• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

What are your thoughs on next gen Intel Battlemage GPUs?

playerwhoplayyes

New Member
Joined
Oct 7, 2023
Messages
15 (0.03/day)
Location
Chile
I hope they do well, they could shake the market if the drivers are optimized for almost everything, the price to performances is good, the performance itself is also a good upgrade from Alchemist and technologies are also good, too much to ask, but considering that in my opinion XeSS looks better than FSR(I don't have a game that supports XeSS, so I'm just making my opinion based on YouTube videos and images) maybe they will provide good technologies in some areas.

What are you expecting?
 
A "nothing burger" - technical term, I promise ;)
Medium class card by design that priced as higher end one (or low end one priced as mid range one).
Just another thing from Intel you can buy, but are afraid of consequences IF something goes wrong.
 
My thoughts? Intel is doing what they should. Participating in the GPU market. You can glad hand and preach about any corporation, but having a third player is important.

Will it perform? No; not with a B560. I think a lot of people are too positive, and a lot of people are too critical. Generally to try and rally some kind of echo chamber about the logo they like to fall asleep with at night.

In history. Atleast as I have noticed since before dual cores became real. Is that architecture changes are stair cases, usually next to each other.

The low end B cards will not meet or out perform mid or high end A cards. Mid rage B cards might be slightly better than mid or on par with highend A cards. I dont think mid range will beat an A770. No Just like stairs the B770 will beat the A770 but you will need to wait for it to release.

People that are critical of Intel will "I told you so" and tell everyone the sky is falling, Intel is going out of business and everything else under the sun because they like beating people on the internet.

People that knew this would likely be the case will say nothing.

People that were somehow banking the history of every technology companies portfolio on the chance that Battlemage low/mid range cards will be 2x more powerful than an A770 will be dissapointed and lick there wounds, or convert to the super criticals in point 1.

My prediction? B560 will release first next month, be a little slower than A770 have marginal power efficiency gains and over all be fairly priced on paper, atleast given A770s arent going to be in prod and everyone will scalp an extra $30 because its the new hotness. Afterwards I predict they will be late to release, trailing 5 and 8 series after CES. Then they will release the high end line. It will be single digit percent better than A770. It will carry the same price point of ~350 like the launch of A770 a few years ago.

I am totally fine with this. Its literally there second architecture. It wont be a driver disaster and there ODMs like sparkle etc are now established so I anticipate availability to be better.

For gamers, I dont think this will be a remarkable release. For Intel and the industry? I think it will be. For competition? Not yet maybe in a few generations.

Battlemage will be where Intel tests there white papers. How many shaders nets how much %? etc. With each iteration the road of future design philosophy gets more clear.

Its important, imo that everyone stop thinking its the core2 and brisbane era. Though a lot of those people are older now and maybe its difficult to cement new concepts, but we have passed the threshold some years ago on performance. AMD 8 series wont shatter the 7900XTX. The 5090 isnt going to be worth it over the 4090. The new Ryzens and the new CPUs are going to be nothing to write home about.

For all technology manufacturers atleast when it comes to compute. We are reaching the end of silicon. 30% perf improvement process over process is long gone, look at your Q6600 and say thank you.

For a new player like Intel (and they are new, dont let the IGP cult brain wash you) even licking double digits on a new architecture (important to note this isnt an iteration, they arent making alchemist better) is like moving a digital mountain.

TL;DR?

I'll buy one, cool.
 
Last edited:
If they put their minds together, and maybe use AI constructively they could potentially become a powerhouse.

Nvidia has been using AI to design their chips for sometime now. Maybe others have to and I live under a rock :D
 
I'm looking forward to reviews. Hopefully they do well for their price point.
 
What are you expecting?
I'm optimistic and excited. Here's why:
1. Intel did well with 1st gen ARC. The launch was a bit rough, but better than expected.
2. The experience from the 1st gen will serve them well with the 2nd gen, especially with what they've learned about the drivers and game optimizations.
3. Their pricing and AIB policies have been very workable and good for the market.
4. They have a small but solid following of users who will be first adopters.

TL;DR?

I'll buy one, cool.
Same here. I've installed and toyed around with ARC cards in system builds, but don't have one personally. Will be changing that.

If they put their minds together, and maybe use AI constructively they could potentially become a powerhouse.
As long as they don't force it on everyone, that would be cool.

I'm looking forward to reviews. Hopefully they do well for their price point.
Same here. The rumor is that focus group testing has begun for retail ready units, which means the press/reviewer community will likely be getting review samples soon.
 
