Is Intel done for?

[Kisai]

On 7/16/2023 at 9:25 PM, Man said:

• Their CPU architecture reached its peak with Kaby Lake (i7-7700K) or perhaps the short-lived i9-9900K, which was completely annihilated by Zen2 (Ryzen 7 3700X).

 

Not exactly. The CPU architecture "stagnated" with the entire DDR4 platform. 6700k-11700k, But this is because their foundry bets were off by a lot. The DDR3 platform (first to 5th gen) was roughly the same amount of time.

 

Like I only retired the 4th gen intel because it just bloody died, that's how I got onto the 11th gen.

 

The performance stagnates because the software Intel CPU's are used for largely don't take advantages of new features. Like look at AVX512. You simply can not use this feature because it generates so much extra heat, so what did Intel do? Nerf it. It no longer exists on 12th and 13th gen CPU's. That is a step backwards.

 

It did however incentivize Microsoft to release a new OS instead of incrementally updating Windows 10. Take a note at the alignment of CPU's. 2015 was Skylake, and Windows 10. NVMe SSD's were also introduced with 4th gen Intel CPU's, but it wasn't a standard feature on anything until the 6th gen, probably because the the M.2 standard didn't exist before then, and you'd have to waste PCIe slot otherwise (which is what I did at the time.)

 

Unfortunately Intel isn't really offering anything compelling. Either you need more (eg two GPU's and 4 SSD's, and thus 48 PCIe lanes connected to the CPU) or you need less (x8 + x4) on a SFF/ITX that isn't big enough to have full size GPU in it. Offering 24 lanes means you can't connect two GPU's, and when the "main GPU" requires 4 slots (don't tell me 2.8 slots isn't 3) you aren't fitting another in there because that second x16 slot will be in the wrong physical location. You end up needing a motherboard that has all X16 slots (like when motherboards used to be all 16-bit ISA slots, there has never been a time since when all slots were the same capability. These x1 slots serve no purpose, and a complex mux system is needed to re-direct lanes to slots that actually use more than one lane.)

 

At this point in time the only piece of hardware that is really steering hardware upgrades is the GPU. DDR5? early DDR5 is no better than high end DDR4, just like when DDR4 came out it was no better than high end DDR3. What about PCIe5? Most motherboards only connect the CPU PCIe lanes as DDR5, the Platform controller/southbridge ends up being PCIe3. Not helpful when that loses the reason to buy PCIe5 cards or SSD's.

 

I can pin point every time I upgraded the computer since the 8088. Only the 8088-80286-80386-80486-Pentium was "worth" upgrading as soon as possible because those Chips performance would double every 18 months and that would continue to be true until the Core Duo chips. Then guess what happened? Programmers refused to change.

 

And they still haven't changed. They are never going to change. Take a look at how many popular Linux packages prefer forking new copies of the application over threading. Chrome? Firefox? Single threaded Javascript. They gave up years ago. Likewise look how chrome spawns new browser processes for every iframe. Some of this, admittedly is due to security issues that just aren't solveable. But this also means that WebASM will never be threaded and never be appropriate for any use. This has the knock on effect of Unity (the game engine) also not being threaded.

 

Basically, because we've decided that multithreading is too hard and too insecure, developing more CPU's with more cores, is increasingly not helpful. Why bother having an 8-core CPU when anything that has to be run in parallel, is better off running on the GPU? At some point you're going to see Intel probably wanting to stick special-purpose ASIC on the cpu (eg hardware transformer's) because that's the only way they are eating Nvidia's lunch.

 

 

[Man]

On 7/19/2023 at 4:26 PM, Kisai said:

Like look at AVX512. You simply can not use this feature because it generates so much extra heat, so what did Intel do?

 

Actually, you're misinformed!

 

The reason Intel nuked AVX512 is because of big.LITTLE. The little cores don't support AVX512 instructions, just the big cores. As you can imagine, that created a lot of headaches for users so instead of fixing the issue by forcing AVX512 to the big cores, Intel downright amputated it from their CPUs. 

 

Kind of a... d!ck move, if you ask me!

On 7/19/2023 at 4:29 AM, tkitch said:

If someone wanted to disable some cores on a 12600k to get a 4c/8t chip, I'd be curious to see how a (nearly) 5GHz chip compares to a 7700k 5ghz.

