Intel Arc G3 CPU family officially released for handheld gaming PCs

[-]

IsometricRain@reddit

Looks great, hope this ends up cheaper than the 358H laptops we've seen so far.

[-]

I wonder about that 2+8+4 die. Why create a custom die for this line when you could have slapped the 12Xe GPU onto the 4+0+4 die that was already lying around, essentially creating a beefed up Lunar Lake (which was already a good handheld chip)? Or might these be binned 4+8+4 dies? Seems unlikely.

Regardless, I’m very curious, it should be a big leap over the Z2E and really bring the fight to AMD in this niche space.

[-]

IsometricRain@reddit

2+8+4 will be a lot better for productivity workloads, and anything multi-core.

Sure, lots of people will buy these just for gaming, but in today's market, it'd be nice to have just one machine for both gaming and, say, software development instead of having to buy a separate >$1000 laptop in addition to a likely very expensive handheld.

An 8 core will be noticeably slower than a 14-core when compiling/building any native app, or even a slightly more complex web-app.

[-]

Verite_Rendition@reddit

I wonder about that 2+8+4 die. Why create a custom die for this line

Are we sure it's even a new die? The TPU article doesn't mention so.

A new die has a relatively high development cost. Whereas Intel can just fuse off CPU cores here from their high volume 16 core die, presumably as part of their binning and salvaging process.

[-]

HavocInferno@reddit

Problem at low power budget is probably the P core draw. So they cut those to give the iGPU more budget.

With the IPC of E cores these days, I wonder if a 0+8+x config would eek out even more GPU power. It's not like a handheld is likely to be doing much high refresh in demanding titles, so you wouldn't necessarily need the best CPU ST.

[-]

Affectionate-Memory4@reddit

With the 8-30W TDP range, I'll be curious to see how good the performance scaling is in that range. Lunar Lake handled very well at very low TDPs.

[-]

div033@reddit

Panther Lake scales down very elegantly. Dell's XPS 14 is capped at 25W and only loses 10-15% performance down from 45W. These will be the kings of handhelds, but they'll be prohibitively expensive.

[-]

RandomGenericDude@reddit

Counterpoint: doubling the TDP for 10-20% gain means the design isn't targeted for that envelope. That's worth a different design...

[-]

div033@reddit

Yeah, but that didn't stop certain tech reviewers for calling out what a "travesty" it was for Dell to cap the power like that, instead only recommending 45W models that you couldn't actually buy yet and didn't release until well after everything dramatically went up in price. Just glad I was able to see it for what it was and save $1000, despite Dell's awful quality control.

To be clear, I'd greatly prefer having more control over the noise/thermals/power, but I think I like $1000 more.

[-]

HavocInferno@reddit

Iirc Core 7/9 Panther scales badly below 20W, specifically regarding iGPU performance. The P cores are eating too much budget at low power target.

Which is why these G3 sacrifice 2 P cores and some boost clocks.

[-]

StunningPush8421@reddit

Yeah intel dies have been pretty small like lunar lake and pantherlake die sizes are smaller than some mobile chips so it makes senses that they can scale down so well.

[-]

AreYouAWiiizard@reddit

80w turbo?? Seems a bit high for handheld.

[-]

frostygrin@reddit

If you use a framerate limiter, as you should, the bursts will only minimize dips in framerate, e.g. from heavy effects.

[-]

AreYouAWiiizard@reddit

Sure but having to set that up every game would be tedious, especially on handhelds where not every game will be able to hit consistent 60 and some games you might want to run quite a bit higher. Also, limiting framerates to something under 60 will make it a less smooth experience in easier to run areas.

[-]

Strazdas1@reddit

You can just use global settings to do it for every game, then modify it for specific games if you want.

[-]

frostygrin@reddit

Doing it once per game that you'll be playing for hours isn't tedious. And you need to do it on a handheld anyway, to limit power consumption. Even regular GPUs require higher voltage at higher clocks, so capping them just a little lower can result in significant increase in battery life.

[-]

Poopincheese@reddit

I’m thinking this isn’t going to be the jump we’re hoping for. 2P cores/8 e cores. And 4 LP ones. It has cut down ram speed from ones already out 8k to 7467mhz. But intel should have better ram usage. Will be interested to see the comparison from 890m/8060s/arc B3 when it releases.

[-]

Danthemanz@reddit

6 months too late. Now no one will be able to afford to buy one for 2 years.... Hopefully they can get a product out the door and get drivers and things sorted now. By the time we get our hands on the next gen maybe components will be a reasonable price again.

[-]

Johnny_Oro@reddit

If $1000 steam deck sold out, there's probably still a market for it.

[-]

sunkenwhalebones@reddit

I don't think so. The Deck is the only handheld with any sort of hype behind it. Most people don't even know other PC handhelds exist. There's plenty of much stronger handhelds (Claw, Legion, Ally) that are still collecting dust on the shelves, often cheaper than the Steam Deck.

[-]

HavocInferno@reddit

What's weird is the 1k Deck sold out, but those same people were not interested in the variety of AMD Z1E/Z2E handhelds that were on the market for 1k previously.

[-]

LonelyNixon@reddit

I imagine it's because before everything else was cheaper and you could get a Steam Deck for less than $500 .Sure, the Steam Deck did sell out, but it was still listed at its original price on the webpage. So I imagine there was a large cohort of people holding out hope.

Now, as for why there are people who still bought the Steam Deck when there are other handhelds that are more powerful, that are still holding the line at the more unfortunately realistic $1,000 price tag, that, who knows, I imagine a good chunk of them were bots and scalpers, but I imagine some people are just buying for the ubiquity and the default Steam OS experience.

[-]

BighatNucase@reddit

It's people who were interested in it, see the price hike and go "man I better just pull the trigger now before it rises again". I think there are a myriad reasons to get a Deck over one of the windows handhelds, price was the big one but not the only one.

[-]

Gamiac@reddit

It's the default handheld now. Everything supports it, it was cheap, and it's easy to use.

[-]

animeman59@reddit

Nope. Those who wanted a Steam Deck, but couldn't get one officially before are the ones who bought the new stock. Instead of paying the same amount of money to online scalpers.

And that's it. No one else wants a handheld at this price. Unless the Intel handheld has a massive performance jump at the same price, i don't see any handheld (new or old) being sold.

[-]

imaginary_num6er@reddit

Maybe this can encourage AMD to implement FSR4 to those RDNA 3.75 APUs

[-]

Fritzkier@reddit

what do you mean encourage? isn't it's already announced for RDNA 3 and 2 just early this month?

[-]

jigsaw1024@reddit

Considering that FSR 4 has been out for the better part of a year, and they are just now announcing release for older products, is not promising. I can totally see AMD half-assing it just to point a finger and tell the community that this is why they didn't want to release on older products.

It also seems to have taken a lot of community push back, as initial reports had that FSR 4 would initially not be coming to older products.

Also given the release schedule (2026 for RDNA 3, early '27 for RDNA 2!), a lot of those products are going to start aging out before FSR 4 even gets a chance to release.

[-]

Seanspeed@reddit

None of this had anything to do with 'community pushback' or anything like that. They were always working on this.

[-]

SuperNanoCat@reddit

They specifically said RX 7000 series, not RDNA3. Presumably, it'll include the APUs, but who knows with AMD.

[-]

Seanspeed@reddit

Well we know that something like the 'RDNA4' in the new Exynos 2600 doesn't have any AI acceleration, so that's probably why AMD was careful with their wording.

It's pretty obvious that if there's enough AI performance, they'll support it. AMD doesn't need 'encouraging', this is about hardware limitations.

[-]