IPC:30% what
Clocks:Yea.
32% SPECint rate n IPC aka how AMD tends to guide their pre-PRQ'd server parts (and how they've measured Zen4 14% IPC, kinda, see their HC'23 Zen4 session endnotes).
Mobile like STX will still boost to upper 4s/lower 5s GHz.
Love to hear how it's betterBut still better than the stupid Zen 5.
ughhh. yeah brother.But still better than the stupid Zen 5.
Most amazingly Zen 5 will have SMT4 and CMT!At this point of hype train I expect that Zen 5 architecture will dance and sign, apart from being able to put on make up.
Cost increase?30% what
if they are going to be as dominate a position as I/We think they are , we can only hope they are so kind ......Cost increase?
Like 15-20% aka the usual.they are saying 'Lion Cove' which is in Arrow Lake which is to launch this fall has 20-25% IPC improvement
Because whatever Intel has is nowhere near enough.Why worry about price increase of Zen 5 then.
Shame that the IPC uplift is getting nullified by the clock regression.Like 15-20% aka the usual.
Because whatever Intel has is nowhere near enough.
No it's stock.The leaked 125W is the maximum configurable TDP I believe?
It's not made for existing laptops at all.The issue with 125W is that it's too high to fit in some existing laptop designs (even ones with dGPUs) like the XPS 16 for instance, which tops out at just 80W device TDP.
It clocks the same or slightly faster than MTL so not really a regression per se.Shame that the IPC uplift is getting nullified by the clock regression
I know, I know. I was giving an example.It's not made for existing laptops at all.
Yea.I know, I know. I was giving an example.
Those can have 240W now, glory be USC-C PD EPR.I'd like to see some Dell XPS-like Ultrabook laptops with Strix Halo, but if the TDP is 125W, then it isn't fit for an Ultrabook, which have a total device TDP of 100W at best.
The point was not the power supply but the cooling capacity.Those can have 240W now, glory be USC-C PD EPR.
No you just stick a vapor chamber and roll.Once you go past 100W TDP, you are going from an Ultrabook to chonky gaming laptop territory.
Sure sure, but it’s a 20% regression relative to Raptor Lakes 6 GHz, thus ST performance isn’t much different than Raptor Lake overall. Power consumption ought to be better due to lower clocks and better process of course, but in the desktop space consumers are already conditioned to 170W TDP. The appeal of Raptor Lake performance in a 135W TDP pales in comparison to +30% at the same 170W.It clocks the same or slightly faster than MTL so not really a regression per se.
It's not really 20% at all.Sure sure, but it’s a 20% regression relative to Raptor Lakes 6 GHz
High single digits better.thus ST performance isn’t much different than Raptor Lake overall
I guess you're right in that Intel HAS to market some kind of advantage of ARL over RPL, otherwise what's the point?It's not really 20% at all.
You're still getting your 1t wins, they're just not huge.
High single digits better.
Not good enough in this comp env but it is what it is.
no you don't.I'm totally cool with AMD gaining more market share for a few generations only because I know that in the grand scheme of things
You have to understand that Intel maintains their MSS by spending money in opex.the Intel-AMD market share needle is still very far from 50-50
AMD's 15+% IPC vs "The second Conroe"Aaaaah... it begins, the cycle of over expectations and forced disappointment when AMD "betrays the promises".
Apple already has done magic with their humongous ROBs.A radically different way to built the ROB buffer? That would allow a way bigger OoO windows in some scenarios (there have been some papers on that).