That'd be a little... unfair, seeing that the i5-12600K has a pretty sizeable 20MB L3 cache, almost twice as much as Alder/Raptor Lake i3s. 

 

On 7/19/2023 at 4:37 AM, -rascal- said:

THEY need to sort out their power consumption of the RX 7000 series vs nVidia's RTX 4000 series.

Umm... no?!

 

The 7900XTX's power consumption is less than 50W more than the 4080 which is like what, +15%? That's a reasonable trade-off 'cause you're also getting ~10% performance bump in bandwidth intensive games. 

 

Let's not forget that the 7900XTX has a much wider bus (384-bit vs. 256-bit) with 24GB vRAM (12 vs. 8 GDDR6 chips), and a massive L3 cache (96MB vs. 64MB) so it's actually more efficient, especially when you consider that we are comparing a chiplet with a monolithic die. 

 

 

On 7/19/2023 at 4:37 AM, -rascal- said:

For comparison, look the the Moore Threads MTT S80 China-made GPU

 

Sure, Intel's doing okay, if you move the goalpost!

 

But the thing is, these unknown Chinese GPUs don't have Intel's money. I mean, Intel nabbed Raja Koduri, the fellow behind AMD's GCN + early 1st Gen. RDNA architectures. That's some serious pedigree. 

 

So yeah, I don't think Intel is doing great... with the kind of money and expertise they've (probably) thrown into Arc. 

 

On 7/19/2023 at 5:10 AM, da na said:

AMD hit their stride with the K6, K7, and K8 architectures but sharply fell off with the introduction of the Core 2 Duo, almost vanishing from the market for a solid ~8 years

Well, first-gen. K10s (Phenom vanilla) were marred by silicon issues (GloFo's 65nm was a mess) so they couldn't 'rev' up all that much higher than 2.5 GHz.  But Phenom II's, fabbed on 45nm, were actually quite competitive. Not top dogs, but competitive. 

 

I mean, AMD was taking pot-shots at premium Core 2 Duos with tri-core Phenom II X3s and quad-core Athlon II X4s with smaller cache pools than full-blown Phenoms. And then, of course, there were hexa-core Phenom II X6s for productivity tasks to compete with premium Core 2 Quads. 

 

Frankly, I regretted getting a Core 2 Duo E8400 instead of a Phenom II X3 or perhaps an Athlon II X4!

 

In any case, it was Bulldozer where things went down the sh!tter. But on the bright side, AMD only had two generations worth of 'duds.' They quickly learnt their lesson and made all the right choices with Ryzen, unlike Intel who stuck with Netburst for like what... 7+ years?!

[NF-A12x25]

On 7/17/2023 at 12:25 AM, Man said:

I've closely followed CPUs for the past two decades (I'm 34, by the way) and, honestly, I believe Intel has never been in a worse position.

• Their consumer CPU business is in shambles, and even in the datacenter market, their market share is steadily declining.
• They're currently in the midst of a CPU rebrand, desperately retiring the iconic i3/5/7/9 in an attempt to change consumer perception.
• Their CPU architecture reached its peak with Kaby Lake (i7-7700K) or perhaps the short-lived i9-9900K, which was completely annihilated by Zen2 (Ryzen 7 3700X).
• Their feeble attempt at "big.LITTLE" has failed miserably, lacking any redeeming qualities.
• Their foundry is falling far behind the likes of TSMC and even Samsung, becoming a cash sinkhole.
• Their newly established GPU division has struggled to gain any traction, securing less than 1% market share.
• Their foray into the discrete GPU industry has proven to be a major misstep, now that crypto mining is dead for the foreseeable future.

Despite all these challenges, Intel seems unwilling to change their ways. They persist in changing the socket every two generations, locking clock multipliers, restricting user voltage tweaks, limiting overclocking to high-end chipsets, and just screwing over users by doing things like disabling AVX-512.

 

The only thing they're willing to change is the brand, like that's going to help!

 

Given Intel's deep-rooted bureaucracy and anti-consumer practices, I simply don't see how the company can survive AMD's relentless assault.

Now, some might argue: "What about Netburst, a.k.a. Pentium 4 and Pentium D?"

 

Well, at least back then, their foundry business was a force to be reckoned with. They were rolling out 65nm CPUs in early 2006, at least a year ahead of both TSMC and Global Foundry. So, at least their foundry business was at the top. But now... well, I'm sure you're already know!

this is an amazing shitpost, please keep it up.

