The few laptop questions and answers were interesting to me, thanks! It does seem Intel wasn't as aggressive with defining power limits on laptops, though it does seem most manufacturers have a better handle on it compared to the initial Kaby Lake R launch.
That's because it was in Intel's interest not to fix that. This way they could claim higher and higher performance for their mobile chips each year, with most people being none-the-wiser that Intel's "trick" to obtain the higher performance was to raise the real power consumption of the chips.
If they were aggressively limiting the TDP, they would've also needed to aggressively limit the performance of their new CPUs.
First, that person didn't mention any AMD bash or any. Second, if you move ahead of the usual reddit and other mainstream and step ahead into the AT forum you'll see that "IntelUser2000" posts and valuable input. You don't even need to be a part of the forum for so long to recognize it btw since I'm pretty new.
Day by day AMD fangirls are way too much riling up, BS at Der8aur and GN Steve next is what ? Get over it. Let me burst your AMD BS, does AMD processors or chipsets have a Datasheet ? NOPE. Nothing at all and its BS because you can never get such detailed explanation of the technology from AMD, they are a CPU Core corp and they don't have that documented.
Only one which I was able to find (2017 Ryzen Master and OC guide) that's it and fcking no info on ant chipset or CPU...absolutely nothing.
Ryzen 3000 is an excellent value and perf product except that it sucks all fun out because of zero headroom in any customization inbefore people bark.
I would actually interpret the very last sentence, "So other CPU manufacturers might have different quality goals than we do." as slight AMD bashing, hinting that their CPUs aren't as high quality as Intel's. He also never mentioned AMD by name. Maybe he's not allowed to?
Well, its difficult to answer without sounding that way when someone asks you "how are you better?"
I wouldn't call it bashing.
Anyway, Guy clarified lot of points that might have been unclear and revealed the motives on how him and his team does things. That's called being open and honest.
There's no part of being an Intel shill that necessitates "bashing" AMD or being incapable of making useful input, so your "rebuttal" here isn't much of one.
Between the weird off-topic ranting and referring to the people who like things that you don't like as "fangirls", it's quite obvious that you're not approaching this rationally.
"Guy Therien: Well certainly the enthusiast crowd is a significant focus that we go after – you’ve heard about our creator PC effort, and our entire K SKU line is orientated towards "
You know speaking a language and seeing it written usually ends up as two different things. There are a few people I know that use irregardless in place of regardless which I'm sure if they saw it written would think that is weird.
I was initially very skeptical about the IPM tool, but he won me over in the end. As I get older I tire of constant tweaking and just want the damn thing to work 100% of the time. Being able to overclock and still have confidence that its going to work, while not just doing a "token" overclock just to say I've done one, sounds ideal.
I agree about your desires. I haven't used IPM, but on my last build (4790k) I found that even with a week of Prime-95 torture in a hot room (~82-84F vs mid to upper 70s normally during the summer); that with the system at near 100% continuous load (gaming is actually my light use case since I suspend distributed computing workloads and don't have everything pegged at 100%) I was still getting a random crash every 2 or 3 months which were preventable with a 100 MHz backoff.
If IPM's edge case evaluation and backoff can get me into that state with less work I'm all in (assuming my 2021-2 box is Intel anyway); especially if it makes retesting and gradual backoff as the hardware ages easy.
Is he trying to somehow imply that Ryzen CPUs won't last as long? Wait, is this their new damage control system; "We can't Compete on raw performance, or performance per dollar, or performance per watt, but we can compete on performance per... year? ... decade?"
Cutress said "But most enthusiasts just want to use an off the shelf processor and still have the best performance".
Ironic, because enthusiast comments so often reflect an obsession with manual overclocking. Most amazing are those who complain that Ryzen parts - because they auto-clock close to the limits of the silicon - are "lousy overclockers"! This even appears in some tech reviews, and not only on the websites dedicated to OC. It's usually dragged in as a final "con" against Ryzen. Con indeed.
