Ajay
Lifer
- Jan 8, 2001
- 15,448
- 7,858
- 136
His net worth is something like $100B. I'd of gotten up there and done that act for less.
Well macs are 10% of the computer market so even if apple isn't a threat to x86 thats still a sizeable chunk of money not going into intel anymore.
Intel is supply constrained and has been for some time. Apple not buying Intel CPUs doesn't cost them anything because they will still have full fabs and still sell everything they can make.
At most since Apple has always chosen more costly SKUs (because they wanted the lowest power CPUs possible) it costs Intel a bit if the orders that replace Apple's are at a lower ASP. On the other hand Apple was a pretty demanding customer, and Intel spent millions of dollars meeting their demands for special SKUs, special packaging, etc. which they won't have to deal with anymore.
If there was ever a time for Apple to drop Intel without Intel feeling any pain, this is the time. Not only was Intel supply constrained due to their self-inflicted wounds delaying process transitions before coronavirus hit, WFH and remote schooling spiked PC demand up for the first time since the market went into decline a decade ago. Intel's traditional PC customers no doubt have enjoyed Apple moving the bulk of their Mac sales (volume wise at least) to ARM already and freeing up Intel's capacity for their needs.
Please share what you think the probability is it launches in October. WCCF said it won't, I don't think there is any other source.
Ajay
Lifer
- Jan 8, 2001
- 15,448
- 7,858
- 136
dullard
Elite Member
- May 21, 2001
- 25,059
- 3,410
- 126
The rumors are for an announcement in late October and availability in Nov.
Announce Oct 27, Available Nov 19 rumor: https://wccftech.com/intel-12th-gen...z690-motherboards-launching-on-19th-november/
Nov 4 rumor: https://www.tomshardware.com/news/msi-spills-possible-alder-lake-launch-date
Edit, another Nov 4 rumor: https://videocardz.com/newz/lenovo-...t-gaming-pc-with-12th-gen-core-alder-lake-cpu
Announce Oct 28 rumor: https://videocardz.com/newz/intel-finally-shows-real-alder-lake-cpus-for-the-first-time
Last edited:
People overhype Apple anyways. They don’t see to understand that the whole reason Apple chips appear to perform so well is that Apple uses a hybrid design similar to ADL-S (among other things). Apple has an efficiency advantage, but most of it comes from being on TSMC 5nm. My iPhone 13 Pro Max scores 1740/4809 in Geekbench, which is roughly where a Core i7-1185G7 would score, the Apple chip having 2 big and 4 small cores, and the Intel chip having 4 big cores.
For what it is worth, Intel does have some mobile chips with only 2 big cores planned. Perf/watt should be interesting.
Not even a dent in revenue. Apple has 10% of the marketshare currently, but most of those machines have Intel inside. At our company, for example, we are evaluating the M1, but thus far we have no incentive to use it. Future laptops may end up being Linux machines on x86. In our case, it is the desire to use the same platform for development as production. Apple doesn’t make server hardware, so…
The announcement is Oct 27th, the embargo and launch is early November.
Last edited:
Ajay
Lifer
- Jan 8, 2001
- 15,448
- 7,858
- 136
On the plus side - it appears that the answer includes the phase "real soon now"; so, not much wait time left.
Then we'll really have something to talk about for the 4th quarter
Any takers for the bet that AVX-512 might become a licensed feature in consumer CPUs?
Ajay
Lifer
- Jan 8, 2001
- 15,448
- 7,858
- 136
No. That would be stupid. Future HEDT - maybe.
Any takers for the bet that AVX-512 might become a licensed feature in consumer CPUs?
Need more info about what Intel's actually trying to do, or why this needs to be in the Linux Kernel.
AVX-512 isn't something that everyone needs. For those who do, something like $50 might be worth it.
Ajay
Lifer
- Jan 8, 2001
- 15,448
- 7,858
- 136
It doesn’t make any sense to me for consumer based systems. But, when Intel offers ADL based Xeons - then there maybe be opportunities for enabling features for low end workstations and servers. That makes sense to me.