[ImorallySourcedElectrons]

On 7/17/2023 at 6:25 AM, Man said:

I've closely followed CPUs for the past two decades (I'm 34, by the way) and, honestly, I believe Intel has never been in a worse position.

You'd be quite wrong about that.

 

On 7/17/2023 at 6:25 AM, Man said:

• Their consumer CPU business is in shambles, and even in the datacenter market, their market share is steadily declining.

I'll let you in on a nice industry secret, TSMC and Samsung don't have enough capacity to fill every order Intel is filling, even if their CPUs were garbage (and they're not) they'd still be selling like hot cakes. How do you think AMD survived a decade of being irrelevant?

 

On 7/17/2023 at 6:25 AM, Man said:

• They're currently in the midst of a CPU rebrand, desperately retiring the iconic i3/5/7/9 in an attempt to change consumer perception.

This is more an effect of crappy reporting in the media than anything else.

 

On 7/17/2023 at 6:25 AM, Man said:

• Their CPU architecture reached its peak with Kaby Lake (i7-7700K) or perhaps the short-lived i9-9900K, which was completely annihilated by Zen2 (Ryzen 7 3700X).

The 12000 series CPUs are doing pretty well, they're also consistently beating AMD for single threaded performance. Additionally AMD still hasn't really been able to move into the professional laptop market that has a lot of money in it.

 

On 7/17/2023 at 6:25 AM, Man said:

• Their feeble attempt at "big.LITTLE" has failed miserably, lacking any redeeming qualities.

Works well, everyone else is copying it.

 

On 7/17/2023 at 6:25 AM, Man said:

• Their foundry is falling far behind the likes of TSMC and even Samsung, becoming a cash sinkhole.

May I suggest you actually stop reading garbage reporting on electronics manufacturing processes? The reality is that Samsung, TSMC and Samsung are about on par with each other, and each has different accents on their processes. Furthermore, the density figures cited are always for highly repetitive structures like RAM and are by no means a realistic representation of what you can achieve logic-wise in a technology, like for ages Intel was able to make far denser memory than anyone else, but that has relatively little use unless if you're solely entering that market, practical performance is limited by power per surface area. And foundries are always cash sink holes during construction and process development, I'm not sure why everyone is suddenly focussing on that so much.

 

On 7/17/2023 at 6:25 AM, Man said:

• Their newly established GPU division has struggled to gain any traction, securing less than 1% market share.

Please note that you're casually neglecting to include industrial customers looking for accelerator cards.

 

On 7/17/2023 at 6:25 AM, Man said:

• Their foray into the discrete GPU industry has proven to be a major misstep, now that crypto mining is dead for the foreseeable future.

Which has been supplanted by the AI craze, for which Intel is surprisingly well equipped.

 

On 7/17/2023 at 6:25 AM, Man said:

Despite all these challenges, Intel seems unwilling to change their ways. They persist in changing the socket every two generations, locking clock multipliers, restricting user voltage tweaks, limiting overclocking to high-end chipsets, and just screwing over users by doing things like disabling AVX-512.

The change in socket has consistently been necessary to hit signal integrity requirements on new and updated communication and interface standards, this is why Intel's USB, PCIe, etc. all manage to hit their spec, while AMD's implementation is rather questionable at best until a couple of years in. And the reality is that most users don't really care about the latter things you mentioned. 

 

On 7/17/2023 at 6:25 AM, Man said:

Well, at least back then, their foundry business was a force to be reckoned with. They were rolling out 65nm CPUs in early 2006, at least a year ahead of both TSMC and Global Foundry. So, at least their foundry business was at the top. But now... well, I'm sure you're already know!

