Exactly this.A 12 Core part is obviously more efficient than an 8 Core part, because the load is distributed over a higher number of Cores which can clock lower. So you can't take this and say "ZEN5 is x% more efficient than ZEN4. We need to wait for July 30th when 3rd party Reviews should drop.
Much cheaper than N3B/E.
Flavors less important than the base node. It isn’t N6 cheap, but for the premium market the N4 family is probably the highest bang for your buck right now. N3E plausibly better depending on what you’re doing and with enough volume, but like N4 and N4P are being used in midrange or entry level premium smartphones now, so.True but still not cheap.
People have forgotten everything, but on purpose because there are no other options.This is hilarious to watch.
Do none of you remember the slow creep up of IPC from Sandy Bridge to Skylake, and the complete stagnation after that for years? 15% increase is fine, good grief.
It's OK, but not good considering it's launching pretty close to 2 years after Zen 4. Will be lucky to maintain the status quo against Intel's next gen and ARM CPUs are definitely going to eat some of AMD and Intel's lunch.This is hilarious to watch.
Do none of you remember the slow creep up of IPC from Sandy Bridge to Skylake, and the complete stagnation after that for years? 15% increase is fine, good grief.
It is good. Excluding the hype, many of us were expecting a larger increase as Zen5 should have been a tock (every other zen iteration). It's not so compelling going from Zen4->Zen5, but ppl (such as me), on Zen2/3 will be seeing substantial gains when upgrading to Zen5 and AM5.This is hilarious to watch.
Do none of you remember the slow creep up of IPC from Sandy Bridge to Skylake, and the complete stagnation after that for years? 15% increase is fine, good grief.
No, Geekbench is fair too. Wider code footprint in some ways? Both are fine. No diss to Specint tho.
View attachment 100347
Bruh?
Look above — not Yo/Y but this year. And Arm has been barreling forward for a while now. X4 was a 10-12% IPC improvement and really close — mid single digits — to Firestorm IPC, closer than Zen 5 for sure. X3 was around 5-8% gains I think and even then was doing rouuughly mobile Zen 3 ST at 5W total power in 8 Gen 2 or 9200, and ofc at less frequency.
It’s honestly strange people are so ignorant about Arm’s cores.
Matching Apple is a high bar on IPC, but even more so with power as the X4 showed, it’s true they’re not A+ cores and probably Qualcomm will have some power leads I can see too.
But “even” being a Firestorm Walmart/generic core like the X4 was (hitting 1600-1690 GB5 for ex, or at 2300-2350 in GB6) with worse power characteristics than Apple — but still doable in phones — is self evidently better than what AMD/Intel could offer in PCs for a similar perf.
The sneering about Arm’s Cortex IP is fair vs Apple and arguably QC, but not vs AMD and Intel anymore save probably some server stuff.
~15% is in line with my expectation, but there were few here boasting >30% and at $999. I wonder these AMD hype is done on purpose? This tend to happen every AMD release. For me the real disappointment is, this is the 4th generation with 16 cores. Time to move on from it.This is hilarious to watch.
Do none of you remember the slow creep up of IPC from Sandy Bridge to Skylake, and the complete stagnation after that for years? 15% increase is fine, good grief.
I do think it’s partially being pumped by the 50% increase in SIMD registers to be fair, but that’s still quite valid. And only partially.Their own SPECint numbers show much lower IPC. GB6 is not a bad bench, but there are multiple reasons SPECint is considered the standard as far as generalized benchmarks go. If SPECint is the standard for everyone else, it spoils be for Arm as well.
Not new for Zen either. At the end of the day Zen 2 being 300MHz slower than hyped never killed anyone.Well, well, well, my friends.
The Zen5 hype train crashed spectacularly.
I do think it’s partially being pumped by the 50% increase in SIMD registers to be fair, but that’s still quite valid. And only partially.
They had like a 34% IPC lead in SpecInt over Phoenix with the X4 (8 Gen 3).
View attachment 100351
View attachment 100352
And I don’t think AMD got 16% IPC in SpecInt 1T did they with Zen 5? Was maybe slightly higher than Arm’s, like 10% vs 5-6%.
225mm² for Strix Pointshould be 220-230mm^2?
I'm still waiting for confirmation (Xino), but rumor mill says that Kraken isn't that smaller than STX actually. Circa 190 - 200mm² somehow.But they’ll have Kraken (4+4) as a separate part later to get the entry premium and cut down on that area.
And we all know what happens with AMD Hype Trains.
I guess it comes from being promised 30-40% IPC increase .This is hilarious to watch.
Do none of you remember the slow creep up of IPC from Sandy Bridge to Skylake, and the complete stagnation after that for years? 15% increase is fine, good grief.
You gotta look who is promising next time.guess it comes from being promised 30-40% IPC increase .
Don't believe the rumors spread by... Intel engineers simply.You gotta look who is promising next time.
I think there is a level of justifiable disappointment, though, because it is the weakest Zen yet regardless of the hype train.
Yeah. For example that’s 32% more area than SDXE (170mm^2 ish) which has 12C of similar performance, almost certainly much better battery life, similar NPU power, similar IO and media engine (actually tbf AMD likely has an encode lead on frames?). GPU though with RDNA3.5 is a raw perf and compute win for AMD and worth same area except I’m not sure how well it’ll scale down.225mm² for Strix Point
Yea. Well, what matters also is what they charge for it. Markups for the top parts usually higher.I'm still waiting for confirmation (Xino), but rumor mill says that Kraken isn't that smaller than STX actually. Circa 190 - 200mm² somehow.
hmmm.NPU is the same, so that's a fixed area cost to be paid. But KRK sports 4 Zen 5 cores with 4 less Zen 5C cores and a half-wide GPU. So something doesn't add up.
3700-4000 GB6 1T around 11W platform is honestly ludicrous. I will bet everyone here Lunar Lake comes in 2800-3200 range at peak and uses more power than that by probably a factor of 50% if not 100%.
Apple M4 supports the SME, but all other platforms do not. As a result, you can't compare those scores. That's what Geekbench says in their release notes for 6.3. From my perspective, GB6 is a terrible benchmark. The previous version is closer to the truth.
I think AMD Ryzen HX 370 will be close to M3 Max in terms of performance and power consumption.
Well, AMD/Intel also get a boost in that subtests due to AVX-512, so if you want to exclude Apple SME from the comparison, you'll have to exclude AVX-512 from x86 parts to keep the comparison fair.Apple M4 supports the SME, but all other platforms do not. As a result, you can't compare those scores. That's what Geekbench says in their release notes for 6.3. From my perspective, GB6 is a terrible benchmark. The previous version is closer to the truth.
I think AMD Ryzen HX 370 will be close to M3 Max in terms of performance and power consumption.
Well, AMD/Intel also get a boost in that subtests due to AVX-512, so if you want to exclude Apple SME from the comparison, you'll have to exclude AVX-512 from x86 parts to keep the comparison fair.
Intel has no avx-512 from alder lake on. Unless you are talking about server chips. And excluding avx-512 is fair ? Not for those of use that use it, and need it.Well, AMD/Intel also get a boost in that subtests due to AVX-512, so if you want to exclude Apple SME from the comparison, you'll have to exclude AVX-512 from x86 parts to keep the comparison fair.
Intel has no avx-512 from alder lake on. Unless you are talking about server chips. And excluding avx-512 is fair ? Not for those of use that use it, and need it.