I don't necessarily buy this narrative. Computers are the hot rods of the 21st century. I wouldn't call the "car people" of today any less "car people" just because their work has changed from tuning carburetors and ignition timings with screwdrivers to flashing an EPU. The enthusiasts want performance. Manual OC has been a necessary means to that end. Now that the benefit is finally going away we will see many fewer people doing it. Of course there will always be the people who argue the "quality benefits of the old ways" (eg vinyl records) but that is a very small group.
Great interview. Extremely interesting and informative. But like many of the articles I read on this site, it's littered with grammatical errors. I realize there are a million things more important than this, but still. Some of them would've been caught by simply clicking the "spell check" button.
It's an interview - a transcription of what was said. People don't talk in complete sentences, surprisingly enough. Especially non-PR trained engineers. This is actually the cleaned-up version, with plenty of Guy's own idiosyncracies coming through.
Here's a few I think are genuine typos, I'm guessing Guy didn't describe a head popping through the laptop, maybe that's reserved for a future product :P
"and can turbo at that high freuqency for longer."
"and that the head maybe doesn’t come through and become uncomfortable to touch"
Oh yeah I totally get that. I didn't mean that he was speaking in incomplete sentences or not being super formal. There are just several typos throughout that aren't related to the fact that it's an interview piece. Another user already pointed out a couple of them below. It's something I've noticed pretty frequently on this site, regardless of the topic. It's not a cardinal sin or anything, but i imagine one that could be fixed relatively easily.
There being some irony as well that if you look at the whole lifetime of a product, Intel's have been losing performance through vulnerability patches quite noticeably. Unless Ryzen CPU's all start dying after three years of use, I'd think Intel should best be quiet regarding longevity.
In my experience, my Intel hardware has been Very reliable. 1) Dell 8500 laptop from 2003 works fine, but its hard drive got tired. 2) Compaq Presario desktop from 2008, on which I'm writing this. 3) a cheap destop purchased in about 1998 lasted my son about 15 years, when its chipset lost a few marbles and the sound hardware and the motherboard USB quit. Worked-around with add-in cards. Never had an intel processor die. The only thing threatening my otherwise good old hardware is Microsoft, by torpedoing XP, Vista, and Win 7, and even the earlier versions of Win 10, and the collusion of the apps and browser companies that now require the latest OS, which can never be loaded onto my otherwise adequate and reliable hardware.
We're still running a couple dual-P3 systems at work (fax server with ISA fax card and PCI modem). And several of the original Opteron CPUs.
We also have dozens of AMD Sempron, Athlon64, Athlon X2/X3/X4 desktops still in the schools, along with some of the earlier Pentium G-series.
We run systems until the hardware dies (motherboard or cpu), basically. A 10-year life isn't odd at work.
My work laptop even has an i5 2700-something in it. Was going to replace it this year, but swapping in an SSD and more RAM made it tolerable to use again, so well keep it for another year or two. :)
My work desktop is an Athlon X2 (8 GB of RAM and a HDl that's being replaced with a Ryzen 5 2600 (16 GB, NVMe) that's redirected to last at least the next 5 years+.
It would probably be impossible for them to provide a useful figure, as it depends so much on load and the specific silicon. I've had 3 Sandy Bridge CPUs go faulty on me since 2011 - the first went faulty around 3 years from new, the the next two died between 4 and 6 years of age. In all cases, the faults were unusual - the CPU would still POST and the system would load windows, but running applications would stutter, halt, and eventually crash without any obvious cause. The one that died first wasn't overclocked, while the other two were. So it seems like even if they provided such a number, it wouldn't be worth much!
My '11 launch i7-2600K is still crankin' away. It's been in 24/7 operation since day one, with the vast majority of the time spent at 4.8 or 5GHz and 1.4-1.45v on an air cooler. Gonna be sad to see it go.
" So motherboard manufacturers and ODMs are investing in their power delivery and thermal solutions to allow them to maximize performance or get a certain about of turbo duration without throttling – to the maximum extent possible."
