NaturalViolence:

Do you not understand how hard it is to achieve a 10% IPC gain on a modern uarch? It takes billions in R&D since all of the "low hanging fruit" have already been picked (branch prediction, macro/micro op fusion, register renaming, etc.). Which is why 10% is the standard. If you look at samsung and AMD it's the same. The optimizations needed to do this stuff are insanely complex. People here seem to think that because Zen has a 61% higher IPC than it's predecessor that means AMD is improving faster than Intel but what they don't realize is this rate of improvement is actually consistent with Intel. There were 5 years of time between the release of piledriver and Zen. At 10% improvement per year that works out to....yup, 61%. But somehow 10% per year from Intel is "pathetic" while 61% after 5 years from AMD is "amazing". Intel CPU development is "dead" and "lol what are they doing with all that money?" while AMD is "leading the charge". Now Zen+ is out and it's 10% faster than Zen just as we would expect. So why is it that in the review thread the same people that complain about Intel's "small" annual performance improvements are praising Zen+? Why are these people still calling these performance improvements from Intel pathetic when AMD has now shown us a development roadmap that seems to be the same slow iterative annual release tick/tock cycle that Intel uses? This is not a fair opinion to hold, this is hypocrisy plain and simple. I'm sorry to be an asshole but there is no nice way to say this. You guys need to take some electronic engineering courses at a university to get an understanding of just how hard this stuff actually is. Or at the very least read some relevant books. I highly recommend starting with Jon Stokes "Inside the Machine" for people with limited understanding on existing trends in cpu design. I'm tired of seeing people scoff at what could possibly be the most complicated field of engineering that exists in the modern age. Even worse is that scoffing is only directed at one company it seems.

Anandtech found Intel's direct IPC gains to be about ~5.7% per generation (I wonder how this looks post spectre/meltdown) which isn't necessarily per year. I agree with you that I see a lot of people expecting AMD's massive 60% increase to continue to some extent - which it won't.. I've wrote several times here that most of what AMD caught up with in Zen was stuff Intel was already doing.. so I don't expect Zen 2 to do much over 6-8% IPC over Zen+. That being said, isn't it a little sad that Intel, a company with significantly more capital, isn't significantly outperforming AMD, a company that's been teetering on the verge of bankruptcy? I know designing these things is hard, I studied computer engineering at Rochester Institute of Technology - I'm fully aware that people sitting here expecting 50% performance increases year after year are delusional.. but there are definitely techniques and methods for improving computing performance/cost. Zen for example, as Kaarme said above, changed the core topology to allow a more cost effective solution to expanding to higher core counts - something Intel has had but never brought into the consumer space until after AMD did. Techniques like Speculative Multithreading could provide significant IPC improvements to workloads and yet we haven't really seen Intel work on that at all. Optical Interconnects and silicon photronics is something Intel has been kind of dabbling in, but again they don't really seem to be pushing a serious investment in it, let alone bringing it to a consumer product. Instead they spent their R&D money on stuff like their gadgets division, which is now closed and yielded nothing of interest. Now after 5+ years of stagnation, AMD caught up and suddenly Intel is going "oh maybe we should start investing in our engineering, close these useless divisions and actually provide a new product (aka ocean cove)". As a fairly informed customer how am I supposed to receive that? Because my take is that Intel has been resting on its laurels due to lack of competition in the market and now they are challenged and it's all "oh fuck we need to do something quick". I don't want to give them money anymore. My current machine is Intel but every custom desktop/HTPC computer I've built since Zen launched I've purposely gone out of my way to build with AMD, even when Intel offered a better deal. My next upgrade to my computer will also be with AMD. And unless Intel comes out with some product that's so impressive I can't resist, AMD will continue to get my support - they took massive risks with Zen topology/HBM/InfinityFabric/LLAPIs/etc in order to stay in the game and I want to continue to see them do that and reward them for it.

Well, eventually this race will be like trying to achieve absolute zero temperature.

you figure all the money intel spent on a dgpu they would have something in the 480/1060 range by now. would it be a good idea to make a vga that out produces a cpu lol it would be a race if they had a mid range they would be dangerous nv amd be like omg is intel going to put out a gpu legend? we need more power!!!!!~ gpu's would be 50 cents again I hope they do it halfway decent this time if its "gaming" cash they are after just our luck it would be a 20k pro user card

NaturalViolence:

Long winded post

I'm sorry that you don't understand when something is completely dead or not.

