@@Dark.Syndicate At the high end without a doubt, but similar to their GPU's they still have some good value options at the budget end. GN's testing showed the i5-12400 did a lot better than the 5600x in terms of performance regression. I'm not suggesting any kind of bias on Steve's testing, just adding another CPU is 50% more work, but I would like to see his analysis of the two platforms in upcoming content.
I really like the way these benchmarks are arranged, I'm sure it makes the "test on realistic hardware combos" crowd somewhat happier (maybe?). Since we see some scaling on several GPUs (with ARC showing the most) maybe this is the best way to do benchmarks going forward? I can appreciate if there's too much work involved but the presentation is just what I want.
It's unnecessary most of the time because each upscaler has different visuals, which is subjective to whether a person will put up with the drop in image quality, and multiple modes and changes over time. It is pretty good here, though. The situation kinda warrants it a bit.
@@user-pw7ii4px4d I definitely happy about offering sanity over insanity. We are the viewer and the consumer we are here for real life usable data if we buy we need data that can be used in face value and not after we compare calculate and guess how it will perform in our different rig with different cpu memory etc. Usually eliminating bottleneck is a good thing but it become evidential that not the case by the current method that current reviewers used so far, there are more variable at play that crush the "pick the best cpu for the test" mentality for GPU testing that yet to be admitted to the point that changed the whole testing methodology on every product that tech reviewers test until than every test is half assed evidentially with the wrong data with the might be correct conclusion. If some data might be right that is not the correct data with correct testing methodology.
@@riven4121 It seems to affect AMD and NVidia GPUs like the 4060 and 7600 by up to 10 percent on some games, sometimes higher. You can see some examples in this video.
I would love the most fair benchmark comparison between Nvidia, AMD, and Intel and that would be: Nvidia Geforce GTX 570 and 580 vs AMD Radeon RX 570 and 580 vs Intel Arc B570 and B580
@@budgetking2591 and the size of chip in the compure affects you as a computer how exactly? Really does not matter is the performance made with engineering magic or using more silicon; the cheaper product with same performance is better to buy
True that's how PC component pricing works in my country. But I also checked stores in my country, only one place sells it online so far, and they sell it for (not kidding) $900. I guess this store is trying to milk the fact that nobody else sells it yet, although it's quite interesting they think someone will be stupid enough to buy it for this price.
@@WireHedd it is when competitors dont have this issue. Specially when this is priced again so poorly in Europe that you can just buy the 4060 for almost same price and not to worry about some cpu overhead issues
Love seeing the extra CPU being used for this, hopefully we can see some more of that in future reviews as well. It paints a much fuller picture of what we can expect from this card if we're actually buying it, and helps catch any weird overhead issues in the future.
That's the neat part about the upscaling data. When you enable upscaling, you are in effect rendering at a lower res (eg. 1080p). So it gets closer to those numbers. But I"m sure it'd be even more with full esports titles.
and also don't forget that 220 will be 320 due intentionally faking good will with the good price and proceeds to hold stocks in the warehouses and delivering it on weekly basis to inflate the price artificially admitted by intel the weekly part the other facts are evidential. Plus in eu get taxed and the price will be a joke, DOA, since you can buy faster cards used, in this price point nobody worrying about it being used gpus.
Whether we like it or not MooresLawIsDead was right this was just a paperlaunch made for Intel to get good press out of reviewers by "offering" great value that is not actually available to most of the public. The CPU overhead issue is just the last nail in the coffin. Hope I'm wrong and they get their shit together when it comes to supply and drivers but time will tell.
@@Josh-cw8by if you compare before tax prices you will find that they aren’t significantly overpriced. The VAT rates aren’t Intels fault and they apply to everyone in the market equally.
If B570 would have 12gb of vram then $220 msrp would be reasonable, you pay 13% less and get 13% less performance. But when you also have less vram, there's no point to save these $30, so msrp should be at least $200 to make some sense
@@DragonOfTheMortalKombat and then we'd have all set props glitching, stretching and collisions lauching everything all over the place... that would be 5070 for $2000 scenario
Thank you so much for including two CPUs! This gives so much more insight! I would love this format for any GPU review because this gives so much insight on whether a specific game will benefit from a "better" CPU in combination with certain GPUs!
Those with a Zen2/3 or Intel LGA1200-ish CPU who want a $250-ish upgrade from a 1060/1070 or 2060/2070 or a 5700XT are far better off exploring the used market. A used 2080Ti with 11GB of GDDR6 can be had all day for $250, and it will outperform the B570 by ~20%. Yes, it's a used card, but so is the one in your machine right now.