Their foundry business is not falling behind. Actually pay some attention to what's really going on:

• Many of TSMC's new process nodes aren't really scaling anything, ten years ago some of these would have been a minor PDK update instead of a new process node.
• TSMC is really good at marketing, but most of the marketing is complete meaningless wank from an engineering point of view.
• Intel is a favourite target of particular institutional investors who are creating a narrative that requires Intel to divest it's fab business, which would result in a large payout for them. The same happened with AMD and Global Foundries. It'd be really bad for the industry over all if Intel were to be split up, because Intel the hardware manufacturer is propping up Intel the Fab at times, and the other way around.  It's why Intel is a financially healthy company compared to AMD.
• Intel has some yield problems, but at the same time they're also using multiple patterning heavily compared to TSMC's and Samsung's focus on EUV, this has advantages and disadvantages, especially because EUV machines have a tendency to have a pretty shitty throughput. So it can actually be economical and preferable for your overall throughput to let your yield slip and switch to multiple patterning because your throughput becomes far higher. So the actual pros and cons are far more nuanced than most reporting makes it out to be.
• Global Foundries is a prime example of why Intel should hold onto its fab business.

Basically, stop taking the 5 minute summary from YouTube videos and actually familiarize yourself with the subject.

[Mihle]

Even if everything OP said was true(it's not) none of those things would prove Intel is "done for". Maybe it would prove that they have/will have a dip, but it doesn't prove that they won't do better in 7 or 10 years time again or whatever.

[tkitch]

2 hours ago, Man said:

That'd be a little... unfair, seeing that the i5-12600K has a pretty sizeable 20MB L3 cache, almost twice as much as Alder/Raptor Lake i3s. 

Or you're wrong?

 

As @RONOTHAN## pointed out:
 

 

Overclocking a 12100 using B-Clock overclocks (on the few boards that supported it) to 5 GHz got it 30-40% gains over non OC'd.

 

So, your argument is hella moot.  

 

12100 is waaaay better than the 7700k and 9900k.  End of story.  Your argument is bad, and you should quit before it gets even worse.

[cmndr]

On 7/18/2023 at 4:21 PM, Man said:

I can see some point of big.LITTLE in mobile space but desktop?

 

I'd much rather take grown-up P-Cores than baby E-Cores which are more or less worthless. Personally, Intel mostly bothered with big.LITTLE to skew multi-core synthetic benchmark resuls in their favor!

 

Plus, they nuked AVX-512 because of big.LITTLE so...

As touched on, if you NEED big cores, it's often only a few. After you have "enough" big cores, there's not much of a benefit to having even more. From there trading 1 big core for 4x small cores (similar die space) improves overall performance AND energy efficiency. Remember the small cores are still similar to Zen 2 and Skylake in terms of IPC, just no SMT. The 13900k basically has a 7800x and a 3900x bolted together when it comes to performance. Which isn't bad. If I could bolt a 7800x to my 3900x I wouldn't feel an itch to upgrade. 

Intel basically traded a bit of short to mid-term jankiness for the ability to nearly double MT performance without hurting ST performance. 


The AVX-512 complaint DOES matter in some cases. With that said, hardly anyone was making a big deal out of it when comparing AMD vs Intel in the past. 

[-rascal-]

9 hours ago, Man said:

Umm... no?!

 

The 7900XTX's power consumption is less than 50W more than the 4080 which is like what, +15%? That's a reasonable trade-off 'cause you're also getting ~10% performance bump in bandwidth intensive games. 

 

Let's not forget that the 7900XTX has a much wider bus (384-bit vs. 256-bit) with 24GB vRAM (12 vs. 8 GDDR6 chips), and a massive L3 cache (96MB vs. 64MB) so it's actually more efficient, especially when you consider that we are comparing a chiplet with a monolithic die. 

 

 

The difference between a RTX 4080 and 7900 XTX can be up to 200W.

 

Additionally, the 7900 XT / XTX experience high transient power spikes.

For example GN reported the 7900 XTX reach up to 725W in Doom Eternal.

This may not be a problem for some, while it will be for others.

 

9 hours ago, Man said:

Sure, Intel's doing okay, if you move the goalpost!

 

But the thing is, these unknown Chinese GPUs don't have Intel's money. I mean, Intel nabbed Raja Koduri, the fellow behind AMD's GCN + early 1st Gen. RDNA architectures. That's some serious pedigree. 

 

So yeah, I don't think Intel is doing great... with the kind of money and expertise they've (probably) thrown into Arc. 

 

What are you talking about?

This is DX11 / DX12 rasterization performance (no FSR, DLSS), at the same resolution, same PC configuration.

 

A 1st Gen Intel ARC GPU is able to perform at the level of a RTX 3060.

As of the RTX 3060 release, nVidia has 27+ years of GPU assets, patents, and experience.

