I feel like it's a little telling that they're saying 50% faster than rivals chips. Only a couple years ago they wouldn't even mention rivals, they'd only be comparing Apple chips to the previous Apple chips.
This was a post I just by a semiconductor analyst. Make of it what you will
If this is true, it would almost certainly affect the Mx too. Could be the reason we're yet to see an Apple Silicon 27" iMac or 15" MBP. However, I wouldn't quite use the doom and gloom wording the analyst did, "i.e. the future looks dim", since one missed speed bump in a line of 20 bumps isn't really much of a hiccup, and certainly no reliable indication of the future.
Considering the reports coming out of Apple that it's becoming a less attractive place to work, they really need to change tack if current policies are resulting in the loss of their best engineers. Apple has some incredibly talented people, and HR should be bending over backwards to keep them - especially now Apple's 100% self reliant on their own engineers to keep the Apple Silicon ahead of x86 and other ARM vendors. The analyst claims "Apple’s Chief CPU Architect, Gerard Williams, as well as over a 100 other Apple engineers left to join Nuvia". If this is true, Cook needs to stop being so stubborn with crappy employee policies, like WFH, health benefits etc - it's very much a worker's market right now especially in engineering, and there are a lot of companies offering a lot of very generous perks.
Ben Bajarin poured some cold water on that theory, and had a couple other interesting comments in the thread.
Unstated is that Apple increased the number of transistors to 15B, where the A14 was at 11.8B. That's a better than 25% increase. That might equate to a larger, more expensive die, or maybe just evolution of the architecture. We'll know, mostly, fairly soon when there is a die shot available on anandtech.
I agree with Ben; there are plenty of options to increase performance outside of process node advances, and no, this isn't any indication of Apple "failing, or flailing".
I feel like it's a little telling that they're saying 50% faster than rivals chips. Only a couple years ago they wouldn't even mention rivals, they'd only be comparing Apple chips to the previous Apple chips.
This was a post I just by a semiconductor analyst. Make of it what you will
If this is true, it would almost certainly affect the Mx too. Could be the reason we're yet to see an Apple Silicon 27" iMac or 15" MBP. However, I wouldn't quite use the doom and gloom wording the analyst did, "i.e. the future looks dim", since one missed speed bump in a line of 20 bumps isn't really much of a hiccup, and certainly no reliable indication of the future.
Considering the reports coming out of Apple that it's becoming a less attractive place to work, they really need to change tack if current policies are resulting in the loss of their best engineers. Apple has some incredibly talented people, and HR should be bending over backwards to keep them - especially now Apple's 100% self reliant on their own engineers to keep the Apple Silicon ahead of x86 and other ARM vendors. The analyst claims "Apple’s Chief CPU Architect, Gerard Williams, as well as over a 100 other Apple engineers left to join Nuvia". If this is true, Cook needs to stop being so stubborn with crappy employee policies, like WFH, health benefits etc - it's very much a worker's market right now especially in engineering, and there are a lot of companies offering a lot of very generous perks.
Gerard Williams and a number of senior Google and Apple processor architects did leave to form NUVIA.
NUVIA was then bought by Qualcomm to implement their high performance Phoenix cores in their SoCs. They are set to debut first in their laptop SoCs in 2022.
NUVIA's performance claims:
If this core is as described (which seems quite credible given the history of work done for Apple, the industry partners that initially invested in NUVIA and the $1.4 billion purchase by Qualcomm), they will likely be an industry leader at launch.
"industry Leader" at launch seems improbable, but in a year, we will have that answer. Until then "performance claims", not performance delivered.
I have been buying apple products since the first iPhone and this year it’s really flat. The upgrades are more suitable for the Tik Tokers or influencers out there. For joe public then it’s not really worth it. We need to see a total new redesign like a foldaway phone etc. Why not look at Samsung. They evolve each year
Wasn't expecting much at all. Then the iPad Mini happened and it was a real nice surprise. I really want one - and I just bought the M1 iPad Pro a couple of months back!
Again, wasn't expecting much with the iPhone. The iPhone 13 seemed okay as an upgrade. Then they started talking about the cameras on the iPhone 13 Pro and that hooked me. My fingers are twitching to click on the "buy" button on the online Apple store for the iPhone 13 Pro Max Sierra Blue 1TB on Friday!