That would be true if I had a GPU. My current CPU is a Ryzen 5 4600G. The used market in my country is almost nonexistent (and what is available you can only find it in FB Marketplace and it's quite pricey considering it's used)
A used Radeon card really just makes more sense to me with all the constraints and potential driver issues with Intel. I helped a friend build a budget gaming PC and we got her a 5700x3d for $129 (USD) new at Aliexpress and I bought a 6650XT on Ebay for $155 shipped. $38 for 32GB of DDR4 3600hhz ram. and she has a very nice little gaming PC.
I own a 7900xtx but I mainly use it for AI (it's just sitting and doing LLM and embeddings stuff). My main computer I work on still has a rx6600 in it and I recently finished cp2077 on it. I could have swapped in the 7900xtx to have a better gaming experience, but really once I tweaked the settings on 1080p, rx6600 performed really well. 6650xt is even faster than that. Great little GPUs.
Even despite the overhead, the question still remains if the rumors are true, and Battlemage supply is very low, and these will remain GPUs that most people will never be able to buy.
DX12 and Vulkan are API:s where the purpose was to minimize driver overhead while OpenGL and DX11 are older API:s which are further away from hardware and I was thinking that by now the games wouldn't hurt that much by driver overhead because newer API:s have minimized the amount of work driver needs to do.
Great review! I really like the extensive on screen graphs, but not sorting them highest to lowest makes them quite difficult to compare. I'd recommend sorting highest to lowest FPS with the card that is being reviewed having a different colored text or bar graph color, rather than the dark highlight. Keep up the good work!
Also difficult to remember the percentage differences as you speak them. I'd add a percentage at the end of each bar graph line, on the far right, with a percentage that is the difference from the card being reviewed. Red text for percentages indicating slower than the card being reviewed and green text for percentages indicating faster. Just my two cents.
Good, useful review. One other thing I would add - there are 4060s selling for the same price as some 7600XTs - the 16GB VRAM is worth considering when deciding which to buy. Given that neither the 4060 nor 7600/7600XT are candidates for RT gaming, for the money there may be better value going forward in the 7600XT.
There are so many tests-native resolution, with upscaling, on different CPUs, and with four different GPUs in various scenarios-that I even have trouble keeping up, and that's amazing! 😂
A budget gpu that doesn't work well with a budget gpu isn't anything to be excited about. You could make the argument "people need to build on modern platforms" but what about people with old pcs? If intel fixes the issue these gpus are definitely worth it but we still have to see it.
I just checked the prices where I live. The difference between the B570 and B580 was under 5$. So maybe give it a little more time for prices to stabilize.
B580 seems to be around 260-270 when it's in stock at Microcenter. Something I have seen one time with 4 units in stock - tho I do not check constantly, particularly since my nearest is 3 hours each way.
Still rocking an RX 5700 XT and playing games in 4K and loving it. No need for an upgrade unless I get a 4070 or higher but no way am I going to pay the inflated prices for that.
This is actually a good way to test the lower end GPUs. Using two CPUs at different ends of the price range. Even if there is no overhead it will clearly show any potential CPU limitation. It may not be the same across all GPUs (AMD vs Nvidia vs Intel).
The Australian pricing is insane, if you use the US msrp ($250) then the final price AUD is $375! I do not have no idea what Intel or the vendors here are thinking.
Current exchange rate puts 250usd at 402aud. Also, keep in mind that US pricing DOES NOT include sales tax while Australian pricing DOES. That said, there appears to still be about a $35AUD markup once you include sales tax (using 8.25%) and convert USD to AUD
@@jeffotaylor Thanks for letting me know, I didn't realised our currency got even lowered lol. There is a 10% tax on all imports here, which was something I completely forgot.
This kind of comparison of the benchmarks is the best.Nobody is going to buy a better CPU just for the sake of that "overhead". Keep up the good work,guys 🍻
They purposely charge it. In my country, most of the dealers are Nvidia's and AMD's bootlicker so if you ask for B570 or B580, they will charge higher.
Why not also throw i5 and i7/i9 into the testing mix? They are still popular and Intel might have a slight improvement there, like AMD does with their own CPUs
AMD needs to learn this simple lesson: set the MSRP to a low value to get a good review, then sell at much higher prices in the real world. When will they ever learn?
AMD has no control of the price the stores set. They only set MSRP. They don't benefit by higher store prices either. So, if they set the MSRP too low, they end up screwing themselves by having to sell the cards to retailers with barely any margin for themselves.
Intel needs to fix this issue asap. Nobody that even considers this GPU, is mad enough to pair it with a high end CPU. Its a budget card meant for budget systems and this much performance shouldn't be left on the table.
Awkward differences. From a production point of view with this amount of units, why not just produce more B580s for like 10-15 dollars cheaper? I'm not quite sure what the goal was?
