Krizby:

Yeah I remember AMD with all the "poor Volta" trash talk failed miserably when their best GPU in 2019 (Radeon VII) couldn't beat 2nd best Pascal from 2017 (1080Ti) LMAO, and Volta wasn't even a gaming uarch. Today AMD is behaving much more mature, they just keep quiet and release their products.

Although I haven't been following marketing materials and ads over the years as well as some people, but haven't AMD's CPU and GPU divisions behaved quite differently? The GPU div is trying to be more flashy because it markets more purely to gamers, I guess. Intel hasn't had anything but CPUs (in things relevant in this conversation), so I suppose it makes sense Intel would be bolder with their talks about the CPUs, even though a really big portion of CPUs never see a single game being run. Of course this year nobody wouldn't have needed to do anything to market graphics cards (which doesn't mean they haven't been marketed, it's just not really needed). Miners even stole trucks to their hands on them.

CEOs should be very careful about statements like that, it might bite them in the ass later...

I hope AD has at least 10% IPC and another 15% with DDR5 and PCIe 5 (~20-25%) with the same power consumption. When you consider all these things together, I expect them to be faster than Zen3D. Zen3 was supposed to come with 3d cache, so 3d-part is really like a band aid. And if Intel can't beat Zen3d, 1 year old arch,...

k3vst3r:

3D cache could appear on zen3+ which apparently pumps up performance by 15%, so if AL is only 10% zen3+ could potentially still take the lead again.

3D cache is ONLY for high end CPUs, 5950x and maybe 5900. Too expensive for everything esle

Horus-Anhur:

Intel without competition became stagnant and lazy. If it wasn't for AMD putting some pressure on them, we would still be stuck with 14nm++++, 4c8t for an i7, and no IPC gains for a whole decade. It's good to see that Gelsinger is able to turn Intel around. We need a strong Intel and a strong AMD, otherwise the market will stagnate again.

Intel had 10n on their roadmap and so is more core CPUs, do you think the 8700K came out because they needed a CPU to compete with AMD? in less then one year of development?

One day, Gelsinger might actually announce a firm ship date for Alder lake--where are the PCIe5 motherboards? Where is the DDR5? Gelsinger also forgot that AMD has had Zen4 under construction for many months now. He likes talking up Intel's future products, but hates to talk about AMD's...! Too funny. Gelsinger is sort of a clown these days. For someone who says he wasn't "fazed" by Apple dropping Intel CPUs, Gelsinger sure has done a lot of talking about Apple and a lot of running down Apple, even made commercials about Apple's dumb "lifestyle" nonsense(!)--far more than he has ever said about AMD. One thing AMD management has down pat that Gelsinger is lacking, is that AMD knows that shipping products turn heads--not vaporware promises with mystery ship dates. And everyone knows that all Intel has done for the past few years is compare Intel's latest CPUs with AMD's last-gen CPUs...! Looks like more of the same. How desperate can you be to start talking like this when you can't ship what you are talking about--can't even provide a firm ship date! The man doesn't even seem to realize it. Guess he's getting tired of people asking him about AMD when he doesn't want to even talk about AMD. AMD has sure "fazed" him--you can take that to the bank...:D

MegaFalloutFan:

Intel had 10n on their roadmap and so is more core CPUs, do you think the 8700K came out because they needed a CPU to compete with AMD? in less then one year of development?

The 8700K is just a Skylake with more cores. It's not a new arch. And those 10nm where delayed for over half a decade.

MegaFalloutFan:

3D cache is ONLY for high end CPUs, 5950x and maybe 5900.

no it isn't.

Blah blah blah , i am waiting for hh 's review till then they can promise what ever they want . For what is worth i wish em to be 100% accurate fierce competition is what we need !

I maybe don't see 3d cache on 5600X, but on 5800X, 5900X and 5950X there is definitely a plan to release these products with 3d cache. This will provide 10-20% boost, depending on workload. But Zen 4 is now long in research and development cycle, so if they came out 2022/Q4, ADL better be as fast as leaks suggest. I really hope Intel provides good competition, otherwise AMD will stretch release date as long as possible so production cost / release price will be best for AMD earnings. We need competition. AMD is getting too far ahead.

We need them to compete on price. Doesn't look like that's going to happen.

Horus-Anhur:

The 8700K is just a Skylake with more cores. It's not a new arch. And those 10nm where delayed for over half a decade.

