Discussion Intel current and future Lakes & Rapids thread

[dullard]

Also, suggesting a purposeful mispricing over this time frame is just grasping at imaginary straws

Just try comparing Mindfactory.de prices for AMD products vs say Amazon's prices for AMD products. Then do that for Intel products. You'll find on average AMD products get about a 5% better conversion rate at Mindfactory.

 

[lobz]

Except STEAM data.

No. The same applies to Steam data. Look at the trends and you may get some useful info. I have another good one: JPR market share numbers. All of these are very misleading at first glance, so you have to put them in the proper perspective.

 

[lobz]

Just try comparing Mindfactory.de prices for AMD products vs say Amazon's prices for AMD products. Then do that for Intel products. You'll find on average AMD products get about a 5% better conversion rate at Mindfactory.

Dear God... I'm done arguing with you 🙂

 

[2blzd]

Do we have any ideas, or clues or rumors/best guesses of how Alder-Lake performance will compare to say Rocket-Lake? We thinking single digit % improvement or greater?

 

[coercitiv]

Intel surely posses the resources to make even the Windows scheduler work somewhat correctly with Alder Lake.

Due Intel $$$ it won't be the sad show as with AMD x86 stuff like - Bulldozer CMT scheduling, Ryzen CCX scheduling, Ryzen preferred core scheduling, Threadripper MCM-aware scheduling, Threadripper 128 thread issue, etc.

Remember Lakefield? They even had devices in development with MS using the chip, yet Lakefield scheduling at launch was just as clumsy as the AMD examples above.

If Intel is serious about using hybrid configurations coupled with heterogeneous MT then we're bound to get a decent scheduler as time goes by, but never assume the first generation of chips gets the full "promised" deal. Just like with AMD they may improve the design in further gens in such a way that prior models may never get full benefits in the end.

ADL-S will be weighed and measured based on it's big core performance alone, anything small core related will be the cherry on top.

 

[uzzi38]

Do we have any ideas, or clues or rumors/best guesses of how Alder-Lake performance will compare to say Rocket-Lake? We thinking single digit % improvement or greater?

If I were to guess 10-15% is a safe bet with any more being a nice bonus. I'm not convinced clocks will go as high as 5.3GHz, but even at 5-5.1GHz we should still be looking at closer to the 15% mark.

 

[DrMrLordX]

Just search this very thread and forum for "Mindfactory" and you get things like this (from the same person that started it a few posts above):

And in that context, he was discussing DiY sales. Not OEM sales. So the problem is . . . ?

Yes, Mindfactory tells us useful information. But we shouldn't generalize it.

Nobody did.

What's somewhat interesting about mindfactory data on lower-end DiY CPU sales is that it may be a leading edge indicator of what OEM sellers are examining for future products when factored against vendor agreements (some OEMs stick with Intel because of long-standing relationships and other . . . reasons). For example, the R5 3600 was an exceptional seller for AMD due in large part to its high performance within an 88W power envelope. OEMs are extremely sensitive to price/performance and performance/watt, especially since they want to save money with cheaper cooling solutions and cheaper motherboard components whenever possible. A chip like the R5 3600 is a harbinger of some migration of OEMs to future AMD products (such as an R5 5600 which has yet to appear on the market). It does not mean that AMD OEM systems will outsell Intel OEM systems by 99 to 1, but it does mean that if AMD can beat Intel in perf/watt and price/watt AND sell the heck out of chips on that basis to DiY users that OEMs will start paying attention and shift their product portfolio accordingly. mindfactory's numbers can show a trend, but they won't show exact ratios. Even other DiY retail outlets aren't going to sell at the same ratios as mindfactory. Germany has been somehwat AMD-friendly for awhile in no small part thanks to Fab 1 in Dresden.

Except STEAM data.

Absolutely. Steam survey data is about as bad as a web poll.

 

[Zucker2k]

Absolutely. Steam survey data is about as bad as a web poll.

Quite the contradiction. Chip sales on one website is Germany is indicative of world wide chip sales, but data from the biggest gaming site on the planet is worthless. As if the STEAM app is bundled with Windows or something.

