Intel Arc A380

[waltc]

Good review--it's exactly what I thought it would be. Can only point out that Intel, too, has spent billions of dollars in R&D on architecture scaling--likely considerably more than AMD, for instance, over the past decade. I don't think anyone buying this card would be impressed with it or even keep it, for that matter... Why should they? So Intel threw a few billion $ and 5 + years at designing a GPU--but can only sell it in China for obvious reasons? Nobody here would buy it. Wake me when they have something they can ship into the West. Sorry to sound negative, but I haven't seen anything to get excited about yet, truthfully. Extrapolating future performance from an over-priced, exceptionally low performance GPU with tons of driver bugs that isn't good enough to ship into western markets doesn't seem very worthwhile to me...but that's just me. I remember the three i7xx GPUs I bought from Intel before they dropped out of the GPU businesses the first time (after buying Real3D IIRC.) Took 'em all back... If they get the driver bugs ironed out (big if, for Intel) then selling something like this for ~$100 might give them a spot in the entry-level GPU markets. But not yet...

 

[Imouto]

Kudos to Raja for not disappointing. The guy was promising a turd and delivered handsomely.

 

[W1zzard]

shouldn't the "average fps" and "relative performance" tables be identical, just with different numerical representations? 'Cause at least for 1080p the ordering is different, which seems strange to me.

Yeah it's a bit of an edge case here, I was wondering when someone would ask As you can see from the data, in some games the card with the WHQL drivers failed, average FPS uses "0 FPS" for the average calculation, relative performance estimates a value based on the results from the other results, because using "0" in that calculation won't work, because the other cards would all be relatively infinitely fast in that test

 

[AusWolf]

Quite honestly, Intel needs to subsidize first gen in order to create a large (beta) user base, iron out the issues, and go on to move to competitive successors. I don't really get the "buy it even if it's substandard to help competition" argument.

Intel is not some company struggling to keep afloat and keep developing before closing shop. Intel is not 3dfx before rampage.

It's not going bankrupt anytime soon, it's got billions of dollars, cash, at hand.

They want to succeed, they need to price accordingly. Current pricing on this card is laughable, while the card itself could be an enticing proposition at 110-120 dollars. That's what it should cost, at most, in my opinion.

Don't buy the GPU to help the competition. Buy it only if you want to experiment with something new.

All I'm saying is that we need people with these GPUs in their PCs to submit lots of bug reports so that Intel can deal with the issues either in driver updates or in future GPU generations. If everybody waits for something stable, there will be no future generations (maybe apart from Battlemage as it's already in development).

 

[Assimilator]

I don't understand why every review outfit, including the ones I trust to be objective like GN and TPU, is giving this GPU a pass. Because it's rubbish. Objectively rubbish. Intel/GUNNIR haven't even figured out fan hysteresis FFS!

Saying "RT performance is good compared to AMD" is pointless when we're talking about double or even single-digit RT framerates. And when the generation-old RTX 2060 is twice as fast.

Don't buy this rubbish. Don't pay Intel to be beta testers for them.

 

[PapaTaipei]

I don't understand why every review outfit, including the ones I trust to be objective like GN and TPU, is giving this GPU a pass. Because it's rubbish. Objectively rubbish. Intel/GUNNIR haven't even figured out fan hysteresis FFS!

Saying "RT performance is good compared to AMD" is pointless when we're talking about double or even single-digit RT framerates. And when the generation-old RTX 2060 is twice as fast.

Don't buy this rubbish. Don't pay Intel to be beta testers for them.

I don't think anyone is buy it, it's meant for prebuilt PC I guess. I don't know maybe there is a market for that in developing countries?

 

[InVasMani]

HTPC card and not much else.

 

[QuietBob]

Big thanks to @W1zzard for this thorough review. I especially appreciated the minimum settings tests, since the card clearly isn't meant to handle full detail. Regardless, the 6500XT currently starts at $150 pre-tax here in Europe, making it a much better offer.

BTW, p. 37 has the wrong card model in the temperature table

 

[GreiverBlade]

HTPC card and not much else.

yeah, if it was less power hungry and performing a bit better ... while i am a bit surprised about the A380 (and Intel Xth atempt at GPU, which is still to be seen if they fail, like previous atempt, or not) , the more i read the review in details the more i think, even at same price, a RX 6400 (preferably LP no additional power connector) would be better ...

Big thanks to @W1zzard for this thorough review. I especially appreciated the minimum settings tests, since the card clearly isn't meant to handle full detail. Regardless, the 6500XT currently starts at $150 pre-tax here in Europe, making it a much better offer.

