Hawk Point (8040 series) is rightfully a dud since it's just a 7040 series refresh. I have to question why they bothered pushing it through the door. I am barely seeing any show up at stores. It's been what? 6 months since they announced it and I still only see a few 8945HS equipped laptops compared to the numerous Core Ultra 5/7/9 laptops (Some SKUs, like 135H still MIA).
I look forward to Strix Point, X Elite, and Lunar Lake stirring up the market.
7040 series also pretty much had a paper launch. With some Shenzhen brands (rightfully, probably) throwing tantrums because they didn't get the amount of 7040 chips they bought.
So I am expecting to see Strix being paper launched in Computex, with availability around CES (lol), just in time for Lunar Lake's arrival.
Also curious about the X Elite, another paper launch that by the time it arrives, will compete with a chip 2 refreshes newer than the M2 baseline on the original announcement...
which stores? what region? recent earnings had client up so could that mean oems are buying more and the resulting laptops will trickle into stock for H2?
seeing numerous laptops in stores doesnt necessarily mean they're all selling, if physical retail then it's likely sponsored shelf space due to the low store margins (well it's not shopping season this time of year anyway, though i did get a ryzen laptop in may of a different year)
and some benchmarks have shown hawk point has more of a gain over phoenix than expected
It's really kind of amazing that AMD's desktop marketshare never got anywhere close to a majority. There were several years where AMD had a clearly superior design on a superior process. Of course, Intel's margins contracted a lot during that period, so I guess AMD had an impact in that way.
I suspect AMD is near peak marketshare in all segments, though. Intel moving to 20A while AMD is stuck on TSMC N4 is going to hurt.
intel is also losing hyperthreading, and when will they have 3d cache?
process node isnt everything, a mash of performance + power + price (+manufacturing cost) are what matter
diy desktop likely has been majority for a while, but prebuilts, oems, institutions? that's a lot of stock that needs to be guaranteed mixed with old contracts and momentum, i would never expect majority yet
Losing hypterthreading is a feature not a bug. Intel has efficiency cores to help with MT performance -- AMD doesn't have those (although they are adding their 'c' cores).
Process node isn't everything, but it's a lot and absolutely affects performance and power. The only way for AMD to beat Intel by using an inferior process is if AMD has a much better design. That was the case with Athlon vs. Netburst. I doubt it will be the case moving forward.
The real problem with hyperthreading now is how easily it has been exploited for security vulnerabilities, and the protections required along with the fact most pipelined 'hyperthreads' aren't hits for multiple factors (bad queuing, under-prioritized branch prediction, collisions requiring discard, etc) don't make it a compelling performance improvement compared to other options. While its kind of "free" performance, it's not as big as it used to be two decades ago when Jackson technology was announced.
To reach that conclusion you would need to have both designs manufactured on the same process. Thanks to TSMC, AMD has been using a superior process since 2019.
I doubt there is any meaningful performance/watt difference between designs, holding all else constant.
The E-cores are shit. They're just revamped Atom cores! They might be a power saver compared to the P-cores but they lose a lot of performance in the trade. I'd take AMD's ZenC any day.
Look again. E-cores are way more powerfull than Atom ever was. Current E-cores outperform big cores from the Atom era (like Sandy Bridge). But even if they weren't, their point is perf/watt. If you have let's say 100W power budget, you will get more performance from E-cores than from P-cores in that power budget (as long as the task can be split to multiple cores). P-cores are only for those tasks, that are not well parallelized and for such usecase 8 P-cores are usually more than enough.
Of course the E-cores got better, the Atoms were really bad for their time even. The perf/watt is only slightly better than the P-cores simply because the perf is still shit.
E-cores have nothing to do with Atom, and E-cores are quite potent. The problem is the P-cores are too weak to compete with AMD's Zen4 cores head-on so Intel needs to combine the performance of both P and E cores to match them, which is only effective in multithreaded applications.
