9800X3D vs. R5 5600, Old PC vs. New PC: Intel Arc B580 Re-Review!

[@FNLNFNLN]

"This is why we test GPUs with an overkill CPU" Intel > I'm about to ruin [tech reviewers] whole day.

[@JamesLee-mp2qz]

My R5 5600 is already "Old PC"? Noooooooooo

[@Andrey543100]

Just buy r7 5700x3d and relax.

[@gaiahunter3863]

It is a frigging dinosaur...

[@Xamiakass]

imagine playing on 3900x, then :DDD

[@AnEagle]

@@gaiahunter3863 It's still quite powerful, and still expensive, so not really. It's a dinosaur when even the second hand market doesn't want it

[@grintcoalways]

We should give up. We have boomer CPU now.

[@vesopsd]

thanks steve

[@BonJoviBeatlesLedZep]

Back to you Steve

[@hackfleisch7424]

The more you Steve, the more you Steve

[@Shadow_banned_by_YouTube]

stonks thief

[@CrypticWritings89]

Thanks (Australian) Steve.

[@Qalibrated]

@@CrypticWritings89 Upside down Steve?

[@thevindictive6145]

People with a 9800x3D are not gonna get a B580 to pair it with. Lucky this issue was brought forward so those that have a budget CPU can now make a better decision. Thanks hardware unboxed.

[@Reeper-x6x]

This informs Intel CPU users nothing, it was a half baked test

[@mark.082]

​@@Reeper-x6xBut it informs people that plays games alright. If you own an old Intel cpu, chances are this video will be helpful enough, and if you own a modern Intel cpu... Well, you are not even interested in value and best performance in the first place lol

[@pedroferrr1412]

@@mark.082 AHAHAHAAH, nice!

[@raawesome3851]

The issue is, there's still not a better option than Intel right now. Both Nvidia and AMD have abandoned the lower end realm/given people intentionally bad products at the lower end

[@nk-dw2hm]

@@mark.082 tbf Intel hasn't competed in value for over a decade

[@dethskullcrusher]

The sheer ammount of dedication and effort Steve puts into every video is commendable. Never change Steve.

[@AlexanderBukh]

Amazing work, yep.

[@BlindlyNostalgic]

Yes….. but I worry he doesn’t sleep.. or spend time with his family.😂

[@butmunchass]

yeah but some of his statistics are regarding prices that don't match reality. i.e. you can't buy intel Arc 580 anywhere at normal prices

[@Trucking4Jesus]

you do know he has an entire team that does the tests? he just spearheads the narration because he IS the brand lol

[@jeremytine]

too bad he bad mouthed previous requests for real world testing like this...

[@GewelReal]

"5600 is old PC" Me that just got 5600:

[@Terry.W]

And me..

[@FlyTimeRC]

Still a great cpu.

[@ravijotsingh7831]

Old doesn't mean bad. Enjoy bro...

[@dvornikovalexei]

I'm still saving to buy a 5600🤣

[@kg4wwn]

I had a 4500 arrive in the mail yesterday. Granted, that's not for my system, but one I'm building for a friend, but it is still going into a "new" system.

[@Akasen1226]

Steve you gotta go even further. Bring out the 4790k

[@cIappo896]

Nah, get out the Phenom II x4 955.

[@EBMproductions1]

The i7 4790k is the OG king.

[@JamisDuanne]

Get the 2600K in there too

[@asdf_asdf948]

I dont think the phenom can use pcie gpus, those are agp slot only

[@sagee9355]

i will do u one better....bring out the pentium g3258

[@mintia6314]

Thanks for the hard work! And sorry for your loss of touches to new gpus presented in Vegas.

[@funtaril]

Thanks for the second half of the video, it's very showing

[@ovemalmstrom7428]

I just built a new system using the R5 7600 paired with the Arc B580 LE, so far very satisfied. It runs cool, quiet, and stable. I got a new 1440p monitor to go with it, and so far, every game I have tested on it runs great. The R5 7600 seems to work fine with the B580 in 1440p for me. I really hope Intel will work extra hard on this driver problem to find a fix before it is to late.

[@Ad_Dystopia]

The question is, if the problems are driver related or architecture related.

[@Dominus_Potatus]

You have better architecture than 5600. For new system built, I think B580, if sold at its price, is still worth the money. But it is not worth if you have old system especially since RX 6700 XT is sold cheaper, significantly cheaper with better overall performance

[@Remmaster]

great GPU and cpu!

[@watchlover7750]

Not sure it's fixable pal

[@howardyeah]

The overhead problem isn't as much of an issue at 1440p

[@FM-kl7oc]

I guess some form of overhead testing will be a new standard test to be included in reviews from now on?

[@Hardwareunboxed]

Well for Arc GPUs sure, overall doesn't change much for Radeon and GeForce.

[@kg4wwn]

@@Hardwareunboxed Might want to do a small one for new GPU generations from everyone. Just one or two games, enough to say "It looks like it works on older CPUS" or "stay tuned for the next video while we see if these two games are a fluke".

[@andyderp6473]

​@@Hardwareunboxedmore work POV... Better man then most of us Steve

[@raggedcritical]

Sure makes the case for a basic sanity check for all new GPUs at least.

[@houdini329]

@@Hardwareunboxed But it does sometimes for the RX 7600. And who knows, what cards affected as well.

[@coladict]

So in trying to isolate the GPU performance using the very best CPU for gaming, it actually masked the problems that real users would experience. No one buying a 9800X3D would settle for a B580, but someone on an old PC might.

[@simdaydreamer5239]

And b580 will not perform well on that old pc

[@turtleneck369]

@@simdaydreamer5239 still would choose it over 8gb VRAM models with something like 5600 just cuz u can upgrade to 5700x3d and games will just eat more VRAM they already do more than 8gb at 1080p 1440p most demanding ones in recent 3 years

[@GizmoTheGreen]

@@simdaydreamer5239 right but do you plan on staying on that cpu for 2-3 more years? or is a cpu upgrade next in line? if so the b580 is still a good choice as you'd see higher uplift after cpu upgrade than if you bought a "new" 4060 for example and then after cpu upgrade you end up gpu bottlenecked again.

[@katv1195]

It doesn't help that the analysis is only 2 CPUs.

