News Intel prepping at least five new Battlemage GPU models — new PCI IDs added to Linux 6.11 kernel

[Admin]

Intel has listed five new Battlemage PCI IDs that will be integrated into the next update cycle for the Linux 6.11 kernel. The lack of many IDs hints that Intel's next-gen GPU launch could have far fewer SKUs than its outgoing Arc A-series lineup.

Intel prepping at least five new Battlemage GPU models — new PCI IDs added to Linux 6.11 kernel : Read more

 

[Giroro]

Battlemage is going to be at least a year late by the time it launches. Intel is not in a position where they get to phone it in and release after, or even near their dug-in competition. They needed to beat Nvidia/AMD to market, by a wide margin. They completely blew that window. They won't ever win on overall performance, and having the better price/performance only matters if customers have a chance to consider the value of your cheaper option *before* they've bought the expensive thing from the brand they already know and trust.

I don't get the Impression Intel is willing to fight for market share. If Intel only aspires to remain in a distant last place, then why even release a new product?
Intel needed to move hard, and fast. They needed to make big waves. Their only choice to compete in discrete GPUs was to be disruptive. But launching a midrange products, possibly several months after the competition, will not be disruptive.
What it looks like we are going to get is the contented defeatism I would expect from a lazy executive like Raja Koduri, who just wants to be safe, sit back, and keep re-releasing the same old product over, and over (and over, and over, and over).... but I thought Intel rightfully got rid of him from their GPU team awhile ago.

 

[Deleted member 2731765]

Phoronix reports that these five PCI IDs are confirmed to be five Battlemage GPUs, further proving that Battlemage discrete gaming desktop cards are just around the corner.

Battlemage should launch either at the tail end of this year or at the start of 2025.

You are contradicting your own paragraphs/text here. These GPUs are definitely NOT around the corner for sure. Most likely we can expect them to launch early next year, 2025, as also mentioned by you.

We all know the next-gen GPUs have been delayed for sure. I don't even expect AMD to launch their next-gen RDNA4 cards this year. Nvidia might release Blackwell by end of this year, but this is just speculation for now.


Also the title and heading of the article is a bit misleading.

No, Intel is not prepping FIVE new GPU models. An ID entry doesn't necessarily imply that it belongs to a certain GPU-class model. Because some models can also share the same ID.

MOST importantly, just because a DRM pull-request has spotted these IDs, does not confirm that they are going to be used in any end product, or find their way inside an actual Battlemage GPU.

There have been few other ID entries before as well for BMG or Battlemage Xe2 architecture, but those were just preliminary entries (I will post the linux kernel patch links soon).

So IDs certainly don't directly correlate to the number/count of variants/models. We can expect several similar device IDs to be spotted as we approach the official launch date.

5 ID entries don't necessarily mean that there are 5 GPU models in the making. That's not the case.

Intel prepping at least five new Battlemage GPU models , five PCI IDs are confirmed to be five Battlemage GPUs

 

[Deleted member 2731765]

Before Arc Alchemist launched, we saw as many as 32 PCI IDs reserved for Intel's Arc A-series GPUs.

That's a wrong link embedded by mistake. It takes me to the same Phoronix page. This is the correct Tom's link:

https://www.tomshardware.com/news/intel-may-be-prepping-32-dg2-skus

 

[Eximo]

Gaming aside, Intel is also making some decent improvements in productivity applications. A770 hardware is quite suitable for some jobs and is cheaper then the competition. If that continues with Battlemage, and they manage to extract more of the hardware's performance then they managed with Alchemist, we could see decent sales as accelerators.

And they seem to be really into Linux support, which is nice to see.

I am probably going to get the top spec Battlemage card just for fun. If it is even close to the promised 4070 like performance, it will be an entertaining side-grade.

 

[thisisaname]

Just around the corner, is that the same corner that after which I will ascend to Godhood?

 

[Deleted member 2731765]

Just around the corner, is that the same corner that after which I will ascend to Godhood?

Yes sir, the blessing is just around the corner ! 👼


BTW, just wondering what happened to the "Arc Alchemist refresh" discrete GPU lineup which was rumored to launch this year itself ? No news or any update till now, so most likely the idea was dropped by Intel.

The rumored "refresh" lineup was expected to be referred as the 'Alchemist+' family, and was expected to include two new SKUs, the ACM+ G21 (successor to the current ACM-G10) and the ACM+ G20 (successor to the ACM G11).