Meteor Lake was supposed to have Adamantine, but it didn't happen. If Arrow Lake also skips Adamantine, that would suck. I want to see it in action, and in large amounts (anything above 128 MB would be an improvement on the L4 they used in the past).
itis just the inertia. Most people just don't know that the #1 in the list of TOP 500 exascale supercomputers last few years was called Frontier based on AMD EPYC, it's more than 2x faster while consumes same power as the next one in the list - Aurora - based on Intel. Look at Passmark high end processors benchmarks - all 16 first places are AMD. Out organization also switched to AMD as well as many my colleagues.
AMD may have gained a little share so far, but its revenue is the lowest and decreasing, at risk to lose income, or loss; while Intel is moving to product rapidly with lower nm, gaining on TSMC, so will gain on AMD. I can't see there is any good news to AMD, I am sorry.
The wider problem isn't so much an AMD-only thing, but a PC industry issue. This article does rightly point out a piece-of-pie increase for AMD, but it doesn't (and shouldn't) note the decline in overall sales numbers or total units purchased which has been a troublesome drag on both Intel and AMD for a few years now. Some of that may begin to change as lifecycle replacements of pandemic-era purchases roll in over this and next year, but the other problem is that it's getting harder to justify a dedicated computing platform due to consolidation of compute into datacenters and cloud host physical locations alongside ever more credible "phone-replaces-PC" and/or "Timmy-only-owns-a-phone" situations that appear to be influential on business and consumer buying habits.
Mercury Research knowingly presents flawed data to present a false narrative immersing unknowing members of this comment post, and the Anadtech audience, in a syndicate operation's invented reality.
The AMD (and Intel) and OEMs and Mercury objective is to conceal racketeering and corporate tax evasion.
Mercury so said AMD share and growth statement specific q1 are not correct and far from on the nuances left unreported in a reversal of the facts by Mercury and report outlets. Ultimately Mercury is concealing generations run end supply elastic price slide, PC price deflation into second half and AMD Client division's $467 million loss in q1 2024 relying on TR7K as a price at cost margin support.
My counter point report 'draft' has been emailed to Ryan, Gavin and Anton for publication here at Anandtech. I offered up the report on my own name or with Anton as a point counter point piece also available to Ryan and Gavin.
I've had it with the industry analyst lies to conceal actual x86 volumes, actual AMD and Intel desktop, workstation, mobile and server category shares.
Mr. Shilov has connections, back in 2009 as the voice and transmitter of the Intel stock market rigging supply signal cipher in its fourth generation disintegrated math story problem format.
Since 2009, I have complimented Mr. Shilov seemingly escaping nefarious forces, technology essentially a mob 'associate network' warring business of clans and all over the world. Mr. Shilov somehow rising above to become what appears to me the only generalist PC tech reporter capable of investigative journalism today. When he puts himself to that task.
I applaud Mr. Shilov's rise above nefarious forces but with this AMD q1 share piece have concerns and I will email Mr. Shilov again today on aspects of his report, from Mercury Data, that do not math.
Mr. Shilov is also aware and from me directly, that I have a spot reserved for his personal appearance at the Intel Corporation United States Congressional investigative hearing on the matters of Intel Corp. caught up in a 30 year inter nation economic espionage subject 18 USC 1956, 1962(c) and 1831 and 2382.
This matter in EUCC currently is the Intel C-240/22-P appeal matter addressing bundle deal sales close incentives of all types; payoffs essentially for access and allocation, this is not about first dollar discount. I have advised AMD pay attention to C-240/22 participating in today what they complained about back in 2005 pursuant EUCC v Intel 37.991 subject AMD v Intel.
Note Mr. Shilov during his stint at X-bit Labs is not the only expert witness United States Congress is interesting in conferring on my recommendations. Specific Intel Supply Signal Cipher in its third generation product category by quarter percent supply format the Barret men Mr. Ryan Shrout of PC Perspective, Mr. Fred Abazovic of Fudzilla, Ms. Monica Chen of Digitimes among others. For Signal Cipher In its first generation format Mr. Linley Gwennap and Mr. Kevin Krewell.
The hot line for witness immunity, well known and for over a decade as proceedings progress, application into the USDOJ cartel amnesty program is joyce.branda@usdoj.gov.