[@fredsas12]

Good thing there are 5800X3Ds for AM4

[@houdini329]

So, i guess from now on GPUs have to be tested with high end and low end CPUs? Because it seems, that the ARC GPU is not the only one. RX 7600 suffers from this issue sometimes. And who knows which cards do as well?!

[@lizardx6504]

No, he set the settings to medium so that the FPS was playable. Also the problem is only in 2 wahammer warthunder games. Moreover, in the second this is fixed by switching to DX12

[@lizardx6504]

And so video cards are tested at ultra settings and more often at a higher resolution than 1080p. There are rarely exceptions.

[@xjax8519]

"RX 7600 suffers from this issue sometimes" In games from AMD like Warhammer, Last of Us and Starfield ))) They want sell you new CPU Ryzen )))

[@mademedothis424]

High low and mid. For as much as I appreciate a more comprehensive review, I wish this had spent less time fiddling with settings to find the pitfalls on the low end 100 buck CPU and had instead tested a mid-range 200-ish part (and maybe an Intel part) to help determine what CPU makes sense to pair with the B580. As it is this re-review warns me to reconsider pairing the B580 with a low end part, but doesn't tell me what is a good CPU pairing. The B580 is definitely still a very cost-effective GPU in the right system, I wish the review helped figure out what the cheapest system that makes sense is.

[@zanasLTU]

​@lizardx6504 in Spider-Man remastered also b580 suffers 14:44

[@timinski_i]

Also keep in mind these are US prices. Over here in germany the B580 is 300€, the RX 7600 is 265€ and the RTX 4060 is also 300€. I wouldn't wanna be part of the experiment for that pricing.

[@dvornikovalexei]

Here cheapest was 4060 for 280 so i got that. 7600 is 310 and B580 is being scalped

[@karambiatos]

@@dvornikovalexei imagine scalping mid range GPUs from intel lmaoooo.

[@V3RAC1TY]

@@karambiatos its for its AI promormace

[@firerocman]

Another win for the "real world tests" crowd.

[@jimdob6528]

This was an insane amount of work. Thank you Steve for your dedication and hard work! I still am excited for their b380 card to put into a server I am building with the Ryzen 9 7900

[@moldyshishkabob]

Wait, have there been leaks for a B380 that I didn't hear about?

[@TheGreenReaper]

@@moldyshishkabob Low-profile 1-slot Battlemage when?!

[@ZSM4747]

This was really interesting and informative, thanks HUB team.

[@Reeper-x6x]

An Intel CPU may have worked beautifully, all we truly know is perhaps the drivers need revision on low performance AMD CPUs or the GPU

[@larkinc]

Great range of testing. I do recommend using a chart that shows outliers, e.g. box or violin chart. The B580 may still have a cost per frame advantage even with its reduction for 5600, but that is fairly meaningless when the range of performance is so wild from good to terrible depending in the game.

[@lucazani2730]

Also because it is not real pricing, which is nowhere to be seen close to the announced msrp

[@atthelord]

Man these are some very interesting results! Legit interesting findings Steve.

[@sdnnvs]

tltr, "To make the most of the B580, it is necessary to pair it with a CPU of at least a Ryzen 5 7600 or Core i5-13600K."

[@javier3108]

I think I'm subscribed to around 80+ KZbin channels, but yours is one of the 3 or 4 that I have the bell notifications on and I can confidently like everyone of your videos at the very start without having watched it yet. Thanks for all the great content, Steve and Tim! Looking forward for the eventual arrive of the RTX 5000 and RX 9000 GPU testing!

[@External2737]

Same. This is a great channel. I too limit the bell

[@lucazani2730]

Yeah this is by far my favourite KZbinr at all. Tech KZbinr for sure at least. The thing I love the most about HUB is that they NEVER try to be funny. The entratainment comes from the information side of the video. If there is humour, it is delivered naturally from the circumstances alone. This is way different from GN. Great channel as well, but GN delivers jokes which are very close to my sense of humor so most of the time I enjoy them, but sometimes they are a little forced and hurt how much I enjoy the video. Both channel are great btw

[@UM_88-s2c]

​@lucazani2730GN has worse gpu reviews

[@lucazani2730]

@@UM_88-s2c nah GN GPU reviews are fine. HUB GPU reviews are better because the graphs are clearer and sharper tho

[@AntonioLexanTeh]

That is why they showed only 1440p in their slides... I think this cpu overhead is a driver issue, it could be resolved in the not too distant future. Intel has been fast with updating and fixing their drivers, gpu is still good for $250

[@vuurdraak-]

Even though I understand why reviewers normally want to test with the fastest CPU, these benchmarks shows how important it is for individual consumers/pc-builders to see data that is more closely aligned with the current hardware they got, as I had no clue before that not only the Arc B580 but also Radeon 7600 can have performance issues in this case with War Thunder. When I plan to buy new hardware I look at big channels like Hardware Unboxed, but immediately afterwards go on a deep dive search for benchmark data of my favorite games and/or benchmark that more closely match my hardware. As a rule of thumb don't buy new hardware and wait at least three months for stuff like whats in this video to be found. TY for this video.

[@RAZGR1Z]

Thanks for putting in all that time for this issue, Steve. Looking forward to all the reviews coming in the next few months!

[@beaten_tech]

Im still blown away by how good it is for there second gen

[@Warron75]

Wooow, I was planning to upgrade my Ryzen 5 5600, but I realized there's no point in doing so. The 5600 still handles 1440p gaming perfectly with my RX 7800 XT. You saved me money, awesome, thank you!

[@JPgreekgaming]

The 5600 is a fine CPU but this benchmark only says that the system is gpu bound, not that the cpu doesn't make a difference. The 7800xt is faster and will be more prone to being cpu limited in certain scenarios.

[@prosecanlik4296]

I wanted to swap out my 5600G just so I can utilize PICe 4.0 for my RX 6600. But I saw benchmarks, identical results, so I stayed on 5600G

[@Yerinjibbang]

same specs here really enjoying 1440p gaming on mine even 4k is doable!

[@The_Noticer.]

Do note that the 7800XT is significantly faster than the 7600. So dont use this data for your application. You will probably see a decent boost going to a 5800X3D, but since it's EOL its no longer worthwile to do.

