Image: Intel

After only two years on the job, Bob Swan has been ousted from his role as chief executive officer of Intel. Swan’s removal was confirmed by a press release shared by Intel today, which noted that his last day would be February 15. Replacing Swan is VMware CEO Pat Gelsinger, whose 40-year experience in the industry will hopefully be enough to right what some critics say is a slowly sinking ship.

“I am thrilled to rejoin and lead Intel forward at this important time for the company, our industry and our nation,” said Gelsinger. “Having begun my career at Intel and learned at the feet of Grove, Noyce and Moore, it’s my privilege and honor to return in this leadership capacity. I have tremendous regard for the company’s rich history and powerful technologies that have created the world’s digital infrastructure. I believe Intel has significant potential to continue to reshape the future of technology and look forward to working with the incredibly talented global Intel team to accelerate innovation and create value for our customers and shareholders.”

“My goal over the past two years has been to position Intel for a new era of distributed intelligence, improving execution to strengthen our core CPU franchise and extending our reach to accelerate growth,” said Swan. “With significant progress made across those priorities, we’re now at the right juncture to make this transition to the next leader of Intel. I am fully supportive of the board’s selection of Pat and have great confidence that, under his leadership and the rest of the management team, Intel will continue to lead the market as one of the world’s most influential technology companies.”

Swan had been CEO at Intel since January 2019. We are hearing that investors are cheering at the news of his removal, which is expected, being that Intel had slipped behind various competitors such as AMD and NVIDIA during his tenure. In addition to chip production troubles, the company is also stressing with the loss of major customers such as Apple, which is dumping Core processors for its in-house, ARM-based M1 SoCs.

Recent Posts

Join the Conversation

35 Comments

  1. Not sure everything that has happened at Intel is his fault – there’s been a lot wrong over there for a while, and a ship that big takes a bit to turn around. But I don’t know enough about what happens in the boardroom to play too much armchair QB over it — I will say this was the loudest announcement anyone made for CES though.
  2. Not sure everything that has happened at Intel is his fault – there’s been a lot wrong over there for a while, and a ship that big takes a bit to turn around. But I don’t know enough about what happens in the boardroom to play too much armchair QB over it — I will say this was the loudest announcement anyone made for CES though.

    Agreed. One of Intel’s issues, their "diversity" initiative, started in 2015 before Swan took over.

  3. I think their biggest issue was complacency. For the past 14 years they told people what they needed instead of giving people what they wanted. AMD came along and gave everyone what they wanted. Regardless of what Swan did there was no righting that ship in 2 years.

    Gelsinger going come in and sell you a 16 core CPU, but only allow you to use 4 cores and make you buy licensing for the remaining 12.

  4. Regardless of what Swan did there was no righting that ship in 2 years.

    Yeah, this is my take too.

    Intel’s problem is twofold.

    1.) Once they crushed AMD with their mix of illegal business practices and the Core architecture, they became complacent. Their business culture forgot how to be competitive, innovate and succeed. They essentially turned their business into a cash cow, misleading themselves into thinking it would last forever. Now both AMD and ARM are slowly turning into real threats.

    2.) Spectacular failures in process node development. This was one of the ways they traditionally stayed ahead, with process excellence, and they just dropped th eball completely. I don’t know what the root cause here is. Poor Communication? Not hiring the right people? Not funding it sufficiently for the increasing complexities of smaller process nodes? Who knows.

    What I do know is this. Process node development is a several year long effort costing billions. There is no way that could be fixed in two years. Changing a corporate culture is even more difficult. That is unlikely to be successful in two years either. Whatever Intel’s board does they need to put the right person in place, and empower them to do whatever is necessary to fix the problems, and then be patient while they execute, because this is going to take a while.