Seems pretty cool, especially cinematic mode, macro and extra battery life. However, IMO the wide angle lens has very little utility vs telephoto. I can't remember the last time I wanted a wider angle view but I regularly zoom in. You can usually step back, but you frequently can't get nearer to something. Anecdotally I regularly see people zooming in, and they just step back if they can't fit everything in the frame.
You can't just step back to get wider scenery shots (pano is cool, but definitely not the same), and it sounds like you've never taken group photos indoors, like at home. It's kind of strange that you say this, because I've personally witnessed the struggle to squeeze people together literally hundreds of times over the years. I suspect most people have experienced this. The wide angle lens also allows you to get more spur of the moment shots where you don't have time to compose and get people to pose, it's just situational. You can crop later, but you can't add viewing angle later.
Telephoto obviously has great utility as well, but the farther you zoom the more you need to worry about things like stability and light with a tiny camera. The wide angle just simply works, and also adds great utility.
I can't recall ever not being able to take another step or two back in order to squeeze everyone in a photo, but have wished that I could get closer to something or someone a million times. There's a reason a lot more zoom lenses than wide-angle DSLR lenses are sold.
Playing with the cameras and lenses a whole bunch, you might find that the telephoto isn't actually used as often as you think when you have selected telephoto. The AI in the software is good enough that sometimes it will take a better shot with the standard lens instead of going for the telephoto. Appleinsider did an article on this a while back.
The wide angle lens allows for a lot of additional creative shots that you don't get with the standard and telephoto too. Also much easier group selfies.
I have been buying apple products since the first iPhone and this year it’s really flat. The upgrades are more suitable for the Tik Tokers or influencers out there. For joe public then it’s not really worth it. We need to see a total new redesign like a foldaway phone etc. Why not look at Samsung. They evolve each year
It really must be getting harder to hire people for low-paid jobs. There is always flurry of new or suddenly reanimated accounts just after a big announcement, just turning up to try to start a negative narrative, but these attempts are really pathetically transparent.
Seems pretty cool, especially cinematic mode, macro and extra battery life. However, IMO the wide angle lens has very little utility vs telephoto. I can't remember the last time I wanted a wider angle view but I regularly zoom in. You can usually step back, but you frequently can't get nearer to something. Anecdotally I regularly see people zooming in, and they just step back if they can't fit everything in the frame.
You can't just step back to get wider scenery shots (pano is cool, but definitely not the same), and it sounds like you've never taken group photos indoors, like at home. It's kind of strange that you say this, because I've personally witnessed the struggle to squeeze people together literally hundreds of times over the years. I suspect most people have experienced this. The wide angle lens also allows you to get more spur of the moment shots where you don't have time to compose and get people to pose, it's just situational. You can crop later, but you can't add viewing angle later.
Telephoto obviously has great utility as well, but the farther you zoom the more you need to worry about things like stability and light with a tiny camera. The wide angle just simply works, and also adds great utility.
I can't recall ever not being able to take another step or two back in order to squeeze everyone in a photo, but have wished that I could get closer to something or someone a million times. There's a reason a lot more zoom lenses than wide-angle DSLR lenses are sold.
The zoom just refers to variable focal length. There are wide angle zooms after all.
The iPhone 13 uses a Lightning connector which has USB 2 bandwidth. USB C is 20 times faster. 20 times! Just think about that for a few seconds. A 10 GB 4K movie takes 2.7 minutes to transfer to the iPhone 13 while the same movie takes 8.3 seconds on current Android phones. What reason did Apple have to refuse to add USB C for the past few years? It's on most of their other devices now including their iPads.
The wife likes her original SE, and i suspect she will go for a 12 mini when the time comes mainly due to the colour options!
For the lack of real cpu fanfare, i am guessing they have focused on the macbook pro 14 & 16 switch - they will need the volume there and the highest performance improvements. I reckon an October reveal will surprise all 😎. The one big commitment Apple made was the Intel to Asi switch deadline and i cannot see them messing that up. I suspect the iPhone 14 next year will get a much more dramatic update.
The iPhone 13 uses a Lightning connector which has USB 2 bandwidth. USB C is 20 times faster. 20 times! Just think about that for a few seconds. A 10 GB 4K movie takes 2.7 minutes to transfer to the iPhone 13 while the same movie takes 8.3 seconds on current Android phones. What reason did Apple have to refuse to add USB C for the past few years? It's on most of their other devices now including their iPads.