If i were intel, figuring out what is causing the overhead issue would be my number 1 priority. The b570 could one of the best bargains in a long time if they can just iron out the overhead.
I feel like I own one of the only B580s in the wild. Somehow snagged one of the Intel Limited Edition cards. So far on a 13600K it's fine, I anticipate that a B570 would be fine for anyone with a 12th gen Intel CPU and up.
I paid 319EUR for a B580, it's now powering my nas / media server, handling any transcoding job better and faster than anything I had before. Running decently fast Mistral 7B local LLM is also nice. Even if it was a bit more expensive than the 4060, was still worth for better out of the box open source support, better transcoding quality and speed.
Thank you for the 5600 testing! I am the owner of one lol. Yes I can upgrade to 5700x3d but Im scared to tamper with that upgrade, not confident enough with it yet (and yes I built the system myself but that I found easier than taking off the cooler, unsocketing, and socketing in the new CPU, not comfortable with that).
So basically: If all GPUs are at MSRP the Arc cards are still the best value (cost per frame) off all the cards not matter the CPU at 1440p. I hope they can somewhat fix the issues with driver updates, but it's still a decent card for the money (if you can get one for the advertised price)
Rx 7600 is the joke tbh .. not powerful enough,isn't power efficient, has 8gb ram, bad RT, bad upscaling & costs around same as 4060. Overall all 4 gpus are similar performance, won't notice any difference from going one to another
The vast majority of gamers who have slower CPU's would still see an uplift in performance considering most don't have a GPU that can handle 1440p gaming at all.
That's great that by Steve's idea nobody buying this GPUs for new build but only upgrading old outdated computers. Nobody will ever think of buying 150USD CPU like 13400F and 250USD GPU.
I am a budget gamer with an old RX 470, and I upgraded my motherboard/CPU/RAM a year or two back to Alder Lake because GPUs were too expensive. I'm thinking there could be others out there like me who didn't want to overpay for a GPU so they put that money into the rest of their system instead. (My system upgrade felt like a bargain compared to any GPU I might want.)
In terms of performance per dollar, the B570 certainly doesn’t make a lot of sense, but depending on the usage case scenario, 30 dollars is 30 dollars. For a workload/gaming mixed-use scenario where that 30 dollars might be better spent on a better CPU, the 570 makes a lot of sense.
Your calcs weren't that bad anyway it's still more. Anyway it's not retailers gouging, they pay more for stock. Margins in this industry are below 10% most of the time, especially for internal components.
This could be still decent for somebody who mostly does productive work (programming etc.) and wants to do occasional gaming. But that is of course provided it is available at a decent price.
Thanks for the comparison. Some things i would've liked to see tested is these battlemage cards with the top of the line AM4 CPU, the R7 5800X3D (or the 5700X3D) to see if the cpu overhead problem persists. Also I would've liked if you included the settings used for the upscalling benchmarks, i hope you didn't use XeSS at quality mode since it has a better picture quality compared to DLSS and FSR at the cost of lower framerates, if you use FSR at quality then you must use XeSS at balanced mode to achieve similar picture quality.
@ArchieBunker11 Hopes, mostly at this point. But more seriously, the RX 7000 generation was very weird at the entry level. A RX 7600 barely hung there with the previous gen RX 6600XT and there was a big gap to the RX 7700XT, which was around RX 6800 performance. The RX 6700XT had no real equivalent in the RX 7000 series. AMD talked specifically about that gap a couple of weeks ago and how the RDNA 3 was not flexible to target that level of performance. I think they will aim exactly at a replacement for the RX 6700XT with the entry level RDNA 4. And the RX 6700XT-like performance and 12GB VRAM would be a preferable option than the B580, IMO
@@konstantinlozev2272 Why do people look back on the 6700xt with such rose tinted glasses? The 6700Xt had an msrp of $479. $80 more than the 3060Ti, and $20 less than the 3070. And according to TPU’s 2025 GPU test suite with modern games its lost ground to the 3060ti, in spite of VRAM. The 7700xt was a low performance increase over the 6000, but its msrp was slightly better. If they want to repeat that, they’ll get beat down by nvidia again. If that model is going to work, you’re right. It would have to be cheap
Here in the Philippines, the b580 is priced 300usd and stores are all out of stock, much to no one's surprise. The 4060 here is priced the same or just 20-30usd more
Shaking my head at the quad charts. That overhead issue is really making GPU testing a pain for you. I hope Intel has all hands on deck to improve this. I mean a CPU load issue, implies this is a driver problem that is fixable.