 

[Abwx]

Test of a TGL based Nuc at NBC :

Intel NUC11PAQi7 Panther Canyon Mini PC Review: Tiger Lake Done Right

Intel's implementation of the Core i7-1165G7 is better than most laptops. Raw processor performance is faster than the ill-fated Core i7-8809G released just a couple of years ago while being much more power efficient in the process.

www.notebookcheck.net

At 40W or so perfs in Cinebench R15 are close to 10% lower than a 4C/8T Renoir, this amount to roughly 20% lower perf/watt at same perf.

For comparison :

AMD Ryzen 3 4350G im Test

Der AMD Ryzen 3 4350G ist die kleinste Zen-2-APU. Er löst nahezu alles, was bisher da war, ab und macht Vorfreude auf neue Athlon.

www.computerbase.de

 

[DrMrLordX]

Quite the contradiction.

No. @Hitman928 already explained it. There have been entire threads on the poor quality of Steam survey data.

 

[Asterox]

Test of a TGL based Nuc at NBC :

Intel NUC11PAQi7 Panther Canyon Mini PC Review: Tiger Lake Done Right

Intel's implementation of the Core i7-1165G7 is better than most laptops. Raw processor performance is faster than the ill-fated Core i7-8809G released just a couple of years ago while being much more power efficient in the process.

www.notebookcheck.net

At 40W or so perfs in Cinebench R15 are close to 10% lower than a 4C/8T Renoir, this amount to roughly 20% lower perf/watt at same perf.

For comparison :

AMD Ryzen 3 4350G im Test

Der AMD Ryzen 3 4350G ist die kleinste Zen-2-APU. Er löst nahezu alles, was bisher da war, ab und macht Vorfreude auf neue Athlon.

www.computerbase.de

Huh, Core i7-1165G7 for 426$ in this NUC configuration is very bad price=performance ratio.This will be minimum 700-800$ NUC.

Intel® Core™ i7-1165G7 Processor (12M Cache, up to 4.70 GHz, with IPU) - Product Specifications | Intel

Intel® Core™ i7-1165G7 Processor (12M Cache, up to 4.70 GHz, with IPU) quick reference with specifications, features, and technologies.

ark.intel.com

If you buy from China, you can get R3 4350G for 170-180$.

Here is what you can get for 430$. This is final retail price for this combination, R3 4350G + Asrock X300+local currency(Croatia) conversion in dolars.If you need more CPU power, you can buy R5 4650G for 237$.

ASRock Deskmini X300, barebone , HDMI//DP/D-Sub USB3.1 | ADM

ASRock Deskmini X300, barebone , HDMI//DP/D-Sub USB3.1, 90BXG3T01-A10GA0W- pogledajte najbolju ponudu uredskih računala po najpovoljnijim cijenama!

www.adm.hr

Procesori | AMD Ryzen procesori | ADM računala

ADM Računala Web Trgovina, prilaz V. Brajkovića 4, 10000 Zagreb - prodaja vrhunskih računala i opreme po najboljim cijenama!

www.adm.hr

Procesori | AMD Ryzen procesori | ADM računala

ADM Računala Web Trgovina, prilaz V. Brajkovića 4, 10000 Zagreb - prodaja vrhunskih računala i opreme po najboljim cijenama!

www.adm.hr

 

[Zucker2k]

No. @Hitman928 already explained it. There have been entire threads on the poor quality of Steam survey data.

His argument is about data collection methodology (steam survey), which in global terms is actually more representative than online sales by a German e-tailer. Are you arguing against this (little inconvenient) fact?

 

[Hulk]

I am going to make some predictions regarding Alder Lake based on what I have read.

Performance will be better than Rocket Lake across the stack at various price points. It's hard to quantify percentages and IPC due to the different nature of RL vs AL but suffice it to say I think it will be a very good competitor for Zen 3. In addition, I think Intel is going to to make significant strides in efficiency for the desktop when you consider the move to 10SF and the Big/Little core strategy.

My thought is that Intel wouldn't be pushing AL out the door so soon after RL unless they had something they believe is really good. With the astronomical prices of computer parts now I'm putting off my next build until after the summer.

 

[jpiniero]

My thought is that Intel wouldn't be pushing AL out the door so soon after RL unless they had something they believe is really good.