Anton, Ryan and Gavin have all the data to reassess the AMD q1 situation where Anandtech was at the end of this AMD tail wagging on the thousands media outlets who reported AMD's false narrative before AnandTech across WW industry, electronics, PC, business and financial press. Anton, Ryan and Gavin make this right welcome to ur rocket ship ride. mb
What are you smoking and can you share it with the rest of us? Also who are you and what makes you think your opinion is important/informed enough for us to read a text wall?
In fact, nevermind, no one cares what your response would be. Go back to your mom's basement kiddo.
Heh! Totally fair and no argument from me, but it is hard to resist the low hanging fruit of a decent portion of the PC enthusiast sub-group. They're so...well you know what I mean. You're lucid enough to get it, obviously.
Mt response here and to all of PC Media land who got caught up in the AMD, Mercury, Shilov q1 2024 puff piece where not ONE single assessment of the Mercury data specific AMD is true, in fact misrepresented can be found here;
I've asked Anton to rewrite the correction and Ryan the Gavin have the same opportunity. Anton's take is very much misrepresented presenting basically an invented assessment. Original AMD q1 desktop and server said "AMD UP / GAINS" specific q1 on the actual data was also very much misrepresented and I have communicated with AMD corporate investigations and Chief legal on this topic who follow all my data not just this data. The fact is AMD concealed a down quarter in server and desktop subject run end volumes in the quarter, the server and desktop generation transition ahead and a deflationary price slide for existing PC generations into q1 2025. AMD is also concealing that AMD has a product laundering issue on the amount of unreported revenue units going out as bundle deal presenting AMD channel 'product category' shares well in excess of that which can be determined from the financial alone. This has been known for over 5 years and yes in q1 someone attempted to conceal that who are AMD and the beneficiaries of the unreported revenue units as sales close incentive in volume procurements. There is also a suspect internation extort tariff question for country access, commerce operation and political leave, TC I tend to agree with you but its more than solely a doing business in Russian issue, all over Asia Pacific, South America, even in the United States where OEMS are engaged in brokerage resale of AMD and Intel components to down stream channels. mb
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
44 Comments
Back to Article
meacupla - Friday, May 10, 2024 - link
The mobile market is getting spicy.Hawk Point (8040 series) is rightfully a dud since it's just a 7040 series refresh. I have to question why they bothered pushing it through the door. I am barely seeing any show up at stores. It's been what? 6 months since they announced it and I still only see a few 8945HS equipped laptops compared to the numerous Core Ultra 5/7/9 laptops (Some SKUs, like 135H still MIA).
I look forward to Strix Point, X Elite, and Lunar Lake stirring up the market.
Captain obvious - Friday, May 10, 2024 - link
7040 series also pretty much had a paper launch.With some Shenzhen brands (rightfully, probably) throwing tantrums because they didn't get the amount of 7040 chips they bought.
So I am expecting to see Strix being paper launched in Computex, with availability around CES (lol), just in time for Lunar Lake's arrival.
Also curious about the X Elite, another paper launch that by the time it arrives, will compete with a chip 2 refreshes newer than the M2 baseline on the original announcement...
Blastdoor - Friday, May 10, 2024 - link
By the time it ships the best argument for X Elite would be if qcom throws in a free 5g modem.kn00tcn - Friday, May 10, 2024 - link
which stores? what region? recent earnings had client up so could that mean oems are buying more and the resulting laptops will trickle into stock for H2?seeing numerous laptops in stores doesnt necessarily mean they're all selling, if physical retail then it's likely sponsored shelf space due to the low store margins (well it's not shopping season this time of year anyway, though i did get a ryzen laptop in may of a different year)
and some benchmarks have shown hawk point has more of a gain over phoenix than expected
meacupla - Friday, May 10, 2024 - link
Amazon, Bestbuy, Walmart, newegg, etc. online stores. North America.I just checked today. The only Ryzen 8000 anything that shows up is 8945HS.