Denial:

every custom desktop/HTPC computer I've built since Zen launched I've purposely gone out of my way to build with AMD, even when Intel offered a better deal. My next upgrade to my computer will also be with AMD. And unless Intel comes out with some product that's so impressive I can't resist, AMD will continue to get my support - they took massive risks with Zen topology/HBM/InfinityFabric/LLAPIs/etc in order to stay in the game and I want to continue to see them do that and reward them for it.

I'm with you, I built 5 Ryzen systems last year along with several A series systems. Even supplied a couple of AMD laptops, Intel don't get a look in anymore unless a customer specifically requests it. Building a Ryzen 2200G system next week, quite looking forward to that 😀

Denial:

...

call 911, theres been a murder.

I wonder why Keller and some friends aren't opening a new company to design a brand new CPU architecture from scratch, something way better than x86 and move the world into a new era of computing. And he doesn't need expensive foundries to do that, since there's plenty of them already who are more than willing to produce the new silicon. Considering his track record, I'm sure there will be plenty of investors willing to put a few hundred millions of $ into such a project, considering the potential outcome !

^ That would be interesting 😎

Oh thank god someone here actually knows what they're talking about and doesn't respond to everything but just throwing out ad hominem's.

Denial:

Anandtech found Intel's direct IPC gains to be about ~5.7% per generation (I wonder how this looks post spectre/meltdown) which isn't necessarily per year. I agree with you that I see a lot of people expecting AMD's massive 60% increase to continue to some extent - which it won't.. I've wrote several times here that most of what AMD caught up with in Zen was stuff Intel was already doing.. so I don't expect Zen 2 to do much over 6-8% IPC over Zen+.

Correct, I meant per core performance. Sometimes it's due to IPC, sometimes clock rate, but usually a combination of both.

Denial:

That being said, isn't it a little sad that Intel, a company with significantly more capital, isn't significantly outperforming AMD, a company that's been teetering on the verge of bankruptcy?

Well they have focused more on per core performance which as you well know with your background is A LOT harder than increasing the core count and provides more consistent performance advantages across a broader range of software.

Denial:

I know designing these things is hard, I studied computer engineering at Rochester Institute of Technology - I'm fully aware that people sitting here expecting 50% performance increases year after year are delusional..

Exactly, and yet I get attacked every time I bring it up.

Denial:

Zen for example, as Kaarme said above, changed the core topology to allow a more cost effective solution to expanding to higher core counts - something Intel has had but never brought into the consumer space until after AMD did.

Agreed.

Denial:

Techniques like Speculative Multithreading could provide significant IPC improvements to workloads and yet we haven't really seen Intel work on that at all.

This is where I disagree. They've purchased companies in the past solely for the reason of doing R&D into this. They likely are working on it but as you well know it's quite complex and not fairly well proven at this point. Just because it's not out in a product yet doesn't mean that it isn't being worked on. And while it might work well in theory they may have had unforeseen issues.

Denial:

Optical Interconnects and silicon photronics is something Intel has been kind of dabbling in, but again they don't really seem to be pushing a serious investment in it, let alone bringing it to a consumer product.

Same as above. Assuming this is even practical to implement it would be so difficult to develop that there is no real reason to do it until we hit the point where further silicon development is more expensive. People often assume that throwing money at it will fix any technical issue but this might not be the case here.

Denial:

Instead they spent their R&D money on stuff like their gadgets division, which is now closed and yielded nothing of interest.

Yes that was legitimately stupid on their part. However I would like to point out to the other readers here that stupid is not the same thing as evil.

Denial:

Now after 5+ years of stagnation, AMD caught up and suddenly Intel is going "oh maybe we should start investing in our engineering, close these useless divisions and actually provide a new product (aka ocean cove)".

This I disagree with. "Start investing"? They've been spending billions every year on this and their investments have been increasing by 30% year to year consistently for years now. Nor has it ramped up more than usual in response to Zen. Hell I hate to say this but there is no evidence at this point that Zen's launch had any effect on them at all.

I see the migration continues ...