Is there any info about idle power? I think that's what interested me the most from this product. I expect worse than Strix Point but much better than the Range ones.
First gen product in a new swimlane, lots of low hanging fruit.It's an unimpressive demonstrator.
Cost is similar to similar performance parts with weaker CPUs. And yes RDNA3.5 is only so-so, RDNA4 would've made it a class leader.It's barely better than the preceding solution and comes with many downsides (cost, Radeon GPU).
And it will only get better.But at least it can scale up and down more efficiently.
Except it lacks a lot of Medusa IOD features.It is not promising. Fancy new IOd didn't save Zen 5
We shall see, probably but clocks should finally get out of the 5.1Ghz slump.and Zen 6 is an even smaller IPC upgrade.
The Z6 CCD is general is way faster and N2 is no joke vs N4P.And unlike Strix Point to Medusa Point there isn't going to be a big change from this CCD to Zen 6 CCD.
100W isn't "over 50% more" than 70W.It's consuming 70W while matching in gaming performance the 13900H + 4070 mobile that consume over 50% more power, and there's no 8GB VRAM limit.
The closer integration also enabled Asus to put in there a 25% larger battery while reducing the weight by 80g.
And I'm pretty sure the 13900+4070 aren't even going to work at 25W total package, let alone performing above the level of a Strix Point.
This means you can comfortably take the Z13 Halo on a bus/plane trip and play games for over 2 hours before needing a wall socket, whereas the older 4070 model wouldn't last 1 hour.
It makes sense from a performance per watt standpoint. Makes zero sense from an economic sense.
The Z6 CCD is general is way faster and N2 is no joke vs N4P.
Yes it will be against other N2 parts, so it depends on how others execute.
Upto 9hrs battery life on a 70whr battery using custom power profileHere is
At least for the ROG Flow, it seems to have a significantly cheaper launch price than the Intel+4070m model it's replacing while being lighter weight, having a slightly faster screen, and having a larger battery and better battery life. Unless I'm missing something. . .
View attachment 117483
View attachment 117484
Why do you say that? There's no reason why AMD can't put cut down Strix Halo in $1000-$1500 laptops.What you are missing is that Strix Halo will only be used on very expensive devices.
So it will be put on 2000$+ laptops that will compete against 1000-1300$ 4070 laptops.
A 13900H is old as all heck and on a significantly older node, that's a very low bar to beat. It'd be much more appropriate to compare this to the many Strix Point + 4060/4070m implementations that are no doubt floating around.It's consuming 70W while matching in gaming performance the 13900H + 4070 mobile that consume over 50% more power, and there's no 8GB VRAM limit.
The closer integration also enabled Asus to put in there a 25% larger battery while reducing the weight by 80g.
And I'm pretty sure the 13900+4070 aren't even going to work at 25W total package, let alone performing above the level of a Strix Point.
This means you can comfortably take the Z13 Halo on a bus/plane trip and play games for over 2 hours before needing a wall socket, whereas the older 4070 model wouldn't last 1 hour.
On the contrary, this chip makes full sense for larger handhelds like the AYANEO KUN or the Legion Go.
People were expecting this to perform badly at 40W, but at 25W it's already outperforming Strix Point and at 35W it performs well above anything we've seen.
Look at Returnal benchmarksA 13900H is old as all heck and on a significantly older node, that's a very low bar to beat. It'd be much more appropriate to compare this to the many Strix Point + 4060/4070m implementations that are no doubt floating around.
To only have noticeable gains at power limits that are at the edge of what's acceptable for handheld sizes with battery life (I. E 50-60Wh batteries) for significantly higher costs most likely kills any gaming handheld implementations with this SKU.
Things have changed, N2 is cheaper than expected, timing and yields are good.I thought the consensus was that it will use N3P?
I thought the consensous was that it will use N3P?
Because Strix Halo, by virtue of its die area + multiple die packaging probably costs at least twice what Strix Point does for AMD? And we know Strix Point SKUs are by and large not cheap. And that's before the additional margin AMD will need to compensate for the significantly reduced volumes this will have versus SP, AMD does not put in R&D for such a specialized product to get lower margins than their more 'vanilla' dies in SP&HP.Why do you say that? There's no reason why AMD can't put cut down Strix Halo in $1000-$1500 laptops.
Things have changed, N2 is cheaper than expected, timing and yields are good.
Basically N3 is just like 20nm, the last gasp of an xtor tech with too limited benefits, N2 is better perf/$ vs N3, nothing else has to be said.
What you are missing is that Strix Halo will only be used on very expensive devices.
So it will be put on 2000$+ laptops that will compete against 1000-1300$ 4070 laptops.
This chip is expensive, or at least AMD will sell it very expensive + the motherboard needed and all that stuff. There is no realistic market for 1500$+ handhelds, outside very niche people.
The fact that Asus out of all brands is selling this much cheaper than their old Intel+Nvidia combo shows that Strix Halo has a decent price point. That Z13 carries both form factor and Asus tax, yet it still hits the $2000 mark.What you are missing is that Strix Halo will only be used on very expensive devices.
So it will be put on 2000$+ laptops that will compete against 1000-1300$ 4070 laptops.
Because Strix Halo, by virtue of its die area + multiple die packaging probably costs at least twice what Strix Point does for AMD? And we know Strix Point SKUs are by and large not cheap. And that's before the additional margin AMD will need to compensate for the significantly reduced volumes this will have versus SP, AMD does not put in R&D for such a specialized product to get lower margins than their more 'vanilla' dies in SP&HP.
So ultimately, I doubt Strix Halo is more expensive for partners than a Strix Point + NV solution at the same performance level.
CPU makes a real difference in cyberpunkYou also get 16 full Zen 5 cores versus 4 Zen 5 + 8 Zen5c cores in Strix Point.
Blackwell mobile is like 15% faster per SKU, it is a very weak upgrade.Presumably Hawk/(Arrow?) + 5070 Ti/5080 would be much faster in games (and although also very pricey) comparable in price.
There will be cheaper strix point devices. Strix Halo with one CCD shouldn't cost appreciably more to produce than a 4070 mobile plus CPU.Because Strix Halo, by virtue of its die area + multiple die packaging probably costs at least twice what Strix Point does for AMD? And we know Strix Point SKUs are by and large not cheap. And that's before the additional margin AMD will need to compensate for the significantly reduced volumes this will have versus SP, AMD does not put in R&D for such a specialized product to get lower margins than their more 'vanilla' dies in SP&HP.
And those solutions are far higher BOM than Strix Halo.
It does seem like it.Doesn't seem like it.