• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD Zen 6 CPU Specifications Leaks – Big Boost Unveiled

It is actually something I've wondered about for hybrid core designs: Could they make an X core that was big, hot, and inefficient, but the sort of thing you toss a heavy game thread on. Realistically the limit is that core design is costly and there wouldn't be the reason to design a core just for that (the E cores Intel uses are in wide use in other CPUs) but I like the idea.
I don't think you'll see this on desktop/laptop much. But some phones have 3 core types: one, maybe two for big stuff, 4ish medium cores, and two small cores for background/low power stuff.

If AMD were to do a three tier chip, it would probably be a new smaller, slower, cooler wimpy tier rather than a bigger, faster, hotter top tier. ZenC is all about the design efficiency you get with a hard ceiling of ~ 4 GHz instead of 5+Ghz. You might get additional design efficiency if you limit clocks to 2 GHz? Maybe not that much though.
 
I don't think you'll see this on desktop/laptop much. But some phones have 3 core types: one, maybe two for big stuff, 4ish medium cores, and two small cores for background/low power stuff.

Panther Lake laptop has 3 core types and the upcomiing desktop version has well (P, E, LP-E).

Has for putting 2 really big L1-L2 cache core for games main loop and render thread or something, maybe they prefer using extra silicon on the gpu instead when it come to console, consistency is valuable for modern game engine job system, they will take the small core for the os-background core heteroginity instead .
 
I don't think you'll see this on desktop/laptop much. But some phones have 3 core types: one, maybe two for big stuff, 4ish medium cores, and two small cores for background/low power stuff.

If AMD were to do a three tier chip, it would probably be a new smaller, slower, cooler wimpy tier rather than a bigger, faster, hotter top tier. ZenC is all about the design efficiency you get with a hard ceiling of ~ 4 GHz instead of 5+Ghz. You might get additional design efficiency if you limit clocks to 2 GHz? Maybe not that much though.
Ya it was more Intel I was thinking about since they already have lorge, smol, and very smol cores on their chips. My thought was "What about extra lorge?" Like an X core that is the size of two P cores and isn't even used for normal multi-threaded, apps, just things like games that really peg out a core and you'll take the increased heat and slower speeds on the other cores for having the one massive core really fly.
 
Ya it was more Intel I was thinking about since they already have lorge, smol, and very smol cores on their chips. My thought was "What about extra lorge?" Like an X core that is the size of two P cores and isn't even used for normal multi-threaded, apps, just things like games that really peg out a core and you'll take the increased heat and slower speeds on the other cores for having the one massive core really fly.

Yeah, I think the chonky cores are already pretty close to max chonk. Been a while since I looked, but Apple cores had more execution units, but ARM memory model gives you more room for OOO execution and lower clockspeed makes it easier to feed from memory and makes clocking across the core easier.
 
Ya it was more Intel I was thinking about since they already have lorge, smol, and very smol cores on their chips. My thought was "What about extra lorge?" Like an X core that is the size of two P cores and isn't even used for normal multi-threaded, apps, just things like games that really peg out a core and you'll take the increased heat and slower speeds on the other cores for having the one massive core really fly.

The problem I'd think then is as certain things grow/come about/need less/more compute etc - what's (needed to be) big and what's small would keep shifting back and forth (like workloads that maybe benefit from multithreading etc).

Better to just make (for the most part, overall) big general purpose cores, shift the workloads and power up/down 'from behind' as AMD is supposedly doing. Maybe playing around with this can also help them with low idle/low usage energy consumption, which they still have issues with against Intel (and against Nvidia in GPUs).

PCs are not phones and phones are not PCs, as close as they may be. Hence why even Intel is looking into unified cores again.