 

[Eximo]

I imagine launching such cards at current prices wouldn't go over too well with investors. And people wouldn't be lining up to buy them at higher prices. Probably best to readjust with an all new lineup.

 

[DS426]

Battlemage is going to be at least a year late by the time it launches. Intel is not in a position where they get to phone it in and release after, or even near their dug-in competition. They needed to beat Nvidia/AMD to market, by a wide margin. They completely blew that window. They won't ever win on overall performance, and having the better price/performance only matters if customers have a chance to consider the value of your cheaper option *before* they've bought the expensive thing from the brand they already know and trust.

...

AMD and nVidia are starting out with mid-high end and high-end respectively, with mid and lower end (probably not nVidia once again) coming several months later. This means anyone looking at the lower end of the market will be waiting, which probably gives Intel enough time to finally release Battlemage.

 

[thestryker]

The only thing I'm really hoping for with Battlemage is that they've resolved the architectural issues that held Alchemist back. The A770 has the wildest performance swings I've ever seen which weren't related to missing features as it can go from slower than a RX 6600 to faster than a RTX 3070 and it's not just driver related either.

LNL will give us an idea, but until discrete GPUs are in the wild it won't be totally apparent.

Being late will hurt Battlemage unless Intel were to sell them at firesale pricing. They really need to be early and cheaper if market impact is the goal since most people just look at what's out when they want to buy.

 

[Eximo]

AMD and nVidia are starting out with mid-high end and high-end respectively, with mid and lower end (probably not nVidia once again) coming several months later. This means anyone looking at the lower end of the market will be waiting, which probably gives Intel enough time to finally release Battlemage.

Perhaps.

When Nvidia or AMD does a fall launch they tend to release the mid-range cards the following Summer. If the rumors of Battlemage launch are correct they would almost certainly be running right into middle of those launches.

 

[Deleted member 2731765]

The only thing I'm really hoping for with Battlemage is that they've resolved the architectural issues that held Alchemist back. The A770 has the wildest performance swings I've ever seen which weren't related to missing features as it can go from slower than a RX 6600 to faster than a RTX 3070 and it's not just driver related either.

LNL will give us an idea, but until discrete GPUs are in the wild it won't be totally apparent.

Being late will hurt Battlemage unless Intel were to sell them at firesale pricing. They really need to be early and cheaper if market impact is the goal since most people just look at what's out when they want to buy.

Couldn't agree more on this !

 

[Deleted member 2731765]

Instead of hunting for ID entries which provide zero hint/clue on any development status, it is better we focus on something concrete.

Intel's ARC Battlemage BMG-G31 discrete graphics processor codename spotted.

This happens to be the first real evidence that Battlemage G31 is in the pipeline, besides G21. Intel design tools webpage has the GPU listed under testing. BGA package with 3283 pins.

I expect this GPU to sport 32 Xe2 cores. Previous two entries spotted were, BMG-G10 and BMG-G21. It appears Intel has apparently cancelled the G10 variant.

BGA3283-BMG-G31 VRTT Interposer - Prototypes (Internal only).

https://designintools.intel.com/ptt...vrtt-interposer-prototypes-internal-only.html

 

[Deleted member 2731765]

There are already tools which were being used for verification and testing purposes for Battlemage discrete GPUs though.

Two test tools for Battlemage GPUs were listed on the Intel DesigninTools webpage. BGA2362-BMG-X2 and the BGA2727-BMG-X3-6CH (this link is currently not working).

The second chip is of a slightly bigger package size than the top Alchemist chip. The X2 tool features a 2362 BGA array, and the X3 tool feature a 2727 BGA array.

For reference:: ACM-G10: BGA 2660 (Alchemist).

 

[TechyIT223]

If the performance of Lunar lake version of Battlemage can be replicated on the desktop version, then AMD should be concerned.

It already destroys AMD's offering in performance and more importantly battery efficiency if used for handheld PCs.

I remember Intel's Tom Petersen saying that 30% of the Arc GPU team is working on Battlemage software while the hardware team has already fully switched to Celestial, during an interview to PCWorld back in January. For now they hope to have Battlemage out before CES 2025. We will probably learn more during Intel's next Vision event.

 

[Deleted member 2731765]

Intel shouldn't hurry just yet to start working on the 'next-next gen' CPU lineup, aka CELESTIAL, Xe3 arch. They should instead first focus on Battlemage architecture/lineup, and fully fine-tune and optimize the silicon as much as possible.

 

[TechyIT223]

I imagine launching such cards at current prices wouldn't go over too well with investors. And people wouldn't be lining up to buy them at higher prices. Probably best to readjust with an all new lineup.