Or OEMs are unhappy with Rocket Lake because of the power draw.

 

[lobz]

Or OEMs are unhappy with Rocket Lake because of the power draw.

It's supposedly the same as CML, and I'm pretty sure Intel sells every single chip they can possibly produce.

 

[LightningZ71]

I suspect that Intel is pushing AL because they are already capacity constrained on 14nm, and really want to get a return on their investment on 10sf, 10esf. It's not like they won't be able to sell every RL they make for the next two years.

 

[Hulk]

Or OEMs are unhappy with Rocket Lake because of the power draw.

I wonder if OEM's care at all about power draw? Most desktop systems can supply the electricity and dissipate the heat. And the OEM's aren't paying the electricity bill to run it. Also they don't advertise power as a selling point nor to most consumers who buy these systems know or care about it. All they need to put on the box is "NEW! 11th Generation from Intel. Up to 5.3GHz" and they will sell them.

But eventually the masses will wise up and the build your own community I think is more disgusted with the inefficiency of CL/RL than OEM's. RL will hold back the tide for a while but AL is necessary before Intel digs a hole so deep they can't climb out.

 

[Hitman928]

I wonder if OEM's care at all about power draw? Most desktop systems can supply the electricity and dissipate the heat. And the OEM's aren't paying the electricity bill to run it. Also they don't advertise power as a selling point nor to most consumers who buy these systems know or care about it. All they need to put on the box is "NEW! 11th Generation from Intel. Up to 5.3GHz" and they will sell them.

But eventually the masses will wise up and the build your own community I think is more disgusted with the inefficiency of CL/RL than OEM's. RL will hold back the tide for a while but AL is necessary before Intel digs a hole so deep they can't climb out.

OEMs care because if power draw goes up significantly they then have to spend more on higher power PSUs and cooling the component. It can also limit the form factor they can use (at least within the cost they are willing to spend). Many times what ends up happening is higher power components just get power limited by the OEM and you get high noise and lower performance than you think you are getting based upon the system components.

 

[dullard]

Or OEMs are unhappy with Rocket Lake because of the power draw.

The thing is, Intel mainstream desktop chips only draw 200+W of power, when (a) there is thermal head room and (b) if you set it to do so. With most OEMs (Dell, HP, Lenovo, and similar), there is little to no incentive to (a) pay for the extra thermal head room and (b) to set it to use more power. Why would an OEM pay more for a better heat sink when there is no need to do so?

So, unless you have some evidence from OEMs, I'm going to go with the idea that they do not care. Especially when users might just go ahead and pair it with a 300+W video card, showing that power consumption on desktop often isn't high up on user's checklist.

Now with custom build OEMs (Origin and similar), there is incentive to do use more power for more performance. But again, that buyer usually isn't too concerned with desktop power consumption.

 

[jpiniero]

So, unless you have some evidence from OEMs, I'm going to go with the idea that they do not care. Especially when users might just go ahead and pair it with a 300+W video card, showing that power consumption on desktop often isn't high up on user's checklist.

Amps maybe? Intel might be charging more for Rocket Lake because the die is bigger? More expensive cooling?

If they are unhappy with Rocket Lake, then perhaps the Q4 sequential desktop volume increase is OEMs stocking up on Comet Lake.

 

[scannall]

His argument is about data collection methodology (steam survey), which in global terms is actually more representative than online sales by a German e-tailer. Are you arguing against this (little inconvenient) fact?

I don't think either one is definitive. But, if looked at over a long period of time it can show where the market is trending to.

 

[lobz]

I am going to make some predictions regarding Alder Lake based on what I have read.

Performance will be better than Rocket Lake across the stack at various price points. It's hard to quantify percentages and IPC due to the different nature of RL vs AL but suffice it to say I think it will be a very good competitor for Zen 3. In addition, I think Intel is going to to make significant strides in efficiency for the desktop when you consider the move to 10SF and the Big/Little core strategy.

My thought is that Intel wouldn't be pushing AL out the door so soon after RL unless they had something they believe is really good. With the astronomical prices of computer parts now I'm putting off my next build until after the summer.