Translation: TDP is and will continue to be a worthless number for Intel chips, as stock board settings will violate the piss out of it while somehow remaining "in-spec". Tell Intel Guy Fellow Therien he needs to cough up a separate "Turbo TDP" and make the CPU respect it regardless of what the motherboard demands, EXCEPT when *actually* overclocking (running non-stock settings, even if that just means throwing a switch in the BIOS). Otherwise his suggestion of testing with THREE motherboards with different configs sounds like something I'm sure you'll enjoy implementing from now on!
I mean seriously AMD is mauling the hell out of Intel when it comes to real-world power, and I feel like it has barely received a passing mention in most of the tech press. Meanwhile for years many outlets harped on and on about AMD's high power consumption and poor efficiency when the shoe was on the other foot.
TDP/PL1 is the same regardless, the manufacturers manipulate PL2 and Tau based on their design. Your Turbo TDP is likely already covered by PL2. The issue is that OEMs never ever list what they set PL2 to be.
The issue is that BY DEFAULT the CPU should be in control and have some modicum of respect for the TDP. You should need to throw a switch to enable board-dictated limits. There's no reason an end user should have to guess at what the results will be when they drop a "95W" CPU into X or Y board. If Intel wants to sell 125, 150W chips, let them label them as such.
AMD has a noticeable advantage in power efficiency, and I feel like this is going to be glossed over continuously by the mainstream press (unlike the 'dozer days) - at least until Intel gets its new process ironed out. Then I predict it will suddenly become an issue that needs a spotlight.
Agreed entirely here. Saying "it's in spec" isn't much use when we don't know how the spec is being implemented for any given CPU.
Laptops are definitely the worst for this, though. It seems like between Intel's TDP, their boost specs and the manufacturer's interpretations of those specs, we've ended up in a situation where you *have* to read a review of a device just to have some idea how it will perform.
@DrIanCutress: Embarrassing that you have an article like this and succeed in ignoring completely the performance losses due to Intel vulnerabilities. Are you desperately trying to ignore this serious issue to avoid upsetting Intel? Your lack of independence and timidity entirely undermines your credibility with readers. This place used to be a go-to site for technical information, now it's just a sham.
Not embarrassing at all, because the security vulnerability stuff isn't what Guy works on, and wasn't what I wanted to ask him about. Not everything has to be about security. Getting clarity on Intel's position on TDP and Turbo was the absolute key critical goal here in the time allowed. Getting detailed answers about key critical questions requires time, finesse, and interviewer technique. If you dive in the deep end straight away then the interviewee clams right up and he get nothing.
Ian thanks for your non-answer. The point is Anandtech find the time for this article, and others, but we're still waiting for the article on the latest Intel vulnerabilities - so where is it? We were informed months ago you were "waiting on Intel" so are they still stalling you, or have you spiked it by now in the hope we will forget (hint: we won't)? The simple fact of the matter is that while you may not think "everything has to be about security" it's an important topic for your readers particularly when it impacts performance by up to 40% and when you ignore these security/performance issues that - mainly - affect Intel it tends to make you look... biased.
The CPU review articles have stated that systems are patched for Meltdown, Specter, etc. The performance losses will be reflected in the benchmarks. What more should they be doing?
> The CPU review articles have stated that systems are patched for Meltdown, Specter, etc.
No they weren't, because Anandtech botched the Ryzen reviews by using older Windows versions for Intel that lacked the latest mitigations giving Intel an unfair advantage and skewing the results significantly, rendering the benchmarks null & void - this cockup has been covered extensively with Anandrech claiming the mitigations were not available even though they were, which shows a surprising lack of knowledge.
> The CPU review articles have stated that systems are patched for Meltdown, Specter, etc.
No they weren't, because Anandtech botched the Ryzen reviews by using older Windows versions for Intel that lacked the latest mitigations giving Intel an unfair advantage and skewing the results significantly, rendering the benchmarks null & void - this cockup has been covered extensively with Anandrech claiming the mitigations were not available even though they were, which shows a surprising lack of knowledge.