Arkaign
Lifer
- Oct 27, 2006
- 20,736
- 1,377
- 126
Something I think would be cool to see as this progresses is an enthusiast/workstation CPU branch for Intel 'small' socket (1700 etc).
Something that has bothered me for the past decade or so with Intel is their insistence on putting IGP in ALL of their non HEDT SKUs, whether it's enabled or not. One of the mostly fair criticisms of Intel during the whole run honestly from Lynnwood/Sandy all the way to current day was their fairly modest IPC and cache improvements at each die shrink or new release. Yet over that same time they seemed to blow almost the entire budget for new transistors on a package to the IGP.
On the plus side, this makes for totally usable onboard graphics for even 4K panels and multi display setups for non gamers. And similarly some pretty effective hardware based acceleration for a few things with Quicksync, as Gamers Nexus has pointed out.
HOWEVER, it really put progression on the back burner for the folks who used dGPUs, and created an even worse comparison vs Ryzen than it needed to be. AMD didn't saddle mainline Zens with a huge amount of die space and transistors for graphics a lot of their customers didn't need. No, they did it right with a few APU products on that side of things, and didn't cripple their entire lineup like Intel did.
Look at this Coffee Lake for example, and imagine what they could have done without gimping it with an IGP :
I really believe that in most instances, we're better off with IGP on the Chipset vs on die. Leaves room for OEMs to optionally dedicate some GDDR for premium IGP options, and moves a heat-generating component off die so it can have less concentrated heat generation. And now that Intel is getting even more into dGPU offerings, it could enable an entire range of onboard options for Mobo builders to utilize, from none, to entry level basic stuff, to basic gaming/workstation stuff with some GDDR, perhaps up to 35-45W options on the high end that could make for fairly effective 1080p + FSR/Xess upscaled gaming.
Something that has bothered me for the past decade or so with Intel is their insistence on putting IGP in ALL of their non HEDT SKUs, whether it's enabled or not. One of the mostly fair criticisms of Intel during the whole run honestly from Lynnwood/Sandy all the way to current day was their fairly modest IPC and cache improvements at each die shrink or new release. Yet over that same time they seemed to blow almost the entire budget for new transistors on a package to the IGP.
On the plus side, this makes for totally usable onboard graphics for even 4K panels and multi display setups for non gamers. And similarly some pretty effective hardware based acceleration for a few things with Quicksync, as Gamers Nexus has pointed out.
HOWEVER, it really put progression on the back burner for the folks who used dGPUs, and created an even worse comparison vs Ryzen than it needed to be. AMD didn't saddle mainline Zens with a huge amount of die space and transistors for graphics a lot of their customers didn't need. No, they did it right with a few APU products on that side of things, and didn't cripple their entire lineup like Intel did.
Look at this Coffee Lake for example, and imagine what they could have done without gimping it with an IGP :
I really believe that in most instances, we're better off with IGP on the Chipset vs on die. Leaves room for OEMs to optionally dedicate some GDDR for premium IGP options, and moves a heat-generating component off die so it can have less concentrated heat generation. And now that Intel is getting even more into dGPU offerings, it could enable an entire range of onboard options for Mobo builders to utilize, from none, to entry level basic stuff, to basic gaming/workstation stuff with some GDDR, perhaps up to 35-45W options on the high end that could make for fairly effective 1080p + FSR/Xess upscaled gaming.
Attachments
dullard
Elite Member
- May 21, 2001
- 25,059
- 3,410
- 126
1) As a person who's computer fried in a power outage this year, in the middle of a graphics card shortage, having integrated graphics in a new CPU was wonderful. I didn't have to wait months or pay hundreds of dollars more than I wanted. I would never have purchased a rocket lake chip otherwise, but with no GPUs available, rocket lake it was. A computer with an Intel 11700 was $300 cheaper than the cheapest AMD computer that I could get immediately.
