News Intel GPUs - waiting for B770

[DrMrLordX]

Just look at the MSI Claw why mobile Arc makes no sense right now. Perf/watt is not there.

Is that really due to the graphics IP/iGPU?

 

[mikk]

Is that really due to the graphics IP/iGPU?

A lot of the struggle came from a suboptimal Bios/software. MSI made big improvements with Bios 1.06: https://www.msi.com/news/detail/The...ents-Through-New-BIOS-and-GPU-Drivers--143544

Meanwhile Bios 1.08 further improves performance in combination with a new MSI Center update. We need a retest with Bios 1.08.

MSI Center M v1.0.2405.0603 Released

New version can be downloaded on product support page, or via Live Update in MSI Center M.

New Features:
• Adds “Over Boost” toggle in User Scenario, which optimizes BIOS settings for in-game performance for an average of 12% boost in frame rate.
o Requires pairing with BIOS 108 (or newer)

https://www.reddit.com/r/MSIClaw/comments/1cood82/update_msi_center_m_v1024050603_released

 

[Heartbreaker]

Good news. Author fixed his comment, only mobile discrete Battlemage is dead, desktop part still alive.

https://twitter.com/x/status/1788811592353906728

You really need a great part to succeed in laptops. AMD seems to have even tougher time selling laptop GPUs than they do desktop parts.

So Intel will have to spend it's time in the wilderness before it would make much sense to chase laptop wins.

Desktop remains the crucible to improve their GPU prowess.

 

[Tup3x]

It's a complete NVIDIA monopoly. Unless you have superior efficiency and performance, it would be just a waste of money to even try to compete there.

 

[igor_kavinski]

It's a complete NVIDIA monopoly. Unless you have superior efficiency and performance, it would be just a waste of money to even try to compete there.

Intel had a great chance with A770M 16GB. $999 laptop with 16GB VRAM GPU could've disrupted the gaming laptop market in terms of perf/$ but their drivers were in awful shape at the time. Even the Chinese company Machinist didn't put out laptops with that GPU, despite there being a pre-order page for it.

 

[Mopetar]

They'd probably be best of breaking into the data center and professional markets first. The margins there are high enough to support the costs of development. They'll still need GPU tech for all of their desktop/laptop CPUs and they can spend the time getting drivers ironed out there before starting to go after the consumer market for discreet cards.

 

[Aapje]

You really need a great part to succeed in laptops. AMD seems to have even tougher time selling laptop GPUs than they do desktop parts.

An issue is that you really need to knock it out of the park. If you produce a video card that you need to discount to move, you only sacrifice margins on the GPU parts. If you produce a laptop with a video card that people dislike, you need to discount the entire laptop, so you also sacrifice margins on all other laptop components, like the CPU, chassis, screen, motherboard, etc.

 

[DAPUNISHER]

It's a complete NVIDIA monopoly. Unless you have superior efficiency and performance, it would be just a waste of money to even try to compete there.

Few believed AMD would get back in the CPU game, but they did. Intel can do this. They just need a "good enough" product for many more gamers to adopt them. Nvidia has fostered a lot of ill will the last couple of years among PC gamers. Many are looking for a reason to dump them.

 

[Kepler_L2]

Few believed AMD would get back in the CPU game, but they did. Intel can do this. They just need a "good enough" product for many more gamers to adopt them. Nvidia has fostered a lot of ill will the last couple of years among PC gamers. Many are looking for a reason to dump them.

Most PC gamers just want AMD/Intel to be "competitive" so they can buy cheaper NVIDIA GPUs.

 

[DAPUNISHER]

Most PC gamers just want AMD/Intel to be "competitive" so they can buy cheaper NVIDIA GPUs.

That's what made this quote so funny -

Very bad news if true...
...Since I was waiting for Intel to drive some competition so that...
...I can buy any future AMD GPU cheaper of course!

I do think from reading YT comments, reddit, and forums, that many are ready to dump Nvidia if the right product shows up. Some will not buy AMD no matter what. Others are long time repeat Intel customers. There's an opportunity there. While I don't think there is any blue ocean for Intel to explore with GPUs, my not being able to envision it may just point to my lack of imagination and acumen.

 

[Tup3x]

Few believed AMD would get back in the CPU game, but they did. Intel can do this. They just need a "good enough" product for many more gamers to adopt them. Nvidia has fostered a lot of ill will the last couple of years among PC gamers. Many are looking for a reason to dump them.

Desktop is another matter entirely. With cheap(er) good enough product they might have some success.

 

[DavidC1]

You really need a great part to succeed in laptops. AMD seems to have even tougher time selling laptop GPUs than they do desktop parts.

It's unfortunate but not wholly unexpected. Desktop GPUs are bought based more on individual opinions, since an individual can go buy them. Laptops however are at the "mercy" of the manufacturer. So the collective decisions for laptops arise from choices of a smaller few. Also as a manufacturer you risk damaging your reputation.