They *weren't* fully patched, though; only for issues fixed by mid-2018. The articles said patches for MDS weren't available, when in the most important cases (boards supporting the recent-gen Intel CPUs) there were new BIOS images available coordinated with Microsoft's update released in May.
Dear Ian, if I interpret your discussion correctly, you wanted Intel to explain how consumer-side testers should benchmark the real world performance of new Intel parts? I feel that the Intel’s response was very clear, although you may have a problem accepting it. Intel does not seem to have an interest in guaranteeing predictable performance when their CPU’s are combined with third party motherboard configurations. I think you should just accept the fact that getting an objective real world performance can no longer be determined from the cpu purchase. You need to test for real world oddities with a variety of third party motherboards.
For one example, a motherboard’s software configuration may not work optimally with the cpu, which may lead to sub-optimal power behaviour. My first coffee lake build seemed to oscillate between two power limits unpredictably, and at the time I blamed the motherboard software or hardware for this behaviour. I quickly sold the cpu and motherboard and I have not seen this problem anywhere since then. I see nothing wrong with Intel trying to direct your attention to problems when some motherboards try to boost performance beyond hardware capabilities. In the real world 100w power budget might not be as simple as asking Intel what their cpu can and cannot do.
Seems like just the right guy to ask about how Intel decides which thermal interface material to use on different product offerings and what impact they expect.
WIntel should be more transparent on such details, even more when they come as UEFI settings for us. Instead of publishing such info on exclusive interview, they should make a didactic article on their own sites.
They should also enforce naming standards on settings, as each mobo manufacture seems to use different names for the same thing. In example, on my Aorus Ultra there's setting for disabling Turbo, for Turbo TDP (which the name makes seem to be the base TDP, we need to read the description on manual to understand it), and for Turbo max current. There doesn't seem to be a setting for Tau duration.
I suppose that WIntel isn't also clear enough on these regards with mobo manufacturers, so each one use settings and names they suppose to be the appropriate.
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
53 Comments
Back to Article
jeremyshaw - Thursday, July 25, 2019 - link
The few laptop questions and answers were interesting to me, thanks!It does seem Intel wasn't as aggressive with defining power limits on laptops, though it does seem most manufacturers have a better handle on it compared to the initial Kaby Lake R launch.
Krysto - Thursday, July 25, 2019 - link
That's because it was in Intel's interest not to fix that. This way they could claim higher and higher performance for their mobile chips each year, with most people being none-the-wiser that Intel's "trick" to obtain the higher performance was to raise the real power consumption of the chips.If they were aggressively limiting the TDP, they would've also needed to aggressively limit the performance of their new CPUs.
IntelUser2000 - Thursday, July 25, 2019 - link
I love this article. Among the few I didn't just skim through and actually read most of it.Guy Therien also sounds open and honest about what he's doing.
B3an - Thursday, July 25, 2019 - link
"IntelUser2000" .. "I love this article" .. "Guy Therien also sounds open and honest "I see the Intel shills are already here on the shill article. At least come up with better names.
Quantumz0d - Thursday, July 25, 2019 - link
First, that person didn't mention any AMD bash or any. Second, if you move ahead of the usual reddit and other mainstream and step ahead into the AT forum you'll see that "IntelUser2000" posts and valuable input. You don't even need to be a part of the forum for so long to recognize it btw since I'm pretty new.Day by day AMD fangirls are way too much riling up, BS at Der8aur and GN Steve next is what ? Get over it. Let me burst your AMD BS, does AMD processors or chipsets have a Datasheet ? NOPE. Nothing at all and its BS because you can never get such detailed explanation of the technology from AMD, they are a CPU Core corp and they don't have that documented.
Only one which I was able to find (2017 Ryzen Master and OC guide) that's it and fcking no info on ant chipset or CPU...absolutely nothing.
Ryzen 3000 is an excellent value and perf product except that it sucks all fun out because of zero headroom in any customization inbefore people bark.
ajp_anton - Thursday, July 25, 2019 - link
I would actually interpret the very last sentence,"So other CPU manufacturers might have different quality goals than we do."
as slight AMD bashing, hinting that their CPUs aren't as high quality as Intel's. He also never mentioned AMD by name. Maybe he's not allowed to?
