Horizen - ZEN

RTX 4080, RX 7900 XT, Arrow Lake vs Zen 5, 2022 Shopping Advice | Tech YES City | Broken Silicon 179

YouTube video

Bryan from Tech YES City joins to discuss Nvidia RTX 4080 Performance, RDNA 3, and more!
SPON: Get 10% off Healthy, High Protein Ramen w/ code “brokensilicon”: https://bit.ly/3wKx6v1
SPON: Atlas VPN BLACK FRIDAY https://atlasv.pn/MOORESLAW is $1.70/month + 6 mo FREE!

0:00 Who is Bryan?
4:28 What has changed in the Techtuber space over the last 10 years?
8:14 Why aren’t there Low End CPUs anymore? Are budget builds dead?
28:14 RTX 4080 16GB Performance
37:02 Does Nvidia need to drop the 4080’s price? Is AMD hiding something?
47:41 RX 7900 XT 20GB Performance Expectations
55:45 How good does the 7900 XTX need to be to take market share?
1:03:41 Is the 7900 XT overpriced? Does RDNA 3 have a clock speed flaw?
1:12:37 Would the 4080 12GB been ok if it was called a “4070”?
1:17:56 Will RDNA 4 & Blackwell hit a rasterization performance wall?
1:25:31 Are Zen 4 & Raptor Lake pointlessly good at gaming?
1:31:21 Intel Meteor Lake Desktop – Should Intel even bother?
1:38:53 Arrow Lake vs Zen 5 – What do we actually want from them?
1:48:37 Who should buy Zen 3 now? Who should wait for Zen 4 X3D or ARL?
1:56:14 Best Used Professional GPUs
1:59:53 Buying Used Parts in Japan vs Europe vs United States
2:05:43 Is Japan’s DIY community growing?
2:10:34 Is the DIY market down overall?

Support MLID by getting a FREE SSD from Microcenter: https://micro.center/brokensilicon

Cheap Windows & Gaming Keys:
25% software discount code: brokensilicon
Windows10 Pro ($16): https://biitt.ly/shbSk
Windows11Pro ($22): https://biitt.ly/shZBP
Office2021($53): https://biitt.ly/vWd96
3% EVERYTHING discount code: dieshrink

Get 10% off Vite Ramen w/ “brokensilicon” at the following link: https://bit.ly/3oyv4tR

Patreon: https://www.patreon.com/MooresLawIsDead
US Amazon Affiliate Link: https://amzn.to/2KpXaTm
AU Amazon Affliliate Link: https://amzn.to/34ZkLqS
UK Amazon Affliliate Link: https://amzn.to/33kllix
DE Amazon Affliliate Link: https://amzn.to/3A7FO5W

Address Fan Mail to:
Moore’s Law Is Dead,
PO Box 60632
Nashville, TN 37206

Broken Silicon [A PC Hardware Podcast] is available on ALL platforms.

Twitter: https://twitter.com/mooreslawisdead

All Music by SAHARA:

Broken Silicon Logo by Tom.
Broken Silicon burning graphic by Tom & Sulphurous:
2021 Video Intro by Tom & abstractrealism on the MLID Discord

Tech Yes City: https://www.youtube.com/channel/UC9Tn-atYOt8qZP-oqui7bhw
MLID Panther / Lunar / Arrow / Meteor Lake Leak: https://youtu.be/8NFc1bumTwc
MLID Early RTX 4070 Leak (including pictures): https://youtu.be/GZ3Lw0WW99A
$500 Nvidia A4000 16GB: https://www.ebay.com/itm/374216243085


ArabicChinese (Simplified)DutchEnglishFrenchGermanIndonesianItalianJapanesePortugueseRussianSpanishTurkish

Related Articles


  1. Re: the the "4080 12gb" having the same die size as the 1080,

    I mean, even with their cherry picked benchmarks at the 4000 series preview, the 4080 12gb was only beating the 3080 by maybe 20-25% on average. The 3080's gains over the 2080 were around 40%. Even if they had priced the 4080 12gb at $700, there might not have been such a huge backlash, but would have been pretty disappointing. Just like how the 4080 16gb is disappointing.

  2. 40:00 Good point but they could also be a tactical move to make sure NVIDIA doesn't last minute lower their prices when they realise AMD is doing better.