    As much as I hate what has become of MSI these days, Intel needs to do what MSI did in 2012. At that point MSI was slowly failing, failing to have adapted to the changing market. The founder, Henry Lu announced in a company wide meeting that they would be transitioning to a gaming focused company, because that was where the market was headed, and if anyone wasn’t fully committed to this, they should leave the company. I hate the "gaming" aesthetic that has taken over th eindustry, but it was unquuestionably what they needed to do the business climate as it was and still is.

    Now, I’m not suggesting that Intel needs to make gaming their focus, but they need to make it very clear that they are in crisis, that they are setting a new course (whatever that is) and that those who aren’t completely on board should either leave, or they will be fired.

    In an organization the size of Intel there are always middle managers and directors more focused on their internal power struggles and building their own little fiefdoms within the company that they have lost track of what it takes to make the entire organization successful.

    Intel needs to set a clear course, clear objectives, and they need to make it clear that anyone who gets in the way will be out the door faster than you can say 14+++, and then they need to follow up on this and not make exceptions, regardless of how long Director X has been with the company or who they know or are related to. They need to cut out the cancer and they need to do it mercilessly. That is the only way.

    This is often why organizations in turnaround mode hire outside CEO’s. Existing company guys have often made too many cozy relationships and aquired too many pet projects along the way and can’t bring themselves to ruthlessly do what needs to be done. Now Bob Swan was by no means a long term Intel guy, having only joined the company in 2016, but maybe that was enough time that he already had too many existing relationships in the organization. A complete outsider may be the best choice Intel has, and is probably what they should have done two years ago, rather than promote Bob Swan. I know nothing about him as an executive. He may be a great business leader, but turnarounds require a special subset of talents and lack of regard for "friends" inside the company, and maybe Bob Swan just didn’t have it.

  5. Yeah, this is my take too.

    Intel’s problem is twofold.

    1.) Once they crushed AMD with their mix of illegal business practices and the Core architecture, they became complacent. Their business culture forgot how to be competitive, innovate and succeed. They essentially turned their business into a cash cow, misleading themselves into thinking it would last forever. Now both AMD and ARM are slowly turning into real threats.

    2.) Spectacular failures in process node development. This was one of the ways they traditionally stayed ahead, with process excellence, and they just dropped th eball completely. I don’t know what the root cause here is. Poor Communication? Not hiring the right people? Not funding it sufficiently for the increasing complexities of smaller process nodes? Who knows.

    What I do know is this. Process node development is a several year long effort costing billions. There is no way that could be fixed in two years. Changing a corporate culture is even more difficult. That is unlikely to be successful in two years either. Whatever Intel’s board does they need to put the right person in place, and empower them to do whatever is necessary to fix the problems, and then be patient while they execute, because this is going to take a while.

    As much as I hate what has become of MSI these days, Intel needs to do what MSI did in 2012. At that point MSI was slowly failing, failing to have adapted to the changing market. The founder, Henry Lu announced in a company wide meeting that they would be transitioning to a gaming focused company, because that was where the market was headed, and if anyone wasn’t fully committed to this, they should leave the company. I hate the "gaming" aesthetic that has taken over th eindustry, but it was unquuestionably what they needed to do the business climate as it was and still is.

    Now, I’m not suggesting that Intel needs to make gaming their focus, but they need to make it very clear that they are in crisis, that they are setting a new course (whatever that is) and that those who aren’t completely on board should either leave, or they will be fired.

    In an organization the size of Intel there are always middle managers and directors more focused on their internal power struggles and building their own little fiefdoms within the company that they have lost track of what it takes to make the entire organization successful.

    Intel needs to set a clear course, clear objectives, and they need to make it clear that anyone who gets in the way will be out the door faster than you can say 14+++, and then they need to follow up on this and not make exceptions, regardless of how long Director X has been with the company or who they know or are related to. They need to cut out the cancer and they need to do it mercilessly. That is the only way.