IntelUser2000 - Thursday, July 25, 2019 - link
Well, its difficult to answer without sounding that way when someone asks you "how are you better?"I wouldn't call it bashing.
Anyway, Guy clarified lot of points that might have been unclear and revealed the motives on how him and his team does things. That's called being open and honest.
Spunjji - Friday, July 26, 2019 - link
There's no part of being an Intel shill that necessitates "bashing" AMD or being incapable of making useful input, so your "rebuttal" here isn't much of one.Between the weird off-topic ranting and referring to the people who like things that you don't like as "fangirls", it's quite obvious that you're not approaching this rationally.
peevee - Thursday, July 25, 2019 - link
"Guy Therien: Well certainly the enthusiast crowd is a significant focus that we go after – you’ve heard about our creator PC effort, and our entire K SKU line is orientated towards "Somehow I doubt he actually said "orientated".
Ian Cutress - Thursday, July 25, 2019 - link
You'd be surprised how many people say orientated rather than orientedFreckledTrout - Thursday, July 25, 2019 - link
You know speaking a language and seeing it written usually ends up as two different things. There are a few people I know that use irregardless in place of regardless which I'm sure if they saw it written would think that is weird.nathanddrews - Thursday, July 25, 2019 - link
English vs. English.Orientated is at least a word, but irregardless - *shudder*. No excuse. XD
peevee - Friday, July 26, 2019 - link
Outside of UK?peevee - Friday, July 26, 2019 - link
And more importantly - did he?peevee - Friday, July 26, 2019 - link
Google:"oriented" 1.7B results
"orientated" 25M results
The first is 68 TIMES more popular. No wonder the 2nd sounds so annoying.
EdgeOfDetroit - Thursday, July 25, 2019 - link
I was initially very skeptical about the IPM tool, but he won me over in the end. As I get older I tire of constant tweaking and just want the damn thing to work 100% of the time. Being able to overclock and still have confidence that its going to work, while not just doing a "token" overclock just to say I've done one, sounds ideal.DanNeely - Thursday, July 25, 2019 - link
I agree about your desires. I haven't used IPM, but on my last build (4790k) I found that even with a week of Prime-95 torture in a hot room (~82-84F vs mid to upper 70s normally during the summer); that with the system at near 100% continuous load (gaming is actually my light use case since I suspend distributed computing workloads and don't have everything pegged at 100%) I was still getting a random crash every 2 or 3 months which were preventable with a 100 MHz backoff.If IPM's edge case evaluation and backoff can get me into that state with less work I'm all in (assuming my 2021-2 box is Intel anyway); especially if it makes retesting and gradual backoff as the hardware ages easy.
AshlayW - Thursday, July 25, 2019 - link
Is he trying to somehow imply that Ryzen CPUs won't last as long? Wait, is this their new damage control system; "We can't Compete on raw performance, or performance per dollar, or performance per watt, but we can compete on performance per... year? ... decade?"It's a joke :)
AshlayW - Thursday, July 25, 2019 - link
In all seriousness; great article and interview, thanks. I enjoyed reading it, some good technical insight. :)Oliseo - Thursday, August 1, 2019 - link
No, that's something your inferring.I've leave you guessing what I infer about you....
Arbie - Thursday, July 25, 2019 - link
Cutress said "But most enthusiasts just want to use an off the shelf processor and still have the best performance".Ironic, because enthusiast comments so often reflect an obsession with manual overclocking. Most amazing are those who complain that Ryzen parts - because they auto-clock close to the limits of the silicon - are "lousy overclockers"! This even appears in some tech reviews, and not only on the websites dedicated to OC. It's usually dragged in as a final "con" against Ryzen. Con indeed.
