P + E Cores ?

P are "performance" cores and E are "(power) efficient" cores.

 

Can you provide a link? This sounds really weird. Whilst some threads may benefit from running on certain cores - or combination of cores - and can definitely determine which CPU core(s) they are running on, I can't fathom why a DRM implementation would be written in a way as to cause problems.

 

This is the origin of the story
https://gamersgospel.com/denuvo-and-other-drm-might-have-issues-on-intels-new-cpus/

Also, "Data Relationship Management" ?

 

Thanks! Denuvo, why I am surprised. That said there is incredible use of the word "might" in that article which makes me question how likely this really is.

Mix in the retro-definition of 'DRM' makes me question the entire article. It's not a site I've heard of.

 

The Intel quote does specifically mention games and DRM together so it's likely some PR person googling "DRM". Considering the effect Denuvo has on CPU performance in games, I think there's some validity to it. Intel wouldn't be posting this unless there was a potential issue they identified.

 

According to a bunch of discussions, the following was present in Intel's Alder Lake developer guide:

However, it seems Intel took down the document for correction because it mentioned support for AVX512 on the P cores, which they don't have.

I'm all for Alder Lake being a competitive architecture on desktops (it doesn't look like they'll stand a chance in mobile / laptops against against Rembrandt), but Intel mentioning this in their documentation probably means they've run into issues already.

I wonder how Denuvo still manages to get so many high profile clients, though. They're mostly bad publicity at this point.

 

In general the companies that use it don't really care about bad publicity. It's highly likely that the number of sales lost due to Denuvo is far less than the number of sales lost due to piracy.

 

To be fair to Denuvo, when it's first out, it prevented piracy for a few months, which is important for game publishers: the majority of revenues from most games are from the first few months.
If you are not in the software business yet keep claiming that "piracy does not hurt their bottom lines," sorry you have to understand why I can't take your words seriously.

 

It seems to vary, game to game. Resident Evil Village launched in early May but the cracked version didn't appear until early July.

 

Just going to add some more information.

The P Golden Cove cores you can think of basically as iterative updates upon what you think of as Intel's current desktop CPU stack. Golden Cove is the update upon Willow Cove used in Tiger Lake. Willow is an update upon Sunny Cove used in Ice Lake and by extension also Cypress Cove in Rocket Lake (11xxx) on the desktop (why this split happened is another story related to process issues with Intel).

The E Gracemont cores are iterations of what you'd know of as Intel's Atom line.

Now while you might think the E cores are efficiency in terms of power efficiency and therefore lower power usage it's not really that straight forward. There's already being seen with this and I suspect more so once launch/reviews come out as well with people wondering what lower power usage does for the desktop (expect peak power draws to go up). It'll be interesting to see how Intel chooses to communicate this and how at least some of the public respond.

Another note is that the desktop will have 2 different dies. One is 8P+8E the other is just 6P cores. 6 core locked SKUs (12600k/f is unlocked) and down are expected to use the latter die, so these interestingly won't be hybrid.

12900k will be 8+8, 12700k 8+4, 12600k 6+4 and are to be released first over the next week or so.

It's varied a bit over it's lifespan but still it's consistently been much more effective than going without Denuvo. Denuvo is continually updated. It's also worth noting that Denuvo itself is not the only DRM that is present in a game.

There's been times such in 2019 when it was being broken within a week? While I think in 2017 it was often down to hours with more than one person/group able to do it.

Currently I believe there is only one person (or a group using the moniker) that is doing after 2019 and the times ranged in the months again.

Another by product is that game updates are not regularly released with games that use Denuvo either, which given sometimes the post release issues of games (or content) in modern times is also another factor.

