[Rumor] Intel Arc A770 Early October Launch

Wait for the reviews first. :) lol
I don't know if they really manage 3060 or even somewhat close to 3070, the delays will sting.
I'd like to stay optimistic though. Who knows perhaps XeSS is super resolution+. Performance wise though Nvidia does have a lot of 3080/70 stock to work through... and AMD has already been lowering MSRP on the 6k gen.

Just don't be a buggy mess... and let Intels management let the GPU division take the bronze and live to fight the next race.

I’m more concerned about drivers. Intel doesn’t need to compete at the top end of the market. The majority of the market IS in that 3060 or 3070 space, and they have a lot of agreements in place with OEM system builders. If their drivers aren’t a disaster, the product will sell well and they can build a good share position. The problem is that the drivers, currently, appear to be a disaster, and anything running DX11
has terrible performance.
 
If Kyle is correct and current stock is the end of line, doubt they put much effort into fixing any of this.
Ya at the price they are selling these there is no way they fab more of them... that is a pretty big die to sell the best of the best off the wafer for that price. (a price which can only go down right)
Next Intel GPU fab run will be battle mage... perhaps. Even if Intel has the fab space already purchased... I think management is going to need to see some impressive engineering samples before they sign off on a large run. Intel isn't going to be ok with firesales on GPU silicon for years.
 
Just imagine, in a alternate reality where Old Bennett never left Intel, and he was shilling for these GPUs. I bet he would've been rocking a liquid cooled A770 on his open loop destroying games at 1440p.
 
Heard rumors of a working larrabee test station out in the wild, might be the closest thing to one running actual games. To me that’s an interesting story compared to this Arc lineup
1664654506394.png
 
Intel Arc A770 16GB Limited Edition GPU Review & Benchmarks



NVIDIA never saw this coming... A770 & A750 Benchmarks!

 
Let's at least go to the go-to on YT reviews that always get it right without all the pomp and circumstance.


It's much better than what I expected. It's a good baseline for future Intel GPUs.
 
Basically it cost too much for the performance. Now if they had released 6 months to a year ago, these would fly off shelves. Even as gaming GPU's since the market was so scarce. So if Intel is selling current msrp at or near a loss, things are not going to look good.
 
I remain guardedly optimistic, and intend to snag an A750 when time and my budget allow for it.
 
So it's a card that only an enthusiast can really use, but it's absolutely not enthusiast grade. It's better than I thought, but still... Not great.

Or even good.
 
Both the A750 and A770 are around $300, but the 6600 XT which beats both in benchmarks can be found $300 new and $250 used. The RTX 3060 can be found for slightly less than $300 on Ebay used. IF the A750 was $200 and the A770 $250 then we'd have something here. Not a bad choice for a Linux user, but the 6600 XT is still the better choice.
 
Very interesting. Performance seems to be all over the place right now, depending on the game. I've been looking at benchmarks all over the internet and comparing the card against my existing 2080 that I'm going to replace soon. It seems like about 40% of the games show the A770 as being about the same speed as my 2080 or maybe just a hair faster. 40% of the games show the A770 being WAY slower than my 2080. And the remaining 20% have the A770 with a good lead over the 2080. I would assume that this wild inconsistency is due to driver issues, but not sure. It seems to me that the value of this card really depends on what games you play and how this card performs in those specific games. If you want a card that is great at everything, this isn't it.
 
I'm just happy that the cards didn't completely suck. For their first commercial debut, its not that bad. However, they're gonna get their asses whooped and end up in a bargain bin short order. They really needed to take the loss on this one and put them in the price categories they should be, around 200 bucks. They basically said they were targeting the lower end of the market and they should stick to that, and dump them on the market.

A part of me can almost buy into the possibility that Intel is canning the Arc lineup. They were expecting a win and they got a turd instead. It would be a huge shame if they hang it up after this, there is some potential here and these are compute beasts IIRC. There is no way Intel's drivers are mature. All reports indicate that they are a dumpster fire. Given time, the drivers will improve and so too will these chips.... theoretically.
 
I think one possible way to do it, keep the high price and stay fully marginal in the DIY market to keep appearance (where I do not see the point of those cards outside AV1 feature-DP2.0 which would be really niche over a 6650xt at those price).

Sell them through OEM at a massive loss, gaining some market share without destroying the perception. I would imagine that even at a really good price and OEM will want to see nice drivers, specially if those computers in those price range will be bought to play the GTA 5 and easy to run title of the world often DX 11.
 

That is the only way I ever saw this working. Sell them in cheap low end gaming pre-assembled PCs at Costco and similar. Price them to a point where they're an exceptionally great deal. Software/drivers can still annoy people, but if they're essentially getting something better than an integrated GPU for not much more people won't care as much.

Otherwise, even if cheaper, I'd still go with Nvidia/AMD for driver support.
 