willis936 - Thursday, July 25, 2019 - link
I don't necessarily buy this narrative. Computers are the hot rods of the 21st century. I wouldn't call the "car people" of today any less "car people" just because their work has changed from tuning carburetors and ignition timings with screwdrivers to flashing an EPU. The enthusiasts want performance. Manual OC has been a necessary means to that end. Now that the benefit is finally going away we will see many fewer people doing it. Of course there will always be the people who argue the "quality benefits of the old ways" (eg vinyl records) but that is a very small group.ianisiam - Thursday, July 25, 2019 - link
Great interview. Extremely interesting and informative. But like many of the articles I read on this site, it's littered with grammatical errors. I realize there are a million things more important than this, but still. Some of them would've been caught by simply clicking the "spell check" button.Ian Cutress - Thursday, July 25, 2019 - link
It's an interview - a transcription of what was said. People don't talk in complete sentences, surprisingly enough. Especially non-PR trained engineers. This is actually the cleaned-up version, with plenty of Guy's own idiosyncracies coming through.willis936 - Thursday, July 25, 2019 - link
I was curious if "bajillion" was a technical term. For some reason my browser wants to correct the spelling to "bazillion".Pheesh - Thursday, July 25, 2019 - link
Here's a few I think are genuine typos, I'm guessing Guy didn't describe a head popping through the laptop, maybe that's reserved for a future product :P"and can turbo at that high freuqency for longer."
"and that the head maybe doesn’t come through and become uncomfortable to touch"
"The second reaso is SKU seperation."
ianisiam - Thursday, July 25, 2019 - link
Oh yeah I totally get that. I didn't mean that he was speaking in incomplete sentences or not being super formal. There are just several typos throughout that aren't related to the fact that it's an interview piece. Another user already pointed out a couple of them below. It's something I've noticed pretty frequently on this site, regardless of the topic. It's not a cardinal sin or anything, but i imagine one that could be fixed relatively easily.peevee - Monday, July 29, 2019 - link
"This is actually the cleaned-up version"With every word "oriented" he actually said carefully replaced by your annoying "orientated".
notashill - Thursday, July 25, 2019 - link
I see a lot of mentions of reliability and product lifetime, but does Intel actually specify the expected lifetime of their CPUs?gijames1225 - Thursday, July 25, 2019 - link
There being some irony as well that if you look at the whole lifetime of a product, Intel's have been losing performance through vulnerability patches quite noticeably. Unless Ryzen CPU's all start dying after three years of use, I'd think Intel should best be quiet regarding longevity.dwbogardus - Thursday, July 25, 2019 - link
In my experience, my Intel hardware has been Very reliable. 1) Dell 8500 laptop from 2003 works fine, but its hard drive got tired. 2) Compaq Presario desktop from 2008, on which I'm writing this. 3) a cheap destop purchased in about 1998 lasted my son about 15 years, when its chipset lost a few marbles and the sound hardware and the motherboard USB quit. Worked-around with add-in cards. Never had an intel processor die. The only thing threatening my otherwise good old hardware is Microsoft, by torpedoing XP, Vista, and Win 7, and even the earlier versions of Win 10, and the collusion of the apps and browser companies that now require the latest OS, which can never be loaded onto my otherwise adequate and reliable hardware.phoenix_rizzen - Sunday, July 28, 2019 - link
We're still running a couple dual-P3 systems at work (fax server with ISA fax card and PCI modem). And several of the original Opteron CPUs.We also have dozens of AMD Sempron, Athlon64, Athlon X2/X3/X4 desktops still in the schools, along with some of the earlier Pentium G-series.
We run systems until the hardware dies (motherboard or cpu), basically. A 10-year life isn't odd at work.
My work laptop even has an i5 2700-something in it. Was going to replace it this year, but swapping in an SSD and more RAM made it tolerable to use again, so well keep it for another year or two. :)
My work desktop is an Athlon X2 (8 GB of RAM and a HDl that's being replaced with a Ryzen 5 2600 (16 GB, NVMe) that's redirected to last at least the next 5 years+.