Blastdoor - Friday, May 10, 2024 - link
It's really kind of amazing that AMD's desktop marketshare never got anywhere close to a majority. There were several years where AMD had a clearly superior design on a superior process. Of course, Intel's margins contracted a lot during that period, so I guess AMD had an impact in that way.I suspect AMD is near peak marketshare in all segments, though. Intel moving to 20A while AMD is stuck on TSMC N4 is going to hurt.
kn00tcn - Friday, May 10, 2024 - link
intel is also losing hyperthreading, and when will they have 3d cache?process node isnt everything, a mash of performance + power + price (+manufacturing cost) are what matter
diy desktop likely has been majority for a while, but prebuilts, oems, institutions? that's a lot of stock that needs to be guaranteed mixed with old contracts and momentum, i would never expect majority yet
Blastdoor - Friday, May 10, 2024 - link
Losing hypterthreading is a feature not a bug. Intel has efficiency cores to help with MT performance -- AMD doesn't have those (although they are adding their 'c' cores).Process node isn't everything, but it's a lot and absolutely affects performance and power. The only way for AMD to beat Intel by using an inferior process is if AMD has a much better design. That was the case with Athlon vs. Netburst. I doubt it will be the case moving forward.
Samus - Saturday, May 11, 2024 - link
The real problem with hyperthreading now is how easily it has been exploited for security vulnerabilities, and the protections required along with the fact most pipelined 'hyperthreads' aren't hits for multiple factors (bad queuing, under-prioritized branch prediction, collisions requiring discard, etc) don't make it a compelling performance improvement compared to other options. While its kind of "free" performance, it's not as big as it used to be two decades ago when Jackson technology was announced.GeoffreyA - Saturday, May 11, 2024 - link
So, Intel has had, with Skylake, Sunny, and Golden Cove, better performance per watt than their Zen counterparts?Blastdoor - Sunday, May 12, 2024 - link
What point do you think you're making?GeoffreyA - Monday, May 13, 2024 - link
According to the performance-per-watt metric, AMD does have a much better design.Blastdoor - Monday, May 13, 2024 - link
To reach that conclusion you would need to have both designs manufactured on the same process. Thanks to TSMC, AMD has been using a superior process since 2019.I doubt there is any meaningful performance/watt difference between designs, holding all else constant.
GeoffreyA - Monday, May 13, 2024 - link
You're right about the process. I wish we could compare both on the same footing.evanh - Sunday, May 12, 2024 - link
The E-cores are shit. They're just revamped Atom cores! They might be a power saver compared to the P-cores but they lose a lot of performance in the trade. I'd take AMD's ZenC any day.qap - Monday, May 13, 2024 - link
Look again. E-cores are way more powerfull than Atom ever was. Current E-cores outperform big cores from the Atom era (like Sandy Bridge).But even if they weren't, their point is perf/watt. If you have let's say 100W power budget, you will get more performance from E-cores than from P-cores in that power budget (as long as the task can be split to multiple cores). P-cores are only for those tasks, that are not well parallelized and for such usecase 8 P-cores are usually more than enough.
evanh - Monday, May 13, 2024 - link
Of course the E-cores got better, the Atoms were really bad for their time even. The perf/watt is only slightly better than the P-cores simply because the perf is still shit.Foeketijn - Tuesday, May 14, 2024 - link
But when you compare in P and E core ratio, less E cores seem to perform better per Watt. It's silly but true.Blastdoor - Monday, May 13, 2024 - link
X86 is shit — it’s all just based on the processor used in a calculator from the 70s!In fact, humans are shit! They’re just based on some dumb apes that mutated!
charlesg - Tuesday, May 14, 2024 - link
Speak for yourself.Samus - Tuesday, May 14, 2024 - link
E-cores have nothing to do with Atom, and E-cores are quite potent. The problem is the P-cores are too weak to compete with AMD's Zen4 cores head-on so Intel needs to combine the performance of both P and E cores to match them, which is only effective in multithreaded applications.nandnandnand - Friday, May 10, 2024 - link
Meteor Lake was supposed to have Adamantine, but it didn't happen. If Arrow Lake also skips Adamantine, that would suck. I want to see it in action, and in large amounts (anything above 128 MB would be an improvement on the L4 they used in the past).Zoolook13 - Monday, May 13, 2024 - link
Zen 5 will utilize both N4 and N3 it seems.haukionkannel - Friday, May 10, 2024 - link
So AMD market share still is low and income is even less than that... sigh"Buy Intel" still rules the market...