[@Skylancer727]

I mean, I still use a 9900K all these years later. I mainly just want to upgrade for emulation. Love emulation but it's really CPU heavy. Just worry about emulators not being made for AMD. Most of them still are built off the code prioritizing Intel.

[@zet772]

I hate the fact that all new GPUs launched are more expensive in europe then their counterparts, even though the MSRP should be lower than competititors current prices, so in europa, b580 is more expensive than 4060. How are they expecting to even sell those

[@haukikannel]

They don´t expect…

[@dvornikovalexei]

They don't tbf. Intel is making losses on each card, they just launched it with a fake price to satisfy the shareholders

[@lowzyyy]

Yeah it should be much cheaper, and in contrast its more expensive. Zero chance to sell

[@theracingban]

It's not intels fault it's the USA taxes

[@someonespotatohmm9513]

Take into account MSRP prices are before tax and US prices typicaly don't include tax in their list pricing.

[@rise4097]

Thanks for covering this Steve! Interested in pairing the B580 with a 5600X so watching all this closely.

[@KevinSolway]

I doubt the extra 3% CPU speed will help much.

[@everope]

Might as well get a 5700x3d instead

[@TheBenMillard]

All killer, no filler. Great video and an important re-visit given the logical system specs for a budget GPU upgrade or 2nd-hand mixer build.

[@ahnafstaseen]

ive been following your content since 2019 but wasnt a subscriber then... the amount of effort you put into this overhead review made me to subscribe! Great Job man

[@Eric-yv5ex]

Hey Steve, I hope you see this important note about the “Quality” term regarding upscaling. DLSS and FSR “Quality” upscale by 1.5x, however, in order to achieve that same 1.5x scaling with XeSS 1.3 or later, you must set it to “Ultra Quality” (XeSS 1.3 or newer “Quality” actually scales at 1.7x) So if we were to do a true test of all upscalers at the same scaling option (let’s use 1.5x in this case), it would have to be DLSS Quality vs. FSR Quality vs. XeSS Ultra Quality. I’m not sure how this affects testing results, but wanted to bring it to your attention. Thanks for your work!

[@Manicsar1]

I picked up an open box 7600XT from Best Buy for 289 and it's better than I was expecting from all the negative reviews on it. It's paired with the i5 14600 non k and It runs the games I play, which includes Cyberpunk, Eve Online (Multibox with high settings), Marvel Rivals and many other steam games with no issues on an ultra-wide 1440p monitor (3440x1440).

[@The_Noticer.]

Steve, one interesting question about these tests. Did you see full GPU utilization in all these tests? Because that would further support your "CPU doesnt matter at 4K" video. Because that would imply that even at seamingly full saturation you get different results with different CPU's.

[@ivanfisun1414]

Yes, it will be very nice to see some widespread intel cpus tested as well, thank you!

[@dradis1]

Does it make a difference using new vs old Intel GPUs? Standard logic would say no, but here we are with this video, so I am not sure if standard logic applies right now.

[@Matticitt]

The fact that even the 9800X3D appears to be limiting the B580 is crazy.

[@longtutjenkins]

OR the overhead of the B580 is hurting it's own performance on a high-end CPU at lower resolutions. They need to fix that overhead so they can maximize the CPU performance paired with that gpu when used for low settings and resolution.

[@tim3172]

@@longtutjenkins hurting... it... is own? performance? What?

[@longtutjenkins]

@@tim3172 By requiring that much CPU overhead, it is bottlenecking its performance at low resolution. for whatever reason, the B580 requires that much performance from the gpu in order to push those frames and when the CPU is not strong enough to do that and the game tasks together (like everything below a 9800x3d and sometimes even that) it will bottleneck itself. That is caused by the GPU. Enough explaining?

[@cooltwittertag]

​@@longtutjenkinsyou realize thats what the guy you replied to was able to say in a single sentence? We know the B580 is causing a bottleneck and isn't just so fast every CPU bottlenecks its sheer power.

[@danr8011]

The fact that B580 is so bad even with 9800X3D is crazy

[@gerardphilippetuazon491]

This is probably too much to ask for Steve, but would it be possible to test with the Ryzen 7 5700X3D and Ryzen 7 7800X3D? Does the 3D V-Cache significantly reduce the effects of the driver overhead of the B580?

[@funtaril]

look up HUB's previous video on this topic, 5700X3D also showed decrease in performance of B580

[@Amokhahn]

@@funtaril it reduces the fps but 3D Vcache helps

[@TheHalfmanofOz]

Looking fit and healthy, Steve. Thanks for exploring this further.

[@MarkJYule]

Thanks Steve! Running a 5900 (with an AMD 6750) and a 5700x3d with an old RX580 that I'm looking to upgrade - the Arc580 was th prime contender...now going to wait and see what prices do as the new Nvidia and AMD cards come out....

[@DavidEasthope]

I'm noticing that on a large number of the benchmarks, when the RX7600 and RTX 4060 are tied on either CPU, the B580 tends to be slightly more performative than them, but when there's a difference between CPUs on the AMD/NVIDIA cards, the B580 takes a bigger hit overall. Seems like on some games, the graphics scheduling might not be great - especially on Last of Us Part 1, which I'd chalk up to "unoptimized console port" - while on others, the B580 is a clear winner even for an AM4 chip, e.g. Plague Tale: Requiem, where it beats both the RX7600 and RTX4060 by at least a 10 FPS margin even when suffering a CPU hit from the 5600. Overall, seems like if a game is going to have a performance difference between two CPUs, the B580 will multiply the issue. Hopefully it's something that can be fixed with game-specific configurations from Intel. Even before Intel hit the market, there were plenty of examples of games coming out that ran great on NVIDIA but awful on AMD before a patch or driver update, and vice versa.

[@fabiozannettino]

@hardwareunboxed The "Quality" preset of XeSS is more aggressive than FSR or DLSS.

[@Frozoken]

He knows he's mentioned it himself before he definitely would've just matched it it's just too long to explain in text on a graph

[@melexxa]

it's not, the xess quality is xess balanced using dp4a fallback on geforce and radeon gpu, on arc gpu xess quality is same internal res as dlss, fsr quality (66.666% internal res)

[@KrisDee1981]

Steve it would be nice to test B580 vs RTX 3060 on PCIe 3.0 vs 4.0. So many people out there still have B450 platforms, and RTX 3060 will run at PCIe 3.0 x16. Combining PCIe 3.0 x8 with that CPU overhead could be a disaster for Arc GPU.

