You mean basically not at all?

No, as in 20% performance loss. Which admittedly destroys the $$$/FPS metric, but card still functions properly, just worse than it should have. Kinda like buying an AIB card with inadequate cooling.

And no, not shilling for Intel here, just pointing out that hardware isn’t garbage instantly just because a new piece of tech comes out. The PS2 could still play all PS2 games perfectly even when the PS3 came out, after all.


Back in my day rebar was something we used to keep concrete together. :smile:


Please Remember to stay respectful in discourse and on topic


Most people in this situation will have that ancient CPU because they would have bought it from Ebay, and in that case they’re probably getting a used 480, 580, 1060 or 1660 anyway imo.

I really don’t understand this hatred of ReBAR than OP has. It’s something that all vendors support now and is giving you additional performance through more efficient use of the hardware you already have. That’s not what “ewaste” means.


I wonder if the CHIPs ACT helped them decide that they could take the hit

1 Like

Dude, this thing can pull 20 Terabezos!

Even if ARC does eventually get canned, I’m pretty happy that the tech will trickle into their CPUs. If they can integrate the low-mid tier ARC into the same silicon as their H series chips on mobile, they could easily steal back some market share from AMD. The only problem they’d be running up against is their terrible power management relative to Zen.

1 Like

Its definitely one of the factors that makes ARC either viable or non viable, but its so minimal in scale that they probably won’t be convinced.

Intel’s about to have a rough period

rebar isn’t that much of a problem. Intel probably sells the majority of cards via OEM/SI anyway. And having brand new cards requiring ~5 year old tech isn’t that special either.

And people buy Intel because it’s Intel. Most people didn’t catch up on Intel lagging behind. Brand recognition is strong and enduring.

I mean I am getting daily updates about A380’s hitting shelves in the US. Its not just OEM bro.

Can you outline your thinking to any of this, because what you are saying makes no sense. You keep making apocalyptic claims and talking about intel going back on even trying to compete in HPC – which is their actual bread and butter – all because the first generation entry tier early attempt has some driver problems. I cannot wrap my head around what could even lead to that conclusion, and you have not offered a satisfactory explanation.

It’s a shame nobody has bothered to boot up Linux with the Intel Mesa drivers with one of these cards yet (not sure if its supported in GIT yet but it should be).

Everyone focuses on the Windows drivers which clearly are 100% not ready, not even close! BUT the Intel Mesa drivers have been maturing for quite some time now, at least for their iGPU stuff, but that should translate over to the new discrete pretty easily!

Just a little correction here, AMD have NO built-in systems for StorMi, it’s a 100% software only solution that AMD provide a license for use on specific AMD hardware. With very little effort these limitations can be bypassed and it operates anywhere.

Except the documentation… this makes the FOSS’d code almost useless as it’s poking at registers that are given non-obvious obscure names. FOSS code doesn’t make the thing open. Also the PSP core and ATOMBIOS are fully closed and will never be open due to HDCP requirements.

Note the ATOMBIOS is a way to get around the GPL issue NVIDIA have, it’s a binary blob that is parsed and executed by a virtual CPU implemented in the amdgpu module in the kernel. It does the super secret magic initialisation of the GPU, preventing us from doing cool things like making SR-IOV available to the masses simply because of a lack of documentaiton on the registers it’s programming.

Exactly this! The intel GPU source and driver stack is far more open then AMDs, and when there are issues the Intel engineers actually will work with the public to get things resolved. AMD OTOH I had to complain about the reset bugs and get thousands of people to sign on, directly communicate with Lisa Su, and talk to some engineers in the background in their own personal time to just get a “hint” of how to work around this problem.

Still it took three GPU generations (Vega/Navi/BigNavi) before we saw AMD even acknowledge FOSS community on this devcie breaking issue.

Don’t be fooled into thinking that AMD are for FOSS, it just makes sense to get a native driver embedded into the Linux kernel from a sales point of view.

As for Arc… I am still hoping to get one, intel would dominate the workstation GPU market if they play things right. Most PCs don’t need a super uber fast gaming ready GPU for their intended use. While I love the 3090 I have in my PC for my Windows VM, I am still holding out hope for an Arc GPU for the Linux workstation side.

That combined with the Arc’s AV1 support and performance it’s a very exciting card to get a hold of, espesially for anyone interested in live streaming.


I’ve seen some news on 5.19 Kernel and 5.20 (aka Linux 6) and it looks really good for support from what I can see. Question is whether you will run these kernels or are on LTS Kernels. 5.19 is about to enter distro repositories in the coming days.

Id ask how these new cards are but shit doesnt even count unless is comes with bugs and hareware flaws. :frowning:

We need a new GPU player… many more GPU players.


There was some news about an interesting bug where they wouldn’t use the onboard memory, getting about 1% of the performance they should have on Linux since it all went through RAM.

I think the driver situation is more important than the performance, tbh. This needs to be fixed. Having said that, I am very curious as to how it performs on Linux where drivers are handled much differently.

As for the performance… eh. It’s a 1650/6400 competitor, which is more low end than I normally pay attention to outside of the novelty, but it seems to perform about where it should do for the price. I’m more itching for the likes of the A750 to be released, because that’s a performance tier I am actually more interested in.

1 Like

Put Steve in front of a Linux machine with NVidia GPU. He will pretty much say the same things. Cards haven’t been released yet and the public wants bad news, so everyone is jumping on Intel. It’s miles better than DG1 last year.
And I still need a cheap <75W GPU for a VM. Only option I had last year was a 1030 GT for 150€. With Intel, we have already more options <200€ and A380 will be pretty much what I’m looking for.

1 Like

When I say built in I mean it loosely. I think of the whole system as a product. Card, driver package, and additional software to get the performance you want out of the card.

Double correction.

The driver isue IS the performance issue.

At that, everyone keeps saying ‘bit you can run with out rebar, you don’t need rebar, blah’ except to have the card literally work at all you need rebar.

Sorry guys stupid is stupid.