I totally agree. If you are using an iPhone 13 Pro and take ProRes photos, you’ll need the fast transfer speeds of USB-C. For me, I’m not upgrading this year as it’s the first time in a long time that I’ve been underwhelmed by an update, but if they had changed the connector to USB-C, I’d have snapped their hands off - mainly because the 2x iPads and MacBook in my house all use USB-C and I would like to not have different cables all over the place.
I feel like it's a little telling that they're saying 50% faster than rivals chips. Only a couple years ago they wouldn't even mention rivals, they'd only be comparing Apple chips to the previous Apple chips.
This was a post I just by a semiconductor analyst. Make of it what you will
If this is true, it would almost certainly affect the Mx too. Could be the reason we're yet to see an Apple Silicon 27" iMac or 15" MBP. However, I wouldn't quite use the doom and gloom wording the analyst did, "i.e. the future looks dim", since one missed speed bump in a line of 20 bumps isn't really much of a hiccup, and certainly no reliable indication of the future.
Considering the reports coming out of Apple that it's becoming a less attractive place to work, they really need to change tack if current policies are resulting in the loss of their best engineers. Apple has some incredibly talented people, and HR should be bending over backwards to keep them - especially now Apple's 100% self reliant on their own engineers to keep the Apple Silicon ahead of x86 and other ARM vendors. The analyst claims "Apple’s Chief CPU Architect, Gerard Williams, as well as over a 100 other Apple engineers left to join Nuvia". If this is true, Cook needs to stop being so stubborn with crappy employee policies, like WFH, health benefits etc - it's very much a worker's market right now especially in engineering, and there are a lot of companies offering a lot of very generous perks.
Gerard Williams and a number of senior Google and Apple processor architects did leave to form NUVIA.
NUVIA was then bought by Qualcomm to implement their high performance Phoenix cores in their SoCs. They are set to debut first in their laptop SoCs in 2022.
NUVIA's performance claims:
If this core is as described (which seems quite credible given the history of work done for Apple, the industry partners that initially invested in NUVIA and the $1.4 billion purchase by Qualcomm), they will likely be an industry leader at launch.
"industry Leader" at launch seems improbable, but in a year, we will have that answer. Until then "performance claims", not performance delivered.
A14 GB5 per core performance is 1580. M1 GB5 per core performance is 1700. If the A15 is at 3.2 GHz like in the M1, it will hit 1700. With just an improved cache memory architecture, a small frequency bump and TSMC 4nm next year, the M2 and A16 will be in the 2000 range. In the upper half of the Nuvia's target perf bubble. This is assuming Nuvia ships in 2022.
So yup. Wait and see. QCOM has to actually integrate Nuvia and ship a chip in the space of 2 years. Sounds tough. The key will be who gets the first 4nm and 3nm wafers next year. I'm betting Apple will.
And, how about we lay off the conspiracy theories with semianalysis, everyone. All we know about A15 is 1 sentence from the media event. We can all panic next week when GB5 results come out, but as of today, this thing is the thing that the Internet does: find the rage and controversy, then amp it up to 11.
We just had a complete cycle of it just last Friday through Sunday, when the Epic vs Apple decision came out. It went from "Epic wins big, Apple to lose $19b!" to "Epic got crushed, hopefully Apple won't be mean to developers" in the span of 3 days. This story utterly dominated Techmeme for a couple of days, amping up the Apple drama, only to sink down in position so fast that you would have missed it if you went out for hike without your phone. It's now to the point of the usual, where the media will never speak of its mistakes, like Klingons and their foreheads.
I feel like it's a little telling that they're saying 50% faster than rivals chips. Only a couple years ago they wouldn't even mention rivals, they'd only be comparing Apple chips to the previous Apple chips.
This was a post I just by a semiconductor analyst. Make of it what you will
If this is true, it would almost certainly affect the Mx too. Could be the reason we're yet to see an Apple Silicon 27" iMac or 15" MBP. However, I wouldn't quite use the doom and gloom wording the analyst did, "i.e. the future looks dim", since one missed speed bump in a line of 20 bumps isn't really much of a hiccup, and certainly no reliable indication of the future.