It's skylake cores but Kaby Lake improved things a little bit, even if IPC barely changed. The original Skylake overclocked absolutely terribly. Like you were lucky to get 4.5 ghz out of it. Kabies were doing 4.8-5 ghz. Coffee Lake was Kaby with two more cores. But the silicon was better so it was even easier to get 5 ghz.

Waiting on real world benchmarks is more important here than ever. A synthetic benchmark will just run all of the CPU cores (big and little) at full load, which isn't what will happen in the real world (or at least, won't until game devs start properly coding for it). If you throw 4 cores that have one IPCs and clock speed into the mix with 16 that have a completely different IPC and clock speed, frame rates are going to be all over the place. The OS will have to isolate the two sets, and you will end up with an 8C/16T CPU with 4 idle "little" cores, and the result will be a much smaller increase in IPC than the leaks are showing. How Windows and game engines handle an asynchronous CPU is going to play a massive role here. In the end, we still really know absolutely nothing about the real-world performance of Alder Lake, and anyone who says differently is lying (or is an Intel engineer or CEO or a guy under an NDA on beta software and silicon, none of which matters at this point). Between that and the recent Hardware Unboxed video showcasing the massive impact that cache has on game performance vs core counts, which could have massive implications for Zen3+ (or not), the one thing we know for sure is that it's going to be a SUPER interesting Q4.

illrigger:

Waiting on real world benchmarks is more important here than ever. A synthetic benchmark will just run all of the CPU cores (big and little) at full load, which isn't what will happen in the real world (or at least, won't until game devs start properly coding for it). If you throw 4 cores that have one IPCs and clock speed into the mix with 16 that have a completely different IPC and clock speed, frame rates are going to be all over the place. The OS will have to isolate the two sets, and you will end up with an 8C/16T CPU with 4 idle "little" cores, and the result will be a much smaller increase in IPC than the leaks are showing. How Windows and game engines handle an asynchronous CPU is going to play a massive role here. In the end, we still really know absolutely nothing about the real-world performance of Alder Lake, and anyone who says differently is lying (or is an Intel engineer or CEO or a guy under an NDA on beta software and silicon, none of which matters at this point). Between that and the recent Hardware Unboxed video showcasing the massive impact that cache has on game performance vs core counts, which could have massive implications for Zen3+ (or not), the one thing we know for sure is that it's going to be a SUPER interesting Q4.

I just looked at Intel thread director hierarchy and it goes like this: _Pcores _Pcores + Ecores _Pcores + Ecores + Hyperthreading So the usage of Ecores is more preferrable to Hyperthreading, I bet because Ecores would bring higher performance benefit. Let say Ecores improve performance by 50% (vs 30% from Hyperthreading), that would means 12900K would perform as if it had 12 Pcores before Hyperthreading is active (since HT improve performance by 30% per core, that would means 8x0.3= 2.4 extra Pcores). This kinda allign with 12900K cinebench score, 800 ST and 11500 MT (800x 14.4)

Astyanax:

no it isn't.

Yep it is, google it. Due to cache price, they cant inflate the prices of cheaper CPUs, only 5950x getting it and maybe 5900x

Horus-Anhur:

The 8700K is just a Skylake with more cores. It's not a new arch. And those 10nm where delayed for over half a decade.

Its skylake but you cant just glue two more cores and call it a day. what im saying that 8700K was on the roadmap no matter what, zen or no zen

Kaarme:

I expect Alder Lake to perform well in games, significantly above Ryzen 5000, but I hesitate to expect low power consumption. Intel hasn't been overly interested in that for the longest of time, which is best demonstrated by trying to reach very high clocks instead of making the IPC better. ....

I entirely disagree. They are very interested in power consumption. Their problem is they can't do anything about it until they can utilize a more efficient process. All they can do at the moment is add another plus to their process' name and hope for the best.

Horus-Anhur:

Intel without competition became stagnant and lazy. If it wasn't for AMD putting some pressure on them, we would still be stuck with 14nm++++, 4c8t for an i7, and no IPC gains for a whole decade. It's good to see that Gelsinger is able to turn Intel around. We need a strong Intel and a strong AMD, otherwise the market will stagnate again.

I also disagree with this. They have designs they are chomping at the bit to implement but, as noted above, they are stuck at their current processing node. That is not because they are stagnant and lazy. It is because they gambled and lost on their lithography strategy. Samsung and TSMC took a different approach which succeeded and they are progressing, obviously.