So in other words, they gutted Willow in order to actually be able to produce the cores on 14nm, and by that they ended up with a higher IPC than Willow Cove. Mkay.
So in other words, they gutted Willow in order to actually be able to produce the cores on 14nm, and by that they ended up with a higher IPC than Willow Cove. Mkay.WillowCove uses the non-inclusive cache subsystem like Zen3 and Skylake-X which performs better on independent threads because L2 does not contain L1 copies and L3 does not contain L2 copies. And when it comes to CypressCove, the cache system is inclusive like Skylake and SunnyCove, which works better in interdependent threads, i.e. in games. I think it is quite possible that CypressCove will have an IPC comparable or higher than Zen3 at least in games is this chance.
WillowCove can gain compared to SunnyCove, but it can also significantly lose, mainly due to the non-inclusive cache system used.
There is absolutely nothing wrong with Vega on Renoir, so...It's very frustrating, Cezanne and Van Gogh, choose your poison... outdated IGP or CPU
There is absolutely nothing wrong with Vega on Renoir, so...
You should look into this technology called pci it really is quite amazing.Other than it will be the longest lived IGP on the consumer market by the time Cezanne gets replaced and it is now getting outperformed by Intel on notebooks, nothing, there is nothing wrong with Vega, on CEZANNE. We got stuck on quad cores for how long? 6 years from SB to Ryzen? Vega has a chance to be over 4 years old by the time it gets replaced by Rebrandt. Just saying.
Other than it will be the longest lived IGP on the consumer market by the time Cezanne gets replaced and it is now getting outperformed by Intel on notebooks, nothing, there is nothing wrong with Vega, on CEZANNE. We got stuck on quad cores for how long? 6 years from SB to Ryzen? Vega has a chance to be over 4 years old by the time it gets replaced by Rebrandt. Just saying.
oh finally someoneNot even true and that goes to Intel number aswell. You can way more on both systems. I gain with 3900x from stock 3600mhz XMP (CL16) to CL14 with dram calc fast subtimings.. 20-30 fps better low fps in 64 player BFV.
Problem is no reviewer actually test multiplayer scenarios. I had, now sold 3900x, like 130 fps low fps in 64 player BFV. A overclocked 9900k with 4300mhz low timings ram has like 150fps. So it will be interesting to see what my 5900x will give in same scenario.
do we have enough data to say it? and please no cinebenchRocketLake is a gimped SunnyCove core, there is no comaprison with Zen3. Zen3 is ~5-7% faster than latest iteration of Coves (Willow Cove)
so It's ok using an 3 years old gpu arch in Cezanne when they need to defeat Tiger Lake IGP while having a better arch named RDNA?There is absolutely nothing wrong with Vega on Renoir, so...
Do we really know anything about Zen 3 cache hierarchy design changes at this point? It doesn’t seem like we really know much of any thing. We have some idea of performance, which indicates that Zen 3 does very well at gaming, but we really don’t know how any of it is achieved. We do know that AMD considers Zen 3 a new architecture and that the cache hierarchy has been redesigned, but we do not know some of the things that you seem to be making assumptions about. I would suspect that a non-inclusive cache effect on gaming performance is a bit of an assumption also. We don’t know how AMD redesigned the cache hierarchy, but it is probably non-inclusive since they still list total cache. I believe they have indicated that they increased cache bandwidth a lot, so with the much larger, monolithic 32 MB cache per CCD, I would expect communication between threads to be much faster.WillowCove uses the non-inclusive cache subsystem like Zen3 and Skylake-X which performs better on independent threads because L2 does not contain L1 copies and L3 does not contain L2 copies. And when it comes to CypressCove, the cache system is inclusive like Skylake and SunnyCove, which works better in interdependent threads, i.e. in games. I think it is quite possible that CypressCove will have an IPC comparable or higher than Zen3 at least in games is this chance.
WillowCove can gain compared to SunnyCove, but it can also significantly lose, mainly due to the non-inclusive cache system used.
It's still better to have the best IGP along with best CPU both in desktop and laptop.As mentioned many times, AMD is now in the high volume mainstream laptop market, where gaming laptops has seperate GPU, and laptops without seperate GPU are mainly sold to people that doesn't care much about having best possible integrated GPU performance.
Of course they will loose some sales to Tigerlake due to keeping Vega but they will gain much more thanks to the big momentum that have achieved with Renoir. OEM are taking AMD seriously in laptop now since the customers are "screaming" for more Renoir laptops.
So having a well balanced laptop APU for most of the middle and high-end market that is delivered on time and with competitive cost (not too big GPU-part...) is what AMD focus on.
Cezanne will be a big success thanks to the CPU performance.
Memory bandwidth is too low for real gaming APU's anyway.
This bigger IGP would cost only $4 more, which is negligible considering what AMD is selling their laptop chips for.
If you were fighting to get market share before your competition has time to react, would you produce 330 products per day at $30 or 300 products per day at $35?It's still better to have the best IGP along with best CPU both in desktop and laptop.
