The thing says 2x faster CPU/GPU vs x86 equivalent.....X Elite specs apparently.
(FLOPS should be TFLOPS.)
X Elite specs apparently.
(FLOPS should be TFLOPS.)
4.6 TFLOPS is a great number for integrated graphics. More than a plain M2, but less than a pro. Better than current gen Intel/AMD integrated graphics.
Raw TFLOPS is obviously just a small part of graphics performance, but it does show some potential, especially if coupled with the full 136 GB/s memory (and enough of it).
Marketing. But that being said, it's probably competitive.The thing says 2x faster CPU/GPU vs x86 equivalent.....
Say what now?
Even AMD Phoenix Point should be pretty close to that GPU already, and Strix Point 16 CU will curb stomp it, and I don't think Zen5 will be anywhere near 2x slower than Oryon either.
I would say likely so, but it's immaterial at the moment considering the current state of WinARM software, as binary translation of x86 apps will exact a significant toll on whatever the performance of Oryon happens to be unless Nuvia managed to do a M1 type x86 -> ARM maneuver in their first gen µArch to negate some of the bintran impact.Marketing. But that being said, it's probably competitive.
I'd say that AMD do have ARM plans on slow burn, but very much conceptual/early design stage vs full throttle.Intel and AMD should be preparing their own WinARM chips if they want to remain relevant.
I've been thinking for a while this is inevitable. While x86 will stay on top for Windows for a few more years, WinARM is going to keep growing until it's an equal citizen. Then MANY more players will be able to compete for the Windows SoC/CPU business.
Keep in the this is Qualcomm's higher end chip. The M3 is Apples base end. The M3 Pro should still have a better GPUHmm, interesting. Obviously, 12 HP cores should be 8P+4E???
Adreno GPU with 4.6TFlops indicates 1536ALU with 1.5GHz. Then GPU performance should be faster than upcoming M3 unless Apple clocks their GPU higher...
Marketing. But that being said, it's probably competitive.
I am afraid X Elite would be Qualcomm's mainstream SoC for PC platform in the same price range as upcoming M3-based notebook. This is due to the usage of 128-bit memory bus; Qualcomm able to squeeze 8P+4E cores and 1536ALU GPU within 128-bit LPDDR5x memory bus is pretty impressive. Of course, Apple is partly blamed here due to their product segmentation strategy, I am expecting M3 to feature same 4P+4E cores and 10 GPU cores (1280 ALU) with LPDDR5 support. Maybe Apple think the inclusion of ray tracing support is enough to warrant the upgrade which I think would be missing in X Gen1 series...Keep in the this is Qualcomm's higher end chip. The M3 is Apples base end. The M3 Pro should still have a better GPU
Who is excited for Snapdragon Summit 2023?😃
Its 12 GPU cores for M3.I am expecting M3 to feature same 4P+4E cores and 10 GPU cores (1280 ALU) with LPDDR5 support.
John Bruno, Vice President of Enginnering at Qualcomm.
Sure. It would be great if we finally got competitive third player for Windows ecosystem.Who is excited for Snapdragon Summit 2023?😃
😄Who is excited for Snapdragon Summit 2023?😃
The thing says 2x faster CPU/GPU vs x86 equivalent.....
Say what now?
Even AMD Phoenix Point should be pretty close to that GPU already, and Strix Point 16 CU will curb stomp it, and I don't think Zen5 will be anywhere near 2x slower than Oryon either.
Ironically there is a RISC-V based GPU out there.Damn, I forgot GPUs run on x86. Knight's Mill (or whatever it was) from Intel really took off didn't it?
Ironically there is a RISC-V based GPU out there.
No idea, I think the one I was thinking of was more of a research project.I assume you're referring to the Think Silicon IP. It's very low end (and I've only really seen their IPs in one merchant product, the Ambiq Apollo4.)
There are lots of coincidences in life, this isn't one of them.John Bruno, Vice President of Enginnering at Qualcomm.
Or is it just a coincidental nick?
Yes.yo is that the real John Bruno and Gérard Williams? o_0