They could have also discontinued the current Arc Alchemist GPU lineup, and made these new "refresh" cards available at the same price tag as the Arc Alchemist A series GPUs.

The Arc 770 and 750 cards have already seen great discounts these past few years after all. I guess the same PCB of Arc Alchemist could have also been used for the planned Arc Refresh GPUs imo.

Just my two cents...

 

[Deleted member 2731765]

I guess the same PCB of Arc Alchemist could have also been used for the planned Arc Refresh GPUs imo.

Unless some serious design and architectural layout changes were made for Alchemist refresh parts (which seems unlikely though), re-using the same existing PCB would not have been feasible enough.

 

[TechyIT223]

Yeah I know that. But mostly a refresh SKU will have some minor changes like faster clock speed benefits, faster memory speed or bandwidth , or maybe uses a refined process node as well, 😀, so in that case a PCB reuse might be somewhat feasible.

But I'm not an expert on all this so it's just my assumption.

 

[thestryker]

I think there were simply too many fundamental issues with the Alchemist design to justify actually launching a refresh. Timing wise they probably wanted to just to have something new on the market, but without resolving the core issues the performance benefit most likely wouldn't have justified the silicon cost. The Alchemist GPUs are quite big compared to the performance level they reside in. As bad as it is not having something new on the market releasing something new that doesn't really change anything is worse.

 

[TechyIT223]

I hope Intel has learned their lesson with Arc Alchemist GPU lineup. Mainly the architecture and all the early issues related to both hardware and software.

 

[Deleted member 2731765]

If Intel can release highly competitive mainstream/mid-range "gaming" GPUs within a reasonable budget, then they can grab a lot of market share, unlike Alchemist which was a bit late for the party.

Despite that, it's impressive to witness how much effort Intel's driver team has done in the past one year or so.

Current Alchemist A-series ARC discrete GPUs have seen a pretty decent performance uplift in most of the latest AAA/AA games, if not all. We just need more mainstream GPUs in a price bracket that a lot of gamers can afford, and not some highly expensive or a HALO product (thinks of RTX 5090 ) .

The Arc A770 which sports a 16GB VRAM buffer/capacity is still considered a pretty decent gaming GPU based on it's price/performance ratio.

I also look forward to an update on the XeSS up-scaling tech as well, with new iterations, along with the rumored XeSS "ExtraSS" tech, which could be a frame generation technique, aka based on "Frame Extrapolation" instead of "Frame Interpolation".

Based on Intel's paper, Extrapolation method uses information beyond the bounds of the input sample to produce an approximation of the frame.

Although, Extrapolation might produce less reliable results, and add more artifacts, but we have seen similar issues with interpolation as well, so with a few tweaks and optimizations, XeSS "ExtraSS" could be a middle ground in offering good quality with higher FPS.

So I'm curious to learn how this pans out, and see this implement in future games as well, assuming Intel is going for this tech, and it is feasible to use. Nothing is official yet though.

https://asia.siggraph.org/2023/presentation/?id=papers_744&sess=sess155

BTW, the research paper itself also highlights the differences between Interpolation and Extrapolation, which seems obvious.

While Frame Interpolation generates better results, it also introduces higher latency when generating frames which is why NVIDIA and AMD have latency-reducing technologies such as Reflex and Anti-Lag required to deliver a smooth frame-generation experience.

Extrapolation doesn't produce very high latency, but has difficulties due to lacking key information to produce a new frame.

But the paper claims that ExtraSS aims to solve this by using a new "warping" method that can help produce better quality vs the previous frame generation methods and with lower latencies.

More info can be found here::

https://dl.acm.org/doi/pdf/10.1145/3610548.3618224

 

[Deleted member 2731765]

I hope Intel has learned their lesson with Arc Alchemist GPU lineup. Mainly the architecture and all the early issues related to both hardware and software.

Yup, pretty sure they have learned from their past Arc Alchemist GPU lineup.

Is Battlemage going to be better?

"I absolutely expect so, it's a new technology, it's a new architecture, we fixed some known bugs, let's call it architecture bugs (things), that had we already had our first generation we wouldn't have done. Think about how much we have learned this generation and all of that has been pushed into Battlemage so you'll see better scaling on DX12, better ray tracing, things that you'd expect us to do as we move forward on a discrete cadence."

Tom Petersen - Intel Graphics Spokesperson (via PCWorld)

 

[TechyIT223]

Ah, it's TAP again with his bold claims. Hope he's right this time 👍😀