Tested: Intel's Arrow Lake 140T iGPU mostly maintains an edge over AMD's older 880m

These 2 laptop CPUs and gpus are 2 different categories. It should be a laptop with at least AMD Ryzen™ AI 9 HX 370/75 for a fair this comparisment.
 
Is the top Arrow Lake-H iGPU supposed to be identical to Lunar Lake? I forgot.

Too bad many of these will end up with a dGPU, but it will end up in (eventually) $500 media consumption laptops and mini PCs.

Looking to the future, maybe AMD will neglect its mainstream APUs and desktop iGPUs, only to crush everything with Strix Halo, Medusa Halo, etc.
 
  • Like
Reactions: gg83
It's almost like you didn't pay attention to the Asus Zenbook S16 results that have the 370 HX!
Not exactly. That Asus it a different category - ultra slim. They usually are downclocked, in order to keep thermal low. See the input adapter - it is only 65Ws vs 120 for MSI Prestiges.

 
  • Like
Reactions: P1nky and Quirkz
Not exactly. That Asus it a different category - ultra slim. They usually are downclocked, in order to keep thermal low. See the input adapter - it is only 65Ws vs 120 for MSI Prestiges.

There's no way that Arrow Lake MSI laptop is coming anywhere near 120W power use. None at all. Not without a dGPU. The Asus laptops were tested in "performance" mode that basically gets them up to ~45W total, for the chips and rest of the system. 65W is more than adequate. The ARL-H might need more than 65W if you were charging the battery while running it flat out, but I'm skeptical you could even get that to break 80W.
 
There's no way that Arrow Lake MSI laptop is coming anywhere near 120W power use. None at all. Not without a dGPU. The Asus laptops were tested in "performance" mode that basically gets them up to ~45W total, for the chips and rest of the system. 65W is more than adequate. The ARL-H might need more than 65W if you were charging the battery while running it flat out, but I'm skeptical you could even get that to break 80W.
Not sure what are you talking about ... but in that article it is mentioned that this Intel Core Ultra 9 285H MSI Prestige 16 AI E comes with 140W power adapter. Why do you think is that? If it was possible MSI could save few bucks for a cheaper power suply. Btw, MSI Prestige AMD version comes 100w PA(not 120W as I wrote.). This are obvsioly an 'adjusted' or fixed results.
 
  • Like
Reactions: P1nky and Quirkz
These 2 laptop CPUs and gpus are 2 different categories. It should be a laptop with at least AMD Ryzen™ AI 9 HX 370/75 for a fair this comparisment.
According to Just Josh, Intel wanted reviews to focus on the 365 instead of the 370 or 375 as the main competitor. I personally don't think this is entirely fair, but it's what Intel wanted.
 
  • Like
Reactions: P1nky
Why is the AMD model heavier than the Intel (4.19lb vs 3.31lb), when the battery is smaller (82Whr vs 100Whr)?

Per another review, this particular MSI Intel model runs very hot. It also runs at 55W vs AMD's 30W in that review (Zenbook S16 Ryzen 9 370HX).

View: https://youtube.com/watch?v=B-ESfY9uVWM&t=183
Just Josh's review found the same. Apparently it's an MSI problem, not an Intel one. Other laptops with ARL-H do not have this problem.
 
Not sure what are you talking about ... but in that article it is mentioned that this Intel Core Ultra 9 285H MSI Prestige 16 AI E comes with 140W power adapter. Why do you think is that? If it was possible MSI could save few bucks for a cheaper power suply. Btw, MSI Prestige AMD version comes 100w PA(not 120W as I wrote.). This are obvsioly an 'adjusted' or fixed results.
Having extra capacity in a power adapter is the way to play it safe, and you can get transients that spike higher than average power draw. But if you have a fully charged laptop, plugged into a power meter, you'll find most never come anywhere near the power limit of the adapter. (The only exceptions are typically laptops with dGPUs, where you can basically double the total power draw thanks to having a GPU.)