  3. I am surprised not more people have pointed out the real-world benefits of DP 2.1 over HDMI X.anything, given how common multimonitor setups are these days. 8k100-something isn't even a novelty yet/at this point. But 4k sure is. And running 3x 4k120(/144)hz displays, mouse, keyboard, mic, high-res HBR audio, camera, maybe some nvme external storage – All off one single cable from your GPU… Now THAT is interesting. And only DP2.1 can do that. Because _DisplayPort supports daisychaining! DP2.1 doesn't just do display and audio passthrough. It also has dedicated links / "lanes" for USB4 passthrough. Including Thunderbolt 3, and possibly Thunderbolt 4 eventually.

    And for the big simulator gamers out there, you can top this daisychaining and passthrough bonanza off with the fact that AMD doesn't limit their "ordinary" Radeon cards to 4 displays like Nvidia do with their GeForce cards.

  4. There's plenty good budget options still, they just shifted "the model". Over the last 3-4 years demand side if the market has shown that there's more buyers with bigger pockets than what TSMC, Intel and Samsung combined can ever hope to accommodate.

    So instead of AMD, Intel etc spending billions extra on development, fab work, promotion, packaging/assembly and logistics, to make budget models of the latest gen – they just keep making the former gen, and sell it much cheaper. The RnD, all the big one-time sunken costs, the marketing, all the early driver and microcode bugfixing – all of that is already done and paid off.

    The former gen IS the budget range for value-hunting customers now. And overall that working really well. Because you pay a much more comfortable price, but you don't actually make the kind of compromises as you used to with budget parts back in the day. Let me just remind you, that it wasn't until 10th gen barely a few years ago, that you got hyperthreading on an i5. 10600K was the first time any intel chip cheaper than an i7 got more than 6 threads.

    IMO the only reason things are a little bit strange around budget stuff around this time, is the introduction of DDR5. Or more precisely, that some people want a really cheap budget system – but also expect it to be newest gen including DDR5. And thst3not realistic for at least another 18-24 months. But the good news is – a 5800X, 5900X or 5800X3D, a B550 board and some stinking fast but cheap (no rgb) 3600MHz CL14 ram – is gonna be just as fast a year or 2 from now as it is today. The former gens don't really get slower anymore, there's just so much grunt on tap in modern chips vs what modern games and software needs to run well. Well, unless you're buying pentium or celeron parts I guess.

  5. So raytracing is still pointless because the games people really play dont support it or the hit is not worth it. So to say raytracing is a selling point is weird.

  6. I've bought a fair bit of stuff for my Clevo laptops off aliexpress. Parts are cheap and I've never had any issue, not even with shipping damage, but there's the potential there for dodgy. You need to watch for some things, like for screen panels they list a specific model number and hardware revision and bury in the fine print "we will send compatible" (this is also an Ebay thing). You treat them like its gumtree, message them and make clear what you want, let them make the call whether they will send exactly what you want, before you commit to the sale. It's pretty clear which sellers are normal and which dodge the issue and won't answer your question.
    And double-reverse-translate Chinese to check your meaning if you use taobao. Google translate sometimes horribly mangles things and some English sentence structures fail completely.

  7. Don't know why amd used those names, it really is the 7900 XT and 7800 XT, that XTX should have been used if they would have been competing on equal terms with the 4090.

  8. Can confirm, I'm the friend. Gigabyte 3080 for $500AUD, changed the thermal paste and memory pads and dropped 30c. Cant get better value!

  9. Didn't know Brian is such an Nvidia fanboy. He didn't seem to understand that the RX 7900 XTX is aiming for the 4080 instead of the 4090. And it's straight up ignorant to say that AMD must be better in every way with a USD 1000 card to beat the USD 1200 4080.

  10. I heard that AMD separated clock domains on RDNA3. They clocked the front end a lot higher then the back and GPU core because of the chiplet design. Basically they didn't want another VEGA moment. You seen how fast VEGA64 can fly on Vulkan. It has 4,096 Streaming Processors with 8GB of HBM2, so you would think It would just scream. And it does in the Vulkan API. In content creation it does pretty well for entry level editing. I would like to see Intel's ARC A770 go up against a VEGA64, like a Sapphire Nitro+ version of the VEGA64. They both have 4,096 cores. Though, not directly comparable, it would be a cool versus.
    I actually bought a RTX TITAN for $1500 a couple years back cause i was having problem editing 4K60 vids longer than a hr. Regardless, i had it and it was pretty fun comparing it to the Nitro+ 6800XT. Both have 4,608 cores but it wasn't as close as you would think.
    AMD's RDNA3 is amazing. I also, picked up a binned 6900XT TOXIC that does 3GHz (18.6Gbps Vram) and it runs it all at 4K. But, back to the VEGA64 which almost had a 40% gain over a GTX 1080 in Vulkan titles today but it was only on Vulkan titles. AMD made RDNA-3 with VEGA's front end, not in mind. Yeah, a 1700MHz VEGA64 is a boss in Vulkan titles like DOOM 2016 ect. But, the only reason it lost in other DX11-DX9 titles, is its starved front end. So, AMD's RADEON group had to separate the clock domains to make this chiplet based GPU work. I would like to learn more about this but AMD is holding there cards pretty close to there chest. Just getting this information took till now to get released.