2) It isn't like Intel can just simply put on more cores in that space. That takes a major redesign, more core-to-core latency, and gives each core less power to play with (meaning it would kill Intel's main selling point the last few years of being faster in single threaded applications).
3) If you really don't need it, then you can buy the F versions with it turned off. Then the integrated graphics doesn't use power and the chip is slightly cheaper.
Last edited:
So according to Intel E core has same Integer IPC as SKL, that s way more accurate than comparing 4 E cores to 2C/4T SKL.
medium.com
Alder Lake introduces a new client SoC architecture that scales from powerful desktops to…
Author: Marcus Yam, Technology Evangelist at Intel
medium.com
Arkaign
Lifer
- Oct 27, 2006
- 20,736
- 1,377
- 126
I didn't say necessarily have NO CPUs with integrated graphics. Just that I think AMD's approach is better overall, and being able to choose by Mobo would be better still.
And designing multiple CPU lines isn't asking too much. AMD does it. Intel does it. This would just be a different approach. And no, having more cores doesn't immediately mean latency, unless you want to argue 11900k has worse latency than 11600k for example. Even per core makes no sense as a complaint because of the very granular per core clock controllability.
IGP as a die space hog is not without its own drawbacks. And on apples to apples comparison, you can see what those are (eg; Zen 5600X vs 5600G).
Even if you do absolutely nothing with the added die space other than add cache, that's low hanging fruit and pays major dividends. AMD had to cripple the cache in their Zen3 APUs to get it to fit the IGP.
Like everything else in life, nothing is really free, and there are tradeoffs in everything. I think Intel's overly stack wide decision to saddle everything with IGP, particularly at the cost of other potential progress from 2nd to 11th gen consumer SKUs is regrettable.
AMD did it better, and I think that's been fairly well shown by their product stacks. The criticism for AMD has been how poorly some of their products have been made available on that front by shortages and delays/slow to market. The Zen3 APUs were OEM only for far too long, and they also have really ceded value segments to Intel almost in their entirety, which is a risky proposition in terms of abandoning market segments.
And designing multiple CPU lines isn't asking too much. AMD does it. Intel does it. This would just be a different approach. And no, having more cores doesn't immediately mean latency, unless you want to argue 11900k has worse latency than 11600k for example. Even per core makes no sense as a complaint because of the very granular per core clock controllability.
IGP as a die space hog is not without its own drawbacks. And on apples to apples comparison, you can see what those are (eg; Zen 5600X vs 5600G).
Even if you do absolutely nothing with the added die space other than add cache, that's low hanging fruit and pays major dividends. AMD had to cripple the cache in their Zen3 APUs to get it to fit the IGP.
Like everything else in life, nothing is really free, and there are tradeoffs in everything. I think Intel's overly stack wide decision to saddle everything with IGP, particularly at the cost of other potential progress from 2nd to 11th gen consumer SKUs is regrettable.
AMD did it better, and I think that's been fairly well shown by their product stacks. The criticism for AMD has been how poorly some of their products have been made available on that front by shortages and delays/slow to market. The Zen3 APUs were OEM only for far too long, and they also have really ceded value segments to Intel almost in their entirety, which is a risky proposition in terms of abandoning market segments.
dullard
Elite Member
- May 21, 2001
- 25,059
- 3,410
- 126
I think you are confusing AMD's use of TSMC's better process with benefits of no integrated graphics. I don't think many people would agree that AMD is doing well because of their lack of integrated graphics on most of their chips. AMD is doing well since TSMC's current process is far better than Intel's and AMD chips were well designed for that fact.
As for Intel having core-to-core latency issues when they add more cores, please see this: https://www.anandtech.com/show/15785/the-intel-comet-lake-review-skylake-we-go-again/4
Arkaign
Lifer
- Oct 27, 2006
- 20,736
- 1,377
- 126
We have direct examples of Zen products of equal Zen cores with and without IGP. There is a penalty and tradeoff in using potential die space and TDP budget for graphics. If the entire Ryzen lineup was albatrossed with IGPs, there is zero doubt they would have been less performant and effective at reaching higher performance tiers. Things like 12C and 16C might not have even been practical on 7nm to say nothing of previous fabs.