Last edited:
I am extremely biased and think a bit of hopium is in order for someone who mainly cares about the 4k 60fps experience. Some thoughts.

  • The B580 needs to out perform at the level of an A770, but likely needs to perform better. This is how process generations typically work. A next gen midranged GPU should perform on par with a current gen high end (these terms are relative in the age of $2000 halo products). I'd give this a 50-50 of doing so.
  • I really want to buy a new GPU (Check Specs for context) but I'm not interested in anything more than $500. I'm also not interested in Nvidia as I have a poor opinion of them and AMD has a track recrod of running hot. Intel has space to enter and improve the market here.
  • I think Intel will surprise people, but will still have driver issues in some titles.
  • Their release timing is impeccable if they can get cards to etailers mid-december for a Christmas purchase.
  • I will buy one without a second thought if they can meet any or all of these goals and provide a 12 or 16GB model for under $300. I'd ideally love the B770 but the timing doesn't match up for me.
Overall I think if they nail the timing and get positive reviews they will see sales. They will be first to market for this release cycle and may be delivering a last generation performance product, but it might be just good enough if aggressively priced.
 
Would like great improvements over the weaker areas of last gen.

  • efficiency
  • idle power draw
  • introduction of Intel frame-gen solution
  • better than competition price/perf.
Fix most of these things and maybe I'll buy one just to mess around with some new hardware this year.
 
B580 being only 20 Xe cores bodes well for the architecture. Xe2 is easily 30-40% faster than Xe1 at the same core count. B580 has a 60% higher clock at the same max TBP. I'm expecting 80% higher performance for B580, putting it on par with a RX7700XT. Wishful thinking? Maybe.
 
I really want to buy a new GPU (Check Specs for context)
Anything Intel release this year will be a sidegrade best case scenario. B700 series might be enough to prove an upgrade but I'd gladly pass and wait till something at least 200+ % my current GPU becomes affordable. Considering your budget, you might as well turn to the dark side and buy an RX 7900 GRE. If lucky you can squeeze a lot of additional VRAM juice. Coolers are over-engineered in some models, so running hot ain't gonna be your main concern. However, with AMD, you forget about DLSS (not really an issue after getting more than double the performance, just go native) and CUDA (not an issue if you couldn't care less about prosumerism).
What are you expecting?
Frankly, I'm not expecting a kaboom. Just a minor step-up. Intel are very late to the GPU party but not too late. If (and it's a considerable "if") Battlemage provides >40 % over Alchemist at the same wattage and/or at the same MSRP then it's not highly impressive but it's enough to shake up the market... a little bit. 60+ % will definitely be a game changer. Imagine B580 occasionally beating 3070 Ti. That would be great! Very unlikely to happen though.

They're currently anti-closeted about their mess ups which is good news. They have a truckload of them which is bad news. No idea if Intel plan to make a 300+ W flaghsip GPU next year but if they actually do it I'm interested in purchasing it... unless the price is too messed up.
 
Interested in the high end cards of this generation. Depending on performance and price, I may pick one up to play with on one of my other computers (still going to stay with the 3080 FTW3 on my main rig).
 
....

For all technology manufacturers atleast when it comes to compute. We are reaching the end of silicon. 30% perf improvement process over process is long gone, look at your Q6600 and say thank you.

....
That line is GOLD!, single digit performance increases (except for the flop of arrow lake) is the norm & not for long either. Your 100% correct about silicon in nanotech manufacturing reaching its practical threshold. Its' a pity that online reviewers & armchair critics don't get it & rubbish a company like AMD for example with the Zen 5 release & its lack of double digit performance increase over Zen 4 because they don't understand this concept of the hard physical limitations with silicon fab.
 
B580 being only 20 Xe cores bodes well for the architecture. Xe2 is easily 30-40% faster than Xe1 at the same core count. B580 has a 60% higher clock at the same max TBP. I'm expecting 80% higher performance for B580, putting it on par with a RX7700XT. Wishful thinking? Maybe.
B580 is between 4060 and 4060 Ti. It may be able to reach 4060 Ti. The clock is only 40% higher, not 60%. The peak is 2.8GHz vs 2GHz. But it has 20% less render units, so it has to make up for that. Clocks should do that and some. Memory bandwidth is lower but the newer architecture will more than make up for that. It'll perform noticeably better in real world games vs synthetics when compared to Alchemist. It'll also do lot better in lower settings and resolutions.