[@TexasFun]

Intel's Arc Graphics - Desktop Quick Start Guide specifies the mother board must have: a full-size PCI Express 3.0 (or newer) x16 slot to work properly. On a mb with only an PCIE3X8 it may not even work or at least will be likely to have performance issues.

[@DaveTheeMan-wj2nk]

I had a gen 3 board. The performance for the B580 was *Horrible* I returned it, and got a refurb 3070 (nice model to) for same price. And holy molly, my performance went THROUGH the roof. In world of warcraft, I went from 120-180fps in a dungeon, to 200-400!!!! And every game I played, even at 1440p was smoother, and played better with WAY higher fps. And games that had issues like starfield, were not fixed. ALso B580 can't play emulation for switch. Glitches out. So glad I got the 3070. It's been a dream card for me so far :)

[@isaace.6157]

The content I've been waiting on. Thanks for putting this out, we appreciate you!

[@Tomazack]

Can't imagine how many hours of work this took and I am not on the market for a budget GPU either, just interested in learning about new tech releases. Great work, sad reality about B580 as a budget option.

[@Light1c3]

You can't find this card at the MSRP ($250). You wouldn't know this during the original review, but I still see that price being used to calculate value at the end of the video.

[@msolace580]

at local store here in midwest, not that hard to find 249.99 limit 1 per household avail 10+, im getting 5090 tho

[@Dave-oh5pt]

This is why they is a good case for benchmarking with typically paired components. No one with a 9800x3d is going to get a b580.

[@haukikannel]

$500 CPU with $300 GPU sounds good pair to me! So I am sure some people will go for it.

[@Dexion845]

@@haukikannelBut why waste that money on the CPU when you're bottlenecking it with a low-end GPU. This is an Intel issue, hopefully they can fix it.

[@cxngo8124]

Yeah but cpu pricing scales well. The next jump up from a b580 would be a 4060ti which is terrible. The next good value gpu is double the price of the b580. You can also get a 9700x bundle for $400 where I live. And pairing that with a b580 is perfectly fine.

[@ChainsOfSadness]

@@cxngo8124 next more or less reasonable GPU, if you run 5600 or 12400, is 7700XT. So value depends on location and stock mostly. At least, in Eastern Europe atm, most affordable B580 ~325Eur, 7700XT ~415Eur. 20-25% extra price, ofc, but you got +20% fps @1440p. Probably. If 7700xt don't have overhead problem.

[@MightyCats2011]

​@haukikannel have the fastest cpu with a budget GPU? I think more than half of users won't have 9800x3d or if they had it would go for faster GPU like 4070 or more expensive

[@darkfruit3657]

Thanks Steve!

[@mzza]

B580 is 300€+ in the EU. It is and was DoA pricing and has actually seemingly/possibly driven up the RX7600 price as well. For months that card used to be available for under 250€ and now the cheapest models are already 270€ and still trending upwards.

[@tavernadosgamesBsb]

The scenario should be even worse on real world usage when other things are open in background, like browsers, streaming software and video capture things

[@joannaatkins822]

I have a question about PCIE version that might be relevant to the B580 situation, and driver overhead. If an older system has PCIE 3.0, does that affect resizable bar Vs newer PCIE 4.0 and 5.0 systems? Obviously the B580 issue is affected strongly (with AMD at least) by CPU generation, but CPU generation is also hand in hand with PCIE support. So, taking an "old" and new system like the Ryzen 5 5600 and 9800X3D and downgrading PCIE versions, would that affect FPS and 1% lows? My gut instinct is that it won't have any effect, but to assume makes an ass of u and me. Could be interesting, could be a non issue, but most of us don't have the equipment to check.

[@fdgdfgdfgdfg3811]

Ryzen 5 5600 has pcie 4 not 3

[@joannaatkins822]

@fdgdfgdfgdfg3811 I know that, but for 1:1 performance you'd need to test the systems you already have to eliminate other factors. One of the older Ryzen chips like the 3600 and 2600x perform generationally worse than the 5600, and many people upgraded their CPUs but not their motherboards, which were left with PCIE 3.0 instead of the 4.0 that they're capable of. I also, if you check my comment, suggested doing the same with the 9800X3D. The only reason to do that would be to ensure CPU consistency, and a baseline to see if the CPU ability is affected.

[@fdgdfgdfgdfg3811]

@@joannaatkins822 PCIE is only a problem if the gpu has low memory like 8gb then you will get low fps if the vram is exeeded,12gb are not really good either ,so the best option is 7800xt ,no one with 100usd cpu will buy a gpu likle the b580 its like 320+ usd while the 7800xt is a bit more with alot more performance.amd gpu run better with low end cpu

[@yuralmspw4911]

would be nice if you include cpu power consumption numbers in all scenarios. I assume cpu power draw and temps are higher with b580 in any scenario, which is bad...

[@tvcultural68]

Hardware Unboxed is my favorite hardware channel. The video could have ended at 17:45, as the topic was already clarified by then, but they weren’t satisfied and went even further, running more tests with medium settings and active upscaling. Incredible!

[@theveay]

Respectful amount of work, great job! And to see Steve recognize upscaling as an useful metric for reviews, interesting.

[@Perrajajaja]

Thanks for including some upscaling performance! I'd like to see more of that in the future, as for many games (and I think many gamers) this is a real world usecase

[@jimmahT]

So where is the bottleneck starting and ending? Is it a frequency problem? A core count issue? Cache related? At what point does the framerate stabilise? Would like to see if a higher core count CPU from 5000 series makes a difference or x3d on 5000 series. Is this the same with older Intel CPUs? Could DDR4 platform be the problem? This video is great but for me there are more questions than answers. Keep up the good work guys!

[@AndyViant]

The 5800X3D was included in the initial review.

[@jutube821]

Indeed, why not also test the 12400F, one the most used budget CPUs out there. I suspect it might be a bit better than the R5 5600, a single chart would have been enough.