    This is often why organizations in turnaround mode hire outside CEO’s. Existing company guys have often made too many cozy relationships and aquired too many pet projects along the way and can’t bring themselves to ruthlessly do what needs to be done. Now Bob Swan was by no means a long term Intel guy, having only joined the company in 2016, but maybe that was enough time that he already had too many existing relationships in the organization. A complete outsider may be the best choice Intel has, and is probably what they should have done two years ago, rather than promote Bob Swan. I know nothing about him as an executive. He may be a great business leader, but turnarounds require a special subset of talents and lack of regard for "friends" inside the company, and maybe Bob Swan just didn’t have it.

    Gelsinger probably knows more people at Intel than Swan, after working there for a few decades in the past. Hopefully he has enough perspective coming back to be impartial and make the right calls.

  6. 1.) Once they crushed AMD with their mix of illegal business practices and the Core architecture, they became complacent. Their business culture forgot how to be competitive, innovate and succeed. They essentially turned their business into a cash cow, misleading themselves into thinking it would last forever. Now both AMD and ARM are slowly turning into real threats.

    2.) Spectacular failures in process node development. This was one of the ways they traditionally stayed ahead, with process excellence, and they just dropped th eball completely. I don’t know what the root cause here is. Poor Communication? Not hiring the right people? Not funding it sufficiently for the increasing complexities of smaller process nodes? Who knows.

    Honestly I see these two as tightly intertwined. The fact that Intel is still putting out competitive processing power with competent overall performance at 14nm is a testament to the rest of the company not dropping the ball completely.

    But of course their ‘cash cow’ comes from being able to actually produce the stuff that they design.

    AMD came along and gave everyone what they wanted.

    AMD really hasn’t even come close to that. Not that they could actually make the products if they did; they’re more supply-constrained than Intel.

    As examples: AMD could make Nvidia obsolete in the laptop space. The only reason that an eight-core APU with RTX2060-level performance doesn’t exist is because AMD hasn’t made it, when they very well could.

    They could also put a basic GPU on their IO die, so that every Ryzen shipped with graphics hardware. Doesn’t need a lot rendering horsepower, no more than Intel’s IGPs, but it would need up to date support for transcoding video. And you know, the driver and software support to use that hardware, which we know AMD continues to struggle with.

    Oh! And consumer APUs that support ECC. Thanks for going full Intel on that one, AMD. But wait, Intel has i3’s that support ECC, when used with their server chipsets. Figure that juxtaposition out.

    Last, despite AMDs supposed Linux prowess, switching between the graphics on an AMD APU and an AMD dGPU is apparently more assed up than it is with Intel / Nvidia.

  7. Gelsinger probably knows more people at Intel than Swan, after working there for a few decades in the past. Hopefully he has enough perspective coming back to be impartial and make the right calls.

    Ah, I didn’t know he was a former Intel guy. Looks like it has been 12 years since he left Intel to join EMC.

    I have no idea what turnover is like inside Intel. I know it varies greatly from industry to industry and company to company. There are companies I left 12 years ago I could walk into and not recognize a single face or project. Typical turnover in my industry befopre you get a better offer somewhere else is about 2-4 years. No one really ever stays longer than that. I wonce worked for an organization in a different industry, and there it was the polar opposite. Lots of people who had been with the company for 30+ years and had never worked anywhere else…

    Hopefully he has been distant from it long enough that he can make the tough decisions that need to be made.

  8. Honestly I see these two as tightly intertwined. The fact that Intel is still putting out competitive processing power with competent overall performance at 14nm is a testament to the rest of the company not dropping the ball completely.

    But of course their ‘cash cow’ comes from being able to actually produce the stuff that they design.

    AMD really hasn’t even come close to that. Not that they could actually make the products if they did; they’re more supply-constrained than Intel.

    As examples: AMD could make Nvidia obsolete in the laptop space. The only reason that an eight-core APU with RTX2060-level performance doesn’t exist is because AMD hasn’t made it, when they very well could.

    They could also put a basic GPU on their IO die, so that every Ryzen shipped with graphics hardware. Doesn’t need a lot rendering horsepower, no more than Intel’s IGPs, but it would need up to date support for transcoding video. And you know, the driver and software support to use that hardware, which we know AMD continues to struggle with.