That is NOT an impressive achievement?

 

Do you know why there hasn't been a third company to compete with AMD + nVidia?

How can you compete?! It is literally a monopoly between the two. They are TOO far ahead.

If your first generation GPU cannot even compete with a GTX 750 Ti from 2014, a private company would IMMEDIATELY go bankrupt.

 

Moore Thread is FUNDED by the Chinese government.

The Chinese government wants a piece of the silicon pie, so they are heavily funding start-ups.

Plus, the CEO who started Moore Thread was the FORMER VP of nVidia China.

I can certainly bet you, he probably packed a couple of important design documents, etc, when he departed from nVidia.

Not only that, they headhunted a BUNCH of Engineers from nVidia, Intel, ARM.

They have 9+ year experience Senior GPU Arch Engineers from nVidia, former Engineers from Huawei, AMD, Qualcomm.

( MOST of the people working there ARE former nVidia)

 

Spoiler

When the Chinese government gets involved, money does not really become an issue.

When I was visiting Shang-Hai in 2019, NIO (the Chinese BEV company) was literally draining money.

Tencent was planning to heavily invest in them, but for some reason backed out.

Then the Chinese government stepped in; they have been dumping money, in the $ Billions of USD in state funding, since 2020, into NIO.

 

[xg32]

On 7/16/2023 at 9:25 PM, Man said:

 

i own a 5800x3d, 5950x and a 9900k, i would not touch the 3700x with a 10ft pole.

 

intel was more screwed back in the opteron/athlon days before conroe came out, or when amd stock was at 2dollars.

 

The lead amd has isnt that big, it's mainly just the 7950x/3d, 5800x3d and 7800x3d, i don't think intel has an answer atm to informed consumers.

 

i've had 2 friends recently order 4090 pre-builts, they both went intel, simply cause amd chips werent available, or intel still has branding power over amd, they also didnt give a damn about power consumption, which is the majority of ppl.

 

This is coming from some1 who's top cpu choices atm are 5800x3d, 7800x3d,  7950x3d  and have no interest in intel chips. 

 

 

[Kisai]

8 hours ago, Man said:

 

Actually, you're misinformed!

 

The reason Intel nuked AVX512 is because of big.LITTLE. The little cores don't support AVX512 instructions, just the big cores. As you can imagine, that created a lot of headaches for users so instead of fixing the issue by forcing AVX512 to the big cores, Intel downright amputated it from their CPUs. 

 

Kind of a... d!ck move, if you ask me!

That'd be a little... unfair, seeing that the i5-12600K has a pretty sizeable 20MB L3 cache, almost twice as much as Alder/Raptor Lake i3s. 

 

I'm not misinformed. Literately, every software compiler out there defaults to "no AVX512" and when you use tools that can use it, they warn that the performance of AVX512 causes the cores to throttle down, so make sure this is really what you want to do. OneDNN, Pytorch and Tensorflow all basically throw "are you sure??" at you.

https://github.com/pytorch/pytorch/pull/80253

https://github.com/pytorch/pytorch/issues/80252

 

Because of the "big.LITTLE" design in 12th/13th gen, it's not possible to send an AVX512 workload to an E core, and Windows 10 (and likely older Linux Kernels) are not aware of how to manage this. Intel decided to then nerf AVX512 from the 12th and 13th gen CPU's despite the P cores die having the feature.

 

So as a result, 11th gen is the last Intel Desktop CPU that supports Intel AVX512, and as a consequence no desktop software can be written to utilize it with the assumption it's available.

 

It's unfortunate, because that pretty much makes AVX512 dead, and cedes AI progress to it's competitors. Nobody is going to buy a much more expensive Xeon CPU that the AVX512 is enabled on when a GPU can be had for half the price and 10x the performance.

 

 

[dizmo]

Haha, hilarious. I don't think you've closely followed them if you're asking questions like this.

Intel is very, very far from becoming irrelevant. They're just trimming the fat, as any smart business would. As for your specific points?

