What Other Reviewers DON'T Tell You...AMD Ryzen 7 9800X3D VS Intel Core Ultra 9 285K

[@TechArc_]

Hi everyone! 👋 I know using an RTX 4070 for these tests isn’t ideal for comparing high-end CPUs like the Ryzen 7 9800X3D and Intel Core Ultra 9 285K. However, as a small channel working within a tight budget, it’s the best GPU I currently have available. To offset this limitation, I focused on CPU-bound scenarios first (at 720p and 1080p) to highlight processor performance, then included GPU-bound games for a more balanced perspective. As the channel grows, I’m committed to upgrading my setup and testing with higher-end GPUs in the future. Your feedback and support make that possible, and I truly appreciate it! Let me know your thoughts or suggestions below.

[@TooBokoo]

To be fair, your test scenario is a bit more realistic to the everyday user. Most people don't have a 4090, so your tests results are something more common folk would be interested to know.

[@pinakijana9479]

​@@TooBokooMost people who can't afford a 4090 won't buy a 9800x3D or a 285k

[@TooBokoo]

@@pinakijana9479 Why not? I just bought a 9800X3D last week for $439. I wanted a monster CPU to pair with my 7900XTX. That makes a pretty nice combo and costed less together than a 4090 by itself. I could have bought a 4090, but $1,600 isn't worth it for any GPU, in my opinion. The 9800X3D doesn't necessarily have to be paired with the absolute top-end GPU to make sense.

[@BurnMyBread9]

@@pinakijana9479 They absolutely would. Nobody playing on a 4090 is aiming for maximizing 1080p. At 4k, the GPU is the bottle neck and not the CPU so it really almost doesn't matter what CPU you pair with a 4090. Sure at 4k, a 9800x3d might squeeze out a few more FPS but not enough to warrant a $100 or $200 price difference. The real advantage is in that Esports range where you gain those insane 200 to 300 fps differences.

[@pinakijana9479]

@@BurnMyBread9 When you have a 4090, I gues you atleast have a $4000 system. I don't think you'd want an inferior cpu to save 100 bucks.

[@MattCookOregon]

i cant even find a 285k anywhere man

[@wesleybaumgardner3740]

Scalpers snapped up all of the 9800s and 285ks, both companies said they are releasing more.

[@MattCookOregon]

@wesleybaumgardner3740 waiting for release of more and update that fixes everything

[@wesleybaumgardner3740]

@@MattCookOregon I have been looking for more information on why the 285k is doing so poorly. side by side the 285k should out perform the 9800x3d, the two things the 9800x3d has that the 285k doesnt is that hyperthreading and the 64mb of L3 infinityfabric cache. I think I heard on GN that the cache on the 285k was performing way under its expected timings. The bios on a bunch of the motherboards do need updates, the ram was something else I was looking at, because Intel supports the newer CUDIMMS and from the tests I saw it was giving a performance gain, but with some really high CAS. That was in gaming and not some workflow where you might expect it. Maybe the firstword on the CUDIMMs are like 5ns or something. While digging into that, people were talking about the scheduling for the CPU and how its part of the problem. This I wasnt shocked at, as windows 11 is built more around raptor lake. Again with windows optimization I saw that win11 has a feature called core isolation (its off in win10 so I never had to deal with it.) higher end chips are losing 10-25% performance from that setting being on. I havent really went over all of the Wins 11 issues because I am still using a x99 board, I have a 5960x that is OC to 4.3, it was at 4.6 all core, but after 10 years its not that stable anymore. But due to no bios updates for the board, I have been stuck with 16gb of ram and thats BSOD my system. But rather than messing with that old system, I wanted to get a new one. the 285k is in bad shape, so I looked at 9800. But as you know they are both sold out or around 1k for one. Intels 300s will be on this same socket, but the 9800 is on the last gen AM5, so the next chip is the AM6. Sorry for the long post, I just wanted to pass on some info, because we are in the same place sort of. I did spec a whole AMD system out, with most of the top of the line stuff. minus the video card that comes out in jan. But it was about $1600, much cheaper than this old x99 im using, the CPU was $1200 lol, back when I had adult money 10 years ago. I have im broke money now.

[@kingeling]

Tha 4070 is without any doubt an awful bottleneck, even at 1080p...

[@TooBokoo]

Yeah, but it's interesting to see, nonetheless. Not many people have a 4090. Seeing these CPU differences with a card like the 4070 that far more people own is neat to see.

[@pinakijana9479]

​@@TooBokooYou should see Hardware Unboxed video about why they benchmark cpu at 1080p. If your GPU is not powerful you can lower some settings and boost your fps. But if your cpu is bad, it doesn't matter what you do, you'll never improve your fps. Also even if you don't find any differences now, future games will definitely broaden the performance difference between these cpu.