Spunjji - Friday, July 26, 2019 - link
It would probably be impossible for them to provide a useful figure, as it depends so much on load and the specific silicon. I've had 3 Sandy Bridge CPUs go faulty on me since 2011 - the first went faulty around 3 years from new, the the next two died between 4 and 6 years of age. In all cases, the faults were unusual - the CPU would still POST and the system would load windows, but running applications would stutter, halt, and eventually crash without any obvious cause. The one that died first wasn't overclocked, while the other two were. So it seems like even if they provided such a number, it wouldn't be worth much!Slash3 - Saturday, July 27, 2019 - link
My '11 launch i7-2600K is still crankin' away. It's been in 24/7 operation since day one, with the vast majority of the time spent at 4.8 or 5GHz and 1.4-1.45v on an air cooler. Gonna be sad to see it go.Alexvrb - Thursday, July 25, 2019 - link
" So motherboard manufacturers and ODMs are investing in their power delivery and thermal solutions to allow them to maximize performance or get a certain about of turbo duration without throttling – to the maximum extent possible."Translation: TDP is and will continue to be a worthless number for Intel chips, as stock board settings will violate the piss out of it while somehow remaining "in-spec". Tell Intel Guy Fellow Therien he needs to cough up a separate "Turbo TDP" and make the CPU respect it regardless of what the motherboard demands, EXCEPT when *actually* overclocking (running non-stock settings, even if that just means throwing a switch in the BIOS). Otherwise his suggestion of testing with THREE motherboards with different configs sounds like something I'm sure you'll enjoy implementing from now on!
I mean seriously AMD is mauling the hell out of Intel when it comes to real-world power, and I feel like it has barely received a passing mention in most of the tech press. Meanwhile for years many outlets harped on and on about AMD's high power consumption and poor efficiency when the shoe was on the other foot.
Ian Cutress - Thursday, July 25, 2019 - link
TDP/PL1 is the same regardless, the manufacturers manipulate PL2 and Tau based on their design. Your Turbo TDP is likely already covered by PL2. The issue is that OEMs never ever list what they set PL2 to be.Alexvrb - Sunday, July 28, 2019 - link
The issue is that BY DEFAULT the CPU should be in control and have some modicum of respect for the TDP. You should need to throw a switch to enable board-dictated limits. There's no reason an end user should have to guess at what the results will be when they drop a "95W" CPU into X or Y board. If Intel wants to sell 125, 150W chips, let them label them as such.AMD has a noticeable advantage in power efficiency, and I feel like this is going to be glossed over continuously by the mainstream press (unlike the 'dozer days) - at least until Intel gets its new process ironed out. Then I predict it will suddenly become an issue that needs a spotlight.
Spunjji - Friday, July 26, 2019 - link
Agreed entirely here. Saying "it's in spec" isn't much use when we don't know how the spec is being implemented for any given CPU.Laptops are definitely the worst for this, though. It seems like between Intel's TDP, their boost specs and the manufacturer's interpretations of those specs, we've ended up in a situation where you *have* to read a review of a device just to have some idea how it will perform.
CityBlue - Friday, July 26, 2019 - link
@DrIanCutress: Embarrassing that you have an article like this and succeed in ignoring completely the performance losses due to Intel vulnerabilities. Are you desperately trying to ignore this serious issue to avoid upsetting Intel? Your lack of independence and timidity entirely undermines your credibility with readers. This place used to be a go-to site for technical information, now it's just a sham.AV_Stables - Friday, July 26, 2019 - link
aye its been brought up and avoided many times /Ian Cutress - Friday, July 26, 2019 - link
Not embarrassing at all, because the security vulnerability stuff isn't what Guy works on, and wasn't what I wanted to ask him about. Not everything has to be about security. Getting clarity on Intel's position on TDP and Turbo was the absolute key critical goal here in the time allowed. Getting detailed answers about key critical questions requires time, finesse, and interviewer technique. If you dive in the deep end straight away then the interviewee clams right up and he get nothing.CityBlue - Saturday, July 27, 2019 - link
Ian thanks for your non-answer. The point is Anandtech find the time for this article, and others, but we're still waiting for the article on the latest Intel vulnerabilities - so where is it? We were informed months ago you were "waiting on Intel" so are they still stalling you, or have you spiked it by now in the hope we will forget (hint: we won't)? The simple fact of the matter is that while you may not think "everything has to be about security" it's an important topic for your readers particularly when it impacts performance by up to 40% and when you ignore these security/performance issues that - mainly - affect Intel it tends to make you look... biased.Mr Perfect - Friday, July 26, 2019 - link
The CPU review articles have stated that systems are patched for Meltdown, Specter, etc. The performance losses will be reflected in the benchmarks. What more should they be doing?CityBlue - Saturday, July 27, 2019 - link
> The CPU review articles have stated that systems are patched for Meltdown, Specter, etc.No they weren't, because Anandtech botched the Ryzen reviews by using older Windows versions for Intel that lacked the latest mitigations giving Intel an unfair advantage and skewing the results significantly, rendering the benchmarks null & void - this cockup has been covered extensively with Anandrech claiming the mitigations were not available even though they were, which shows a surprising lack of knowledge.