• Their consumer CPU business is far from in shambles. Did you know AMD reported a 64% decline in CPU sales? Intel only reported 36%. So, by your logic, AMD is twice as likely to fail as Intel.
• Rebrands happen. Does it matter? Not in the least. Nerds and those in the space will know what the changes are and begrudgingly accept it, everyone else had no real idea what the numbers and letters meant anyway.
• I wouldn't say the 9900k was annihilated. IIRC it still performed better in gaming.
• If you're just looking at node size, perhaps. However their CPUs, even on higher nodes, still perform just as well. Node size isn't everything. You should know this.
• Of course they struggled to gain traction, they couldn't make enough cards, had trouble with drivers, etc. It's a launch. If you expect them to have massive gains immediately you've lost the plot entirely. It takes years to gain market share. Do you consider AMD to be a failure in the GPU market? Their share is also incredibly low, considering they've been at it for decades.

Other points:

• While they've fallen a bit, Intel's stocks are at some of the highest they've ever been.
• Intel is working with the US Government to design a new power grid. If that's not a license to print money, I don't know what is.
• Intel revenue '22: $63 billion with a net income of $8.02 billion
  AMD revenue '22:  $23.6 billion with a net income of $1.32 billion
  Tell me, which of those companies is doing better?

[CDR_Xavier]

You would think that as the years went by, AMD's platform is now neck-to-neck compared to Intel ones.

Yet to this day, Intel platforms are still more reliable than AMD's offerings by a significant margin. Both on the mobile sector and the desktop sector.

 

Now, I'm not saying AMD is bad, but from what I have experienced I am not convinced they are on equal footing yet.

[Monkey Dust]

Gaming is only a tiny part of the market, so what we think is largely irrelevant. The server market moving AMD will hurt far more, but...

In the bulk of the PC market, Intel still dominate, because they have brand recognition, and quite likely exclusivity deals with manufacturers. If you've been issued a laptop for work, it's probably Intel. The bulk of laptops for sale are Intel, their sales still dwarf AMD.

 

What might put a huge dent in their sales is a move to ARM. The power efficiency versus x86 is insane. Especially important, as for almost all everyone now, us aside, PC = Laptop. ARM allows longer battery life, or for cheaper machines significantly smaller batteries. In 10 years time, I reckon there is a solid chance that x86 will be niche. And if the consoles go ARM next time around, PC gamers will too.

[BabaGanuche]

First Intel is not screwed. They may lose some marker share, but they cannot lose too much just because the rest of the industry does not have the capacity. They do need to be careful as they are no longer ahead.

 

1 hour ago, Monkey Dust said:

The power efficiency versus x86 is insane.

I wish people would stop saying this, as it is not true. The ISA is not very important when it comes to efficiency. Look at the efficiency differences between Intel P and E cores AMD Zen 4 and Zen 4c cores. That alone is evidence that efficiency is very decoupled from the ISA.

 

Here is a case where x86 is more efficient than ARM. The 128 core AMD Bergamo x86 CPU is more efficient than the Altera Amper 128 core ARM CPU. This a good apples to apples comparison between the two as they are both current gen and targeting the same market, cloud service providers.

 

[aisle9]

On 7/17/2023 at 12:25 AM, Man said:

I've closely followed CPUs for the past two decades (I'm 34, by the way) and, honestly, I believe Intel has never been in a worse position.

• Their consumer CPU business is in shambles, and even in the datacenter market, their market share is steadily declining.
• They're currently in the midst of a CPU rebrand, desperately retiring the iconic i3/5/7/9 in an attempt to change consumer perception.
• Their CPU architecture reached its peak with Kaby Lake (i7-7700K) or perhaps the short-lived i9-9900K, which was completely annihilated by Zen2 (Ryzen 7 3700X).
• Their feeble attempt at "big.LITTLE" has failed miserably, lacking any redeeming qualities.
• Their foundry is falling far behind the likes of TSMC and even Samsung, becoming a cash sinkhole.
• Their newly established GPU division has struggled to gain any traction, securing less than 1% market share.
• Their foray into the discrete GPU industry has proven to be a major misstep, now that crypto mining is dead for the foreseeable future.

Despite all these challenges, Intel seems unwilling to change their ways. They persist in changing the socket every two generations, locking clock multipliers, restricting user voltage tweaks, limiting overclocking to high-end chipsets, and just screwing over users by doing things like disabling AVX-512.

 

The only thing they're willing to change is the brand, like that's going to help!

 

Given Intel's deep-rooted bureaucracy and anti-consumer practices, I simply don't see how the company can survive AMD's relentless assault.