[@SlowHippie]

I'm wondering if it's how resizable bar is handled in the older cpus

[@AndyViant]

@@jutube821 ​ It would be great if he could include it in the 50 game benchmark he's going to do. Not all 50 games, obviously, but pick out some of the games he did a deeper dive here on, like Cyberpunk, Warhammer 40k, Spider-Man Remastered, The Last of Us 1 and War Thunder. Those 5 games, or really any 5 games from the second half of this video would do. But choosing different API's and different Direct X versions would certainly help. Even better if we could get a couple more intel CPU's. We've covered every single Ryzen generation (Ok We skipped Zen 1 but have Zen + as per Steve's first review of the issue). If we could do a low and mid 10th gen (say a 10400f and a 10900k), a low and mid 12th gen (eg 12100 and 12400 or 12400 and 12600) and a high end 13th or 14th gen (eg 13700k or 14900k). I wonder if Intel didn't see this in their own testing because they did the great majority of their testing on Intel CPU's. This is the downside of pretty much all the testers moving to 9800X3D test platforms. Right now the 50 game mega test probably makes a bit less sense than finding out at which CPU generation (or CPU power) the B580 just falls apart. You could honestly pick one game and sub in the 7800X3D, 7700X, 7600 and watch the performance tanking with each step to a lower grade CPU. the reality is the B580 is a good product IN THE RIGHT CIRCUMSTANCES. We just need to know what those circumstances are.

[@romanbabucha139]

Old CPU and their chipset have some problems when B580 want direct accessing RAM memory. When it use Ryzen 5xxx or Intel gen 10 CPU, everything is allright. Older CPU degrade power of B580.

[@TechAgama]

Does a Budget GPU require the flagship CPU to show full performance? This really kills the value. Intel needs to solve this issue as fast as possible.

[@cerberus_uwu]

interesting take on it.. when it clearly shows with the 5600 that at 1440p it performs better at native resolution than the NVIDIA and AMD offerings....

[@pikaachoo3888]

@@cerberus_uwu the thing is, most budget gamers game on 1080p, and they are the most likely ones to buy this gpu. Anyone who is going for 1440p gaming will look into the 7800XT/4070S instead.

[@cerberus30700]

@@pikaachoo3888that’s absolute bollocks 🙄😅😂😂 I think the sales of the card prove that!… I play at 1440p and I’m going to it from a 3060 - and know others who have done likewise

[@turtleneck369]

you dont you can pair it with 7500 that is 150 cpu or 14400f is where i live 150€ also and can buy 70€ ddr4 mobo with 32gb ram (60€) and u have great combo for B580 ,5600 before christmas here was only 90€ but yeah if u are on older platforms its def disappointing maybe driver updates will try to help with that ,if u are on am4 platform 5700x3d is also good options although here in EU its not as cheap around 200€+

[@ovemalmstrom7428]

​@@cerberus30700I did the same, moved to 1440p with the Arc B580.

[@AlexMiseroy]

Does this mean that every card in the future have to be tested with lower end or older CPUs?

[@IntelArcTesting]

Thanks for the testing, have been waiting for this. Many people asking me for advice on buying arc and now I know which people I should actually recommend it to.

[@ChrisP872]

I think the nature of the "CPU overhead" for ARC is less CPU gruntwork and more like baton passing in certain games with certain APIs. I've seen videos where people use Vulcan wrappers for Direct X games and the ARC switches from 70% GPU utilization to 98% and gains tons of frames.

[@brandonupchurch7628]

On Spiderman I think you should have also pointed out that on the 5600 upscaling actually performs worse than native on the B580 on the 5600.

[@_oriTech_]

does this data apply to the r7 5700x3d? i mean... if u hanging on older ryzen cpus, u can easy upgrade to the 5700x3d, its still available and has a good price too.

[@haukikannel]

Look the previous video. 5700x3d was in that!

[@andrexskin]

It still does, probably in a lower % but the problem is there as well

[@_oriTech_]

@@haukikannel thx, missed the previous video with the overhead issue.

[@radu1006]

This review opens a new chapter in GPU testing. Now you will have to test each GPU with a similar priced CPU. And now the question? How does the B580 behave with a 12400F?

[@Findmeout8293]

Tons of reviews are there testing it with a 12400F. Overall b580 doesn't do bad, it's "usable" on the 12400F.

[@lizardx6504]

12400f with DDR4 is similar to 5600. No one uses it with DDR5

[@Frozoken]

​@@lizardx6504No it still absolutely is faster bro

[@turtleneck369]

@@Frozoken depends on memory u pair it with also if u pbo or oc 5600 its pretty similiar to 12400f in gaming depends on titles but stock both with same memory 12400f is a bit faster in gaming you right about that

[@Raupe-n2k]

Are you sure 250$ really is the "typical Newegg Price" for the B580 at 9:40 in the Video and Not just the MSRP? When looked it up parallel to watching the video, the cheapest one was 340$, Most around 380$-400$ and the 270$/280$ ones were Out of Stock. Maybe you can show what really is the avarage Price for the B580, I would be interested in such content, but at the pricing I can currently find the Intel GPU would be much worse value than shown in the Video.

[@ETophales]

Great investigation and interesting results. I think it goes to show that choosing a GPU is a lot more complicated than what can be gleamed from testing it with a high end CPU at a native resolution and max or near-max quality. It's worth taking into account when reviewing future CPUs. Some amount of testing with lower settings and upscaling might be worth doing always. We always assume that testing at 1080p is good enough, but if testing at 1440p with upscaling produces different scaling, then that's definitely worth knowing.

[@themusesquad8554]

Doesn't Intel Emulated the Dx9 DX10 and DX11? Maybe that is the reason for the high overhead for thous while DX12 gets much less?

[@Hardwareunboxed]

We're still seeing issues in DX12 games though.

[@themusesquad8554]

@Hardwareunboxed I've noticed, but not as much as the older directX, which they claim to emulate.

[@suiton20]

The way intel handles the older apis is similar to proton on Linux. Essentially dxvk. It is a translation layer that translates api calls into Vulcan. Not emulation. Depending on the game some will either run slower to faster than the native api. The games that have poor cpu threading such as all dx9 games will benefit from forcing dxvk on. It does some translation magic making use of multithreading. On dx11 games it’s up in the air. Only dx11 games that I noticed a benefit on my setup are state of decay 2, subnautica, watch dogs 1 and early access titles. I probably have too much overhead. Nvidia GPUs don’t really need to utilize dxvk unless the game is that badly optimized. Amd GPUs benefits the most and has the least overhead from using dxvk thanks to existence of the steam deck.