Actually as an individual your risks are greater too. Contrary to an A770 going $250 sometimes, a laptop with it is 4-6x the price or some cases greater.

If Battlemage proves to be popular, you can bet laptop vendors to lap them up quickly. Most people are followers, and that is true for companies too. So whether losing mobile wins are temporary or permanent eventually depends solely on Intel.

 

[DeathReborn]

An issue is that you really need to knock it out of the park. If you produce a video card that you need to discount to move, you only sacrifice margins on the GPU parts. If you produce a laptop with a video card that people dislike, you need to discount the entire laptop, so you also sacrifice margins on all other laptop components, like the CPU, chassis, screen, motherboard, etc.

AMD GPU's in laptops are rare (outside of overpriced APU's) in the UK, Nvidia are everywhere and Arc might as well not exist. I was shopping in the £500-750 region and there was literally no AMD/Intel Discrete Options in that price range.

I would have loved a laptop with a A530M/A550M/A570M but you're looking at £1,000+ for even the A350M to shop up. If Intel is more efficient this time around they should move forward with laptop cards, even if they don't move the performance forward.

 

[DavidC1]

If Intel is more efficient this time around they should move forward with laptop cards, even if they don't move the performance forward.

If they are desired people would find a way to use them. Look at people repurposing gaming cards for AI!

 

[del42sa]

https://videocardz.com/newz/intels-...ject-potential-first-battlemage-gpu-to-launch

Intel’s BGM-G21 GPU added to LLVM project: potential first Battlemage GPU to launch​

 

[Hans Gruber]

If Intel can equal a 4070 with 16GB of vram and with 180ish watts of power consumption similar to the 4070. What are we looking at for price on the Battlemage GPU's? Will Battlemage be shooting for 5060/5060ti performance whatever that may be?

 

[igor_kavinski]

I think at this point Intel should just compete with themselves and deliver better than Alchemist performance/watt AND performance/$. If they approach 4070 (not SUPER) performance, I would consider that a bonus. Even something on par with 4060 Ti 16GB is acceptable to me, at this point in time. We must be patient, even though we might be feeding monster 2.0 after Nvidia. As soon as Intel is in a better shape, expect their card prices to go higher than AMD. Their professional cards already show a disturbing trend of being priced wayyy too high.

 

[Aapje]

Key is to have strong competition, so the companies keep each other in check.

But it's way to soon to speculate about Intel beating Nvidia. Let them prove they can make a profit on their GPUs first.

 

[adroc_thurston]

Intel-powered Aurora supercomputer fails to dethrone AMD-powered Frontier on Top500 list, again — claims spot as fastest AI supercomputer with HPL-MxP benchmark instead

Aurora faces stability issues from hardware failures, cooling malfunctions, and operational errors.

www.tomshardware.com

certified uh oh stinky.
not even submitted to Green500 jeezas.

 

[adroc_thurston]

Intel-powered Aurora supercomputer fails to dethrone AMD-powered Frontier on Top500 list, again — claims spot as fastest AI supercomputer with HPL-MxP benchmark instead

Aurora faces stability issues from hardware failures, cooling malfunctions, and operational errors.

www.tomshardware.com

certified uh oh stinky.
not even submitted to Green500 jeezas.

banger.

 

[gdansk]

That may be bad performance but worse: how many billion did they lose on that?

 

[adroc_thurston]

That may be bad performance but worse: how many billion did they lose on that?

not much, it was like a 600m contract in total.
Either way Aurora is half the efficiency of Frontier in HPL. Just wild.

 

[gdansk]

not much, it was like a 600m contract in total.
Either way Aurora is half the efficiency of Frontier in HPL. Just wild.

I read they had to agree to a 2 year no-cost maintenance contract and took a large loss on the contract, overall. But maybe it was just hearsay.

 

[adroc_thurston]

I read they had to agree to a 2 year no-cost maintenance contract and took a large loss on the contract, overall. But maybe it was just hearsay.

Yeah there are some funny clauses to Aurora delivery, but again, not a major loss for Intel.
The system itself being a stinky is, though.

 

[DavidC1]

https://videocardz.com/newz/intels-...ject-potential-first-battlemage-gpu-to-launch

Intel’s BGM-G21 GPU added to LLVM project: potential first Battlemage GPU to launch​

The original specs for G21 called for 40 Xe cores, meaning 25% more than A770. That is very underwhelming in late 2025. If Intel thinks that's going to be the highest SKU in order to save some $100-200 million, they better rethink their plans or cancel Battlemage altogether because it'll need to be sold at $200 again. If they worked on a 56/64 Xe core G10 version, get that out as the leading product.

It is by having a relatively high-end product that allows you to raise margins and have pricing flexibility. Continuing on what is nowadays a low-mid end product is a Business 101 to bankruptcy.