BTW, p. 37 has the wrong card model in the temperature table

switzerland : cheapest RX would be the 6400 @ 170ish $ (still a slightly better offer than the A380 )

 

[Dirt Chip]

I don't understand why every review outfit, including the ones I trust to be objective like GN and TPU, is giving this GPU a pass. Because it's rubbish. Objectively rubbish. Intel/GUNNIR haven't even figured out fan hysteresis FFS!

Saying "RT performance is good compared to AMD" is pointless when we're talking about double or even single-digit RT framerates. And when the generation-old RTX 2060 is twice as fast.

Don't buy this rubbish. Don't pay Intel to be beta testers for them.

From a narrow consumer point of view I agree, I is rubbish you better of with.

But from a wider point of view, of a process at it very beginning, it is firstly very interesting and specifically, from this A380 review, much encouraging. Sure not a rubbish start.

 

[W1zzard]

BTW, p. 37 has the wrong card model in the temperature table

Fixed, thanks!

 

[noel_fs]

absolute garbage

 

[efikkan]

absolute garbage

The card isn't garbage, but it is terrible value.

I think every article/review about low-end cards needs a disclaimer that these cards have many valid uses beyond gaming, especially those wanting to add new capabilities to older machines which still have are working fine.

If Intel priced this card below $100 and didn't market it as a gaming card, the reception would be fine.

 

[Assimilator]

If Intel priced this card below $100 and didn't market it as a gaming card, the reception would be fine.

Except they didn't. What-ifs are about as useful to the GPU market as a new competitor that can't compete.

Ultimately Intel's greed is what has made Arc DOA. I disagree with Wizz that this card should be priced as low as $100 (although that would be a slam-dunk in terms of marketshare); realistically all Intel needed to do was price it lower than GTX 1630 i.e. $135. They simply didn't bother to do that because that would cause them to make a loss, and that's pure greed and pride and refusal to accept market reality.

It certainly doesn't help that Intel decided to overbuild an entry-level card, thus increasing cost, instead of going as simple and cheap as possible. It's like they have no concept of the fact that price is king at the low end of the market.

 

[Valantar]

Except they didn't. What-ifs are about as useful to the GPU market as a new competitor that can't compete.

Ultimately Intel's greed is what has made Arc DOA. I disagree with Wizz that this card should be priced as low as $100 (although that would be a slam-dunk in terms of marketshare); realistically all Intel needed to do was price it lower than GTX 1630 i.e. $135. They simply didn't bother to do that because that would cause them to make a loss, and that's pure greed and pride and refusal to accept market reality.

It certainly doesn't help that Intel decided to overbuild an entry-level card, thus increasing cost, instead of going as simple and cheap as possible. It's like they have no concept of the fact that price is king at the low end of the market.

It's worth noting that in one of the recent YouTube conversations with Ryan Shrout and another Intel GPU guy I can't remember(can't remember if it was one of their GN appearances or that thing they did on LTT), they highlighted that the US MSRP of this GPU would not be $150, but rather something unspecified but lower (and that Chinese pricing was affected by local/regional factors like tax ++). Still, even at $130 this would be rather dubious in terms of value IMO. What they said was that Arc pricing would be aggressively competitive and based on what they call "tier three" game performance, i.e. performance in poorly/unoptimized games running on older APIs, rather than well optimized (tier 2) or DX12/Vulkan (tier 1) games. I guess we'll see how that pans out though.

 

[Assimilator]

and that Chinese pricing was affected by local/regional factors like tax ++

I guess neither AMD nor NVIDIA GPUs get taxed, huh? What a ridiculous non-argument from Intel.

What they said was that Arc pricing would be aggressively competitive

Except it's not, and Intel says a lot of things that are complete bullshit.

 

[Dirt Chip]

Except they didn't. What-ifs are about as useful to the GPU market as a new competitor that can't compete.

Ultimately Intel's greed is what has made Arc DOA. I disagree with Wizz that this card should be priced as low as $100 (although that would be a slam-dunk in terms of marketshare); realistically all Intel needed to do was price it lower than GTX 1630 i.e. $135. They simply didn't bother to do that because that would cause them to make a loss, and that's pure greed and pride and refusal to accept market reality.

It certainly doesn't help that Intel decided to overbuild an entry-level card, thus increasing cost, instead of going as simple and cheap as possible. It's like they have no concept of the fact that price is king at the low end of the market.

