Discussion Intel current and future Lakes & Rapids thread

[jpiniero]

Nobody will buy that trash when AMD's offering from two years ago was already superior in every aspect. There is literally no configuration of RKL that is technically competitive. Price cutting is the only viable play.

Rocket Lake might be faster in games. Might not have to cut prices either if AMD is too busy selling Milan.

 

[dmens]

Rocket Lake might be faster in games. Might not have to cut prices either if AMD is too busy selling Milan.

Yeah it might be faster at a few games at 1080p low detail.

 

[shady28]

Leaked RKL specs :

 

[Hulk]

It doesn't matter that there aren't a lot of consumers in the market for this because the few that are will spend a lot of money which gives Intel good margins. Apple sells under 10% of the worlds PCs and only around 15% of the world's smartphones, yet they make more profit than everyone else because they sell to the part of the market where the margins are the highest.

Intel has a lot of clout in the marketplace. They have been reliably supplying huge quantities of cpu's and other computer related parts to manufacturers for decades. For large vendors plus or minus 10% performance doesn't make a difference with computers being as fast as they are these days. Intel has been coasting (struggling with 10nm) for a few years now.

On the other hand people are noticing the architectural and process lead that AMD has gained on them. Right now it's the techies, but if Intel doesn't respond that will diffuse downward to the average user. It might take a year or two but if Intel doesn't catch up it will happen.

It's a race.

AMD must increase supplies of Zen 3 and produce mobile parts in quantities that matter to the big vendors before Intel can close the gap they have created with Zen. This is tough because no one wants to order 10,000 systems that can't be reliably delivered and put their ass on the line.

On the other hand Intel must take it's billions of dollars of cash and stop living in the past. Their current architecture is beaten and their process tech is far behind from an efficiency point of view.

It's a fantastic example of the free market and capitalism at work and the consumers are the ones who benefit.

 

[mikk]

[Intel-gfx] [PATCH 22/22] drm/i915/adl_s: Update memory bandwidth parameters

Intel added Alder Lake-S IGP code to the Linux kernel back in early December. What caught my eye is this, which suggests that the stock DDR5 speed is probably 6000.

(For comparison, the deprogbwlimit value on Rocket Lake is 20, on Alder Lake-S it's 38, Tiger Lake is 34)

You are a bit off here:

deprogbwlimit = 38, /* GB/s

For bandwidth, other memory manufacturers have quoted that for the theoretical 38.4 GB/s that each module of DDR5-4800 can bring

DDR5 is Coming: First 64GB DDR5-4800 Modules from SK Hynix

www.anandtech.com

Yeah it might be faster at a few games at 1080p low detail.

You are not an expert, you should know that low details lowers CPU load as well. Your troll attempt should be better, max details without AA at a low resolution is the way to go if someone wants to maximize the CPU bottleneck for gaming.

 

[jpiniero]

You are a bit off here:

deprogbwlimit = 38, /* GB/s

It's relative, maybe there's a bottleneck somewhere. I'm going by it being 90% higher than Rocket Lake's 3200, and slightly higher than Tiger Lake's 5400.

 

[mikk]

It's relative, maybe there's a bottleneck somewhere. I'm going by it being 90% higher than Rocket Lake's 3200, and slightly higher than Tiger Lake's 5400.

Tigerlake-U LPDDR4x-4266= 34.1 GB/s

We won't see DDR5-6000 on ADL-S in the first batch. DDR5-4800 2DPC/DDR5-5200 1DPC looks legit for ADL-S.

 

[dmens]

You are not an expert, you should know that low details lowers CPU load as well. Your troll attempt should be better, max details without AA at a low resolution is the way to go if someone wants to maximize the CPU bottleneck for gaming.

Hah. At those setting, whatever difference is bottlenecked by the monitor in actual gameplay so it really is an utterly worthless benchmark. Nice try at a pedantic argument while deliberately ignoring the fact that RKL is DOA.

 

[jpiniero]

Hah. At those setting, whatever difference is bottlenecked by the monitor in actual gameplay so it really is an utterly worthless benchmark. Nice try at a pedantic argument while deliberately ignoring the fact that RKL is DOA.

It does give them a reason to release it rather than just continuing to sell Skylake.

 

[dmens]

It does give them a reason to release it rather than just continuing to sell Skylake.

I gave it like 5 minutes of thought and he is still wrong. Depending on the game, bumping up texture details, geometry work and pixel shading complexity can easily bring the GPU back into the equation even at lower resolutions. He brings up "without AA" as if it constrains the resolution... MSAA just adds extra shading computations after the texture samples are returned, it does not increase the number of pixels being shaded. What he is thinking of is running without super-sample AA, which *nobody* uses anyways.

Also CPU loading drop for decreasing detail is much less severe than the GPU loading drop... so the claim that the *best* way to bottleneck the CPU by drastically *increasing* GPU loading to some magical level where it is just below the GPU bottleneck (as if that GPU bottleneck is a clear line) is bad methodology chasing after some trivial bump in CPU drawcall loading.