  11. The G4560 and 1050ti was a killer combo. I was playing DOOM 2016 and Sniper Elite 4 at 1080p60 maxed out on a Core2Quad Q6600 OC'ed to 3.6GHz on the eVGA 680i SLi Black Pearl Rev.2 MoBo with 8GB's of 1200MHz of DDR2. Basically the fastest DDR2 available. Also, Vulkan/DX12 works wonders on older hardware. Even with Windows 7.
    Vince (K|NGP|N) from eVGA even offered a trade for the Z390 DARK boards with 9900K's to anyone that had the 680i SLI Black Pearl, which was a water cooled version of the eVGA 680i SLI board. That wasn't a bad deal but, i kept mine.
    As far as crypto mining. I had bought a 6pack of RADEON VII's for $325 a piece a month before the mining boom hit and 1 month later sold 2of'em for $4K+.
    Bought a open box 60in 4K120 LG OLED with a binned 3GHz 6900XT TOXIC and 3090. Then soon sold it to get the 3090ti TUF V2 OC for just $200 more.
    I now see 3080ti TUF's for $780 all day long. But MSI's 6900XT Trio X is selling for a unprecedented $600-$660 depending on the damage of the box. Yes i say box because (open box) is all i buy. "Tech Yes got me hip". I recently picked up the ASUS z690-E Strix with a 12700K 5.5GHz all core 4.1GHz on the E-cores. 32GB's of 5600MT/s FlareX 5 dimms that run a 6400MT/s for a cool $700. You could probably get those components for the same now but, this was a year ago. Also, most people don't realize that you still get the manufacturers warranty even if its an open box item. So take that for what its worth.

  12. I saw the 5500's on sale for $120 too. I picked up a couple of them and found motherboards for $70 each. sweet deal. edit: it's down to $100 now and the 4500 is down to $80, oh well.

  13. TechYEsGuy is annoying. He even laughed at his own dim witted rebuttal to an Nvidia critic he had heard. With it, he showed how much he loves using thought terminating clichés, like at around 31:1831:30. Dude, What a m0r0n. Judging how ethically shitty a company is with any particular decision is something that happens all the time, so simply offering the analysis that "any company is going to want to make more money" is not only stupid, and it stupid for more than 1 reason,

    it's also a substance free rebuttal to ….anything one could imagine.
    It also objectively inferrs that all decisions made by all corporations are equally placed on the ethical gradient. Which is obviously just as absurd as techYesGuy sounds when said what he did.

  14. Going back to ATI and drivers, I had a HD4850 512MB and that ran COD:MW2 60fps full HD no issues, black ops came out a year later and it was a stuttering mess, I kept thinking treyarc were to blame and that they were worse at PC ports than Infinity ward but perhaps it was drivers. Man I thought the HD4850 as an amazing card but then soon as Battlefield 3 came out I was like man this game runs like crap and had to buy a GTX 460 1GB to play it smooth at 1920×1080.

  15. I wonder if as GPU performance scales to ridiculous frame rates in 4k, maybe they'll start advertising 1% lows because their averages are already at like, 240 fps ultra settings, lol. Alternatively, for competitive games, I could see super sampling becoming the next "esports seller" in the sense that you can apply SSAA to downscale from 8k to 4k, letting you see finer details in the distance for people who like to snipe from a huge draw distance away. If you don't believe me check out a shooter in 720p versus 4k and see how far away you can see a target. I'm not big on shooters but I could see it, tbh.

  16. WoW wasn't just smoother on Nvidia cards, it was also more stable. Back in Vanilla WoW in the second raid Black Wing Lair, for multiple weeks every time the whole raid did go around the corner into a certain room every single player with an ATI GPU (usually one of the otherwise very good 9000 series) had not just a game crash, but a full-on Windows blue screen. We (including me with my 6800) had to clear that room with about a third of the raid missing and then those people when they got back had to cross it looking to the ground.
    If I remember correctly, it took over a month until ATI finally released a driver that fixed this issue.