I'd say there's a 99% chance that ADL will be faster than Zen 3 in most use cases, especially single threaded or at equal core counts. Of course, by then, Zen 3 will be a well established product line and AMD can make Intel's life pretty hard and the success bittersweet with adjusting the prices properly. Then the ball will be in their court, to bring Zen 4 out in time.

 

[Hulk]

I'd say there's a 99% chance that ADL will be faster than Zen 3 in most use cases, especially single threaded or at equal core counts. Of course, by then, Zen 3 will be a well established product line and AMD can make Intel's life pretty hard and the success bittersweet with adjusting the prices properly. Then the ball will be in their court, to bring Zen 4 out in time.

Single threaded ADL vs Zen 3 comparisons will be measurable as we are used to and well-versed with single core-to-core comparisons.

As I eluded to in my post, multi-core comparisons of ADL to Zen 3 are going to be really interesting and I think problematic. At least at first until we get a handle on how to make these comparisons. What will our performance comparison metrics look like?

8 core Zen 3 vs 8/4 ADL?

Or 12 core Zen 3 vs 8/8 ADL?

What's a little core worth in terms of a big core? How do the big and little cores scale/work in various multithreaded apps?

I think with the CPU market starting to differentiate architecturally now more than ever, clock speed and cores are starting to matter less and less. The end result is simply how does this processor compare with this one, which has been how the non-tech geeks (not us) have been looking at this from the beginning.

For people like us who think about these things during idle moments it's going to be super interesting when this clash get going in earnest. The really fun part will start once RKL hits the ground and we get real benchmark data and analysis from Ian. I'm also thinking Intel may creak open the door just a bit on ADL soon after RKL just to keep our attention on their side of the table.

On the other hand, AMD is sitting pretty thinking, "What me worry?" I've got the fastest architecture, the most efficient process, and I'm selling every single chip I can produce.

 

[blckgrffn]

I think we are going to see a long tail overlap with RKL and everything that comes after especially on the desktop, presuming it is the last stand for 14nm.

On the PC side, for the broad swath of normal users, RKL will simply be good enough. Since it has Xe graphics it should allow for Intel to support it for an extremely long with their driver packages going forward.

And when it comes down to it, neither Intel nor AMD has that much incentive to focus on desktop CPUs with cutting edge nodes going forward. The real money being on the server side and second to that having power efficiency for mobile. Desktop CPUs have to come in a distant 3rd. As others have well said above, the OEMS are building their desktops to handle 65W-95W CPUs. RKL allows them to just keep using the same components with regards to PSUs and the like. If the global silicon shortage continues (why won't it?) how long will we see these sold?

I am not saying there won't be ADL and follow on desktop releases chasing halo positions, I just expect that even a couple years from now Intel is still going to be selling the market a lot of RKL. Especially if it stays remotely performance per dollar competitive. Will we see a 14nm CPU that uses DDR5? Ha...

 

[Exist50]

I think we are going to see a long tail overlap with RKL and everything that comes after especially on the desktop, presuming it is the last stand for 14nm.

On the PC side, for the broad swath of normal users, RKL will simply be good enough. Since it has Xe graphics it should allow for Intel to support it for an extremely long with their driver packages going forward.

And when it comes down to it, neither Intel nor AMD has that much incentive to focus on desktop CPUs with cutting edge nodes going forward. The real money being on the server side and second to that having power efficiency for mobile. Desktop CPUs have to come in a distant 3rd. As others have well said above, the OEMS are building their desktops to handle 65W-95W CPUs. RKL allows them to just keep using the same components with regards to PSUs and the like. If the global silicon shortage continues (why won't it?) how long will we see these sold?

I am not saying there won't be ADL and follow on desktop releases chasing halo positions, I just expect that even a couple years from now Intel is still going to be selling the market a lot of RKL. Especially if it stays remotely performance per dollar competitive. Will we see a 14nm CPU that uses DDR5? Ha...

I agree with the gist of your comment, but do keep in mind that if the sharkbay leak holds, and so far it's looking good, there will be a dedicated 6+0 desktop chip. The 8+8 will probably be focused on halo systems, at least at first, but going into 2022, it's probably reasonable to expect the 6+0 to gradually supplant Rocket Lake for mainstream desktops.