News Intel GPUs - we've given up on B770, where's Celestial already

[moinmoin]

Eww, soft touch matte finish? Is that still the kind that melts and becomes sticky over time? I hate that stuff, affected other plastic in storage as well. Or have they found another way to achieve that in the past decade?

 

[clemsyn]

In your case I wouldn't be surprised if that's the problem...

Mikk is having the same issue as well.

 

[DAPUNISHER]

Eww, soft touch matte finish? Is that still the kind that melts and becomes sticky over time? I hate that stuff, affected other plastic in storage as well. Or have they found another way to achieve that in the past decade?

That's a really good question. I should think new composites are being used. It would be funny but not haha funny if it was that old stuff. Especially if Intel bails on the whole thing in a couple of years, and your collectors item looks like old trick bubblegum.

 

[Insert_Nickname]

Eww, soft touch matte finish? Is that still the kind that melts and becomes sticky over time? I hate that stuff, affected other plastic in storage as well. Or have they found another way to achieve that in the past decade?

I managed to save a few things using a soft cloth, a bit of 93% alcohol and some standard issue vinyl cleaner. No guarantees though. I think the effect depends on the specific compound(s) used.

Just a useful tip.

 

[moinmoin]

I managed to save a few things using a soft cloth, a bit of 93% alcohol and some standard issue vinyl cleaner. No guarantees though. I think the effect depends on the specific compound(s) used.

Just a useful tip.

I actually know how to correctly handle it now (a LOT of stuff I had was affected, took time to sort out, clean and/or throw away all of it). The correct widely recommended treatment to prevent that from happening is regularly massaging it with car cockpit/dashboard cleaning fluid keeping the stupid useless "soft touch" chemical plastic mess well soft. And after the fact you can get it off like old extremely sticky crude oil and hope that whatever is behind that (and was lying around that) didn't dissolve as well.

But I really hate that stuff with every fiber of my body now. Old tech put in storage for any duration is not intended to cause any unwanted chemical reactions there. And that "soft touch" expletive ensures that you'll be regularly touching that expletive for the duration you own it, whether you like it or not.

 

[Insert_Nickname]

The correct widely recommended treatment to prevent that from happening is regularly massaging it with car cockpit/dashboard cleaning fluid keeping the stupid useless "soft touch" chemical plastic mess well soft. And after the fact you can get it off like old extremely sticky crude oil and hope that whatever is behind that (and was lying around that) didn't dissolve as well.

You are correct of course. It's often a right mess to deal with, and it doesn't always do the trick.

That is what is called vinyl cleaner over here. We're talking about the same stuff.

 

[moonbogg]

My opinion has changed about the Intel GPUs. I like them. They should persist and not cancel their GPU program. They look nice, perform good, and are priced fairly. I hope they persist as a third player because the market needs it.

 

[igor_kavinski]

My opinion has changed about the Intel GPUs.

Are you basing your change of opinion on A380 reviews? Bit premature, no?

My thoughts:

if A770 is really the price/performance killer Intel is promising it to be, there might not be enough of them to go around due to being limited editions and it may leave a sour taste for people who really want one.

If it has problems in the games that are not on Intel's list of 100 or so "optimized" games, people will be angry.

I just don't see Intel walking away with a really positive consumer response. But good luck to them!

 

[NTMBK]

Are you basing your change of opinion on A380 reviews? Bit premature, no?

Yeah, I'd wait for the Anandtech review before making up my mind.

 

[coercitiv]

Yeah, I'd wait for the Anandtech review before making up my mind.

but by then the soft touch matte plastic will be all mushy...

 

[majord]

Realistically, looking at what they've put forward , it's an RTX2070- 2070S . Right down to the 200+w TDP (215) , die size balllpark and power requirements, bus width, the lot. It Even looks like an FE!

If it was joining the battle between 5700XT and 2070/S back in mid 2019 , undercutting the $399 price tag the XT launched at (and having RT to boot), it'd be very disruptive. and a real worry for AMD in particular , but that's a scenario near 3.5 yrs old .

I think the hype around its release now for $329-$349 is more indicative of what's happened to the industry since then. The fact this makes it good value next to a 3060 really shows what poor value that card still is, and that it still hasn't fallen to its MSRP (Unlike AMD cars that have falled below launch MSRP ) .

As for the future , well I guess it depends on how they can execute with Battlemage. they'll need to make exponentially large gen on gen jumps, ( A 'Navi moment' or 2) and / or for AMD/Nvidia to naively take their foot off the gas pedal (Which does happen, so don't rule it out) There's now, realistically 2 Generations to catch up on before they can either tackle the high end, or be profitable in the lower/mid end.