Considering the reports coming out of Apple that it's becoming a less attractive place to work, they really need to change tack if current policies are resulting in the loss of their best engineers. Apple has some incredibly talented people, and HR should be bending over backwards to keep them - especially now Apple's 100% self reliant on their own engineers to keep the Apple Silicon ahead of x86 and other ARM vendors. The analyst claims "Apple’s Chief CPU Architect, Gerard Williams, as well as over a 100 other Apple engineers left to join Nuvia". If this is true, Cook needs to stop being so stubborn with crappy employee policies, like WFH, health benefits etc - it's very much a worker's market right now especially in engineering, and there are a lot of companies offering a lot of very generous perks.
Gerard Williams and a number of senior Google and Apple processor architects did leave to form NUVIA.
NUVIA was then bought by Qualcomm to implement their high performance Phoenix cores in their SoCs. They are set to debut first in their laptop SoCs in 2022.
NUVIA's performance claims:
If this core is as described (which seems quite credible given the history of work done for Apple, the industry partners that initially invested in NUVIA and the $1.4 billion purchase by Qualcomm), they will likely be an industry leader at launch.
"industry Leader" at launch seems improbable, but in a year, we will have that answer. Until then "performance claims", not performance delivered.
A14 GB5 per core performance is 1580. M1 GB5 per core performance is 1700. If the A15 is at 3.2 GHz like in the M1, it will hit 1700. With just an improved cache memory architecture, a small frequency bump and TSMC 4nm next year, the M2 and A16 will be in the 2000 range. In the upper half of the Nuvia's target perf bubble. This is assuming Nuvia ships in 2022.
So yup. Wait and see. QCOM has to actually integrate Nuvia and ship a chip in the space of 2 years. Sounds tough. The key will be who gets the first 4nm and 3nm wafers next year. I'm betting Apple will.
And, how about we lay off the conspiracy theories with semianalysis, everyone. All we know about A15 is 1 sentence from the media event. We can all panic next week when GB5 results come out, but as of today, this thing is the thing that the Internet does: find the rage and controversy, then amp it up to 11.
We just had a complete cycle of it just last Friday through Sunday, when the Epic vs Apple decision came out. It went from "Epic wins big, Apple to lose $19b!" to "Epic got crushed, hopefully Apple won't be mean to developers" in the span of 3 days. This story utterly dominated Techmeme for a couple of days, amping up the Apple drama, only to sink down in position so fast that you would have missed it if you went out for hike without your phone. It's now to the point of the usual, where the media will never speak of its mistakes, like Klingons and their foreheads.
Here are some preliminary GeekBench 5 benchmarks for the iPhone 13 / 13 Pro
GeekBench Metal:
iPhone 13 - 10,608 iPhone 13 Pro - 14,216
"On the graphics side, a Metal result for the iPhone 13 (designated iPhone14,5) reveals a score of 10608 for the A15 Bionic in that device, roughly 15% higher than the score for the A14 Bionic in the iPhone 12."
GeekBench 5:
iPhone 13 - Single Core (1724) / Multi-core (4587) iPhone 13 Pro - Single Core (1738) / Multi-core (4766)
I have been buying apple products since the first iPhone and this year it’s really flat. The upgrades are more suitable for the Tik Tokers or influencers out there. For joe public then it’s not really worth it. We need to see a total new redesign like a foldaway phone etc. Why not look at Samsung. They evolve each year
Unadulterated horseshit. All of these products are going to sell in the metric fucktonnes. lol folding phones.
The iPhone 13 uses a Lightning connector which has USB 2 bandwidth. USB C is 20 times faster. 20 times! Just think about that for a few seconds. A 10 GB 4K movie takes 2.7 minutes to transfer to the iPhone 13 while the same movie takes 8.3 seconds on current Android phones. What reason did Apple have to refuse to add USB C for the past few years? It's on most of their other devices now including their iPads.
Wrong on multiple levels, see today's USB spec post for why.
The iPhone 13 uses a Lightning connector which has USB 2 bandwidth. USB C is 20 times faster. 20 times! Just think about that for a few seconds. A 10 GB 4K movie takes 2.7 minutes to transfer to the iPhone 13 while the same movie takes 8.3 seconds on current Android phones. What reason did Apple have to refuse to add USB C for the past few years? It's on most of their other devices now including their iPads.
Wrong on multiple levels, see today's USB spec post for why.
He probably knows better, but trolls have an agenda regardless of facts.
It's interesting that he just ignores the iPad Pro with the lightning connector back in 2015 supporting USB 3.0 speeds… which we know is all due to the Fresco Logic FL1100SX 2-port USB 3.0 Host Controller discovered by iFixit's teardown and then backed up by performance testing.