    Oh! And consumer APUs that support ECC. Thanks for going full Intel on that one, AMD. But wait, Intel has i3’s that support ECC, when used with their server chipsets. Figure that juxtaposition out.

    Last, despite AMDs supposed Linux prowess, switching between the graphics on an AMD APU and an AMD dGPU is apparently more assed up than it is with Intel / Nvidia.

    We’re not talking about GPU’s. Intel doesn’t make a GPU that’s even remotely competitive with AMD. We’re talking about CPU’s. Zen 1 came along and people noticed. Zen 2 has Intel sweating. Zen 2+ started to curb stomp Intel. Zen 3 was a Mortal Kombat "finish him!".

    People wanted more than 4 cores. AMD delivered. Can Intel recover? Sure, but it’s going to take some time, all the while AMD is not slowing down their development.

  9. People wanted more than 4 cores. AMD delivered. Can Intel recover? Sure, but it’s going to take some time, all the while AMD is not slowing down their development.

    Honestly, I don’t think most people care about cores. They just want to know that their **** will run well.

    I happen to use a Threadripper as my main system, but if I were completely honest, I’d choose fewer faster cores than more slower ones.

    Though based on where we are today, we definitely need more than 4.

    I was running Metro Exodus the other day, and in a few of the scenes that game REALLY hammered my CPU. I run it stock. It was boosting to 4.3Ghz with an overall CPU load of 54%… On a 24C48T chip.

  10. We’re not talking about GPU’s.

    You can try not to, but they always enter the conversation where it matters: common desktops and laptops.

    Intel doesn’t make a GPU that’s even remotely competitive with AMD.

    They make plenty, and plenty where AMD could produce a GPU but chooses not to. Sure, Intel’s higher-end Xe products are still ‘on their way’, but we’ve seen enough to know that Intel can play in the ‘not as good as Nvidia’ space just fine alongside AMDs graphics.

    We’re talking about CPU’s. Zen 1 came along and people noticed. Zen 2 has Intel sweating. Zen 2+ started to curb stomp Intel. Zen 3 was a Mortal Kombat "finish him!".

    I guess the passing of events might look that way to someone determined to view corporate strategies and engineering failures as a narrative for a brawling game, but it’s really more that TSMC stumbled on a way forward for smaller nodes and signed up AMD, whereas Intel stumbled. Had either of those events not transpired, AMD would be back where they always are. And where they’re on the verge of being again.

    Zen to Zen 2 has shown pretty solid improvements, some of which have brought competitive advantages, while also pushing AMDs perennial weaknesses back into the light. Thus there are some workloads for which I can’t recommend anything other than AMD, and some for which I can’t recommend any AMD product, and that’s all on AMD.

  11. Honestly I see these two as tightly intertwined. The fact that Intel is still putting out competitive processing power with competent overall performance at 14nm is a testament to the rest of the company not dropping the ball completely.

    But of course their ‘cash cow’ comes from being able to actually produce the stuff that they design.

    This is a big deal right now since TSMC is stretched so thin. Intel has product on store shelves that you can buy right now and is currently cheaper than the AMD equivalents. Despite people making fun of Intel for still being on 14nm and not having as many cores, they still manage to put out the better performing cores.

  12. The thing is, we’ve seen how long problems like what Intel’s facing now take to correct. Intel managed it last time in about five years. AMD took ten. I said it before and I’ll say it again, Intel won’t return to truly competitive form until 2022 or 2023 at the earliest. Why do I say that? That’s how long any real competitive architecture would take to develop and manufacture. That may even be optimistic given their production node issues.