Deathloop - 2021/09/14 - Example of a recent game that hasn't been cracked yet

Far Cry 6 - 2021/10/06 - Ubisoft, likely will also take longer (I think Ubisoft stacks other heavy DRM as well, you can see this with the list below)

I believe these are the last 5 Denuvo based releases by the date they were cracked (game release date - crack date) -

Resident Evil Village - 2021/05/06 - 2021/07/10 (as mentioned)

Yakuza: Like a Dragon - 2020/11/10 - 2021/05/30

Assassin's Creed Valhalla - 2020/10/22 - 2021/03/21

Immortals Fenyx Rising - 2020/12/03 - 2021/02/22

Anno 1800 - 2019/04/16 - 2021/01/05

As you can also see as mentioned there can be additional protections, hence

This is rather dated information so it may not exactly be the case but I would think it still is. From what I remember one aspects that made Denuvo harder to crack and release wide was that it does checks based on the CPU it detects itself to be running on, including possibly having to download code based upon the CPU it detects. In theory this would add an complication for anyone looking to crack it and release it widely as it would need to function across more CPUs.

This in conjunction with the Intel statement suggests that there might be a problem as Golden Cove and Gracemont (the two cores in Alder Lake) would be considered different CPUs and cause issues unless accounted for.

 

Isn't there at least a range of relevance/impact here?

Surely the amount of money lost to piracy isn't zero, but how many people would actually buy a single player game if they can't pirate it within the first 3 months?

Throughout the years I've seen a number of developers and publishers claiming they lost X millions of revenue to piracy due to the game being distributed Y times on torrents, where X = Y*$50.
This is so detached from reality it's shocking IMHO. Those people need to know that a good number of pirates (which I would guess are mostly teenagers and people whose expendable income doesn't allow paying for videogames) would never, ever pay for the game regardless, and if they can't play the game for free then they won't play it ever.


Are there more people willing to buy the game for full price if they can't find it pirated, than the number of people avoiding the game because it's being plagued by Denuvo? If true, is this delta worth it for the amount of negative opinions the publisher+developer will gain from adopting Denuvo?

 

I think its important for Intel and Microsoft and soon AMD (rumored that they will have 3nm Zen 5 with Zen 4 little cores) for them to sort out behavior as soon as possible. For instance it could be a big performance boon on desktop if the additional efficent cores could just handle windows while the other 8 performance cores are used for the game or rendering your video or whatever it is your using it for.

On the portable side i hope windows scheduler gets smart enough to give as much to the efficent cores as possible.

 

I see gaming (and similar) work loads to be the main challenge with a hybrid architecture. However to a large extent there's going to be limitations on what the IHV and OS and to some extent other middleware that sits in between (API, etc.) compared to the game (or other software) itself being aware. The ADL developer guide (that was removed) from what I remember had sections in their suggesting that game developers flag priority threads that are the most time critical to ensure they are prioritized for the P cores.

Rendering (non real time 3D) and to some extent video encoding (although newer codecs scale wider) would actually be more suited for the E cores. The E cores are essentially aimed at lower time critical threads. As in the case of say 3D rendering and video encoding it's of low importance how fast each individual thread finishes as long as you can split up the workload evenly enough, which is relatively trivial for those types of tasks (especially if compared to gaming).

It seems like Intel's design philosophy (although judging by ARM based hybrids it certainly isn't an unique thought) for it's hybrid architecture is that the E cores are more optimal (not just in terms of power but other factors such as transistor/die budget, etc.) for work loads that scale wide and where performance is measured by throughput (as opposed to latency). While the P cores are more so for time sensitive work loads that tend not scale. Based on this their current roadmap for future generations at the moment point towards scaling up to put in more E cores while leaving the P core count the same and just building them bigger. Raptor Lake (what comes next) is currently rumored as a 8+16 config for example.

So it seems like the advantage with the hybrid design going forward for Intel will be that they can have 2 designs that have different emphasis in design. Presumably E cores will be focused on efficiency in power and transistor cost, basically with an emphasis on putting more of them in a given power and transistor budget.

 

Could this hybrid approach cause frame pacing issues in games?

 

How does the DDR4+DDR5 compatibility work in Alder Lake (or past cross-generation CPUs)? Does that mean the CPUs have separate PHYs for each or are they dual purpose? Not sure how that works.

I couldn't find an official Alder Lake thread yet.