[@no-one_no1406]

Good to see that we're no longer only testing with unrealistic hardware combos! Steve bringing all the data.

[@AutieTortie]

Lolz, you don't understand anything! 😂😂😂

[@Raxiel497]

I realise its extra work that might not change the result much, but I'd be curious to see what the current 'new' budget queen the i3-12100 performs. I think a 'new' budget system is the other use case for a $250 card. Is this an AMD/Intel incompatibility or just a raw horsepower issue?

[@xjax8519]

Intel want you to buy 12400 or 13400 it work perfect with b580 ! Intel GPU for Intel CPU )))

[@JoseDiaz-xm4vv]

Ddr4 vs ddr5 setting

[@AhmadWahelsa]

12100 js equivalent to something like 5600x if im not wrong and yeah would like to see tests since no one doing it on 12100f and its my current cpu (took intel because amd prices here is inflated, r5 4500 costs more than 12100f lmao)

[@tim3172]

Quad cores CPUs aren't "budget queens". They're "poverty spec" CPUs. They get completely flummoxed by most newer games and utterly defeated by Windows Update.

[@AhmadWahelsa]

@@tim3172 hey, it's cheap, atleast in here, AMD (as much as i hate to say it) become what Intel was in here, everyone is after it, even Ryzen laptops is sought after mainly due to iGPU. I was considering r5 5600, but the 12100f is literally at half the price of 5600 in local market here, with exact same performance (or more in some apps)

[@HDJess]

I think the most interesting tests would be pairing the B580 with either i5-13600K/14600K or R5 7600, as those are the go-to options for a new "decent" budget PC nowadays.

[@jasonh9523]

Thanks for the update. Very interesting. I have the sparkle variant. Bought it out of curiosity. Tested it for awhile and then installed it in my daughters PC w/ a 5800x3D. Running well.

[@TechTusiast]

Why no testing with Intel CPUs? Results might be even surprisingly different. I think something like Ryzen 7600 would have been interesting as it is something a budget builder might get for a new build now. People either already have a 5600 or might upgrade some 1000-3000 series CPU to that if they have an existing system, but I would not really buy a new AM4 system anymore. Bad investment.

[@Reeper-x6x]

Exactly, the test is flawed

[@mytech6779]

Testing existing AM4 is the whole point, many millions of AM4 that could upgrade the GPU.

[@Gringomaniac]

Steve adressed that in his "Final thoughts". He's interested in testing this scenario with different CPUs in the future, but basically can't tell when exactly as new GPUs and CPUs are coming that need to be tested too. This particular test was the result of a community poll, so he used the time he has for what the majority wanted. We'll definitely see more of that whenever there's enough time for more testing on this topic.

[@Reeper-x6x]

@@Gringomaniac His final thoughts are of no use , the test is so one sided he should have not released it, but he gave Intel a black eye, I think that was his motovation

[@Gringomaniac]

@@Reeper-x6x Well, you can think whatever you want, doesn't matter whether it's right or complete garbage. But when you care enough to shout it out into the public you should also care to give a reason for why you think that.

[@gluraki4444]

so when building a new pc rn with for example a ryzen 7600 is the b580 a good choice over a 6750XT? considering they are similar in price in the EU

[@AndyViant]

For a 7600 it looks fine. If you're on an older platform I would recommend the 6750XT, but otherwise the B580 has better features than the old 6750XT. Better raytracing, better upscaling etc.

[@emanuelecoppini2023]

The 6750xt it’s better, new games will be more CPU intensive

[@TheArakan94]

This video seems a great argument on why to also benchmark with average CPUs, especially for budget GPUs rather than just the top of the line to remove CPU limitation..

[@lizardx6504]

NO YOU SHOULD NOT. They used low resolution or upscaling and medium settings. When choosing a video card, you need to test the video card, not the processor

[@TheArakan94]

@@lizardx6504 Which is how most people use their budget GPUs.. No one with newest X3D is going to use budget GPU and as seen here, such testing can provide massively skewed results, showing one GPU to be best by far, when it's actually worse than competition when paired with realistic budget CPU. Yes, it goes beyond simple GPU testing where you want to remove CPU limits as much as possible, but it's definitely useful - otherwise HUB wouldn't do it.

[@StarsAtNight1]

@@lizardx6504 You either have Intel stock or you didn't watch the video.

[@gingerbread6967]

Omitting intel was wrong, Intel uses different drivers and they may have performed great

[@Jzwiz]

@@gingerbread6967its covered on other channels, same issues

[@MrSiddhartha99]

Nobody designs GPU chip better than NVIDA...sad but true

[@itsJPhere]

But what did it cost? EVERYTHING!

[@grospoulpe951]

Great and amazing job for re-reviewing the B580, face to face with comparable GPU! Thanks a lot! Have a great week-end!

[@Subarashi_Ningen]

Ryzen 9 9800X3D? anyway thanks Steve, keep sitting comfortably

[@tristanwait4itlegendary]

9950x3d

[@realw1ll]

ARC B580: “runs great on my 9800X3D!”

[@chhandobhihbhushan2742]

My 7600 looked old in the last video, at least in CPU demanding games.

[@AndyViant]

Yeah, feeling the same way.

[@chhandobhihbhushan2742]

@@AndyViant yeah, I was saving up for b580. I'll just wait to see what happens now. Amd and nvidia are also around the corner. If nothing good comes up, I'll just buy last gen amd

[@terry9819]

Thanks for adding the medium quality tests.

[@mrwhat1487]

Steve, you're an absolue madman with all these tests! 🤩And that's the reason why this channel is so valuable; throwin' mass punches, pullin' none. ....and I hate to be that guy, but the initial findings tests have me curious about additional PCIe gen falloff with the B5nn cards. Say, for example if one has an R5 5600 running in a B450 or B350 board, limiting them to PCIe gen 3. Perhaps consider testing 3.0 vs 4.0 when the B570 comes out later this month?

[@dungeonsiege9439]

So Steve, would you rethink on your decision to review all gpus with best fastest cpu now or not

[@mrezkybayusetya491]

Me with a new R5 5500 *chuckles* I'm in danger

[@mytech6779]

Why would you buy that new at the beginning of 2025? Assuming it was purchased for gaming. It may be a fine deal for some other task.