Anyways, in regards to your point, RKL is just hot trash designed to fool investors into thinking 14nm/skylake+++++ is still alive and kicking.

 

[jpiniero]

Anyways, in regards to your point, RKL is just hot trash designed to fool investors into thinking 14nm/skylake+++++ is still alive and kicking.

They can (and are) continue to sell tons of Skylake to Corporate desktop buyers who don't care beyond wanting the IGP. Server is a much more immediate problem for Intel.

 

[dmens]

They can (and are) continue to sell tons of Skylake to Corporate desktop buyers who don't care beyond wanting the IGP. Server is a much more immediate problem for Intel.

Cool you agree then, desktop RKL is irrelevant and embarrassing. On the server front, Icelake server is about to be put down by Milan.

 

[ondma]

Hah. At those setting, whatever difference is bottlenecked by the monitor in actual gameplay so it really is an utterly worthless benchmark. Nice try at a pedantic argument while deliberately ignoring the fact that RKL is DOA.

Of course, those same settings are used to show how much "better" Zen 3 is than Skylake.

 

[ondma]

Cool you agree then, desktop RKL is irrelevant and embarrassing. On the server front, Icelake server is about to be put down by Milan.

You must have selective perception. I didn't see either "irrelevant" or "embarrassing" in his post.

 

[dmens]

Of course, those same settings are used to show how much "better" Zen 3 is than Skylake.

LOL, nice quote marks. Zen 3 is definitively better than Skylake++++ at every metric I care about. From an engineering perspective, it is so far ahead as to put Intel in the technical backwater for at least a year.

 

[dmens]

You must have selective perception. I didn't see either "irrelevant" or "embarrassing" in his post.

Try reading between the lines. Doesn't take much since said irrelevance and embarrassment is so self-evident at this point.

 

[jpiniero]

Cool you agree then, desktop RKL is irrelevant and embarrassing.

The game perf is what makes Rocket Lake relevant. Otherwise they wouldn't bother and just sell Skylake. AMD doesn't technically have much to compete with at this point since they are too busy selling the APUs to mobile, and the IGP is necessary for OEMs to care outside of gaming desktops. And then there's Milan.

 

[dmens]

The game perf is what makes Rocket Lake relevant. Otherwise they wouldn't bother and just sell Skylake. AMD doesn't technically have much to compete with at this point since they are too busy selling the APUs to mobile, and the IGP is necessary for OEMs to care outside of gaming desktops. And then there's Milan.

Hmmm, if gamers are willing to stand in line in the freezing cold outside MC to get an Ampere or Big Navi card, what makes you think they will just settle with Intel trash on the CPU?

Also, RKL game perf will be trivially better in most cases, worse in a few cases with a massive increase in power. Bank on it.

 

[scineram]

Milan is little volume for months. Consoles, Renoir and Vermeer are the hot shit now.

 

[mikk]

Of course, those same settings are used to show how much "better" Zen 3 is than Skylake.

Yes exactly this applies also to Skylake-->Zen3 when it comes to gaming performance, but I guess his is too ignorant and won't understand this.

 

[dmens]

Yes exactly this applies also to Skylake-->Zen3 when it comes to gaming performance, but I guess his is too ignorant and won't understand this.

Heh, apparently because some reviewers posted benchmarks at potato resolutions, that means anyone who states Zen 3 is superior to Skylake++++++ must believe in the relevance of those potato resolution benchmarks!

Nice strawman guys.

 

[Ajay]

Hmmm, if gamers are willing to stand in line in the freezing cold outside MC to get an Ampere or Big Navi card, what makes you think they will just settle with Intel trash on the CPU?

Also, RKL game perf will be trivially better in most cases, worse in a few cases with a massive increase in power. Bank on it.

Come on man. Millions of gamers will buy Intel for the same reason they buy Nvidia. AMD has disrupted the market, and kudos to them for finally doing it. AMD can't remake the market in a few years and they don't have the die capacity to speed up the change either. AMD's technical advantage aren't lost on us hardware enthusiats, but they are on many gamers.

 

[dmens]

Come on man. Millions of gamers will buy Intel for the same reason they buy Nvidia. AMD has disrupted the market, and kudos to them for finally doing it. AMD can't remake the market in a few years and they don't have the die capacity to speed up the change either. AMD's technical advantage aren't lost on us hardware enthusiats, but they are on many gamers.

There is a difference between people buying for better performance, versus people buying an inferior product out of ignorance.

If the argument is Intel can simply out-supply AMD (which is a legitimate argument), then RKL is utterly unnecessary. From a technical point of view, RKL is a step backwards... a step that I have not seen from any company in my entire 15+ years working in high-speed digital semiconductor design.

 

[jpiniero]

Hmmm, if gamers are willing to stand in line in the freezing cold outside MC to get an Ampere or Big Navi card, what makes you think they will just settle with Intel trash on the CPU?

That remains to be seen, but OEMs do sell gaming desktops now.

 

[scineram]

If RL power ends up bad that could dampen OEM adoption.