I have a 7600X3D and have been trying to buy the b580 watched it go out of stock while I was putting in my info. Looking at the lower cpu stuff from several other videos including yours it seams it just comes down to if a CPU bottle neck happens it is more prominent. Which, to me, means content creators need to discuss how important a balanced system is and how you in general want to be GPU bound. So if you are on those older CPUs it might be time to upgrade your CPU instead of your GPU. Or even both
Hats off for including both CPU for GPU benchmarks. This is actual and honest hard work!
Everything he does is actual and honest hard work :)
Only AMD CPUs tho, for some reason.
@@Oozaru85 the reason being that intel is lagging behind right now. dont be an intel fanboy
@Dark.Syndicate Sure, AMD fanboy.
@@Dark.Syndicate At the high end without a doubt, but similar to their GPU's they still have some good value options at the budget end.
GN's testing showed the i5-12400 did a lot better than the 5600x in terms of performance regression.
I'm not suggesting any kind of bias on Steve's testing, just adding another CPU is 50% more work, but I would like to see his analysis of the two platforms in upcoming content.
The new layout showing Native/Upscaling is freaking awesome!!
It is great. Anybody who wants 1080p Data can simply look at upscaling results and the 2 different CPU data is also much appreciated.
Performance wise yeah but you should consider visual artifacts too
Agreed 🙂
Don't get used to it 😅
Still weird rooting for Intel GPUs while against their CPUs.
Except for 12400f
IKR 😂
Really weird ngl 😂
except for 12th gen
AMD in 2013:
I really like the way these benchmarks are arranged, I'm sure it makes the "test on realistic hardware combos" crowd somewhat happier (maybe?). Since we see some scaling on several GPUs (with ARC showing the most) maybe this is the best way to do benchmarks going forward? I can appreciate if there's too much work involved but the presentation is just what I want.
I think the 5600 fall into that category, the 5600x is 120 dollars at Amazon.
I mean, outside of Arc, it's basically irrelevant because AMD and Nvidia don't have driver overhead this severe.
Arc is the exception, not the rule.
It's unnecessary most of the time because each upscaler has different visuals, which is subjective to whether a person will put up with the drop in image quality, and multiple modes and changes over time. It is pretty good here, though. The situation kinda warrants it a bit.
@@user-pw7ii4px4d I definitely happy about offering sanity over insanity. We are the viewer and the consumer we are here for real life usable data if we buy we need data that can be used in face value and not after we compare calculate and guess how it will perform in our different rig with different cpu memory etc. Usually eliminating bottleneck is a good thing but it become evidential that not the case by the current method that current reviewers used so far, there are more variable at play that crush the "pick the best cpu for the test" mentality for GPU testing that yet to be admitted to the point that changed the whole testing methodology on every product that tech reviewers test until than every test is half assed evidentially with the wrong data with the might be correct conclusion. If some data might be right that is not the correct data with correct testing methodology.
@@riven4121 It seems to affect AMD and NVidia GPUs like the 4060 and 7600 by up to 10 percent on some games, sometimes higher. You can see some examples in this video.
I would love the most fair benchmark comparison between Nvidia, AMD, and Intel and that would be:
Nvidia Geforce GTX 570 and 580 vs AMD Radeon RX 570 and 580 vs Intel Arc B570 and B580
🤣
Intel looking good there
perfectly balanced
Can't argue with that logic
Nah, b570 and Switch 2 on the same day is wild.
Oh shit, I didn't know
as if b570 has any hype, its a technical flaw, the chip is much bigger then its competitors for same performance.
@@budgetking2591 and the size of chip in the compure affects you as a computer how exactly? Really does not matter is the performance made with engineering magic or using more silicon; the cheaper product with same performance is better to buy
it really isn't
@@budgetking2591idiotic comment
Meant to cost $220, costs $320.
And depending on your country add a 10-30% tax to that
The EU classic
True that's how PC component pricing works in my country. But I also checked stores in my country, only one place sells it online so far, and they sell it for (not kidding) $900. I guess this store is trying to milk the fact that nobody else sells it yet, although it's quite interesting they think someone will be stupid enough to buy it for this price.
@@canyongoat2096For the novelty? Bizarre
@@aziaufa418 Trump here to change that, US join the club
Has Intel commented yet on the driver overhead issue?
maybe in a year or so and then change their story every week
The official Intel_Support account on reddit said they are looking into it. Idk why they won't release an official statement though.
They have found the issue, its a scaling issue at a software level, a fix is number one priority rn but unknown date for fix
Oh yes the overhead "but" still exists..
it seems to be actually better here, the gap between b570 and b580 is lesser with cpu overhead.
...and that's a big ol' "but" to have hanging over a new release.
@@WireHedd it is when competitors dont have this issue. Specially when this is priced again so poorly in Europe that you can just buy the 4060 for almost same price and not to worry about some cpu overhead issues
@@OneDollaBill depends where in Europe you are...i bought b580 for 290€, cheapest 4060 was 310€..
