Intel CPU, you could enable the IGP and perhaps move the decoding to that "engine" or split the workload up further (assuming an app allows choosing which hardware decoder to utilize).
What gpu are you using? As long as your gpu has support for the video decoder that should never be an issue.
Re-test this using games and apps actually known for their multi-threadedness.
Then also test default scheduler vs. manually adjusted core affinity.
I'm genuinely surprised, but the method looks sound, and conclusions are solid. Thanks Steve!
good stuff, more of this pls. maby add even more loads.
this testing i like
As a Vtuber I actually run dual PC's because how intensive the multi-tasking is when streaming... But that's because a lot of things are running: OBS (10-15%), Vtube Studio (20-30%), audio player (2-5%), Web Browser (5-10%), even with a second PC that is doing this so the other system can game you sometimes add other things as well like a video capture program (used for lighting the model) or the Twitch Integrated Throwing System. I have run this all on one PC, but the experience was horrible.
Why don't you just show yourself instead of some 3D model? Never got this "Vtuber" thing. I suppose I get it if you're too shy to do so, but you'd be great, I promise :3
Might wanna a include some hardware specs instead of just cpu usage....
Web Browser at 5-10%, i call BS, I have 30-60 tabs open at any giving time and I only see at most 5%, and this is with a laptop processor, i9 10980hk and 64gb of ram. I do movie editing for part time work.
@@crylune Because Twitch is ostensibly marketing yourself, and being “normal” doesn’t stand out when you want new people to join your stream.
@@crylune Because that's not an option if I value my security. Showing my face means any crazy (including all to many of my neighbors) can find out way to much information about me with enough time. Don't like Vtubers? Don't watch. That's on you, but I'll take being able to be myself online and not needing to worry about getting SWATed or harassed.
2 things i love about this channel
1- mando action figure.
2- listening to audiences.
love you guys.
Amazing work Steve. Although I expect nothing less of you I didn't expect you are this fast doing benchmarks. In the meantime I'm still waiting for your RX 6800 vs RTX 3070 Ti 50 games in 2022 benchmark blockbuster video!
Thank you for this. I feel that for streaming having an intel with an igpu helps. As the igpu handles the encoding the gpu can handle gaming or other tasks. That would be an interesting test to see.
I know it's too much, but as a long term project, it'd be great to see the Ryzen 1000/2000/3000 added to the sample. Great video! (:
How about flipping the question on it's head; how low on CPU power do you need to go for this to be an issue? 1st gen Ryzen? Older?
1600x 6 core here
I7 2600k here (to be replaced soon).
bet that all 4 core and lower processor gonna have harder task to handle all those things
@@jamzqool Even worse, I'm running at 4K on a 2070S ;)
The effect of Discord is negligible though - and I'm usually in a group voice chat.
I have done cpu based recording/streaming in obs (I think I did 720p 60fps fast preset) while gaming on my old 4770 and it used about 20%. If your just doing discord and youtube I doubt it would make a noticeable difference unless you are already on the edge of your fps target. My guess would be you would have to go back to a 4 thread cpu like the old i5s to see a big difference from discord and youtube (but those barely run modern games anyway so does it really matter?)
Best tech channel ever, nice content and relaxing ambient.
Holding up the coffee cup in the thumbnail had me laughing. Thanks for that. :-)
As long AMD don't fix their GPU encoder, anyone with a quest 2 and playing with oculus link or virtual desktop will need Nvenc or more CPU cores. Niche case? Yes. Just gaming? Also yes.
How about.. what difference does it make to use NVENC to record, stream on discord or on twitch, compared to not using it with various cards? Since it’s the GPU that does the work here, it would be really interesting to see how it scales across all cards. Of course AMD uses something different but it would be good to see that too!
This video is a CPU test not a GPU
The results were expected (for me at least), but thanks for testing and providing the proof!
The best content on UA-cam!! You guys are the best.