Intel Arc B580 GPU Re-Review: Old PC vs New PC Test

[Shirley Dulcey]

I think you need to go back and do yet another round of testing, this time with a couple of Intel CPUs. It's possible that part of the problem is that Intel's drivers are poorly optimized for Ryzen, which wouldn't be a shock since they probably used their own CPUs for development.

 

[Lew Zealand]

I think you need to go back and do yet another round of testing, this time with a couple of Intel CPUs. It's possible that part of the problem is that Intel's drivers are poorly optimized for Ryzen, which wouldn't be a shock since they probably used their own CPUs for development.

Done by a number of YTers already. Spoiler:

R5 5600 and Core i5 12400f performed similarly with the B580.

 

[manowar51]

I saw the black screen bug with my 5600 XT for a short time after I got it. And noticed it correlated with having Afterburner running. Turn off AB and zero black screen, easy fix to a problem caused by unnecessary 3rd party software.

Considering the number of AMD GPUs I've owned in the past 5 years (7 different ones in different machines in regular use), I've had zero corruption and no Windows reinstalls. My Nvidia cards have been similarly trouble-free, so they seem the same to me.

So far the B580 has been fine but this is still early days. One game gives a small amount of the CPU-overhead slowdown (Horizon Zero Dawn) and we'll see what happens in the future with other games.

I did not fix my black screen bug. But at the time I didnt know it was a radeon problem on my 5700XT. I replaced display cables and even my monitor and then I happened upon a video on youtube where Gamers Nexus was demanding answers for the "black screen bug" from AMD at a trade show somewhere. Then it all clicked for me, further research online at the time found that I was far from alone with the issue. It's bizarre, there's endless reddit threads and support pages about it and even amateur youtube videos with hundreds of thousands of views but the mainstream youtubers tended to avoid the subject. Had I known about it, I would not have purchased that card. I ended up replacing the card just 18 months into owning it, whilst its performance was technically still fine.

But this is just the tip of the iceberg for me, there have been so many other problems I have had with Radeon over the years (like the dam flickering issue that chased me across multiple Radeons but did eventually get fixed). For me, this is all usually justified in the price. Radeons are certainly cheaper than Geforce cards per frame. But these days where I have a bit more money and quite a lot less time, I can't really go for Radeon anymore.

Doing some research now, this video claims to have a fix. Its filled with positive comments and has a significantly elevated amount of views for a smaller youtuber. I cant verify it myself as I have better things to do today then crack out an old GPU and try and get it working again:

I don't see these sorts of things with Geforce myself. Although im not trying to say they are completely without issues. My personal experience has been far better with them. To the point where I would say they are bit boring lol.

Sorry to go into an essay about this but it literally cost me so much time and money. It's what made me swear off Radeon for good or rather at least for a few years. This comes after a personal history of predominantly AMD (and ATI) cards. In fact my first ever GPU that I purchased for myself was an X800XT back in 2005.

 

[BadThad]

Wow Steven! That's a serious amount of raw data. Interesting test, thanks.

 

[BadThad]

I think you need to go back and do yet another round of testing, this time with a couple of Intel CPUs. It's possible that part of the problem is that Intel's drivers are poorly optimized for Ryzen, which wouldn't be a shock since they probably used their own CPUs for development.

You missed the whole context...it was for folks with AMD CPUs considering an upgrade.

 

[hwertz]

I'd love to see a review (maybe not as comprehensive, but a few games and cards) in Linux.

For the simple reason that, the Nvidia (binary) drivers are very similar between Windows and Linux (your DX9/10/11 games are being converted to Vulkan by dxvk, and the DX12 ones by vkd3d... but games that already use Vulkan will be using many of the same code paths in both.)

But for both AMD and Intel GPUs, the Mesa Gallium 3D drivers were essentially developed independently of their respective Windows drivers. So with some Vulkan games, you'd have side-by-side comparison just testing the drivers themselves, since in both Linux and Windows they have native Vulkan support. My understanding is the Arc Windows drivers are DX12 and Vulkan only, and DX9/10/11 games are run under a graphics translation setup, as they are in Linux. DX12 games would be native in Windows and translated to Vulkan in Linux.

This might give some interesting data points; my understanding is the Windows Alchemist/Battlemage drivers are "from scratch", while in contrast Linux has a new kernel-mode driver for Xe, Alchemist, and Battlemage (I guess Xe and Alchemist can also work with the previous kernel driver); while the user-mode stuff that actually implements OpenGL, Vulkan, OpenCL, etc. on top of that is the same surprisingly stable and performant intel 'iris' driver that supports about 10+ years of Intel GPUs, with of course support added for the new functionality of Xe/Alchemist/Battlemage. You'd of course want to update 'distro of choice' to most recent mesa drivers since this is new hardware. But testing a few games and cards would add some interesting data points.

 

[mat9v]

This test makes me wonder how much the results are impacted by only 6 cores being available on 5600 and would the test be better if it was conducted on 8 core vs 8 core. I know it is a side issue as we are talking about "insanely overpriced" vs "fairly cheap" cpus, still ryzen 5800X is not so much pricier than 5600 and if the difference in performance would be much smaller?....

 

[ewitte12]

I paired a b580 with a 12600k and got pretty good numbers at 4k but still added 10-15% upgrading to a 14700k. Actually got like top 1% in timespy for that card lol.

 

[Watzupken]

While this is very CPU dependent, I still think it’s very good in value if one can get it at USD 250 MSRP. For people who are using this on older systems, there’s potential for the card to perform quite a lot better after upgrading the CPU at some point in time. You can buy a RTX 4060 which performs pretty much the same and lacking in VRAM for future game titles.