[@TooBokoo]

@@pinakijana9479 I've watched all their content. I can't take anything they say seriously. They test everything at 1080p low, which does make sense, however, at the same time, they say they don't test higher resolutions because they don't know what kind of hardware combinations people are using, so it wouldn't make sense to run tests at those higher resolutions. Well, using that same logic, you could argue that most people don't play at 1080p/low, so why test like that? You could also get even more ridiculous and ask them why they don't test their CPUs at 720p or 480p and really let the CPUs stretch their legs. And then add, well, you don't test at 720p/480p because nobody plays like that! The exact same argument you could make about 1080p/low testing. It just ends up being a circular argument with them. LOL! There are other KZbinrs who will put up higher resolution benchmarks because there are games where it does matter. Hardware Unboxed just seems to like picking the games where the CPU doesn't matter, which is kind of a dumb way to do it because there are many games that are CPU intensive, and it does matter, even at higher resolutions. As far as this video goes. It could still be useful for some people. Let's say there are viewers who have an 8-year-old CPU paired with their 4070 or equivalent AMD card. This video would show them what they could expect out of either CPU if they upgraded from their old one. That's a very specific scenario, but I guarantee there are thousands of 4070 users who currently have it paired with an older CPU, and might be interested in one of these two options. The video won't help anyone who wants to see the best raw numbers, but it will help people with these specific combinations of hardware.

[@killerbsting1621]

My 4080 Super and 9800x3d is far better than what I was using..7950x3d and 4080 super. I love it

[@pinakijana9479]

@@TooBokoo Lol! It's not about 1080p or 720p or 480p. It's about avoiding GPU bottlenecks. The highest resolution you can play all the games without any fear of GPU bottleneck is 1080p. Although GPUs like 4090 do not bottleneck at 1440p in most games, but in some games they might. That's why they do it at 1080p just to be safe. When GPUs become so powerful that they don't bottleneck at 4k in any game, then they'll benchmark at 4k. You wrote all that without even understanding the basics. They are just avoiding GPU bottlenecks so that they can see how far these CPUs can go. They also talk about what should you do when you already have a GPU and trying to buy a cpu. You should first see if the settings/resolution you play at, is GPU bottlenecked. If it's GPU bottlenecks then you don't need a new cpu. If it's cpu bottlenecked then just buy a good enough cpu, that will take your performance just upto the point where it's GPU bottlenecked again. The 9800x3D is for those people who only want the best of the best or planning to buy a high end GPU in the future.

[@SinTM]

Bro, your GPU is huge bottleneck

[@mcarthur05]

Then give him GPU hahahaha he already said he's out of budget..

[@tripd4949]

No it's not.... 4070 is fine.

[@KoseChris]

one of the few reviews i see with 0.1% lows

[@philwithcheese]

Testing in a GPU bottleneck proves nothing.

[@OptimizingNetwork]

Looks like to me that your 9800x3D is not an issue, but you have some serious issues with how you configured a system I don't even drop into 30s with 7950x3D and it is supposed to be weaker than 9800x3D in gaming and better or equal to 7800x3D (as long as you schedule 7950x3D right) And I am talking about Red Dead Redemption 2 Something screwy going on there, 100%. Low of 4 fps? Yeah...something is really wrong, like a broken GPU, badly configured system or dying PSU. I am replacing GPU and PSU, because both failed and it would drop to 0 fps in same game when playing with any CPU Your issue is deeper than the CPU, unless CPU is failing already

[@kingeling]

No, it's just poorly timed polling or a bug in the software. The frame time graph seems perfectly fine. If the percentiles were consistently that bad, we would be seeing consistent hitches in the graph. Btw it's .1% lows, not 100% lol

[@whosdr]

@@kingeling You misread part of their comment.

[@selohcin]

@@kingeling What do you mean by "poorly timed polling"?

[@kingeling]

@@selohcin That they started the OSD logging before the level loaded fully. Loading screens and moments when shaders are being loaded at the start have no frame time or terrible frame times, they shouldn't be included in the measurements.

[@kerotomas1]

Bottlenecking the 9800X3D with a crappy 4070 just to claim close to parity with the more expensive and more power hungry 285K lmao

[@oysterhead5150]

The 285K needs its price down to 500 to be slightly competitive

[@12Music72]

The only reviewer where the 285k beat the 9800x3d in FH5. He also measured the 285k to be basically on par in CP2077. Whereas everyone else measured the 9800x3d to be north of 30% faster in that title. Which also is just really odd. Which makes me doubt TechArc's review. @TechArc. If you're the only one experiencing this, then it's more than likely your configuration is the issue. "What other reviewers don't tell you." is more like "I'm having an issue others are not having"

[@WiseOldDudee]

4:24 I dunno dude.I'm a mouse and keyboard user.If I had that frametime I would definitely notice the jutter caused by those consecutive small frametime spikes :/

[@astralpurpose1344]

9800x3d has better 1%lows with 6400cl32 ram.

[@Bass.Player]

The memory choice is way too low for the intel CPU and it looks like it might have been throttling based on those spikes? I have never seen that before with this CPU...

[@kerotomas1]

throttling has nothing to do with memory, also many reviewers tested that above 6400Mhz games do not benefit of higher speed memory at all if anything the performance is worse due to higher timings

[@Bass.Player]

@@kerotomas1 No one said throttling had anything to do with memory! Do you own a Core Ultra? No I bet you don't! The memory controller runs at the same speed as the memory so yeah it matters. I'm running 8000 CL34 on mine...