He may actually have a case for why Apple doesn't use a USB 3.x controller in the iPhone if he didn't relegate his argument to foolish comments about about the port interface.
The iPhone 13 uses a Lightning connector which has USB 2 bandwidth. USB C is 20 times faster. 20 times! Just think about that for a few seconds. A 10 GB 4K movie takes 2.7 minutes to transfer to the iPhone 13 while the same movie takes 8.3 seconds on current Android phones. What reason did Apple have to refuse to add USB C for the past few years? It's on most of their other devices now including their iPads.
Wrong on multiple levels, see today's USB spec post for why.
He probably knows better, but trolls have an agenda regardless of facts.
It's interesting that he just ignores the iPad Pro with the lightning connector back in 2015 supporting USB 3.0 speeds… which we know is all due to the Fresco Logic FL1100SX 2-port USB 3.0 Host Controller discovered by iFixit's teardown and then backed up by performance testing.
He may actually have a case for why Apple doesn't use a USB 3.x controller in the iPhone if he didn't relegate his argument to foolish comments about about the port interface.
I feel like it's a little telling that they're saying 50% faster than rivals chips. Only a couple years ago they wouldn't even mention rivals, they'd only be comparing Apple chips to the previous Apple chips.
This was a post I just by a semiconductor analyst. Make of it what you will
If this is true, it would almost certainly affect the Mx too. Could be the reason we're yet to see an Apple Silicon 27" iMac or 15" MBP. However, I wouldn't quite use the doom and gloom wording the analyst did, "i.e. the future looks dim", since one missed speed bump in a line of 20 bumps isn't really much of a hiccup, and certainly no reliable indication of the future.
Considering the reports coming out of Apple that it's becoming a less attractive place to work, they really need to change tack if current policies are resulting in the loss of their best engineers. Apple has some incredibly talented people, and HR should be bending over backwards to keep them - especially now Apple's 100% self reliant on their own engineers to keep the Apple Silicon ahead of x86 and other ARM vendors. The analyst claims "Apple’s Chief CPU Architect, Gerard Williams, as well as over a 100 other Apple engineers left to join Nuvia". If this is true, Cook needs to stop being so stubborn with crappy employee policies, like WFH, health benefits etc - it's very much a worker's market right now especially in engineering, and there are a lot of companies offering a lot of very generous perks.
Gerard Williams and a number of senior Google and Apple processor architects did leave to form NUVIA.
NUVIA was then bought by Qualcomm to implement their high performance Phoenix cores in their SoCs. They are set to debut first in their laptop SoCs in 2022.
NUVIA's performance claims:
If this core is as described (which seems quite credible given the history of work done for Apple, the industry partners that initially invested in NUVIA and the $1.4 billion purchase by Qualcomm), they will likely be an industry leader at launch.
"industry Leader" at launch seems improbable, but in a year, we will have that answer. Until then "performance claims", not performance delivered.
A14 GB5 per core performance is 1580. M1 GB5 per core performance is 1700. If the A15 is at 3.2 GHz like in the M1, it will hit 1700. With just an improved cache memory architecture, a small frequency bump and TSMC 4nm next year, the M2 and A16 will be in the 2000 range. In the upper half of the Nuvia's target perf bubble. This is assuming Nuvia ships in 2022.
So yup. Wait and see. QCOM has to actually integrate Nuvia and ship a chip in the space of 2 years. Sounds tough. The key will be who gets the first 4nm and 3nm wafers next year. I'm betting Apple will.
And, how about we lay off the conspiracy theories with semianalysis, everyone. All we know about A15 is 1 sentence from the media event. We can all panic next week when GB5 results come out, but as of today, this thing is the thing that the Internet does: find the rage and controversy, then amp it up to 11.
We just had a complete cycle of it just last Friday through Sunday, when the Epic vs Apple decision came out. It went from "Epic wins big, Apple to lose $19b!" to "Epic got crushed, hopefully Apple won't be mean to developers" in the span of 3 days. This story utterly dominated Techmeme for a couple of days, amping up the Apple drama, only to sink down in position so fast that you would have missed it if you went out for hike without your phone. It's now to the point of the usual, where the media will never speak of its mistakes, like Klingons and their foreheads.