SanX - Friday, May 10, 2024 - link
itis just the inertia. Most people just don't know that the #1 in the list of TOP 500 exascale supercomputers last few years was called Frontier based on AMD EPYC, it's more than 2x faster while consumes same power as the next one in the list - Aurora - based on Intel. Look at Passmark high end processors benchmarks - all 16 first places are AMD. Out organization also switched to AMD as well as many my colleagues.SanX - Monday, May 13, 2024 - link
and what is also important - the price/performance for AMD EPYC is approximately 1.7 times better than with Intel XeonDante Verizon - Saturday, May 11, 2024 - link
It's hard to fight the corrupt system. That's why AMD hasn't overtaken it yetHow long did Intel irregularly retain the right to sell CPUs to Huawei while AMD was barred, without any rational justification?
Jimbo123 - Saturday, May 11, 2024 - link
AMD may have gained a little share so far, but its revenue is the lowest and decreasing, at risk to lose income, or loss; while Intel is moving to product rapidly with lower nm, gaining on TSMC, so will gain on AMD. I can't see there is any good news to AMD, I am sorry.Dante Verizon - Saturday, May 11, 2024 - link
Hahaha sad Intel stockholder. They're losing in every possible market. lolPeachNCream - Saturday, May 11, 2024 - link
The wider problem isn't so much an AMD-only thing, but a PC industry issue. This article does rightly point out a piece-of-pie increase for AMD, but it doesn't (and shouldn't) note the decline in overall sales numbers or total units purchased which has been a troublesome drag on both Intel and AMD for a few years now. Some of that may begin to change as lifecycle replacements of pandemic-era purchases roll in over this and next year, but the other problem is that it's getting harder to justify a dedicated computing platform due to consolidation of compute into datacenters and cloud host physical locations alongside ever more credible "phone-replaces-PC" and/or "Timmy-only-owns-a-phone" situations that appear to be influential on business and consumer buying habits.Threska - Saturday, May 11, 2024 - link
Hence why the server market is so important to everyone, including ARM.Terry_Craig - Saturday, May 11, 2024 - link
TSMC is the last company I see Intel beating.Oxford Guy - Saturday, May 11, 2024 - link
Intel could ship Piledriver-based parts as their main desktop CPUs for general business PCs and Dell would still sell them as the default.GeoffreyA - Saturday, May 11, 2024 - link
Good one! Made me chuckle.Bruzzone - Saturday, May 11, 2024 - link
Mercury Research knowingly presents flawed data to present a false narrative immersing unknowing members of this comment post, and the Anadtech audience, in a syndicate operation's invented reality.The AMD (and Intel) and OEMs and Mercury objective is to conceal racketeering and corporate tax evasion.
Mercury so said AMD share and growth statement specific q1 are not correct and far from on the nuances left unreported in a reversal of the facts by Mercury and report outlets. Ultimately Mercury is concealing generations run end supply elastic price slide, PC price deflation into second half and AMD Client division's $467 million loss in q1 2024 relying on TR7K as a price at cost margin support.
My counter point report 'draft' has been emailed to Ryan, Gavin and Anton for publication here at Anandtech. I offered up the report on my own name or with Anton as a point counter point piece also available to Ryan and Gavin.
I've had it with the industry analyst lies to conceal actual x86 volumes, actual AMD and Intel desktop, workstation, mobile and server category shares.
Mike Bruzzone, Camp Marketing
Terry_Craig - Saturday, May 11, 2024 - link
Sounds like a Russian mafia agent. :')Bruzzone - Sunday, May 12, 2024 - link
Mr. Shilov has connections, back in 2009 as the voice and transmitter of the Intel stock market rigging supply signal cipher in its fourth generation disintegrated math story problem format.Since 2009, I have complimented Mr. Shilov seemingly escaping nefarious forces, technology essentially a mob 'associate network' warring business of clans and all over the world. Mr. Shilov somehow rising above to become what appears to me the only generalist PC tech reporter capable of investigative journalism today. When he puts himself to that task.