There's a big gap between 65W and 120W. If a laptop can potentially exceed 65W (like with maxed out system power draw, plus charging a batter) you need to go to the next level. Usually, that's not going to be 20W higher, so you jump to 120W and now nothing the laptop might do will exceed that amount, or even come close to it.

But let me give a concrete example. The Asus Zenbook S16 with a 65W adapter can pull 36W at the wall, just for charging the battery. Assume 90% efficiency and that means it's using 32W roughly to charge the battery — about half the potential output of the adapter. If I boot it up, it basically hits the 65W adapter limit. I saw 70W at the outlet, which with 90% efficiency would be 63W. But if the battery can take 32W on its own to charge, that means if the laptop needs 50W the battery will have to charge slower and only get 15W. It would take twice as long to charge. So double the adapter to 120W (give or take) and you'd have full speed charging while doing anything else.

Asus didn't take that route, opting for slower charging. MSI has gone the other route. Maybe the ARL laptop draws up to 65W total as well, so it really needs the extra headroom to allow for charging while doing other stuff. It's not that the laptop itself pulls anywhere near 120W, but the combination of laptop plus battery charging can get close to that level. But if you take out battery charging, I'd be surprised if it was breaking 65W, even with a 45W chip.

Keep in mind that the MSI ARL laptop had battery life of over 15 hours with a 100Wh battery. That means it was averaging just 6.7W of power draw during the battery test. Which means the display, RAM, CPU, etc. could all get down to that little of power use.

Also, as another interesting aside, the "fast charging" power draw of a laptop will be much higher than the "finish off the charge and reach 100%" power draw. The Asus S16 for instance gets to maybe 95% charge with fast charging (the 32W I mentioned) and then the last 5% or so happens at a more sedate ~15W rate. (It might even have more charging zones, as I think I saw up to 40W power draw initially before that dropped to 32W.)
 
Can intels igpu catch amds? Or will they be behind one generation or so?
If Intel wants to "go big" like Strix Halo, sure, it could keep up. I don't know that it would be efficient, but for raw performance it's already roughly on par with the around the same number of Xe-cores as AMD has CUs. That's with Xe2, though, and ARL has a modified Xe1 it seems, which holds back performance at times.
Did the title really need the "mostly" and "older"? Such a backhanded compliment like someone pulling your teeth lol.
The problem is that ARL's iGPU performance is inconsistent. At times it's faster than the Lunar Lake IGPU, at times it's (quite a bit) slower. Differences in laptop design make it hard to pinpoint exactly what's going on, and I will say that I didn't do the testing on the MSI laptops so I don't know for certain that it was all done "properly" (i.e. to my standards). I'll vouch for the two Asus laptops, though, even when their results look a bit odd.
 
  • Like
Reactions: Gururu and gg83
If Intel wants to "go big" like Strix Halo, sure, it could keep up. I don't know that it would be efficient, but for raw performance it's already roughly on par with the around the same number of Xe-cores as AMD has CUs. That's with Xe2, though, and ARL has a modified Xe1 it seems, which holds back performance at times.

The problem is that ARL's iGPU performance is inconsistent. At times it's faster than the Lunar Lake IGPU, at times it's (quite a bit) slower. Differences in laptop design make it hard to pinpoint exactly what's going on, and I will say that I didn't do the testing on the MSI laptops so I don't know for certain that it was all done "properly" (i.e. to my standards). I'll vouch for the two Asus laptops, though, even when their results look a bit odd.
Anything small form factor must rely on cooling potential? Could that be the limitation? Power delivery would be second?
 