When AMD had a bad CPU, then the focus was on IGP, because CPU was sh*t, now when they have a good CPU(Zen1 and Zen2) and will release a great CPU(Zen3) a great IGP no longer matters both to AMD and to most of It's fans.
[...]
If I increase the size to 162mm2 I will end up with 297 good dies and 51 defective ones, that's $34 per good die.
This bigger IGP would cost only $4 more, which is negligible considering what AMD is selling their laptop chips for.
Do they care that much about the lost money? If they really cared about having the best profit per wafer, then why produce SoCs for the consoles? We know the margins are very low there and the SoC are big(197mm2 Xbox S and 360mm2 for Xbox X), so a lot of wafers are used for them. Here they lose a lot more money per die.But as they are supply limited that would mean ~40 less chips per wafer - at @ $150 price per chip that would mean $6000 loss per wafer. That's a considerable amount of lost money. Or other way around, they need to charge at least $15 more per chip from little bit better IGP to break even, are customers ready to pay for that?
There's already signs of that in the undisclosed slides on 5800X and 5600X. AMD claims the 5800X is almost 10% faster in games over 10700K, while the 5600X is 13% faster over 10600K. We're looking at different clocks combined with different core counts so it's hard to reach a conclusion based on these slides (even if we commit the sin of ignoring they're marketing driven), but at least we can more easily understand why AMD is confident in their product, especially as we move down the product stack.AMD's numbers are using 2080ti, if they do in fact have the fastest cpu for gaming, that lead will increase with rtx 3090 or maybe some big Navi derivative.
We could well be looking at a solid 10%
If you want to gain market share with limited number of allocated wafers to you, why would you produce big SoCs for consoles with low margins, when It can be used for Zen3 CPUs?If you were fighting to get market share before your competition has time to react, would you produce 330 products per day at $30 or 300 products per day at $35?
Oh, so your solution for AMD's future includes bigger chips and also ditching semi-custom on a whim. So much for serious conversation.If you want to gain market share with limited number of allocated wafers to you, why would you produce big SoCs for consoles with low margins, when It can be used for Zen3 CPUs?
Oh, so your solution for AMD's future includes bigger chips and also ditching semi-custom on a whim. So much for serious conversation.
And to what end, MS and Sony would likely use TSMC capacity anyway.Breaking just a small little contract or two as well. The penalty clauses would no doubt be quite 'amusing'.
Yes, you are right, serious conversation with you is useless, I won't bother anymore.Oh, so your solution for AMD's future includes bigger chips and also ditching semi-custom on a whim. So much for serious conversation.
And where did I say anything about breaking contracts? I just said, if they wanted to maximize produced chips and profit, making SoCs is not a good strategy. Ok, when they got the contract for consoles they didn't know what will happen in the future, so It was a safer bet.Probably
Breaking just a small little contract or two as well. The penalty clauses would no doubt be quite 'amusing'.
And where did I say anything about breaking contracts? I just said, if they wanted to maximize produced chips and profit, making SoCs is not a good strategy. Ok, when they got the contract for consoles they didn't know what will happen in the future, so It was a safer bet.
The same they have to do to cool 10400F ??
People buy 10400F today over 3600
Intel will have no problem cooling their value RKL-S SKUs, and maybe even the flagship (assuming they use the same tricks as with 10th gen). Power per core goes up, number of cores go down. Zero sum.
Lots of attacking the messenger around here. Sigh.
Well, if Zen 3 is 5-7% faster than Comet Lake in most games, maybe a tad less when OCed... you'd think Rocket Lake would be able to do better than that.
RocketLake is a gimped SunnyCove core, there is no comaprison with Zen3. Zen3 is ~5-7% faster than latest iteration of Coves (Willow Cove) so whatever Cypress Cove( the thing in RocketLake) brings will be too little, too hot. Pun intended.
Do you have a rocket lake-s chip?
AMD designed those chips so In the end they need to be involved in the production.I have to think Sony and MS have their own arrangement with TSMC. It has to be "AMD's wafers" due to the agreement with Intel but AMD doesn't really need to be involved.
The point is the stall, getting stuck in the same arch for too long is never good and this is getting closer to the 4C stall every month and that is bad.As mentioned many times, AMD is now in the high volume mainstream laptop market, where gaming laptops has seperate GPU, and laptops without seperate GPU are mainly sold to people that doesn't care much about having best possible integrated GPU performance.
Of course they will loose some sales to Tigerlake due to keeping Vega but they will gain much more thanks to the big momentum that have achieved with Renoir. OEM are taking AMD seriously in laptop now since the customers are "screaming" for more Renoir laptops.
So having a well balanced laptop APU for most of the middle and high-end market that is delivered on time and with competitive cost (not too big GPU-part...) is what AMD focus on.
Cezanne will be a big success thanks to the CPU performance.
Memory bandwidth is too low for real gaming APU's anyway.
You should look into this technology called pci it really is quite amazing.