    Skylake has been around forever, but Intel only started to pay attention to what AMD was doing with its Ryzen 1000 series CPU’s. That was three years ago. We have at least two more to wait before Intel could even remotely possibly create another Core 2 type of situation. Again, that’s being very optimistic. None of that is Bob Swan’s fault, nor would he have been capable of righting the ship as it were in two years. That being said, we don’t have a full picture of what’s going on internally, and therefore he might not be doing a good job anyway. It’s hard to say.

  13. Looking back, it actually feels like Swan was a placeholder CEO. Intel raced ahead, then put its feet up, kicked back on its couch stuffed with money and said, "Let’s see what everyone else is doing" before deciding it was time to get up, take a piss, and get back to work.
  14. The fact that x86 has poor prospects, and everyone and their mamas are making their own cpus, is another problem for Intel. Intels actual market is s shrinking one, and so is AMDs potential market for that matter…, but AMDs can still be an expanding company within the shrinking sector being that they are way smaller.
    Intel announcing outsourcing is most likely a short sighted disgrace. They should go the opposite direction, increase production massively and open their factory, give the manufacturer division independence, and let the design division just work with them as any other privileged client. In this manner one can ebb and flow up and down semi independent of each other.
    If the outsourcing was announced in another context along with other plans (such a split of sorts), it could be a good thing, but it seems is in the context and beliefs that manufacturing is just not that worth it.
  15. The thing is, we’ve seen how long problems like what Intel’s facing now take to correct. Intel managed it last time in about five years. AMD took ten. I said it before and I’ll say it again, Intel won’t return to truly competitive form until 2022 or 2023 at the earliest. Why do I say that? That’s how long any real competitive architecture would take to develop and manufacture. That may even be optimistic given their production node issues.

    Skylake has been around forever, but Intel only started to pay attention to what AMD was doing with its Ryzen 1000 series CPU’s. That was three years ago. We have at least two more to wait before Intel could even remotely possibly create another Core 2 type of situation. Again, that’s being very optimistic. None of that is Bob Swan’s fault, nor would he have been capable of righting the ship as it were in two years. That being said, we don’t have a full picture of what’s going on internally, and therefore he might not be doing a good job anyway. It’s hard to say.

    Intel’s problem is not just the architecture, its the foundry. Intel lost its manufacturing process advantage. I won’t say its behind TSMC or Samsung as the processes are not comparable, but its so behind schedule it doesn’t even have a roadmap anymore.

  16. Thinking about it though, Intel has known about this process issue for some time.

    It was sortof made public 5 years ago when they sneakily abandoned the Tick-Tock model without indicating there was anything wrong. While we may not have realized it at the time, hindsight being 20-20 that was the first reall indication other than some late broadwell and skylake releases we had something was amiss.

    Question is how long they knew about the problems internally. A year before that? I mean every broadwell and skylake release was about 6 months late, so it seems pretty clear Intel have been aware of these issues since at least early 2015.

    They have had 6 years to fix this issue thus far and have not succeeded. That’s not a flattering picture.

    It’s not surprising heads are rolling.

  17. They should go the opposite direction, increase production massively and open their factory, give the manufacturer division independence, and let the design division just work with them as any other privileged client. In this manner one can ebb and flow up and down semi independent of each other.

    They already doubled production in the last 3 years orso and still can’t keep up with demand how are they supposed to take on even more production?

    We are at a strange point where demand is far greater then production capabilities but by the time you get more production facilities this demand may be gone and then your factories are ideling which noone wants.

    I also would not be suprised if intel already has some impproved chip designed, but they just can’t make it yet, hence the backporting they did now which is a less then ideal solution and it may hhave most some of it’s luster by the time they can get it out.

  18. I also would not be suprised if intel already has some impproved chip designed, but they just can’t make it yet, hence the backporting they did now which is a less then ideal solution and it may hhave most some of it’s luster by the time they can get it out.

    It shouldn’t be a surprise, because they do. And have. But yeah, they can’t make them, so the point is rather inarguable!

    It was sortof made public 5 years ago when they sneakily abandoned the Tick-Tock model without indicating there was anything wrong. While we may not have realized it at the time, hindsight being 20-20 that was the first reall indication other than some late broadwell and skylake releases we had something was amiss.