Still better than a RTX 4060.
Love seeing the extra CPU being used for this, hopefully we can see some more of that in future reviews as well. It paints a much fuller picture of what we can expect from this card if we're actually buying it, and helps catch any weird overhead issues in the future.
But where is the extra Intel CPU?
Esports 1080p will show a big driver overhead.
Exactly what i was thinking
That's the neat part about the upscaling data. When you enable upscaling, you are in effect rendering at a lower res (eg. 1080p). So it gets closer to those numbers. But I"m sure it'd be even more with full esports titles.
we got a 10gb card for $220 while on the green side we have a 60 class card on 8gb
We all know it's gonna be 8GB but I still hope Nvidia has the sense to make it at least 10GB.
dont forget the 500$ cpu you need for it to actually be any faster
and also don't forget that 220 will be 320 due intentionally faking good will with the good price and proceeds to hold stocks in the warehouses and delivering it on weekly basis to inflate the price artificially admitted by intel the weekly part the other facts are evidential. Plus in eu get taxed and the price will be a joke, DOA, since you can buy faster cards used, in this price point nobody worrying about it being used gpus.
@@keno5839 no you don't. A 180$ ryzen 7600 or 7500f is more than enough lol.
@@DragonOfTheMortalKombat most people buying cheap gpus are still on old gen motherboards
People aren't getting B580s and man has a B570
Whether we like it or not MooresLawIsDead was right this was just a paperlaunch made for Intel to get good press out of reviewers by "offering" great value that is not actually available to most of the public. The CPU overhead issue is just the last nail in the coffin. Hope I'm wrong and they get their shit together when it comes to supply and drivers but time will tell.
@@enmanuel1950 the suppliers and distro says no moronslawisdead ego hates all intel and ego is big but balls is small
What are you talking about. Cards a re available for decent prices here in Germany. People are getting them. Its just some people who cant get them.
@@Oozaru85 They're overpriced cards in the EU. Obviously nobody is going to buy them and they're going to be in stock.
@@Josh-cw8by if you compare before tax prices you will find that they aren’t significantly overpriced. The VAT rates aren’t Intels fault and they apply to everyone in the market equally.
If B570 would have 12gb of vram then $220 msrp would be reasonable, you pay 13% less and get 13% less performance. But when you also have less vram, there's no point to save these $30, so msrp should be at least $200 to make some sense
Sitting Steve. Oh, the relief!
the cliché like me post!
that's okay i guess, he could be:
- standing
- floating
- floating and T-posing
- floating, T-posing and spinning
@@1Grainer1 don't forget the ragdolling
@@DragonOfTheMortalKombat and then we'd have all set props glitching, stretching and collisions lauching everything all over the place... that would be 5070 for $2000 scenario
So weird comment
Thank you so much for including two CPUs! This gives so much more insight! I would love this format for any GPU review because this gives so much insight on whether a specific game will benefit from a "better" CPU in combination with certain GPUs!
Arc Battlemage: the intel CPU that sells AMD 9800x3D
Fail
**Intel GPU
Arc Battlemage: for those who bought 9800x3D and ran out of money :-)
Those with a Zen2/3 or Intel LGA1200-ish CPU who want a $250-ish upgrade from a 1060/1070 or 2060/2070 or a 5700XT are far better off exploring the used market. A used 2080Ti with 11GB of GDDR6 can be had all day for $250, and it will outperform the B570 by ~20%. Yes, it's a used card, but so is the one in your machine right now.
Really? I can't find anything on eBay that is priced reasonably. Lowest I find is like 10% below MSRP, highest I find is 200% higher than MSRP.
@@connivingkhajiit Wait, what? What are you actually looking at on eBay?
@@Navi_xoo Everybody makes his or her own choice; that's the beauty of having options.
@@Navi_xoo No it isnt. Maybe in your country. That doesnt make it a paper launch.
That would be true if I had a GPU. My current CPU is a Ryzen 5 4600G.
The used market in my country is almost nonexistent (and what is available you can only find it in FB Marketplace and it's quite pricey considering it's used)
The only one to blame for the more expensive Arc prices in Australia is the Australian government.
isn't the msrp identical to the us price? 439AUD (with sales tax) -> 399AUD (without sales tax) = 248.88USD
That doesn't explain why some parts have a bigger markup compared to MSRP than others.
@@aapje volume. Low volume means the overhead is higher.
A used Radeon card really just makes more sense to me with all the constraints and potential driver issues with Intel. I helped a friend build a budget gaming PC and we got her a 5700x3d for $129 (USD) new at Aliexpress and I bought a 6650XT on Ebay for $155 shipped. $38 for 32GB of DDR4 3600hhz ram. and she has a very nice little gaming PC.