[@johnreeves1257]

@@mytech6779 There are budget ballers out here. The Ryzen 5 5500 is like 80 bucks, whereas you're spending minimum $130-150 to get on AM5, not to mention the motherboard and RAM will be more expensive.

[@dvornikovalexei]

5500 sucks kinda. I had one and was choking on Ratchet n Clank and went back to my 3100 and immediately got 70+ fps. Realized 5500 doesn't support PCI E 4.0 but the 3100 does😂

[@johnreeves1257]

@@dvornikovalexei That's abnormal. The Ryzen 5 5500 offers about 20% more single-threaded performance and about 40% more multi-threaded performance than the Ryzen 3 3100.

[@mytech6779]

​@@johnreeves1257 Yes, and the reason AM4 can't command the same market price because it is both a dead end (fewer years before replacement) and it is somewhat lower performing than AM5 options. Free market prices communicate essential economic information. There is an old saying in my area: only wealthy fools can afford cheap junk. I have 12 years from my current desktop and while I will build a fresh machine later this year this one will still be useful for basic tasks another 5 years. It was only a mid-cost build at the time, but it was latest generation with reputable brand components beyond the bare minimum. So it still has reasonably compatible protocols and features and performance (With one modest GPU upgrade in 2018, and some added storage drives.).

[@Anderson_LS]

Steve, the b580 MSRP is fake. Please include a real world price in your charts.

[@Tom_Quixote]

In my country, it costs the equivalent of $345

[@Jtq-y4n]

@@Tom_Quixote In the US it's basically not even available outside of 350+ USD scalpers. Paper launch.

[@TheHalfmanofOz]

I'm not sure why but it seems to be selling close to msrp in Australia (accounting for gst)

[@AndyViant]

@@Jtq-y4n Where I am, they're US$270, including taxes. So, without the tax that works out to $251 for a ASRock Arc B580 Challenger 12GB OC. About $257 for the Intel B580 LE. Still available at some retailers. Your problem is America, and American behavior. Everyone trying to find a way to make money by scamming each other, or rushing to buy the newest most hyped thing. Order from another country, don't support scammers.

[@pieflies]

@@TheHalfmanofOz maybe not seeking as well in Australia?

[@Sunlight91]

Do you finally understand why testing with non-top CPU isn't wrong? Testing with the top CPU is just one part of a complete test. Don't complain about other reviewers doing different parts of GPU testing.

[@nonamenameless5495]

Wow, this must have been a lot of work, epic comparison- thanks a lot for the enormous effort!

[@TheBeastsHere]

surprised there is no rx 6800 in these results as you can get one for around 270-280.

[@bit8793]

Bro where can u get that then 😂😂

[@TheBeastsHere]

@@bit8793 Ebay

[@PixelatedWolf2077]

But that's used 😂

[@TheBeastsHere]

@@PixelatedWolf2077 All my GPU's have been used and never had a single problem, I would rather buy a more powerful used GPU than a slower new one.

[@TexasFun]

This is a better, much more informative video than the one with the 2600 or HC's "it's broken" video... Thanks Steve...

[@KattarAthiest]

Test old games on Arc b580 .

[@AutieTortie]

Why?!

[@Jtq-y4n]

They don't work lmao

[@dvornikovalexei]

DF tried two games, Infinite Warfare and something else and both didn't launch

[@riven4121]

@AutieTortie Intel has issues in games that aren't DX12 based.

[@TheGreenReaper]

@@dvornikovalexei Makes sense, Intel has past issues with miscalculations, perhaps it tried to load Infinity plus one...

[@Nestor.555]

Good stuff Steve, thanks :) It'd be interesting to further investigate a possible overhead issue with AMD. The data of some CPU intensive titles seem to suggest that? Moving on, I would like to see testing of new GPU's also being done with more "real world" test setups, to sniff out possible overhead issues right from the start.

[@truetierra]

would love to see older cpu's through the generations with these 3 cards - even as far back as sandybridge would be really neat

[@lairbox]

I suggest also taking an intel platform to test these Arc cards like intel 12400F. Most of the buyers will try to have intel based system. I believe that with that there will be a more balanced outcome. I also would see a re-review of the intel core ultra 285k, seems that another bios update has been released after 0x114.

[@HodgePodgeProducts]

The 12400F is not going to be any better here. It has very similar performance to the 5600.

[@lairbox]

@@HodgePodgeProducts I'd love to see anyway. Seems 10% more faster than amd 5600. My serious bet is on the 5700x3d or 5800x3d because someone that builds an entry level pc ever considers a x3d processor.

[@AndyViant]

That's one thing I was a bit sad about as all the main review channels released their new year, new test rig setups. We used to sometimes pick up on this kind of behaviour when Hardware Unboxed was running the 7800X3D test rig and Gamers Nexus was running the 13700KF test rig. But now all the major channels seem to have shifted to 9800X3D setups.

[@lizardx6504]

@@lairbox Nobody pair 12400f with ddr5. With ddr4 12400f=5600

[@raulitrump460]

@@HodgePodgeProducts asrock b660 and b760 oc can give you up to 30% extra performance so it would compete even with 5700x3d

[@stevenjamesvillasan9674]

how about pairing it with intel processors?

[@libaf5471]

What the hell is going on in the reviewer space? 250 dollars is NOT the "typical newegg price" for b580.

[@AndyViant]

Most people don't buy on newegg.

[@lowzyyy]

Its 400$ here

[@fabrb26]

@@AndyViant I just got home from local store in France, price tag is at 439€ for the As Rock steel legend. Only that particular model on shelves... But still a better deal that the 999€ Asus Tuf 4070 Ti Super next to it 🙄🙄🙄

[@libaf5471]

@@AndyViant It's not that price anywhere.

[@libaf5471]

@@fabrb26 Finland here. B580 not in stock anywhere. Cheapest guesstimate price is 300. But for 400 you can pick up a 4060ti today. Or a 7700xt for 430. If b580 was actually in stock, that's where the price would probably fall too. And it does not compete against those cards.

