-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
AdoredTV: Zen 3, Zen 4 rumors
- Thread starter Leonidas
- Start date
I Love Rock 'n' Roll
Member
I'm ok for Zen 4. Well done AMD.
ResilientBanana
Member
My next build might be 2021 or 2022 for sure.
thicc_girls_are_teh_best
Gold Member
FOH. The guy's a great content creator, up there with guys like Coreleks, NX Gamer, MLID etc. And they've gotten a lot of things right as well, how dumb is it to throw them out because they got a few things wrong on Ryzen 3000? No one with source connections gets everything right 100% of the time.
It's like the people who were hard-shitting on the insiders for getting PS5 specs wrong, or some of the console warrior holdouts who kept trying to discredit Komachi, Rogame etc. because the Github info and testing data wasn't 100,000,000% aligned with the actual announced specs (even though anyone with two brain cells can figure out they matched up exactly with slightly different conditions i.e XSX GPU being full-chip enabled like the X's GPU was in devkit testing, PS5's GPU getting an upclock on Oberon C0/E0 revision most likely from 2.0 GHz to 2.23 GHz).
Seriously, why is everything so extremely one-way or the opposite with some of you? Someone gets something wrong by a little bit, doesn't mean they're a fraud or were trolling or don't have valid input into topics.
Now this is speculation on my part, but I'd also assume some of the people on any sort of AdoredTV hate train are doing so because of their take on the next-gen systems, i.e he didn't align with their confirmation bias and preferences in that regard. Which is such a lame-ass reason to dislike or suddenly turn on someone. I don't agree with Corelek's take, but I still like their content and they're still very informative to listen to.
Talk about the hyper-politicized culture environment seeping into all aspects of online culture as annoyingly as possible.
Talk about the hyper-politicized culture environment seeping into all aspects of online culture as annoyingly as possible.
Perfect proof of the stupidity I'm talking about, right here.
Last edited:
jonnyp
Member
I'm stupid for calling him out on his BS about Navi? How was he not spouting absolute nonsense for months leading up to Navi being revealed? He wasn't a little bit off - he was totally off.
Take a good look at yourself before making such characterizations, pal.
Last edited:
Elektro Demon
Banned
thicc_girls_are_teh_best
Gold Member
Don't need to, I know how good my abs look but thanks
Also you've said nothing to dispel what I was speaking of: no one's going to get things 100% right all the time. Adored mentioned aspects to Navi's performance that, in some ways, were verified with high clocking of RDNA1 cards. I.e, poor IPC gains at super-high clocks.
It's actually something worth taking into consideration with PS5 because the 50% PPW gain claims of RDNA2 over RDNA1 don't translate into same for IPC. Possibly 20% IPC gains at most, and that's with EUV. Which, we already know neither system is using; they're on DUV with improvements.
It's most likely on that process the sweetspot is probably between 1.8 GHz (maybe 1.85 GHz) to 2 GHz. So it's questionable if Cerny's claims of better-than-linear scaling improvement at high clocks beyond the sweetspot will actually happen. This is probably regardless of how elaborate their cooling system will be, tbqh. That will be interesting to watch, in any case.
Anyhow, the performance levels and pricing on RDNA1 (Navi) were big things he was speaking with concern over and both of those turned out to be true taking the performance delta compared to Turing (at that time) and the higher-than-expected price on the RDNA1 cards into consideration.
Your insistence on them being discarded or seen as irrelevant are based out of nothing substantial whatsoever. It's okay to disagree with people on things but that doesn't mean they need to fade away. I disagreed with a lot of MLID's perspectives on the next-gen consoles, and they've been wrong on things too...so what? They still have perspectives I want to listen to.
It's like trying to cancel someone on Twitter over a mistake, such a petulant thing to do.
Last edited:
McHuj
Member
it’s not necessarily slower in real performance.
The latency in terms of cycles is slower, but there is nothing indicating the bandwidth, how many outstanding transactions, the operating clock, etc. The actual performance of a cache depends on many factors with latency just being one component.
Also for an on chip memory like a cache, latency usually increases with size. This is to be expected if they are indeed doubling cache sizes.
All else equal a larger cache will be a slower cache. This is why you have L1, L2, L3 caches on CPUs instead of just one giant L1 cache.
jonnyp
Member
Don't need to, I know how good my abs look but thanks
Also you've said nothing to dispel what I was speaking of: no one's going to get things 100% right all the time. Adored mentioned aspects to Navi's performance that, in some ways, were verified with high clocking of RDNA1 cards. I.e, poor IPC gains at super-high clocks.
It's actually something worth taking into consideration with PS5 because the 50% PPW gain claims of RDNA2 over RDNA1 don't translate into same for IPC. Possibly 20% IPC gains at most, and that's with EUV. Which, we already know neither system is using; they're on DUV with improvements.
It's most likely on that process the sweetspot is probably between 1.8 GHz (maybe 1.85 GHz) to 2 GHz. So it's questionable if Cerny's claims of better-than-linear scaling improvement at high clocks beyond the sweetspot will actually happen. This is probably regardless of how elaborate their cooling system will be, tbqh. That will be interesting to watch, in any case.
Anyhow, the performance levels and pricing on RDNA1 (Navi) were big things he was speaking with concern over and both of those turned out to be true taking the performance delta compared to Turing (at that time) and the higher-than-expected price on the RDNA1 cards into consideration.
Your insistence on them being discarded or seen as irrelevant are based out of nothing substantial whatsoever. It's okay to disagree with people on things but that doesn't mean they need to fade away. I disagreed with a lot of MLID's perspectives on the next-gen consoles, and they've been wrong on things too...so what? They still have perspectives I want to listen to.
It's like trying to cancel someone on Twitter over a mistake, such a petulant thing to do.
Haha, this is incredible! Me saying his channel should've died is the same as twitter cancelling someone? In which dimension do you live in?
This isn't about "not getting it 100% right" but pretty much getting it 100% wrong while claiming to have leaks and insider info in video after video. And he did it for subscribers. I don't care much for bullshit artists, Mr. "Abs".
He was wrong about the TDP, the price, the number of CUs, the series naming etc etc. He has since deleted most of those videos and he deleted his subreddit when questioned and criticized - what does that tell you? He was also wrong about Zen 2 and not just "a little bit off". If you make big claims which fail to deliver, then expect to be criticized and don't whine about it when that happens.
I have no idea how anyone could defend this guy. If he had said all along that it was his personal speculation based on information publically available I would have had zero problem with him.
Last edited:
llien
Member
AVX 512 was first implemented in Skylake X, which was released in sommer 2017.
Leonidas
Member
Yeah, and AMD will finally have it many years after Intel, and two more gens in the future for AMD if rumor is correct.
Hopefully AMD will adopt new features quicker in the future.
Last edited:
sonomamashine
Member
?
llien
Member
Hopefully AMD will focus on things that matter, and address new instruction sets with marginal uses, only if there are spare resources.
I'm not getting the number crunching pushon CPUs, when you have something that is massively parallel, that is what GPUs are for.
Entroyp
Member
What is AVX 512 used for today?
Elektro Demon
Banned
I know that a lot of dx12 games use it, since it's part of the API.
PhoenixTank
Member
That'd be AVX/AVX2 not AVX-512.
Edit: Unless something has changed recently?
Last edited:
Elektro Demon
Banned
Yeah, you're right. My bad.
But what the hell is avx512 then used for? I gotta google that a bit.