I own a 7900xtx but I mainly use it for AI (it's just sitting and doing LLM and embeddings stuff). My main computer I work on still has a rx6600 in it and I recently finished cp2077 on it. I could have swapped in the 7900xtx to have a better gaming experience, but really once I tweaked the settings on 1080p, rx6600 performed really well. 6650xt is even faster than that. Great little GPUs.
Even despite the overhead, the question still remains if the rumors are true, and Battlemage supply is very low, and these will remain GPUs that most people will never be able to buy.
thanks for the AUD comparison as local pricing really is a different picture. Great review as always
DX12 and Vulkan are API:s where the purpose was to minimize driver overhead while OpenGL and DX11 are older API:s which are further away from hardware and I was thinking that by now the games wouldn't hurt that much by driver overhead because newer API:s have minimized the amount of work driver needs to do.
😮
Well yes that's why Intel GPU prefer vulkan and dx 12. But Intel being still very new requires more work even with these APIs.
The price is really good if stock is available!
Great review! I really like the extensive on screen graphs, but not sorting them highest to lowest makes them quite difficult to compare. I'd recommend sorting highest to lowest FPS with the card that is being reviewed having a different colored text or bar graph color, rather than the dark highlight. Keep up the good work!
Also difficult to remember the percentage differences as you speak them. I'd add a percentage at the end of each bar graph line, on the far right, with a percentage that is the difference from the card being reviewed. Red text for percentages indicating slower than the card being reviewed and green text for percentages indicating faster. Just my two cents.
Good, useful review. One other thing I would add - there are 4060s selling for the same price as some 7600XTs - the 16GB VRAM is worth considering when deciding which to buy. Given that neither the 4060 nor 7600/7600XT are candidates for RT gaming, for the money there may be better value going forward in the 7600XT.
I wonder if this will lead to the "fine wine" effect, where the Arc GPUs end up increasing their lead as CPUs get faster.
it will but no one will care because you're not purchasing a brand new CPU when you can't afford a decent graphics card.
Finally, GPU for the proletariat!
There are so many tests-native resolution, with upscaling, on different CPUs, and with four different GPUs in various scenarios-that I even have trouble keeping up, and that's amazing!
😂
A budget gpu that doesn't work well with a budget gpu isn't anything to be excited about. You could make the argument "people need to build on modern platforms" but what about people with old pcs? If intel fixes the issue these gpus are definitely worth it but we still have to see it.
Marvel rivals in the benchmark suite🔥🔥
Steve's gotta find time to look at Squirrel Girl somehow.
The benchmark should be Strange opening a portal 😂
@ I thought of that too strange portal into a huge fight drops fps by so much lol
ngl 570 gotta end up as a daily driver E-sport card including Marvel Rivals.
Love this review format
He's sitting guys
sad copy paste attention post
@ScotsmanGamerSad 1 brain cell reply 😂
@ Stick to sonic games kid on your switch!
I just checked the prices where I live. The difference between the B570 and B580 was under 5$.
So maybe give it a little more time for prices to stabilize.
B580 seems to be around 260-270 when it's in stock at Microcenter. Something I have seen one time with 4 units in stock - tho I do not check constantly, particularly since my nearest is 3 hours each way.
Still rocking an RX 5700 XT and playing games in 4K and loving it. No need for an upgrade unless I get a 4070 or higher but no way am I going to pay the inflated prices for that.
Thank you for adding newer games like marvel rivals and Indiana Jones It is much appreciated
This is actually a good way to test the lower end GPUs. Using two CPUs at different ends of the price range.
Even if there is no overhead it will clearly show any potential CPU limitation. It may not be the same across all GPUs (AMD vs Nvidia vs Intel).
The Australian pricing is insane, if you use the US msrp ($250) then the final price AUD is $375! I do not have no idea what Intel or the vendors here are thinking.
Current exchange rate puts 250usd at 402aud.
Also, keep in mind that US pricing DOES NOT include sales tax while Australian pricing DOES. That said, there appears to still be about a $35AUD markup once you include sales tax (using 8.25%) and convert USD to AUD
@@jeffotaylor Thanks for letting me know, I didn't realised our currency got even lowered lol. There is a 10% tax on all imports here, which was something I completely forgot.
in my country the price is $263
This kind of comparison of the benchmarks is the best.Nobody is going to buy a better CPU just for the sake of that "overhead".
Keep up the good work,guys 🍻
Thorough but simple, thanks
In my country even RTX 4060s are cheaper than B580s
They purposely charge it. In my country, most of the dealers are Nvidia's and AMD's bootlicker so if you ask for B570 or B580, they will charge higher.
@@rahuldey8539 Sure...