    To note: Skylake would have been the last 14nm product under Tick-Tock. All of the 7000-, 8000-, 9000, and most of the 10000-series are revisions of Skylake. Rocket Lake is the first real revision, and it’s still on 14nm.

    What should be notable is that Intel is appearing to maintain per-core performance leads with Rocket Lake, and logically since Rocket Lake is a dialed-back version of an architecture intended for a smaller process, Intel really does have the architecture lead on paper.

    It’s very hard to overstate just how very lucky AMD and TSMC are that Intel took the wrong route to <14nm!

  19. They already doubled production in the last 3 years orso and still can’t keep up with demand how are they supposed to take on even more production?

    Hmm…

    Well, the PC market had been in long decline – largely due to new work at home and school at home, it finally had a positive growth year last year, growing about 10%.

    Enterprise/Server market has grown, but not consitently over the past 3 years. According to Gartner and IDC, 2018 was 32%, 2019 was only about 7%, 2020 was back up to around 20%. But all were positive.

    So that’s the total possible sales. It isn’t quite double, but not too far off – maybe 70% total growth in the market spaces… so I can’t really dispute "Doubled production in the last 3 years" – as the market almost follows that.

    I’m not even factoring in AMD. I know they are outselling Intel in some markets now, but over the past 3 years, it probably isn’t hugely consequential, so I’ll just conveniently ignore that as a rounding error and say that yeah, @Denpepe has a decent point about production.

    That said, I don’t think Intel’s biggest issues are production. They are having culture issues keeping talented staff around – that’s big. They’ve had a lot of turnover at high profile positions, that creates a lot of stress on the organization. And they’ve been stuck on 14nm and Skylake derivatives for… a long time now. So you have stagnation at the innovation level. And there’s been a large lack of focus: Intel has long been trying to diversify away from x86, they just can’t seem to hit anything else in stride. x86 (between Client and Enterprise) represents the vast majority of Intel’s revenue, and they’ve tried everything from mobile to storage to anti virus to Internet of Things to graphics to drones to … they are clearly just throwing everything at the wall to see if anything sticks, and so far none of it has

    None of those issues are isolated in a vacuum – I think all of them are interrelated in some way. Just a mess that will take a while to unwind, no single silver bullet to fix all those problems. Honestly, I think Intel is headed the same way IBM is — they won’t ever die because they have too much invested into patents, but they will wither into mostly insignificance in everyday culture. If they ever start selling off patents – that’s when you know they are getting spun off into dissolution.

  20. It’s very hard to overstate just how very lucky AMD and TSMC are that Intel took the wrong route to <14nm!

    Yeah, I’ve been saying this ever since AMD fanboys started salivating over the first Zen chips.

    The current Zen chips are great, but Intel is still a juggernaut in CPU design. Once they can manyfacture again, either through manufacturing outsourcing or fixing their process AMD will need to be ready, or we are back to AMD lagging behind again.

    But as others have pointed out, there is a real chance alternative architectures like ARM may make inroads. The x86 wars may not be as relevant any more.

    I’m a little disappointed AMD abandoned the ARM K12 arch. They were ahead of the curve when they started. Unless internal work has continued behind the scenes, it’s a shame to see them give that up.

  21. Hmm…

    Well, the PC market had been in long decline – largely due to new work at home and school at home, it finally had a positive growth year last year, growing about 10%.

    Enterprise/Server market has grown, but not consitently over the past 3 years. According to Gartner and IDC, 2018 was 32%, 2019 was only about 7%, 2020 was back up to around 20%. But all were positive.

    So that’s the total possible sales. It isn’t quite double, but not too far off – maybe 70% total growth in the market spaces… so I can’t really dispute "Doubled production in the last 3 years" – as the market almost follows that.

    That’s not all that Intel fabs, though, right?