At these performance levels Alchemist has probably, what 6 months? before it's a low end card, battling the "4050" and "7500" class GPUs , and if These competitors pull their head in, and are able to price them as low end cards they are then I don't think pricing such large, expensive to build GPU's to compete with them will be sustainable.

Until Reviews of this , and Ada/RDNA3 are out , we won't have an entirely clear picture what they're up against. Lots to digest in the coming ~ month or so

 

[beginner99]

they'll need to make exponentially large gen on gen jumps, ( A 'Navi moment' or 2) and / or for AMD/Nvidia to naively take their foot off the gas pedal

I don't think Intel needs to compete at the top. In my opinion they would be better off with targeting the mass market and efficiency (laptops) as let's not forget the dGPUs would mainly focus to increase volume to make the factories worth the costs. Therefore their focus should be on stability, compatibility (with old hardware and old games!!!) and efficiency.

For me Alchemist is fail due to the rebar requirement excluding older builds, driver stability and compatibility. performance/$ seems to be okish but given the potential issues with stability (dropped in bios) and risk of issues with older games make me choose a 6000 or 3000 series product for a higher price any day.

First fix the basics and then try the halo product.

 

[LightningZ71]

Alchemist, especially the upper end ones, brings something important to the market: cheap VRAM. From the A580, at around $250 or less, on up, you get at least 8GB VRAM, and 16GB VRAM can be had for around $350. That amount of vram can bring a lot of longevity for those cards. We talk about these cards competing with the 1650/5500XT/3050, but, they are all 4GB cards, some with restricted PCIe bus sizes. To get 8GB cards from anyone else, you need the 6600 at around $250 which doesn't run away from any of Intel's 8GB cards but may be consistently slightly better than the A580, or the 3060 12GB or the 3060ti which start in the $355+ range.

What you are suggesting Intel do is the same thing that almost sank AMDs video card business and put them firmly behind Nvidia for years. In the era that AMD could only push out the Rx480/580/590 on their top end, Nvidia was well ahead in performance and raking in sales on competing cards just due to mind share. Yes, Intel needs to get the fundamentals right, and yes, they absolutely need to prove that they can both produce a quality video driver/hardware combination and also NOT abandon it in about a year or less from introduction like they have with other graphics products in the past.

I don't particularly like Intel from their past actions. I don't particularly trust Intel from their past actions. We do, however, NEED them in this market to put downward pressure on prices in the mainstream. For that, I am cheering them on.

 

[DAPUNISHER]

What you are suggesting Intel do is the same thing that almost sank AMDs video card business and put them firmly behind Nvidia for years. In the era that AMD could only push out the Rx480/580/590 on their top end, Nvidia was well ahead in performance and raking in sales on competing cards just due to mind share.

Polaris was selling to miners as fast as they could produce them. Crypto has been a huge boost to AMD, who knows where their graphics division would be without it.

I think we all want a 3rd player in the market. But there will be no pity points for Intel either. Every day they slip further behind because the cards are not available, and become less relevant as a consequence. They'd better get it going soon, or the metaphorical boat will have so much water in it, they will have to abandon ship.

 

[blckgrffn]

Polaris was selling to miners as fast as they could produce them. Crypto has been a huge boost to AMD, who knows where their graphics division would be without it.

I think we all want a 3rd player in the market. But there will be no pity points for Intel either. Every day they slip further behind because the cards are not available, and become less relevant as a consequence. They'd better get it going soon, or the metaphorical boat will have so much water in it, they will have to abandon ship.

Yes, exactly. I also feel like if they thought they could get another year with the products under wraps while they work on the drivers and iterate, iterate, iterate, they most certainly would. Their latest, best window for relevance is inching shut given how they are structuring their slide decks. Clearly landing like a year ago would have been even better, even if volume had been tight they would have had every excuse available for why that was.

Which doesn't mean people won't choose them. Hands in the air for people who chose AMD or (more rarely) NVIDIA when the other card was "clearly" the "better" solution! I'd have my hand up, surely. They only reason I bought an 8500 back in the day was that I found it in the bargain bin in Best Buy and I had to swap drivers for different games to work.

That's not all doom and gloom, how many products are launched fully baked these days? Not many.

 

[Harry_Wild]

Intel’s Arc graphics cards have a warning label!

 

[Heartbreaker]

Intel’s Arc graphics cards have a warning label!

Is it just about Re-Bar?

 

[gdansk]

Is it just about Re-Bar?

Yes, I checked the transcript. Needs ReBAR

 

[linkgoron]

Realistically, looking at what they've put forward , it's an RTX2070- 2070S . Right down to the 200+w TDP (215) , die size balllpark and power requirements, bus width, the lot. It Even looks like an FE!

