What does AI have to do with laptops?No, it means absolutely nothing outside of premium (>$999) laptop space.
PC means laptop those days.What does AI have to do with laptops?
... where would be best to place {NPU}? On the CCD or IOD?
Call it what you like. You won’t be able to use all Win12 versions without it. That’s a huge drawback, especially from marketing perspective.It's fully Win12 compliant, just not for premium(tm) AI PCs.
Technically possible. Could even be that they’ll sell dedicated NPUs as PCIe expansion cards.Does this mean cheap low profile PCIe GPUs with decent AI acceleration for OEMs so they can advertise their desktops as Premium AI PCs?
No one cares, that marketing is only ever relevant for laptops.You won’t be able to use all Win12 versions without it. That’s a huge drawback, especially from marketing perspective.
Arrow Lake could possibly get it if they are able to squeeze one onto it.But a much better and cheaper option is to just include the NPU on the CPU chip.
yeah twice the FMA munch-munch.and Zen 5 is presumably going to "optimize" that feature set in one way or another.
FMA3 you mean?yeah twice the FMA munch-munch.
I mean anything with AVX2 support has 3-operand FMA support.FMA3 you mean?
Yeah.Or Zen 5 will execute two 512-bit FMA per cycle?
Oh weeeeeeee!!!!!!Yeah.
FADDs too.
Are you … are you an AI post?Microsoft is turning itself over as a corporate raider. The AI tools will be used to size people up regarding productivity. It will identify the definition of productivity, decypher how these high performers function, then attempt to replace them synthetically through software. But when 75% of the workforce is clowns, it really will emulate clowns in the end. It will always be hampered trying to emulate creative people because when push comes to shove the real talent will simply pack up and leave. Bring in the clowns!
On a more serious note, your current trend is multi-processing through parallelism. Microsoft already has mapped out every workplace and home for 'efficiency purposes', helping itself to your bandwidth, storage, computing resources, and electricity. If user A wants AI to work on a problem then the PC will soak up neuroprocessing capacity on nearby machines. Chances are it can find a lot of idle power in any workplace or home. And it will convenientlystealborrow from others it identifies around it irregardless if its user A's secure network or not. Because it can until regulation says it cannot.
Idk... quite a funny and Orwellian read.Are you … are you an AI post?
Only in a very-very narrow subset of math-bound FP apps.I think FP will be a larger IPC gain than INT.
Zero leaks about FP perf still.
Makes sense, considering the biggest markets for CPUs are heavily INT/frontend bound. I do wonder which version of Turin has the biggest gains when compared to said versions of Genoa?Only in a very-very narrow subset of math-bound FP apps.
My 2600k system had a TPM, so did my Pentium G2030. I don’t remember if they were TPM 2.0 (probably not) but you can buy a TPM 2.0 module for under $20.Yeah, MS happily left alot of users behind with that. So, they are still all set to 2025. Kind of wish I didn't move over to Win11, but some family members wanted it - so the family tech support technician upgraded . It's not terrible, and I actually like the UI changes. Hopefully 12 will be better.
It kind of freaked me out. I was working on a project in Visual Studio 2022 the other day and it wrote an entire large block of code for me. The code was a loop to load/parse data in a proprietary file format. I thought for a moment that Microsoft had added copilot to visual studio or something.More off topic, Visual Studio 2022 already have AI built-in even without Copilot. It analyses current project and does help a lot when programming. It can understand your coding style and able to do code suggestion and completion accordingly. After getting used to it, it is bit annoying to work without it.
I guess these types of workloads can be accelerated by built-in NPU. Also I remember AMD saying that all future CPUs will have some sort of AI. Not sure about Zen5 though.
Turin-D vs Bergamo.I do wonder which version of Turin has the biggest gains when compared to said versions of Genoa?
I mean they're just not public.The lack of Zen5 news lately is disheartening.
This is one is actually new and interesting since Zen5 is very set in stone.Bizzarely there was more talk about Zen6, which is like ages away
I heard that there was 2 teams, one that handles the Zen, Zen 3, Zen 5 and the other that handles Zen 2, Zen 4 & Zen 6. Is this true?This is one is actually new and interesting since Zen5 is very set in stone.1
No, AMD doesn't have a fixed definition of a "team" the way Intel does.Is this true?
No, there has been some continuity between Zen 1, 2 and again 5 (and likely 6), as well as K12 ARM, Zen 3 and 4. But they are not fixed teams in any sense.I heard that there was 2 teams, one that handles the Zen, Zen 3, Zen 5 and the other that handles Zen 2, Zen 4 & Zen 6. Is this true?