Alchemist needs 4K resolution for the architecture to fully shine. Even 1440p is not enough.
Sparkle Arc A770 ROC Review
It goes from performing like 6600XT in 1080p to 4060 in 1440p to beating 7600 XT in 4K, and nipping at the heels of 3060 Ti.
The low end B cards will not meet or out perform mid or high end A cards. Mid rage B cards might be slightly better than mid or on par with highend A cards. I dont think mid range will beat an A770. No Just like stairs the B770 will beat the A770 but you will need to wait for it to release.
A770 is nothing special over A580. B580 will do at least as good as A770, realistically it'll do 5-10% better at the minimum.
People that were somehow banking the history of every technology companies portfolio on the chance that Battlemage low/mid range cards will be 2x more powerful than an A770 will be dissapointed and lick there wounds, or convert to the super criticals in point 1.
2x will not be far away for the top G31 die with 32 Xe cores, since the B580 is supposed to be at the 4060 to 4060 Ti level. G31 has 60% more shaders, clocked 5-10% higher. You only need about 60% higher than 4060 Ti to reach RTX 4070 Super performance. That's 83% over A770 in 4K. In lower resolutions the gains need to be greater, but that's because Alchemist has hardware flaws and needs higher resolution/settings to perform fully.

The problem is not performance. G31 is supposed to be sometime middle of next year. So then it'll become x60 class again.
 
Last edited:
Interested in the B780 if it ever comes out.
 
Drivers are the most important thing to improve like with Alchemist, though they've gotten better in the last two+ years. I almost got an A770 two years ago but ended up with a RX 6700 XT instead, they were both priced at 480EUR back then.

Competition is always good anyway.
 
It's not even out so can't say anything yet. But I keep my eye on them, I really tired of what AMD and nvidia did to midrange segment, just look at RX 7600/XT and 4060/Ti they were miserable and badly priced and didn't offer much in terms of improvement over older cards. I wanted Intel to at least performs well in this segment and competitively priced, and improve driver support. I will definitely buy them then.
 
Just saw twitter rumours about higher-end cards coming out in Q3 2025 — wow, that's incredibly late, if true.
 
People are afraid of new gpu player even if it is Intel. Too much problems with drivers at the beggining put many people off. Intel should fight aggressively with pricing but they dont...he wont atract new customers and i bet in some time he will close gpu business.
 
4060-4060ti level of performance at £250 price point is likley to do well assuming they can get the narrative right and avoid the driver woes of early A770. Still have an Acer Predator A770 16gb here and find it to be perfectly capable in comparison to 3060 / 4060 / RX 6600 / RX 7600.
 
I just hope they are good value with proper drivers from day 1. There's nothing else to think while we don't have any real world tests.
 
B580 is between 4060 and 4060 Ti. It may be able to reach 4060 Ti. The clock is only 40% higher, not 60%. The peak is 2.8GHz vs 2GHz. But it has 20% less render units, so it has to make up for that. Clocks should do that and some. Memory bandwidth is lower but the newer architecture will more than make up for that. It'll perform noticeably better in real world games vs synthetics when compared to Alchemist. It'll also do lot better in lower settings and resolutions.
You're absolutely right. I (probably wrongly) assumed 2740MHz as a base clock. In TPU's review, the A580 even reaches 2.4GHz in some games.
If the silicon goes from 6nm to 4nm (just guessing), I would have assumed over 3GHz turbo boost.

Edit: I was just checking out ASRock's website and noticed that they spec a 2GHz engine clock for the A580 but TPU's review has a 2354MHz gaming average for the A580 Challenger. It seems that ASRock specifies a minimum turbo frequency and not peak turbo. The lowest the A580 Challenger hits, is 2050MHz during Furmark which seems to line up with ASRock's 2GHz spec.
In the same way ASRock now specs 2.74Ghz for the B580 Challenger and I think it's safe to assume that it's again the minimum turbo frequency. If it can turbo in the same way as the A580 then average gaming turbo is going to be over 3GHz. If you add in the Xe2 efficiency, that will put it around 45-50% faster than the A580.
 
Last edited:
You're absolutely right. I (probably wrongly) assumed 2740MHz as a base clock. In TPU's review, the A580 even reaches 2.4GHz in some games.
If the silicon goes from 6nm to 4nm (just guessing), I would have assumed over 3GHz turbo boost.
Ah, I didn't know about that being the minimum. Then it makes more sense why 4060 non-ti is mentioned.
 
Back
Top