[@NerdrageHockey]

Thank you Steve. I don't know if it is just me but I'm not really shocked by these numbers. If you have a 'Slow' PC throwing in a new Video card doesn't magically make it a 1440p gaming power house. If your CPU is several generations old (and the MB with it) then I think the expectation is that it is time to upgrade. If anything videos like this are great to dispel the myth as put forward by all the initial videos (i.e. B580 = 100 FPS at 1440 in Cyberpunk). The numbers just show that if you are building a budget rig the b580 is an option. I don't know to many people that are going to drop $490 on the CPU alone but straddle it with a $250 video card (don't get me wrong they are out there). I would love to see more budget friendly or middle of the road setups tested as that is really what this card is marketed for. I'm running a ryzen 5-3600/16gb/6600 and waiting to see if the team red and green will have something for my upgrade later this year now that this is an option.

[@suspencer5269]

Why only ryzen processors why no intel CPUs know it wouldn't change much, it's just weird to have that many tests on ryzen only.

[@Jtq-y4n]

I think intel shipped this out early because they knew new GPU's would come in soon and dominate any possible value this thing had. Intel is spiraling.

[@JohnDoe-yo3ui]

I thought the same exact thing. Intel market stock is low, it makes inferior CPUs, their A series GPUs were a flop. So Intel seeing this inferiority of their product vs Nvidia and AMD tried to gain ground by three ways: Have it be much less efficient than the 4060 and 4060ti. More power, less efficient. More performance because of the added power used. Intel did this with their inferior CPUs vs the 65 watt AMD CPUs of the 5000 series. 7nm AMDs vs 10nm Intel, how do you mask that failure, you increase power consumption. Less efficient, get marginal or large gains vs your competitors, though on an inferior product. When you buy AMD or Nvidia, you are buying the Ferrari or Lamborghini class of technology - sleek, stylist, works of art. They are beautiful to behold. When you buy Intel you are buying a pick-up truck. Intel buyers: "With my pick-up truck, I am gonna run over your Ferrari, gerrrr". AMD and Nvidia owners: "lol". More VRAM. Where the GPU is failing vs competitors, add more VRAM to mask this Intel problem. If Intel could make a good 8GB GPU, then make it better with 12 or 16GB of VRAM, then Intel has a product. The added VRAM is there because their 8 gb version would suck so bad. AMD and Nvidia can still make 8G GPUs because they make good quality products. The added VRAM is the only smart decision Intel makes, and it is used to hide bad performance. Run on the coat tails of better and much more expensive CPUs. Not good with efficient and older CPUs? Know the reviewers test with newer and most expensive CPUs, go this route and fool the public that this is a budget GPU. Have the GPU use the added performance of expensive CPUs to help beat AMD and Nvidia in several games. Also ignore the B580 is unstable. It is not consistent as AMDs and Nvidia GPU. AMD and Nvidia are stable from game to game. Not so with Intel B580, it does fine with a few games, then gets crushed in some. Then knowing Nvidia went the efficiency path with the 4060 and 4060ti, have it released so it would be compared with apple and oranges 4060, not the much more power hungry 5060 which is going to blow the B580 away. Millions of pre-built system were sold with 1650 and 1650 super and 1660 and 1660 supers. These systems had 400-450 watt power supplies. Many of those that buy pre-builts from dell, hp and elsewhere don't want to upgrade the Power supply or MOBO. RAM upgrades are easy. Storage upgrades are easy. GPU upgrades are easy. Power supply is more difficult and add expense to the GPU total upgrade. 4060s and 4060ti are made for upgrading the 1650s and 1660s series. This is what the tech reviewers missed on the 4060. 4060 was a G-d send to those with low power supply pre-builts. The B580 is certainly not for pre-builts with limited power. Fine job Steve. Though I would have wanted the reviews on Ultra/High in 1080 and 1440. Medium Settings should be shown when one of the GPUs drops below 60fps. To show if it is playable on Medium Settings. Other than that, I was happy with the review. Steve ignore the criticism that the review only did AMD older CPUs. Older Intel CPUs were done by several other reviewers. Specifically 9th generation Intel and 12 generation Intel CPUs and the results were nearly identical to your findings Steve. Big drops. Nvidia and AMD make reliable GPUs, consistent is every game. Intel built their GPU, deliberately or not, to fool the tech reviewers and it is good to see you Steve got out of that trap. The best budget GPU for pre-builts the past 2 years has been the 4060ti. The performance for 1660 super watts used is phenomenal. It was $370 on Amazon for months for the 8gb VRAM. If you could afford the 16 GB, then you made your system last so much longer since games are requiring more and more VRAM. Though waiting for the newer series in 2025 should be better depending on price. Those new GPUs should be much better than the B580. That is why Nvidia and AMD charge more, because of quality.

[@basyapupkin3444]

In TLoU B580(5600) same fps in 1080p and 1440p - is that correct? Same in Spider-man.

[@Hardwareunboxed]

Yes, overhead issue.

[@samgragas8467]

That is what happens when the CPU is the only bottleneck. Overhead made the CPU slower and it reached its limit.

[@TheGreenReaper]

@@Hardwareunboxed Good for 4K upscaling then! My monitor only does 60 FPS anyway. ^^

[@riba2233]

Not to mention that they are selling for over their msrp lol, what an epic fail

[@haukikannel]

Every AIB sell above msrp… Plame Asus, MSI, and so on for that!

[@riba2233]

@@haukikannel neither asus nor msi are selling Intel gpus. You can only blame Intel, they set the price policy and made just a small amount of MSRP units because they know they are losing money on them.

[@RyTrapp0]

@@haukikannel No - every AIB *has some models* that sell above MSRP. They also typically have a base model for each GPU that they sell at or near MSRP. And, when they don't, like some of the MSRP shenanigans that Nvidia has played, they get called out for it as they should.

[@wewillrockyou1986]

​@@RyTrapp0The intel models are above MSRP too lmao I has nothing to do with the AIBs it's just that the supply is so meagre

[@haukikannel]

@ Yeah! Nvidia normally say that you have to have x units with msrp and rest you can do what ever… The problem normally is… China and scalpers gets those msrp versions and we rest has to buy those owerpriced AIB models… It was like that woth 30 series and 40series. So that is what I expect this time also.

[@Steamrick]

Thank you for this incredibly comprehensive retesting.

[@Veptis]

as one of the first 200 A750 owners... still on a Skylake i5 - I am really excited about the potential performance upgrades once I can buy ARL in Germany!