I applaud Mr. Shilov's rise above nefarious forces but with this AMD q1 share piece have concerns and I will email Mr. Shilov again today on aspects of his report, from Mercury Data, that do not math.
Mr. Shilov is also aware and from me directly, that I have a spot reserved for his personal appearance at the Intel Corporation United States Congressional investigative hearing on the matters of Intel Corp. caught up in a 30 year inter nation economic espionage subject 18 USC 1956, 1962(c) and 1831 and 2382.
This matter in EUCC currently is the Intel C-240/22-P appeal matter addressing bundle deal sales close incentives of all types; payoffs essentially for access and allocation, this is not about first dollar discount. I have advised AMD pay attention to C-240/22 participating in today what they complained about back in 2005 pursuant EUCC v Intel 37.991 subject AMD v Intel.
Note Mr. Shilov during his stint at X-bit Labs is not the only expert witness United States Congress is interesting in conferring on my recommendations. Specific Intel Supply Signal Cipher in its third generation product category by quarter percent supply format the Barret men Mr. Ryan Shrout of PC Perspective, Mr. Fred Abazovic of Fudzilla, Ms. Monica Chen of Digitimes among others. For Signal Cipher In its first generation format Mr. Linley Gwennap and Mr. Kevin Krewell.
The hot line for witness immunity, well known and for over a decade as proceedings progress, application into the USDOJ cartel amnesty program is joyce.branda@usdoj.gov.
Mike Bruzzone, Camp Marketing
Bruzzone - Monday, May 13, 2024 - link
Anton, Ryan and Gavin have all the data to reassess the AMD q1 situation where Anandtech was at the end of this AMD tail wagging on the thousands media outlets who reported AMD's false narrative before AnandTech across WW industry, electronics, PC, business and financial press. Anton, Ryan and Gavin make this right welcome to ur rocket ship ride. mbPeachNCream - Tuesday, May 14, 2024 - link
What are you smoking and can you share it with the rest of us? Also who are you and what makes you think your opinion is important/informed enough for us to read a text wall?In fact, nevermind, no one cares what your response would be. Go back to your mom's basement kiddo.
Threska - Tuesday, May 14, 2024 - link
You might be talking to AI. That word salad reads like it.**Not that you can throw any stones.
PeachNCream - Tuesday, May 14, 2024 - link
"*Not that you can throw any stones."Heh! Totally fair and no argument from me, but it is hard to resist the low hanging fruit of a decent portion of the PC enthusiast sub-group. They're so...well you know what I mean. You're lucid enough to get it, obviously.
Bruzzone - Tuesday, May 14, 2024 - link
Mt response here and to all of PC Media land who got caught up in the AMD, Mercury, Shilov q1 2024 puff piece where not ONE single assessment of the Mercury data specific AMD is true, in fact misrepresented can be found here;https://seekingalpha.com/user/5030701/comments
Just expand the comment line for all the gory details.
mb
Terry_Craig - Tuesday, May 14, 2024 - link
Definitely sounds like a Russian "Not ONE single"Bruzzone - Tuesday, May 14, 2024 - link
I've asked Anton to rewrite the correction and Ryan the Gavin have the same opportunity. Anton's take is very much misrepresented presenting basically an invented assessment. Original AMD q1 desktop and server said "AMD UP / GAINS" specific q1 on the actual data was also very much misrepresented and I have communicated with AMD corporate investigations and Chief legal on this topic who follow all my data not just this data. The fact is AMD concealed a down quarter in server and desktop subject run end volumes in the quarter, the server and desktop generation transition ahead and a deflationary price slide for existing PC generations into q1 2025. AMD is also concealing that AMD has a product laundering issue on the amount of unreported revenue units going out as bundle deal presenting AMD channel 'product category' shares well in excess of that which can be determined from the financial alone. This has been known for over 5 years and yes in q1 someone attempted to conceal that who are AMD and the beneficiaries of the unreported revenue units as sales close incentive in volume procurements. There is also a suspect internation extort tariff question for country access, commerce operation and political leave, TC I tend to agree with you but its more than solely a doing business in Russian issue, all over Asia Pacific, South America, even in the United States where OEMS are engaged in brokerage resale of AMD and Intel components to down stream channels. mb