Basically with the A770 if you're looking in the 3060 to 3060ti type of performance market then this could be something to keep an eye on, if you're expecting anything more than that... not so much.
 
I am not sure about the much better performance than expected at least if we go to we was expected before the first launch, the talk seem to always have been that it would between in between a 3060-3070, seem to have ended up more between the 3060 and 3060TI.

What felt better than expected after the first terrible launch, number of title with big issues with it (crash, terrible performance for some reason) that seem limited purely to older title now.

While RTX title run much better than on AMD and really good a bit like on the 3060 series, it is rarely relevant those cards not being good enough for many title to be a wise decision image quality-performance versus others options.

Certainly good enough to me result to have some faith for the next ones if they happen.
 
Well, all the big three card builders bailed on Intel this round, so certainly it did not look baked well enough for those companies to put their brands on those cards, and keep in mind PCB designs and coolers were already completed. For all intents and purposes, Alchemist is already over. I am hearing today that many of the folks at Intel are considering today's reviews a "win," if that puts anything into perspective for you.
 
Well, all the big three card builders bailed on Intel this round, so certainly it did not look baked well enough for those companies to put their brands on those cards, and keep in mind PCB designs and coolers were already completed. For all intents and purposes, Alchemist is already over. I am hearing today that many of the folks at Intel are considering today's reviews a "win," if that puts anything into perspective for you.

Hopefully if they consider this a win it means we’ll at least see Battlemage come out and possibly improve. If upper management is actually willing to stick behind this for a few generations they might have something good on their hands.
 
No major AIB partners... Can't imagine why. I'm sure everyone would love to be on the hook for tech support and RMAs for these... Things...
I wonder if Intel didn't build stock of "Founder edition" large enough to be unsure if it will not be hard to sell all of them with a profit margin to start with, let alone adding new AIB to it. I could be completely out of field on that one.
 
I liked LTT's take on this. Someone has to buy it if you really want competition in the market.
 
Well, all the big three card builders bailed on Intel this round, so certainly it did not look baked well enough for those companies to put their brands on those cards, and keep in mind PCB designs and coolers were already completed. For all intents and purposes, Alchemist is already over. I am hearing today that many of the folks at Intel are considering today's reviews a "win," if that puts anything into perspective for you.

Which were those? ASUS, MSI and Gigabyte? Is it likely Intel just throws in the towel in the next year or so?
 
how many does he plan on buying?
Perhaps he could buy a few racks of them.... take the 60 or so screws out of the screwless designed cards, so he can water cool them with his pool.

In all seriousness... people buying them is up to Intel. They think they have the pricing right. We'll see. If they don't move they will have to drop them some more.

They are already on to BM... lets just hope BM is a nutty good second effort sooner rather then later. Everyone wants a third player and BM is our best chance at that clearly... Alchemist is no BM but it is sort of an upper decker of the low-mid tier.
 
Perhaps he could buy a few racks of them.... take the 60 or so screws out of the screwless designed cards, so he can water cool them with his pool.

In all seriousness... people buying them is up to Intel. They think they have the pricing right. We'll see. If they don't move they will have to drop them some more.

They are already on to BM... lets just hope BM is a nutty good second effort sooner rather then later. Everyone wants a third player and BM is our best chance at that clearly... Alchemist is no BM but it is sort of an upper decker of the low-mid tier.

I know a lot of people want them to compete at the highest end, but I don't think it is that big of a deal although should be a goal. If they can compete with the upper mid range that would be great. The number one thing they need to so is get drivers and software right. Doesn't matter how fast a card is if you have problems. The Vista era Nvidia drivers were trash as well. But since then Nvidia has been about as perfect as possible. But the older ATI and Nvidia drivers (Vista era) were a pain. I remember for certain games you would have to downgrade to an older driver for certain games to not have problems. Last decade has been essentially flawless for Nvidia drivers on my end. I'm not willing to go back to that bad driver quality, even if the raw performance is better than Nvidia/AMD offerings.
 
  • Like
Reactions: ChadD
like this
The bottom line is that Intel needs ARC to perform well enough, with enough stability, to be able to push Nvidia out of its laptop designs.

A+A and I+I across mobile would be a crushing blow to Nvidia.

I am not even sure Intel is concerned about the actual dGPU market. It just has to be proven and branded there.
 
I would imagine they also want some presence in the data center type of GPU like system, where AMD-Amazon-Nvidia-Facebook and many other have alternative solution or working on them
 
  • Like
Reactions: ChadD
like this
I would imagine they also want some presence in the data center type of GPU like system, where AMD-Amazon-Nvidia-Facebook and many other have alternative solution or working on them
100%. How is that Foveros tech coming along?

On that front, did you see the PR from Meta on Monday? Huge effin deal in the AI world. Nvidia is not happy.... Again.
 
  • Like
Reactions: ChadD
like this
Back
Top