  17. When you talk about Monitor Refresh Rate and FPS its not only about how many frames you can see (which is around 1000hz-US Pilots tests). That also lowers the System Latency and input lag, improves accuracy. That where 8k mouse polling rate will start flexing its muscles and will be fully utilized.
    BTW Brian tested older 360hz that came 1-2y ago. There are 500hz monitors and depends how the panel is made. New panels are now also more responsive (Latency), not just higher hz (Throughput).
    CPUs still need faster single core performance to avoid bottleneck. Cores are already enough for now.

  18. quake 2 rtx, minecraft rtx, portal rtx still have the best versions of raytracing since they are the only games that have full path tracing compared to a hybrid form of it in cyberpunk that doesn't even use real global illumination since it doesn't have ptgi. These are the only games that I have played besides portal rtx since it hasn't come out yet but hybrid rt is pointless for the most part since it doesn't look much better than screen space reflections unless the world is generated in real time like the new starfield might benefit since it's a very large scale area that would be hard to bake lighting in a random generated world.

  19. In regards to hitting the raster ceiling with the 4090,i think the major use case people are missing is VR. A good looking vr game at 120Hz can bring down even an 4090,add raytracing to that and its a ppt slide show

  20. With all the discussions about what CPU performance is needed for gaming, its interesting that its hardly ever mentioned that for specific use cases, a powerful CPU is actually very important. Examples include: online multiplayer games that are not well optimized like escape from tarkov, and MMORPGs in all cases where alot of players are in the area, like sieges or big group fights and the like. In all those cases the GPU doesnt really matter at all, while the CPU is everything

  21. @2:04:00 Import tariffs are usually only on goods produced also here in EU. So most probably not on graphic cards. What comes on top is ~20% VAT on import.
    Edit: Also shipping cost from e. g. US into EU is hideous.

  22. 1:17:55 this is again an outrage! Tom still has issues pronouncing my name correctly! /s

    Wash-machine-man EnNell (and if you want to get REALLY pedantic: Wash-machine-man Underscore EnNell)

  23. Used am4 and 1151v2 cpu's are insane value, so it makes sense that there arent many good budget picks, and the i3 12100 is still great, so most gamers should be okay imo.

  24. @Moore's Law Is Dead Agree with discussion about monitor refresh rates. Give that a very small number of players can tell the difference between 144Hz and 240Hz, the number of players who can tell beyond 240 Hz is extremely small, the focus on graphics should be on better resolution and frame pacing up to at 144 Hz and 240 Hz.

  25. @1:36:00 Methinks AMD should have 256GB RAM possible on AM5 and presto, HEDT. It is not about PCI lanes, it is not about HD space. It is about accommodating hi RAM apps (that also love #cores). So I'd love to see a higher #core higher RAM variant of AM5.
    Edit: Applications are CAD engineering and whole client, server, DB, CI, developer seat setups in a single machine.

  26. I have an 8700k with a 3080.. its no 5800x3D but its no slouch. especially not for 4k120. I'll be upgrading to zen 4 3D once that comes out (to better ensure a .1% FPS of 120+).

  27. The tech crowd referring to am4 as a dead platform months before am5 was silly. We view pc tech so differently than things like vehicles or even tvs.

  28. 31:11 ……… At yet still, the 4080 12Gb ISN'T an 4070ti also ! At best it is an 4070 ! Ngreedia just lost their minds !
    I'll jus wait for the Radeon 7000 series benchmarks, by real people , and than decide ! But most likely I'm switching to Radeon this time ! ( after 9 years of Nvidia )

  29. The problem with Nvidia 4090 is, once you play your power trump card, you can't go back that easily. AMD still can play the power trump card and has much more space on the die. I think they hold back and play this card together when they've figured out how to make the most use of V-Cache for GPU's. Exciting times 🙂

  30. One thing I don't get is why the conversation revolves around 1080p fps. Yes,that's where the market is right now, but isn't there a slow and steady move towards 1440p? 15% of steam users are on 1440p and that's up 2% from the previous month. Maybe we're getting to the point where 1080p is starting to get phased out like how 720p is now irrelevant

  31. It could be that a chiplet design has more latency, and latency scales with frequency, so a 2GHz AMD has 2/3 the latency of a 3GHz one, a 500FPS setting has latency that pulls it back to 400FPS and 80% "use", whilst a 300FPS setting pulls back to 280FPS and therefore 93% use, so it is less about how many Hz AMD can get from their design as how small they can get the latency for a design that uses chiplets, and it really doesn't matter for raster use where it can get 500FPS because the chiplets can't talk that fast, so the higher frequency is left for making AIBs sweet on them with no headaches for AMD.
    If so, then the monolithic design of the Navi33 should be able to get to 3GHz+

Back to top button