If it was joining the battle between 5700XT and 2070/S back in mid 2019 , undercutting the $399 price tag the XT launched at (and having RT to boot), it'd be very disruptive. and a real worry for AMD in particular , but that's a scenario near 3.5 yrs old .

I think the hype around its release now for $329-$349 is more indicative of what's happened to the industry since then. The fact this makes it good value next to a 3060 really shows what poor value that card still is, and that it still hasn't fallen to its MSRP (Unlike AMD cars that have falled below launch MSRP ) .

As for the future , well I guess it depends on how they can execute with Battlemage. they'll need to make exponentially large gen on gen jumps, ( A 'Navi moment' or 2) and / or for AMD/Nvidia to naively take their foot off the gas pedal (Which does happen, so don't rule it out) There's now, realistically 2 Generations to catch up on before they can either tackle the high end, or be profitable in the lower/mid end.

At these performance levels Alchemist has probably, what 6 months? before it's a low end card, battling the "4050" and "7500" class GPUs , and if These competitors pull their head in, and are able to price them as low end cards they are then I don't think pricing such large, expensive to build GPU's to compete with them will be sustainable.

Until Reviews of this , and Ada/RDNA3 are out , we won't have an entirely clear picture what they're up against. Lots to digest in the coming ~ month or so

It's kind of sad really, that it's supposedly a "nice" release at $325. A ~5700XT card, 3.5 years later, almost 60% bigger die on a slightly better process, or 70% bigger than Navi 23 which is probably more performant. This should have been released a year or two ago. Arc A is just way way too late with RDNA3 arriving November and RTX 4 already here.

Yet again, I'm not really impressed by Koduri. He left AMD five years ago, and created a brand new GPU line with Intel which basically matches the last GPU he made at AMD after a 7nm die shrink (Radeon 7 which was basically 7nm Vega64) on 6nm, and is basically two generations behind the competition.

 

[igor_kavinski]

I would prefer that Intel award loyalty points to people buying their half baked cards, so when they upgrade to Intel's nextgen (and they will want to, for better performance from the refined architecture of Battlemage hopefully), they can do so at a considerable discount by redeeming those points.

 

[DAPUNISHER]

I would prefer that Intel award loyalty points to people buying their half baked cards, so when they upgrade to Intel's nextgen (and they will want to, for better performance from the refined architecture of Battlemage hopefully), they can do so at a considerable discount by redeeming those points.

I am guessing you are not a finance major.

I really think crypto saved AMD's butt. Intel isn't likely to get that kind of sales bonanza in the consumer dGPU space. They have a loyal following that will buy their cards, the question is whether or not it's enough to sustain them through the tough times ahead. Bean counters will want their heads on a pike if they keep draining the coffers with terrible ROI. How long can they dodge the ax?

 

[igor_kavinski]

I am guessing you are not a finance major.

I think Intel needs to give some incentive to early adopters because possibly in six months or so, both Nvidia and AMD may have their nextgen mainstream cards out, which will further dilute the value of ARC cards. Their early customers may feel really bad remorse at that point. While 12GB or 16GB RAM is good, it can't make up for lower performance than the competition. What Intel thinks is good value compared to 3060 may turn out to be pathetic compared to 4060 if that comes out in the next 6 or 7 months.

Also, how come I haven't seen Intel comparing A770 to Radeon cards? Are they pretending they don't exist? Is that coz they make Intel's hard work look poor in comparison?

Arc A750 & A770 Performance per Dollar Slides, available Oct 12th (guru3d.com)

A770 has an 8GB version too? Is that the one priced at $329?

 

[NTMBK]

I am guessing you are not a finance major.

I really think crypto saved AMD's butt. Intel isn't likely to get that kind of sales bonanza in the consumer dGPU space. They have a loyal following that will buy their cards, the question is whether or not it's enough to sustain them through the tough times ahead. Bean counters will want their heads on a pike if they keep draining the coffers with terrible ROI. How long can they dodge the ax?

Their real "loyal following" is the OEMs. I fully expect to see lots of midrange gaming laptops with ARC, and Dell/Alienware desktops.

 

[DAPUNISHER]

Sounds like you are suggesting a return to the days of contra revenue. Remember how that went?

 

[DAPUNISHER]

Their real "loyal following" is the OEMs. I fully expect to see lots of midrange gaming laptops with ARC, and Dell/Alienware desktops.

Maybe. Lots seems overly optimistic still. I wonder what concessions Intel would have to make to them? Big risk for the OEMs given returns and tech support issues could be much higher than the norm.