Nothingness
Platinum Member
- Jul 3, 2013
- 2,750
- 1,396
- 136
Isn't the ASUS ProArt PZ13 @FlameTail mentioned a 2-in-1?I see no 2 in 1s? Actual Yoga, Yoga 7 14 (at least that is what it's called in EU, Lenovo has moronic naming strategy)?
Isn't the ASUS ProArt PZ13 @FlameTail mentioned a 2-in-1?I see no 2 in 1s? Actual Yoga, Yoga 7 14 (at least that is what it's called in EU, Lenovo has moronic naming strategy)?
What a poor sad slide: no mention of models of M3 and Snapdragon; and I didn't know Apple was into making Copilot+ PC.View attachment 100342
Qualcomm appears to be fine. Zen5 isn't the ST performance slayer it was hyped up to be.
Really interested to see how they'll handle gaming as that's a big hurdle to the emulation and also driver maturity.Okay, I just finished watching the Qualcomm Computex Keynote.
Here are the highlights, IMO;
NPU efficiency
View attachment 100335
Solid Web browser performance
View attachment 100336
Gaming
View attachment 100337
Testing 1200+ games
View attachment 100338
Partners
View attachment 100339
Overall, it was a okay-ish keynote. Definitely better than Nvidia's 2 hour slog. They could have done it more concisely (several points were mentioned repeatedly), and they blew 20 extra minutes past their allocated 60 minutes. Still, it was Christiano Amon's first Computex Keynote, and I hope we see better and more exciting keynotes in the years to come.
Nice shoes, btw
View attachment 100340
They're both N4P with similar 1T performance, similar TDP and similar core counts.View attachment 100342
Qualcomm appears to be fine. Zen5 isn't the ST performance slayer it was hyped up to be.
TDP isn't really a clear metric for comparison.They're both N4P with similar 1T performance, similar TDP and similar core counts.
ARM seems to be doing fine. Apple has the undisputed ST performance leadership. Cortex X925 brought 17% YoY IPC improvement (1% more than what AMD achieved in 2 years with Zen5, fwiw) and Oryon V1 seems to be in the same ballpark as X925, bur it's virtue being that it's already shipping in laptops, while X925 might take 6-12 months from now, to debut in PCs.RISC architecture was going to change everything not end up in the same spot as x86 at the same time.
It better have some power advantage. And 12% faster memory too.TDP isn't really a clear metric for comparison.
In terms of raw power consumption and perfoemance-per-watt, I believe X Elite will still be better than Strix.
Probably vs. the slowest X Elite but in any case, the difference is so small that the one who has best efficiency wins in my book.View attachment 100342
Qualcomm appears to be fine. Zen5 isn't the ST performance slayer it was hyped up to be.
Wait until you see Zen 5’s power curves on ST.They're both N4P with similar 1T performance, similar TDP and similar core counts.
RISC architecture was going to change everything not end up in the same spot as x86 at the same time.
That 50 TOPS NPU on Strix is interesting. It apparently supports BF16, something Qualcomm's 45 TOPS NPU doesn't?Wait until you see Zen 5’s power curves on ST.
And the battery life. The fact that they’re not talking about it and that they need 25% more frequency to do so does not bode well.
Yeah probably. Honestly who cares lol. I think local AI is useful contra some others, but also like this isn’t going to make a world of a diff.That 50 TOPS NPU on Strix is interesting. It apparently supports BF16, something Qualcomm's 45 TOPS NPU doesn't?
Only tablet so far is Surface Pro. And it has a fan (while even some Intel ones didn't).QCOM promised that we'll see Snap X Elite in more form factors like PC AIO and Tablets. So I guess the mythical fanless Snap X devices will come in the form of tablets? Or are these going to be fanned too?
If QCOM is able to slot it into fanless tablets, it will make for a nice comparison between Lunar Lake 2X4(9W) and Snap X.
There goes my fanless hope I guess. So it's really only Lunar Lake aiming at the fanless Macbook Air Mx market.Only tablet so far is Surface Pro. And it has a fan (while even some Intel ones didn't).
Yea, I said this up the thread I think.View attachment 100354
If you read the footnotes, they used the 4.0 GHz Snapdragon X Elite X1E-80-100 SKU in the Surface Laptop 15 for this benchmark.
The interesting thing is that, it means Qualcomm's top end 4.2 GHz (which is 5% faster than the 4.0 GHz) should have identical ST GB6 to Strix Point.
Yeah, Qualcomm's fine.
View attachment 100354
If you read the footnotes, they used the 4.0 GHz Snapdragon X Elite X1E-80-100 SKU in the Surface Laptop 15 for this benchmark.
The interesting thing is that, it means Qualcomm's top end 4.2 GHz (which is 5% faster than the 4.0 GHz) should have identical ST GB6 to Strix Point.
Yeah, Qualcomm's fine.
And I noticed this slide from AMD: https://images.anandtech.com/doci/21419/AMD COMPUTEX CLIENT PRESS DECK-01-01 (36).png
View attachment 100357
Your chip has the advantage of not needing emulation, and while running native is only that faster than your competition hindered by emulation?
Not good right?
All of those have ARM64 binaries... so noAnd I noticed this slide from AMD: https://images.anandtech.com/doci/21419/AMD COMPUTEX CLIENT PRESS DECK-01-01 (36).png
View attachment 100357
Your chip has the advantage of not needing emulation, and while running native is only that faster than your competition hindered by emulation?
Not good right?