Everyone does, which is why AMD likes to keep their 1t bumps at >25% per core update.You're incorrectly assuming everyone just cares about ST perf and not MT perf.
Because that's all there it so it.not sure what you mean by "just 8161"
random ebeggars aren't sources.Instead instead see this, where several sources mention 8P + 32E for ARL-S.
It doesn't matter, it'll be much faster at 1t than anything else on the market, period.If AMD releases an 8950X with just 16C at $999 or even $699
Do you even understand what Zen5 is?Just like they had to with 7950X which was released at $699, but soon after release had to be dropped to $569 when market reality hit them in the face.
Eh it can be quasi-relevant but not in your average desktop load.Memory bandwidth is irrelevant for 1t and CB*
Cinebench is irrelevant, if it were truly the metric to go by then 7600X/7700X/7800X3D would all be DOA (which is clearly not the case).Memory bandwidth is irrelevant for 1t and CB*
Specint & Specfp (1-copy) can swing 5-7% from better memory.Man we're so back to 2006.
Unbelievable.
Zen 3 didn't outperform RKL in ST. it was pretty close.Hold on, you're saying that Zen 5 will outperform ARL by a wider margin than Zen 3 outperformed RKL?
Yes.Ok, so "everyone" only cares about 1T performance?
It's a reused server CCD so they really could not even if they wanted to.Then why not release 8950X as a 1C CPU?
None of the things you're quoting are official and I certainly won't ever share Intel's internal use slideware with PRQ dates no less.if you have any official sources confirming ARL-S will be max 8P + 16E as you claim, please provide links to those.
Well that's the only thing Intel cores are good at.Cinebench is irrelevant
Well, they kinda are, outside of 7800X3D.if it were truly the metric to go by then 7600X/7700X/7800X3D would all be DOA (which is clearly not the case).
They can also swing whatever percentage up from better cache.Specint & Specfp (1-copy) can swing 5-7% from better memory.
Cinememe isn't really ST per se.Zen 3 didn't outperform RKL in ST
It did, despite roughly 300-400MHz disadvantage, at Computerbase they have Zen 3 slightly faster in ST than the 5.3GHz higher clocked RKL.Zen 3 didn't outperform RKL in ST. it was pretty close.
LNL is Lunar Lake, what is CWF?there's LNL-M (AMD doesn't compete with that directly) and CWF (which looks kinda promising I guess).
Cleanwater Forest, SRF's successor for 2025.what is CWF?
how big? 128MB?Just a bigger CCD and yea, faster caches too to feed the creature.
And the other 1/2 of the sentence you quoted .....Cinebench is irrelevant, if it were truly the metric to go by then 7600X/7700X/7800X3D would all be DOA (which is clearly not the case).
The same would be true of Zen 5 in that case.Well, they kinda are, outside of 7800X3D.
13600K(F) is just a better deal.
I won’t lie, it’d sting if ARL-S only brings a 1-3% ST uplift over RPL-R. I’d probably just skip that generation completely in that case.They can also swing whatever percentage up from better cache.
I know you want to have hope, but hope is dead.
If you want to talk Intel parts, there's LNL-M (AMD doesn't compete with that directly) and CWF (which looks kinda promising I guess).
Same as Zen4, so 32MB L3.how big? 128MB?
No.The same would be true of Zen 5 in that case.
Why, Zen5 is very fast, unlike Intel slop.I won’t lie, it’d sting if ARL-S only brings a 1-3% ST uplift over RPL-R. I’d probably just skip that generation completely in that case.
Good, very expensive.Turin
Stillborn.
Beyond stillborn.
Well it's just -S and one -H tile and they're all bad.
Bleeding edge nodes are whatever those days.There’s a new node from TSMC (N3E) as well as Intel 3
Exciting for $AMD shareholders and no one really else.It’s shaping up to be an exciting year from an enthusiasts standpoint.
I am not saying i would prefer slower 24C over faster 16C.If you could achieve the same performance with 16 threads as with 24 threads, by having higher single thread performance, the comparison between these CPUs would not even be close. The 16 core one, with hither single thread performance would be far more valuable.
So what is it, again?Do you even understand what Zen5 is?
Exciting for $AMD shareholders and no one really else.
Well it's just -S and one -H tile and they're all bad.
It's a bigger jump in ST than Zen4 and there's no competition.So what is it, again?
Cause unless that 30 percent perf increase is some kind of final piece toward achievement of technological singularity or something along the line, i dont see how it warrants 30 percent increase in price.
Especially if Zen4 provided similar speed bump while actually being cheaper than previous gen.
Not where it matters.Especially if Zen4 provided similar speed bump while actually being cheaper than previous gen.
Who even said that.Cause unless that 30 percent perf increase
I've never said -S was just one tile.AFAIK there were 2 tiles for -S configurations.
eh.I still find new hardware exciting, irrespective of vendor.
Specint & Specfp (1-copy) can swing 5-7% from better memory.
They have no Plan B compute dies on TSMC process? How is this possible? They were supposed to take their 10nm lessons to heart. Actually, I think they were REQUIRED to do that by their investors.Well it's just -S and one -H tile and they're all bad.
All but one are TSMC, kek.They have no Plan B compute dies on TSMC process?
They did.They were supposed to take their 10nm lessons to heart.
They did.
Foundry doesn't look like an issue for Intel per se.
That's an IDC skill issue.You can see that with GC, its performance isn't bad , but how it gets it performance looks quite bad when compared to others, big physical core size , big internal structures , quite wide, YOLO watts per core.
Kinda shocked to see you say that. Figured you’d say it was DOA, hopeless, stillborn or something. That matches what I heard as well but the people I talk to are adamant that the upcoming products won’t be as big of a disaster as your prophesying so who knows. For example, they’re 100% sure GNR will outperform Genoa (which you’ve said it won’t).They did.
Foundry doesn't look like an issue for Intel per se.
That's an IDC skill issue.
Nothing wrong with i7 perf/power wise.
Well foundry is the only place at Intel where any sort of accountability exists.Kinda shocked to see you say that
Intel product guys are the lolcows of this industry.I talk to are adamant that the upcoming products won’t be as big of a disaster as your prophesying so who knows
18A already lost Qualcomm so I gotta temper my optimism a bit.The 18A node is going to be a tailwind that you’re not taking into account
At 500W it better do.For example, they’re 100% sure GNR will outperform Genoa
Well, I was hoping to quote everything, but, What I am trying to say is that at 500 watt, unless GNR totally demolishes Genoa in performance and supports avx-512, its doa. My 9554 does great at avx-512 at 320 watt. and works not much below the 7950x. Its like 2.7x a 7950x (in productivity ) and runs at 3.5 ghz at 100% load of 128 threads. The 7950, even at 142 watt is very close to the 9554 in perf/watt. And a dual 9554 holds quite a few world records, not even considering power draw.At 500W it better do.