1785709720686.png


https://www.techpowerup.com/346645/...no-more-performance-and-efficiency-core-split
 
Last edited:
Yeah, I think the chonky cores are already pretty close to max chonk. Been a while since I looked, but Apple cores had more execution units, but ARM memory model gives you more room for OOO execution and lower clockspeed makes it easier to feed from memory and makes clocking across the core easier.
Even if they aren't, it wouldn't be worth the cost. Core design is expensive. It isn't like you sit an intern down and say "Hey that P core? Make one like that but bigger, have it done by Friday." It takes a lot of work. With the P cores and E cores, it makes sense, because the work is already being done. E cores were made for Atom processors and also for certain Xeons, and P cores for desktops and other Xeons. So since you already have two designs, it isn't huge to knock them both on one chip. However designing a new core just for gaming... that's a lot of money and realistically I can't see it being worth it since it would probably only really be useful in that one situation.

PCs are not phones and phones are not PCs, as close as they may be. Hence why even Intel is looking into unified cores again.
We'll see. I've heard this rumor before and they keep messing with the different sized cores. The reason isn't even so much lower power usage but just how things scale. You can fit about 4 E cores in the space of 1 P core, however a P core is not 4x the power of an E core. So if you have a task that splits in parallel really well, E cores get you more performance per die space. There are actually all-E core Xeons out there for precisely such workloads.

Maybe they'll go back to all-P cores for desktop CPU, but their next gen doesn't look like it. I'll believe it when I see it.
 
Maybe they'll go back to all-P cores for desktop CPU, but their next gen doesn't look like it. I'll believe it when I see it.
It is more (at least the rumoured trajectory) an all e-cores for desktop and laptop CPU that is upcoming from Intel, the e-core won the competition battle and will be the core design of the unified upcoming core.

And Nova lake will be a big first step apparently, coyote cove (p-core), artic wolf( e-core) well being different design will have full ISA parity, making dev and scheduler work easier (https://www.techpowerup.com/350572/...ture-avx-512-on-both-p-cores-and-e-cores?cp=3), once e-core can do everything, you make a wider-more cache version and smallers one like AMD.

However designing a new core just for gaming... that's a lot of money and realistically I can't see it being worth it
For a sony it would make some sense, if design cost ever get down and gaming is not the single task where a couple of threads tend to be the bottleneck (many CADs program are like that), I imagine the current big core are already being near the limit of diminishing return in term of scaling cores bigger and the memory bandwith they have access too, while system with giant memory bandwith that could feed something bigger do tend to love core counts and them being predictable-equal.
 
For a sony it would make some sense, if design cost ever get down and gaming is not the single task where a couple of threads tend to be the bottleneck (many CADs program are like that), I imagine the current big core are already being near the limit of diminishing return in term of scaling cores bigger and the memory bandwith they have access too, while system with giant memory bandwith that could feed something bigger do tend to love core counts and them being predictable-equal.
for consoles, maybe, memory access is more important than core speed ?

if one thread goes fast & other threads are lagging will that result in tearing / stutters 🤔

in this particular zen 6 case, this is about 1% lows (especially on server side). may not be much financial incentive for Sony to boost 1% lows, maybe 🤔
 
The problem I'd think then is as certain things grow/come about/need less/more compute etc - what's (needed to be) big and what's small would keep shifting back and forth (like workloads that maybe benefit from multithreading etc).

Better to just make (for the most part, overall) big general purpose cores, shift the workloads and power up/down 'from behind' as AMD is supposedly doing. Maybe playing around with this can also help them with low idle/low usage energy consumption, which they still have issues with against Intel (and against Nvidia in GPUs).

PCs are not phones and phones are not PCs, as close as they may be. Hence why even Intel is looking into unified cores again.

View attachment 818306

https://www.techpowerup.com/346645/...no-more-performance-and-efficiency-core-split
Intel initially split the cores to meet EU and California idle power consumption requirements. But the node shrinks and improvements to power delivery have likely gotten to a place where they probably don’t need them anymore.

AMD still doesn’t meet those requirements, which is one of the main reasons you see so few AMD processors being offered by OEM’s in business class desktops.

And I don’t think AMD even wants to, it’s a high volume, low margin market and AMD likely can’t afford to entertain it.
 
Last edited:
Back
Top