Here are some preliminary GeekBench 5 benchmarks for the iPhone 13 / 13 Pro
GeekBench Metal:
iPhone 13 - 10,608 iPhone 13 Pro - 14,216
"On the graphics side, a Metal result for the iPhone 13 (designated iPhone14,5) reveals a score of 10608 for the A15 Bionic in that device, roughly 15% higher than the score for the A14 Bionic in the iPhone 12."
GeekBench 5:
iPhone 13 - Single Core (1724) / Multi-core (4587) iPhone 13 Pro - Single Core (1738) / Multi-core (4766)
Nice improvement. More performance at less power consumption. Semianalysis played a kind of trick where you take one part that may be true, and extrapolate out to something dramatic. That's usually noise and actively makes you dumber. There's more to design than just number of execution units or branch predictor performance. Memory performance is arguably the biggest performance impediment we have left. It needs to be continually refined and improved. Turns out improving cache performance improves compute performance.
Comments
I agree with Ben; there are plenty of options to increase performance outside of process node advances, and no, this isn't any indication of Apple "failing, or flailing".
Again, wasn't expecting much with the iPhone. The iPhone 13 seemed okay as an upgrade. Then they started talking about the cameras on the iPhone 13 Pro and that hooked me.
My fingers are twitching to click on the "buy" button on the online Apple store for the iPhone 13 Pro Max Sierra Blue 1TB on Friday!
The wide angle lens allows for a lot of additional creative shots that you don't get with the standard and telephoto too. Also much easier group selfies.
For the lack of real cpu fanfare, i am guessing they have focused on the macbook pro 14 & 16 switch - they will need the volume there and the highest performance improvements. I reckon an October reveal will surprise all 😎. The one big commitment Apple made was the Intel to Asi switch deadline and i cannot see them messing that up. I suspect the iPhone 14 next year will get a much more dramatic update.
So yup. Wait and see. QCOM has to actually integrate Nuvia and ship a chip in the space of 2 years. Sounds tough. The key will be who gets the first 4nm and 3nm wafers next year. I'm betting Apple will.
And, how about we lay off the conspiracy theories with semianalysis, everyone. All we know about A15 is 1 sentence from the media event. We can all panic next week when GB5 results come out, but as of today, this thing is the thing that the Internet does: find the rage and controversy, then amp it up to 11.
We just had a complete cycle of it just last Friday through Sunday, when the Epic vs Apple decision came out. It went from "Epic wins big, Apple to lose $19b!" to "Epic got crushed, hopefully Apple won't be mean to developers" in the span of 3 days. This story utterly dominated Techmeme for a couple of days, amping up the Apple drama, only to sink down in position so fast that you would have missed it if you went out for hike without your phone. It's now to the point of the usual, where the media will never speak of its mistakes, like Klingons and their foreheads.
GeekBench Metal:
iPhone 13 - 10,608
iPhone 13 Pro - 14,216
"On the graphics side, a Metal result for the iPhone 13 (designated iPhone14,5) reveals a score of 10608 for the A15 Bionic in that device, roughly 15% higher than the score for the A14 Bionic in the iPhone 12."
GeekBench 5:
iPhone 13 - Single Core (1724) / Multi-core (4587)
iPhone 13 Pro - Single Core (1738) / Multi-core (4766)
"These scores represent a roughly 10% increase in single-core performance and 18% increase in multi-core performance compared to the A14 Bionic in the iPhone 12 lineup."
Additional Benchmarks Reveal A15 Bionic Performance Improvements in iPhone 13 and iPhone 13 Pro - MacRumors
It's interesting that he just ignores the iPad Pro with the lightning connector back in 2015 supporting USB 3.0 speeds… which we know is all due to the Fresco Logic FL1100SX 2-port USB 3.0 Host Controller discovered by iFixit's teardown and then backed up by performance testing.
He may actually have a case for why Apple doesn't use a USB 3.x controller in the iPhone if he didn't relegate his argument to foolish comments about about the port interface.
https://www.anandtech.com/show/16983/the-apple-a15-soc-performance-review-faster-more-efficient
Nice improvement. More performance at less power consumption. Semianalysis played a kind of trick where you take one part that may be true, and extrapolate out to something dramatic. That's usually noise and actively makes you dumber. There's more to design than just number of execution units or branch predictor performance. Memory performance is arguably the biggest performance impediment we have left. It needs to be continually refined and improved. Turns out improving cache performance improves compute performance.