[@kerotomas1]

@ okay you do you

[@photonboy999]

INTEL spokesperson has talked about this new CPU generation, so that's something people may want to watch... basically, they need to push BIOS updates for power profiles, BIOS updates to push CPU firmware updates and improve Windows drivers. This Intel generation may be ready in HARDWARE but the firmware and software was not. They SHOULD have waited, but they wanted to move on from the 13/14th gen debacle and have now just made things WORSE.

[@tripd4949]

I still prefer the 285k Intel.

[@jaycrusnik6620]

Pretty sure amd will continue to dominate for a while this is not intel's finest work.

[@diagnotics7470]

Bro if u are playing 4k game with 4090 all are same intel or amd.. 9800x3d is only good for 1080p gaming.. and if you have 4070 or 4080 it can cause bottle neck. But think this.. if you have 4090 gpu how many times u are playing 4k how many times 1080p? So it is meaningless

[@Tech-Tide2387]

with CPU bound games 4070 will still be a bottle neck

[@megadooooom6582]

wow intel fanboys crying for a while now dude appreciate the tech not the companies i have both systems and i can tell the difference by my own eyes not with charts My X3D AMD system has so much smoother frame rates and almost no stutter than my intel system

[@aurias01]

Same. My 7800x3d can't keep up with my 14700k when it comes to anything other than gaming. The 7800x3d wins in gaming by like 1-2% but loses by 50% on anything else

[@megadooooom6582]

@@aurias01 another fan boy tell what this video is about?

[@paulboyce8537]

INTEL APO 10-50% gains in supported titles. This needs INTELL DTT. Intel DTT is supported on systems with: 11th Generation Intel Processors and newer Intel Iris Xe MAX Graphics or Intel Arc Graphics So when you see a review claiming that APO does nothing no gain +-5% on 4090 Nvidia GPU they have no clue because the 4090 is not supported. Same goes with E Cores/REBAR on AMD/Nvidia GPU's +-5%. On all INTEL system this is very different. That's why you see no Hyper-Threading Technology on Arrow Lake but very strong E Cores. APO will age like fine wine. Just like XeSS XMX support was very limited but today over 200 titles supported. INTEL you need to understand that 285k won't run it best with AMD/Nvidia GPU. 1. PCIe 5.0 direct access to CPU with 8 pin power. REBAR on steroids that will give a huge performance gain. 2. CUDIMM lifting the CPU speed closer to GPU allowing higher pairing. Memory speed is vital. 3. ARC from Xe to Xe2 giving 50% gain. Plus better VRAM. 4. Paired top ARC you get APO that is 10-50% gain on supported titles. 5. ARC Xe from 225W to Xe2 250-275W for more performance. That's doubling the performance. 6. Arrow Lake PL2 Extreme 295W 7. No Hyper-Threading Technology making way to very powerful E Cores. We get 570W all up working for GPU pairing with architecture giving all good frames with no waiting/queuing where 60 FPS at 4k is better than 120 FPS from AMD/Nvidia where more than half the frames are broken or out of time giving stutter. 5080 is 400W card predicted and on paper the all INTEL system looks as good if not better. But with $1000 less coin. Just note that all of these INTEL GPU's need PCIe 5.0 direct access to CPU Z890 motherboard and CUDIMM fast memory. INTEL is holding back till they get a fix for the Arrow Lake for win11/BIOS's/micro code. This looks like it will be timed at next month with the launch of the Battlemage. Taking in count the difference in architecture based on B580 12gb VRAM 192 bit bus 2??W specs that leaked: A770 = 4070ti (known) B580 = 4070ti B750 = 4070ti super B770 = 5080 Upcoming GPU shortage will limit what you get. Nvidia GPU's are gone and AMD is not that desirable option. This leaves INTEL and upcoming Battlemage. The estimated pricing is $200-500 but here is the kick. Battlemage supports PCIe 5.0 direct access to CPU. So you need Z890 motherboard that has PCIe 5.0 with Arrow Lake and CUDIMM clocked memory for high speed. This is smart from INTEL selling the whole system. So it is not really comparable to AMD when for one it is restricted from supported CUDIMM/XMP enabled to UDIMM that in fact is downgraded to 5600 speed. It's like leaving the handbrake on. Adding to that you are not getting the APO. You are not utilizing the E Cores nor the direct access to CPU via PCIe 5.0. And running win11 that doesn't fully support Arrow Lake yet and BIOS's that are lacking and upcoming micro core update that all in has a potential to level the competition. Yet in all INTEL system you are getting the full potential.

[@Nevertobebad]

Honestly, Mid take.

[@aurias01]

Who is buying a $450+ cpu to play in 720p?

[@wesleybaumgardner3740]

Normally with a highend card they lower it to 1080/720 so the card isnt the factor and they can focus on the CPU, a few channels have full videos explaining why they do it this way.

[@JohnDoe-tv4zd]

and amd half the power

[@aurias01]

Also half the score on anything outside of gaming. In gaming it has a 2-10% lead in best case scenario