    I’m not even factoring in AMD. I know they are outselling Intel in some markets now, but over the past 3 years, it probably isn’t hugely consequential, so I’ll just conveniently ignore that as a rounding error and say that yeah, @Denpepe has a decent point about production.

    In order to say that AMD is ‘outselling AMD in some markets’, you have to get real specific about the ‘market’ in question.

    I’d spitball Intel selling 10 CPUs for every CPU AMD sells. I wouldn’t also hesitate to state that for most of those Intel sales, AMD probably has a better product, but that’s even harder to prove since both companies are selling what they make. Perhaps the biggest indicator is that Intel appears to be keeping their prices fairly competitive.

    And they’ve been stuck on 14nm and Skylake derivatives for… a long time now. So you have stagnation at the innovation level.

    If Intel’s 10nm process had rolled out as planned, i.e., Kaby Lake was a new architecture on 10nm as opposed to a quick Skylake respin, would we be having this conversation?

    Can’t say no, but I’d rate it as not nearly as likely.

    And there’s been a large lack of focus: Intel has long been trying to diversify away from x86, they just can’t seem to hit anything else in stride. x86 (between Client and Enterprise) represents the vast majority of Intel’s revenue, and they’ve tried everything from mobile to storage to anti virus to Internet of Things to graphics to drones to … they are clearly just throwing everything at the wall to see if anything sticks, and so far none of it has

    Mobile didn’t pan out because Intel’s parts weren’t better enough to present a case for a change. Same reason AMD doesn’t suddenly get all the design wins and replaces Intel wholesale. Still, Intel’s work on making their x86 IP lower-power has definitely panned out.

    Storage hasn’t been a loss either: Intel hasn’t tried to ‘out-Samsung’ Samsung, and their Optane technology is truly world-class.

    For anti-virus… no one wins anti-virus.

    For IoT, Intel still has quite a bit to contribute. Not sure how far that’ll go, same for drones, but both these markets are rather undeveloped.

    Overall, it’s hard to have a negative outlook on the company, despite the magnitude of their foundry failures.

  22. Yeah, I’ve been saying this ever since AMD fanboys started salivating over the first Zen chips.

    The current Zen chips are great, but Intel is still a juggernaut in CPU design. Once they can manyfacture again, either through manufacturing outsourcing or fixing their process AMD will need to be ready, or we are back to AMD lagging behind again.

    AMD seems to be in a race to shore up their remaining CPU shortcomings, but the biggest one that they simply cannot easily overcome is their reliance on TSMC.

    In terms of actual performance, it’s more that their platform seems ‘rough’ around the edges than anything else. Much of that can be traced to motherboard manufacturers, where AMD bears the blame for lack of leadership.

    But as others have pointed out, there is a real chance alternative architectures like ARM may make inroads. The x86 wars may not be as relevant any more.

    Apple is demonstrating with the M1 something we already knew: most users do not need the general compute performance that modern x86 CPUs provide. Not by a long shot. Further, those modern x86 CPUs aren’t as fast at the things that most users actually do, so the logical course of action is to build application-specific logic to handle real user workloads, and scale out according to use case or market.

    I’m a little disappointed AMD abandoned the ARM K12 arch. They were ahead of the curve when they started. Unless internal work has continued behind the scenes, it’s a shame to see them give that up.

    They wouldn’t do any better than Intel would. ARM CPUs themselves are nothing special; tuned for lower-power from the ground up and not x86 (well, not CISC), but that’s about it.

    What matters, what sells, is everything around the ARM cores on the SoC, and how well that logic is integrated into operating systems and user applications. We’ve seen what it’s taken for Apple to make that all work, and they control their whole stack!

  23. Yeah keeping Optane but it hasn’t exactly set the world on fire yet.

    Intel generally isn’t interested in the low-margin business where say QLC resides. And they made some pretty good QLC.

    Optane, on the other hand, is a nice high-margin product for which there are no competitors.

