sometimes i feel as though we repeat this a lot but here we go again
Ehhhh.
Depends on the case. For example in 9950x case the gpu will be slower than the cores
but if you have big enough matrix and big enough gpu, then for sure.I guess the question I am posing is: Will next gen iGPU's be powerful enough that they render APX pointless?Depends on the case. For example in 9950x case the gpu will be slower than the cores
AMX
Depends on the SKU i guess. I wouldn't count on that for direct desktop chips replacements, the igpu there is not meant for anything more than displaying desktop and handling hw video decode. The premium mobile parts are supposed to be GPUs with attached CPUs, so there for sure
I suspect that AMD will simply (begrudgingly) price Medusa Premium (with AT4) a bit lower than originally planned if they have to, and that's that.
No, it's a discrete compete part at discrete prices.
Cost cost cost cost cost.
Graphics wank is irrelevant.
PTL is a meme. Forget that it exists.
Medusa Points are gonna be officially launched at CES 2027. With a refresh it's gonna last until Zen 7 APUs' arrival. That's H2 2028 or CES 2029.
That's twice in a week I am forced to adamantly agree with you adroc
.I think that just about everything will be cheaper to make than PTL.
Yep. They kept VEGA around forever too
. I think it really comes down to "its good enough for most users".You have a relatively large SoC. Let's say 150mm2 in N3P. Adding a little bit of Die area for potentially 2x gaming performance (or use less CU with the same performance to save Die area) and including RT and FSR4+ much more than 2x would be a decent win in my opinion. And then the super duper mega important topic of DNN acceleration, where RDNA5 is much more capable than RDNA 3.5. Isn't AMD an AI company now
You think like a consumer.You have a relatively large SoC. Let's say 150mm2 in N3P. Adding a little bit of Die area for potentially 2x gaming performance (or use less CU with the same performance to save Die area) and including RT and FSR4+ much more than 2x would be a decent win in my opinion. And then the super duper mega important topic of DNN acceleration, where RDNA5 is much more capable than RDNA 3.5. Isn't AMD an AI company now
Basic engineering principle "You don't get something for nothing".You have a relatively large SoC. Let's say 150mm2 in N3P. Adding a little bit of Die area for potentially 2x gaming performance (or use less CU with the same performance to save Die area) and including RT and FSR4+ much more than 2x would be a decent win in my opinion. And then the super duper mega important topic of DNN acceleration, where RDNA5 is much more capable than RDNA 3.5. Isn't AMD an AI company now
).
amd stopped discrete gpus for lsptops.Basic engineering principle "You don't get something for nothing".
I can't damn or praise AMD's product mix decisions. I would need lots of information I don't have. I can't deny your claims, but point out that gamers don't account for the majority of sales OR the majority of profits. Not even close. They are a sliver.
AMD has obviously decided that the gamers product would need discrete GPU (which they also sell .... imagine that
In all fairness, 6Ghz for Zen 6 seems quite achievable. 7Ghz for Zen 7 feels more like a pipe dream for several reasons:
wrooooooooong it's a full node shrink.
lmao dawg AMD gives squat about Intel.
It's no different than N4 was from N5. Not a full node shrink, just a refinement of the N2 process that is possible due to process maturity and equipment tweaks to N2.
Of course not. They also don't care about yields or profit. They only "give a squat" about providing the max clock speeds they can attain on a given process and cpu design
.
would be nice but don't see them going from 32GB on consumer straight to 64GB.
maybe not on the discrete gpu (50xt)
Facebook
Twitter