Qualcomm's Nuvia SoC rumours - Hamoa 1st Nuvia based SOC in 2024

exoticspice1

exoticspice1

Site Champ
Joined
Jul 19, 2022
Posts
381
www.windowscentral.com

Qualcomm’s next-gen Nuvia-based CPU claimed to have 12-cores and “extremely promising” performance

Is Qualcomm’s Nuvia acquisition going to be as powerful as they expect? One report says it’s “extremely promising.”
www.windowscentral.com www.windowscentral.com

Looks like a 2024 release date. It will likely be on par with M3. dGPUs are also supported.


https://www.Twitter or X not allowed/i/web/status/1590024193927938048/

The leaker also says that early projections are:
"little performance preview (keep in mind that these are super early projections and subject to change): ST > M1, MT > M1 Max.
 
Hopefully it will live up to the hype. Will be certainly interesting if it does (though the more relevant point of comparison for them will be performance and price relative to Intel/AMD at the time as well as the state of Windows on ARM).
 
exoticspice1 said:
[

The leaker also says that early projections are:
"little performance preview (keep in mind that these are super early projections and subject to change): ST > M1, MT > M1 Max.
Click to expand...

Well, duh. I sure hope so. Otherwise it would be extremely embarrassing for them.
 
exoticspice1 said:
[

The leaker also says that early projections are:
"little performance preview (keep in mind that these are super early projections and subject to change): ST > M1, MT > M1 Max.
Click to expand...

Well, duh. I sure hope so. Otherwise it would be extremely embarrassing for them.

P.S. I think it’s odd to boast about 12MB L2 cache per cluster when M2 already has 16MB L2 per cluster… but now I’m just being grumpy for no reason 😁
 
New leaks concerning Qualcomm's "Homoa", the alleged M-series competitor. Check out this Twitter thread.
 
leman said:
P.S. I think it’s odd to boast about 12MB L2 cache per cluster
Click to expand...

And even stranger to have an 8Mb L3 – although, POWER10 has some kind of unique strategy that just builds the L3 out of L2 cross-linking, so maybe they have something of that sort in the works. Also, it looks like they are punting to dGPU, so a lot of SoC complexity is off the chip.
 
Back
Top