Damm! 16000/17000 Gbps of memory speed on that Arc A770
Why not also throw i5 and i7/i9 into the testing mix? They are still popular and Intel might have a slight improvement there, like AMD does with their own CPUs
AMD needs to learn this simple lesson: set the MSRP to a low value to get a good review, then sell at much higher prices in the real world. When will they ever learn?
AMD has no control of the price the stores set. They only set MSRP. They don't benefit by higher store prices either. So, if they set the MSRP too low, they end up screwing themselves by having to sell the cards to retailers with barely any margin for themselves.
Can you do an intel CPU test please with these GPUs
Intel needs to fix this issue asap. Nobody that even considers this GPU, is mad enough to pair it with a high end CPU. Its a budget card meant for budget systems and this much performance shouldn't be left on the table.
IDK in the UK the RX7600 (£230-249) is cheaper than the B570 (£253) and the B580 is like £300.
Same in germany.
7600 is 270€
570 is 290€
4060 is 310€
580 is 330€
@@nipa5961 Thanks, it kinda makes me feel a little better lol (Misery loves company)
@ It's really depressing. MSPR is 290€ here for the B580 but it actually costs 320-340€. Even the old and much faster 6750XT cost the same.
In India
$277 for Rx 7600
$312 for RTX 4060
Battlemage hasn't even launched here 🙃
Good info. Great work sir.
At this point,retailers ARE the scalpers...
Glad I paired a RX 7600 with my Ryzen 5 3600 last year. Lucky.
Awkward differences.
From a production point of view with this amount of units, why not just produce more B580s for like 10-15 dollars cheaper?
I'm not quite sure what the goal was?
thanks HUB, nice video
Can you try Gen 3 speeds on the B580/B570?
Surely Intel can see where upscaling is not doing much, or in some cases anything at all, from these graphs and start looking into fixes?
If i were intel, figuring out what is causing the overhead issue would be my number 1 priority. The b570 could one of the best bargains in a long time if they can just iron out the overhead.
"Competition is no fun without competition".
This was a great review!
I feel like I own one of the only B580s in the wild. Somehow snagged one of the Intel Limited Edition cards. So far on a 13600K it's fine, I anticipate that a B570 would be fine for anyone with a 12th gen Intel CPU and up.
Thank you, Steve!
Good content. Thank you!
When are the B850 motherboard benchmarks coming out?
I paid 319EUR for a B580, it's now powering my nas / media server, handling any transcoding job better and faster than anything I had before. Running decently fast Mistral 7B local LLM is also nice.
Even if it was a bit more expensive than the 4060, was still worth for better out of the box open source support, better transcoding quality and speed.
New format for Intel Arc Gpu's makes sense
Thank you for the 5600 testing! I am the owner of one lol. Yes I can upgrade to 5700x3d but Im scared to tamper with that upgrade, not confident enough with it yet (and yes I built the system myself but that I found easier than taking off the cooler, unsocketing, and socketing in the new CPU, not comfortable with that).
So basically: If all GPUs are at MSRP the Arc cards are still the best value (cost per frame) off all the cards not matter the CPU at 1440p.
I hope they can somewhat fix the issues with driver updates, but it's still a decent card for the money (if you can get one for the advertised price)
The real AMD show will be on January 22?
The new format slaps hard
Just stopped in to see if Steve was sitting.
4:44 Are my eyes broken? Is the B570 slower on the 9800X3D than the 5600 in that test? WTH?
It's just 3-4 fps difference. However upscaled performance is still better with the 9800X3D
Rx 7600 is the joke tbh .. not powerful enough,isn't power efficient, has 8gb ram, bad RT, bad upscaling & costs around same as 4060.
Overall all 4 gpus are similar performance, won't notice any difference from going one to another
Damn cpus tests for gpus should totally be a thing in the future... I really thought cpus had a big impact... apparently not
The vast majority of gamers who have slower CPU's would still see an uplift in performance considering most don't have a GPU that can handle 1440p gaming at all.
That's great that by Steve's idea nobody buying this GPUs for new build but only upgrading old outdated computers.
Nobody will ever think of buying 150USD CPU like 13400F and 250USD GPU.
looking forward to non-meh reviews in future. way way forward, probably years
Thank you for your service.
16:03 "(...) And really moving forward 12GB should be the minimum for GPU's priced at or above 200$."
12GB for the upcoming 5070 is a joke.
It will be cool to see tests with 5700x3d and 5800x3d
I am a budget gamer with an old RX 470, and I upgraded my motherboard/CPU/RAM a year or two back to Alder Lake because GPUs were too expensive. I'm thinking there could be others out there like me who didn't want to overpay for a GPU so they put that money into the rest of their system instead. (My system upgrade felt like a bargain compared to any GPU I might want.)