    That’s more their ‘speed’ ;)

  24. I always figured that Kicking Pat was destined for the CEO chair. Though, I figure when he moved to EMC/VMWare it was no longer a possibility.
  25. Don’t forget all the security holes Intel left in their chips to increase performance. Now that those have to be mitigated, their Skylake derivatives aren’t as potent as they were originally.
  26. I always figured that Kicking Pat was destined for the CEO chair. Though, I figure when he moved to EMC/VMWare it was no longer a possibility.

    Having worked under him, I can say one of the safest rules is never bet against pat Gelsinger. He doesn’t like to lose. This will be interesting.

  27. Don’t forget all the security holes Intel left in their chips to increase performance. Now that those have to be mitigated, their Skylake derivatives aren’t as potent as they were originally.

    The mitigations decrease performance, but only because they break things.

    The security lapses essentially arose from Intel engineers using a design that was intended as a proof of concept for the production design. Fixing them in hardware requires very little change and realistically could have been done from the start.

    The other part of this is that Intel themselves didn’t expect Skylake to run beyond the 6000-series. Instead, they’ve had to rely on it for four more ‘generations’, and now the entire world runs on it! The reality is that if the architecture had been replaced on schedule, security researchers (and bad actors) would have moved on. Intel’s fab foibles made Skylake the #1 targeted architecture.

  28. Don’t cry too hard for Bob Swan, he got paid 67 MILLION in 2019.

    Also you know he got a giant golden parachute departure package, but don’t see details of how much on the net yet.

  29. I think their biggest issue was complacency. For the past 14 years they told people what they needed instead of giving people what they wanted. AMD came along and gave everyone what they wanted. Regardless of what Swan did there was no righting that ship in 2 years.

    Gelsinger going come in and sell you a 16 core CPU, but only allow you to use 4 cores and make you buy licensing for the remaining 12.

    AMD havent you know. I not only couldnt buy one, I found they dont perform as expected.
    I had no choice but to buy a new CPU around the time of Cyberpunks launch and wanted to get a 5800X.
    They were impossible to find so I was forced to buy Intels similar chip, the 10700K.
    This has proven to be faster running Cyberpunk and I suspect it will be for other games.
    This is despite all the reviews showing the 5800X excels overall.
    Not everything is as it seems.

  30. This is despite all the reviews showing the 5800X excels overall.

    This isn’t an untrue statement though.

    I’d also expect the real-world gaming performance margin between these two CPUs to be around an order of magnitude below what a user would notice unless something was particularly broken with a specific game.

    That being said, my 9900K does 5.0GHz across all cores with AVX, and that’s basically as fast as one can get for gaming without going to extremes, including financial extremes.

  31. AMD havent you know. I not only couldnt buy one, I found they dont perform as expected.
    I had no choice but to buy a new CPU around the time of Cyberpunks launch and wanted to get a 5800X.
    They were impossible to find so I was forced to buy Intels similar chip, the 10700K.
    This has proven to be faster running Cyberpunk and I suspect it will be for other games.
    This is despite all the reviews showing the 5800X excels overall.
    Not everything is as it seems.

    If all you are going to base a CPU on is one game that’s rather sad. The AMD is faster in most instances and stomps the Intel in multi-threaded workloads.

  32. If all you are going to base a CPU on is one game that’s rather sad. The AMD is faster in most instances and stomps the Intel in multi-threaded workloads.

    One game is perhaps a touch too simple, and @AntiQuark mentioned that they were looking for a 5800X first.

    Still, I tend to base my purchase decisions on gaming as well; not because I don’t do other stuff, but because games are the only thing that I really care about how fast I can work.

  33. If all you are going to base a CPU on is one game that’s rather sad. The AMD is faster in most instances and stomps the Intel in multi-threaded workloads.

    I bought what I could and found it to be the best.
    I havent found it delinquent in other games but havent compared their framerate, there was no need.
    I’m sorry you are sad.

Leave a comment