Now, some might argue: "What about Netburst, a.k.a. Pentium 4 and Pentium D?"

 

Well, at least back then, their foundry business was a force to be reckoned with. They were rolling out 65nm CPUs in early 2006, at least a year ahead of both TSMC and Global Foundry. So, at least their foundry business was at the top. But now... well, I'm sure you're already know!

Jesus Christ how am I just now seeing this?

 

Intel could completely walk away from selling standalone CPUs and they'd be fine. They could just wake up one day, say, "Screw you guys, I'm going home," and take their processors off of store shelves never to be seen again. It would not matter to them. Like, at all. They have such a stranglehold on the market when it comes to office desktops, workstations and even budget to midrange non-gaming-oriented home PCs that whatever hit their bottom line took would not be anywhere near a killer. They can package their defective CPUs up and call them i7-13700T, or if the CPU is really f**ked, a Pentium or Celeron or i3 or whatever, then sell those off to OEMs like crazy.

 

I don't really get you saying that the Pointless Lake i7-7700K was the peak of their architecture when it was just another lazy 5% IPC gain chip during Intel's lazy, monopolistic era, dropped as a knee-jerk reaction to Ryzen while the Coffee Lake series was readied for market. Intel continued bothering with consumer CPUs after AMD handed Netburst's ass to it for the same reason they'll do so here: the incremental cost of selling their surplus of CPUs not needed for OEM machines or that perform well enough to be sold at a premium is less than the profits to be made by doing so. If that ever changes, then Intel will just exit the consumer market. They'll be fine, and we'll all be f**ked because our choices will be an AMD facing the same competitive landscape that Intel did ca. 2012-2017 or...ugh...Apple.

 

As long as AMD's "relentless assault" continues to include stupid-overpriced CPUs and minimal disruption in the incredibly profitable enterprise/OEM space, Intel's fine. And before you call me a fanboy, the PC I'm typing this on has a Ryzen 5 5600 powering it. My HTPC has an i7-8086K. I have two mini PCs, one using an R5 5600H, the other using an N95. I do not give two shits whose CPU is in my system. My shits are reserved for how much it costs and how much performance I get for the price.

[Beerzerker]

On 7/18/2023 at 6:21 PM, Man said:

Everyone does that, mate. 

 

i5 CPUs are essentially cut-down i7s!

 

I've known that for a very long time because each and every chip made by Intel is made with the intent of it to be a full blown i7 if possible..... But as we all know there will be flaws in the process of making them.
Actually this is how it is for AMD too and anyone else making chips.

So... Instead of just tossing the silicon out, they are binned to see what's good and what's not per chip and what they finally become is based on the chip's binning results.
That's why Intel makes i5's, i3's and all the rest that's not a full blown i7 or whatever they are calling them these days.

That's also how alot of AM3 AMD chips had hidden cores, it's just that these didn't make the cut to be a fully functional quad core so they instead made them into tri-core or dual cored chips with the ones only having two cores in them (Regor core) being cut down to just one core for the Semprons.
Same also goes for the cache in these, that also affecting what they finally became and was the basis of a few chip models too.

Same process for AM3+ and AM4's too.... In fact it's always been that way so I mean it's like, "NO SHIT Sherlock" because we all know that.

You bring nothing new to the table here.
 

[Uttamattamakin]

Short answer is that these things are cyclic.  Early on in the 1980's there were so many architectures at first, then just 8086... which had clones of it, and AMD making licensed copies due to a deal with IBM.  

Then Pentium came and AMD's CPUs were for some cases better than Intels and vice versa.   Intel was DOMINANT though.  Then AMD Athlon was better than Pentium.  Then Pentium 4 wast best then Athlon 64  then Core for a LONG Time then Ryzen has slowly taken over in a lot of areas. Various ARM chips are making a play but I don't forsee ARM replacing X86.  I see X86 being implemented as hardware level emulation in a chip that is RISC based and has arm like properties.   Whoever does that first will own the market since power efficiency is always important. 

That is also cyclic.  CPU's got more and more Hz for decades up to Pentium4.  Then Athlon 64 and core were more efficient and more performant.  We are due for another revolution that leads to more efficient, more performant chips.  If for no other reason than there are hard physical limits to how much power can be channeled into a microchip before it becomes unmanageable.  (Too much heat, or too much current in small wires that are too close together etc etc.)