In terms of performance per dollar, the B570 certainly doesn’t make a lot of sense, but depending on the usage case scenario, 30 dollars is 30 dollars. For a workload/gaming mixed-use scenario where that 30 dollars might be better spent on a better CPU, the 570 makes a lot of sense.
what a surprise aussie retailers gouging. 250 +60%+10% does not= 469 for a b580
Works out $442 so its $27 more.
@ yes my calculations are off a bit
now corrected😉
Your calcs weren't that bad anyway it's still more. Anyway it's not retailers gouging, they pay more for stock. Margins in this industry are below 10% most of the time, especially for internal components.
You should check it with a budget intel cpu like 12400f
performs like a 5600
Wow, I never realized the Arc A series had memory bandwiths of 16 to 17,5 terabits per second! (01:33)
This could be still decent for somebody who mostly does productive work (programming etc.) and wants to do occasional gaming. But that is of course provided it is available at a decent price.
B580 is 450$ and B570 is 350$ right now in my region. Not competitive at all at these prices. But i guess there's a few of them in stock
Quality upscaling is upscaling from the same resolution for DLSS and FSR, but it’s not the same base resolution for XeSS
Thanks for the comparison. Some things i would've liked to see tested is these battlemage cards with the top of the line AM4 CPU, the R7 5800X3D (or the 5700X3D) to see if the cpu overhead problem persists. Also I would've liked if you included the settings used for the upscalling benchmarks, i hope you didn't use XeSS at quality mode since it has a better picture quality compared to DLSS and FSR at the cost of lower framerates, if you use FSR at quality then you must use XeSS at balanced mode to achieve similar picture quality.
If intel somehow gets the cpu overhead under control. I think people would pretty much recommend these gpus left and right.
AMD should be able to snatch the performance crown in the budget segment in its next generation
Based on what exactly?
@ArchieBunker11 Hopes, mostly at this point. But more seriously, the RX 7000 generation was very weird at the entry level. A RX 7600 barely hung there with the previous gen RX 6600XT and there was a big gap to the RX 7700XT, which was around RX 6800 performance.
The RX 6700XT had no real equivalent in the RX 7000 series.
AMD talked specifically about that gap a couple of weeks ago and how the RDNA 3 was not flexible to target that level of performance.
I think they will aim exactly at a replacement for the RX 6700XT with the entry level RDNA 4.
And the RX 6700XT-like performance and 12GB VRAM would be a preferable option than the B580, IMO
@@konstantinlozev2272 Why do people look back on the 6700xt with such rose tinted glasses? The 6700Xt had an msrp of $479. $80 more than the 3060Ti, and $20 less than the 3070. And according to TPU’s 2025 GPU test suite with modern games its lost ground to the 3060ti, in spite of VRAM. The 7700xt was a low performance increase over the 6000, but its msrp was slightly better.
If they want to repeat that, they’ll get beat down by nvidia again. If that model is going to work, you’re right. It would have to be cheap
Here in the Philippines, the b580 is priced 300usd and stores are all out of stock, much to no one's surprise. The 4060 here is priced the same or just 20-30usd more
here in the uk, the 4060 is probably the best buy at the moment, 4060 is £259 ($315) b570 is £250 ($305) and b580(preorder) is £290 ($350)
I wouldn't buy anything with 8GB, not for $315.
Wheres 1080p ?
The Battlemage cards hardly gains any performance at lower resolutions. For 1080p you want AMD or Nvidia.
Shame
Just look at 1440p upscaled numbers, they're in few % of 1080p as that was base resolution before upscaling.
@@gorky_vkyour right
Gotta love this card is probably gonna cost 2/3 or even 1/2 of what the 5060 will, and it still has more VRAM.
Shaking my head at the quad charts. That overhead issue is really making GPU testing a pain for you. I hope Intel has all hands on deck to improve this. I mean a CPU load issue, implies this is a driver problem that is fixable.
Since you're considering this as part of the value analysis, 10 gb is ~17% less vram than 12 gb (12 gb would be 20% more than 10)
At least it shows the demand and that new 8GB GPUs are DOA regardless of which company releases them this year. Above £150 for 8GB is not on anymore
I have a 7600X3D and have been trying to buy the b580 watched it go out of stock while I was putting in my info.
Looking at the lower cpu stuff from several other videos including yours it seams it just comes down to if a CPU bottle neck happens it is more prominent. Which, to me, means content creators need to discuss how important a balanced system is and how you in general want to be GPU bound. So if you are on those older CPUs it might be time to upgrade your CPU instead of your GPU. Or even both
It is 300$ in my country. And that is the cheapest version of B570...
Off course Steve is greifing with Winter Soldier. 😂