A little exaggerating of you, to DOA a ARC on a basis of 1 review and about 3 sec after lunch...
Give it (or don`t..) a few month to mature before you decide. See how it evolve and than choose whether to buy it, recommend it or not.

Maybe dial down the harsh words, so much fustration in you for intel.
You should also consider yourself to do a 'reality check' becuse intel isn't your friend so thay can't "betray" you. Thay are here, by definition and without any disguise, to do as much mony as they can- thay must so. This is how you do business this days.
Using 'greed' ' pride' and so on look like a childish, affended, reaction (which I'm sure your not).

 

[KLiKzg]

Intel's 1st steps back to GPUs, as AI is on their scope.

 

[eidairaman1]

Failintel

Except they didn't. What-ifs are about as useful to the GPU market as a new competitor that can't compete.

Ultimately Intel's greed is what has made Arc DOA. I disagree with Wizz that this card should be priced as low as $100 (although that would be a slam-dunk in terms of marketshare); realistically all Intel needed to do was price it lower than GTX 1630 i.e. $135. They simply didn't bother to do that because that would cause them to make a loss, and that's pure greed and pride and refusal to accept market reality.

It certainly doesn't help that Intel decided to overbuild an entry-level card, thus increasing cost, instead of going as simple and cheap as possible. It's like they have no concept of the fact that price is king at the low end of the market.

This should of been a half height card sub $70, its nothing but a display adapter

 

[Valantar]

Except it's not, and Intel says a lot of things that are complete bullshit.

The product literally isn't launched outside of China yet, so your basis for that statement is ... guesswork?

 

[Assimilator]

The product literally isn't launched outside of China yet, so your basis for that statement is ... guesswork?

The claim: "Arc pricing will be aggressively competitive".
The reality: In 1 of 1 areas Arc has launched in, Arc pricing has been the exact opposite.
Primary conclusion: 100% of the sample size directly contradicts the claim.
Secondary conclusion: Intel's GPU division is feeding you bullshit, again, still.

Unless you want to claim that Intel said it would be price-competitive everywhere except China?

 

[Valantar]

Unless you want to claim that Intel said it would be price-competitive everywhere except China?

That is pretty much exactly what they said - they were explicitly differentiating US/EU pricing from Chinese pricing, and describing different policies and strategies for the different markets.

 

[AusWolf]

The claim: "Arc pricing will be aggressively competitive".
The reality: In 1 of 1 areas Arc has launched in, Arc pricing has been the exact opposite.
Primary conclusion: 100% of the sample size directly contradicts the claim.
Secondary conclusion: Intel's GPU division is feeding you bullshit, again, still.

Unless you want to claim that Intel said it would be price-competitive everywhere except China?

100% of a sample size of 1 isn't much, is it? Let's wait for the whole product line to be released worldwide before we jump to conclusions.

 

[efikkan]

It's worth noting that in one of the recent YouTube conversations with Ryan Shrout and another Intel GPU guy I can't remember <snip> What they said was that Arc pricing would be aggressively competitive and based on what they call "tier three" game performance, i.e. performance in poorly/unoptimized games running on older APIs, rather than well optimized (tier 2) or DX12/Vulkan (tier 1) games. I guess we'll see how that pans out though.

Yeah, those videos with Tom and Ryan were awkward, they struggled really hard to put the cards in a positive light. It was probably right before they claimed to have the most "design wins"!
I fear that claim of pricing competitively according to tiers leaves them with too much wiggle room; competitive according to which standard? Does "special features" play a role in valuation?
It kind of reminds me of pretty much every car brand out there who all apparently are "best in class"…

 

[Valantar]

Yeah, those videos with Tom and Ryan were awkward, they struggled really hard to put the cards in a positive light. It was probably right before they claimed to have the most "design wins"!
I fear that claim of pricing competitively according to tiers leaves them with too much wiggle room; competitive according to which standard? Does "special features" play a role in valuation?
It kind of reminds me of pretty much every car brand out there who all apparently are "best in class"…

I mostly agree, though at least that part of the claim was pretty clear to me: competitive pricing (i.e. similar to or lower than competitors) when comparing games that are poorly optimized for Arc, with well optimized titles seeing the card punch well above its price class (at least from the basis that the difference between "poorly optimized" and "well optimized" in GN's testing was something like a 50% increase in performance compared to competitors). Of course this still leaves a lot of room for working on drivers before setting the MSRP, and judging by this review that work is definitely having effects of various kinds. For now, it seems they're running the usual "launch as a beta test in China" shtick while stalling/waiting it out in other markets.