Anything small form factor must rely on cooling potential? Could that be the limitation? Power delivery would be second?
Yeah, I'm sure cooling factors in as well. 14-inch and 13-inch laptops routinely offer up a bit less power than otherwise "equivalent" 16-inch and 17/18-inch models. So there's the amount of power the manufacturer has designed the laptop to handle (within given thermal and noise limits). And then you want to have some headroom for charging and just to be safe — running a PSU or power adapter at 100% output all the time is generally a bad idea.
Speaking of THW's review, I find its test methodology antiquated, and its verdict wanting. The test entirely ignores upscaling, which for laptops w/ iGPU should be de rigueur. As well, 140T now uses XeSS2, which would have been a great opportunity to check out Intel's framegen implementation vs FSR 3. No worries, there are plenty of better reviews.
To be fair, the number of games with XeSS2 support right now, including framegen, is pretty limited. I know F1 24 has it, publicly available. I'm not sure if anything else is available versus merely announced that "it's coming soon(ish)!"
 
  • Like
Reactions: gg83
Speaking of THW's review, I find its test methodology antiquated, and its verdict wanting. The test entirely ignores upscaling, which for laptops w/ iGPU should be de rigueur. As well, 140T now uses XeSS2, which would have been a great opportunity to check out Intel's framegen implementation vs FSR 3. No worries, there are plenty of better reviews.
XESS2 support is very poor right now. I think there's 1 game that has it currently.
 
The problem is that ARL's iGPU performance is inconsistent. At times it's faster than the Lunar Lake IGPU, at times it's (quite a bit) slower. Differences in laptop design make it hard to pinpoint exactly what's going on, and I will say that I didn't do the testing on the MSI laptops so I don't know for certain that it was all done "properly" (i.e. to my standards). I'll vouch for the two Asus laptops, though, even when their results look a bit odd.
The 140T in the 285H clocks 20% higher than the 140V in the 258V so I'd bet that's where the anomalies are coming from. I'd assume where it's winning is also where Alchemist comes closest to Battlemage on the discrete card side.
 
  • Like
Reactions: gg83
Did the title really need the "mostly" and "older"? Such a backhanded compliment like someone pulling your teeth lol.
Yes it did. "Mostly" because AMD did better on some of the 720p tests, so it was far from total dominance. And "older" because by the time these Arrow Lake laptops get into the hands of most customers, the competing AMD silicon will have been available for at least 8 months.
 
  • Like
Reactions: JarredWaltonGPU
The 140T in the 285H clocks 20% higher than the 140V in the 258V so I'd bet that's where the anomalies are coming from. I'd assume where it's winning is also where Alchemist comes closest to Battlemage on the discrete card side.
It's probably part of it, and probably the other part isn't just clocks but also lack of driver optimizations for Battlemage. When I tested the B580 and B570, there were a few games that didn't show nearly as much of a performance improvement as others.

Black Myth: Wukong was one of the three games (out of 22) that showed less of a performance increase. And by "less" I mean it was actually slower on the B580 than on the A770. Curiously, Intel even noted the issue, nearly two months ago, and as far as I can tell it still hasn't figured out a fix. 🤷‍♂️
 
Black Myth: Wukong was one of the three games (out of 22) that showed less of a performance increase. And by "less" I mean it was actually slower on the B580 than on the A770. Curiously, Intel even noted the issue, nearly two months ago, and as far as I can tell it still hasn't figured out a fix. 🤷‍♂️
Also remember Alchemist performs proportionally worse in Starfield than almost everything else and that's never really been fixed beyond making it minimally playable. I'd love to know what the situation is for things like that since all of the other performance outliers they've publicly identified have been resolved in a better fashion.
 
I've been an underwhelmed by RDNA3.5 in Strix. The improvements are pitiful IMO, and AMD are lucky Intel gimped Arrow Lake and did not use Xe2. They won't be so lucky with Panther Lake, as that goes straight to Xe3 for the iGPU. Hopefully Panther comes with some higher core counts for the iGPU and we can see a real comparison to Halo.
 
Are ya like me? Are you sick of u tubers with the surprised face saying they're outraged their paper thin laptop gets hot playing something intensive? Oof.I have my nephews 12 year old gaming laptop and it looks like a damn Ferrari and weighs a ton.I hope Intel gets their drivers sorted this time.Also some of their cards are priced for me but I don't have certain features on my Haswell desktop.
Nvidia and AMD sure made GPU drivers seem easy to do. I'm still shocked Intel is having trouble with them.