It's also not a zero sum comparison, of course great AMD Zen cores would impress either way, yet their more complete approach (notwithstanding 7nm value/entry) is definitely advantageous.
Intel's bland insistence on jamming IGP on all consumer socket SKUs has a real limitation that reduces the core and cache budget in a meaningful way, and it would simply be superior to have more choices from them that free some SKUs from that limitation.
As for core/cache, it's not special to IGP vs no IGP, that's just CPU design. They're now doing far more than 6-8 cores with Alder. It's not a valid point to reject even the option of more cores purely for that reason when simply offering additional tiers in a stack doesn't eliminate the potential for the other tiers to continue existing.
To a practical point, there is real value in having a more complete model stack. Of late, that has actually been an Intel strength at the low to mid tier. They've simply lacked competitive options on the high end something which IGP inarguably didn't help whatsoever, although I'll concede that it's not possible to know exactly how good a hypothetical IGP-less optimized Rocket Lake branch could have been. More than 0% improved is basically guaranteed however, even if they went the pure easiest route and loaded much of the wasted space with a cache tier.
It's also not a zero sum comparison, of course great AMD Zen cores would impress either way, yet their more complete approach (notwithstanding 7nm value/entry) is definitely advantageous.
Intel's bland insistence on jamming IGP on all consumer socket SKUs has a real limitation that reduces the core and cache budget in a meaningful way, and it would simply be superior to have more choices from them that free some SKUs from that limitation.
As for core/cache, it's not special to IGP vs no IGP, that's just CPU design. They're now doing far more than 6-8 cores with Alder. It's not a valid point to reject even the option of more cores purely for that reason when simply offering additional tiers in a stack doesn't eliminate the potential for the other tiers to continue existing.
To a practical point, there is real value in having a more complete model stack. Of late, that has actually been an Intel strength at the low to mid tier. They've simply lacked competitive options on the high end something which IGP inarguably didn't help whatsoever, although I'll concede that it's not possible to know exactly how good a hypothetical IGP-less optimized Rocket Lake branch could have been. More than 0% improved is basically guaranteed however, even if they went the pure easiest route and loaded much of the wasted space with a cache tier.
Arkaign
Lifer
- Oct 27, 2006
- 20,736
- 1,377
- 126
I hope not, that woupd artificially cap their potential. I do beliece more IGP options are good to have. More options period.
I've always thought that AMD should have included a basic video block on their chipset. Maybe 3 CU's, and an equivalent to quicksynch. It would have made their initial offerings more palatable to OEM's.
Arkaign
Lifer
- Oct 27, 2006
- 20,736
- 1,377
- 126
Agreed heavily on that front. I do think that was a real limiting factor to OEM success, and their laggardly APU offerings and of course the GPU shortages exacerbated this factor considerably.
coercitiv
Diamond Member
- Jan 24, 2014
- 6,198
- 11,891
- 136
If Zen4 is going the APU route as some leaks suggest, the GPU is going to be situated in the IOD. CPU chiplet area won't be affected. I hope the same will happen for Intel starting with Meteor Lake, in the sense that they'll decouple the GPU block from the CPU. (obviously keeping the iGPU for most consumers).
dullard
Elite Member
- May 21, 2001
- 25,059
- 3,410
- 126
Intel is on the path of decoupling them. They did it in the past in limited ways (for example, the Kaby Lake G chips). But it is also in Intel's long term plans for most chips. I just don't know if Meteor Lake is the start of it. They've been talking about it since ~2017 (https://newsroom.intel.com/newsroom...3/Murthy-Renduchintala-2017-Manufacturing.pdf), so it isn't something Intel is just jumping right into.
Last edited:
TRENDING THREADS
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 10K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 7K
-
-
Facebook
Twitter