CityBlue - Saturday, July 27, 2019 - link
> The CPU review articles have stated that systems are patched for Meltdown, Specter, etc.No they weren't, because Anandtech botched the Ryzen reviews by using older Windows versions for Intel that lacked the latest mitigations giving Intel an unfair advantage and skewing the results significantly, rendering the benchmarks null & void - this cockup has been covered extensively with Anandrech claiming the mitigations were not available even though they were, which shows a surprising lack of knowledge.
GreenReaper - Tuesday, July 30, 2019 - link
They *weren't* fully patched, though; only for issues fixed by mid-2018. The articles said patches for MDS weren't available, when in the most important cases (boards supporting the recent-gen Intel CPUs) there were new BIOS images available coordinated with Microsoft's update released in May.RealBeast - Friday, July 26, 2019 - link
Great article Ian, very informative. Thanks!Mookid - Friday, July 26, 2019 - link
Dear Ian, if I interpret your discussion correctly, you wanted Intel to explain how consumer-side testers should benchmark the real world performance of new Intel parts? I feel that the Intel’s response was very clear, although you may have a problem accepting it. Intel does not seem to have an interest in guaranteeing predictable performance when their CPU’s are combined with third party motherboard configurations. I think you should just accept the fact that getting an objective real world performance can no longer be determined from the cpu purchase. You need to test for real world oddities with a variety of third party motherboards.For one example, a motherboard’s software configuration may not work optimally with the cpu, which may lead to sub-optimal power behaviour. My first coffee lake build seemed to oscillate between two power limits unpredictably, and at the time I blamed the motherboard software or hardware for this behaviour. I quickly sold the cpu and motherboard and I have not seen this problem anywhere since then. I see nothing wrong with Intel trying to direct your attention to problems when some motherboards try to boost performance beyond hardware capabilities. In the real world 100w power budget might not be as simple as asking Intel what their cpu can and cannot do.
Mookid - Friday, July 26, 2019 - link
And thanks for very information article 👍sor - Sunday, July 28, 2019 - link
Seems like just the right guy to ask about how Intel decides which thermal interface material to use on different product offerings and what impact they expect.bestbingoonlinesites - Monday, July 29, 2019 - link
What a Great Sites of talking tdp turbo and overlocking an interview with intel fellow guy therien!HikariWS - Thursday, December 5, 2019 - link
I sum up on the awesome explanations.WIntel should be more transparent on such details, even more when they come as UEFI settings for us. Instead of publishing such info on exclusive interview, they should make a didactic article on their own sites.
They should also enforce naming standards on settings, as each mobo manufacture seems to use different names for the same thing. In example, on my Aorus Ultra there's setting for disabling Turbo, for Turbo TDP (which the name makes seem to be the base TDP, we need to read the description on manual to understand it), and for Turbo max current. There doesn't seem to be a setting for Tau duration.
I suppose that WIntel isn't also clear enough on these regards with mobo manufacturers, so each one use settings and names they suppose to be the appropriate.