"The cooling system has been used by Intel for CPUs with an 84 W TDP since at least 2013, which makes the whole recall bizarre."
No its not bizarre. Even anandtechs own reviews show Intels "TDP" ratinh is worthless now. They keep twisting what it means so much its next to worthless.
At least AMD admitted its CPUs were power hungry during its bulldozer days.
everybody scales the CPU as far as they can so initial thing is always a bit different. bulldozer were bad so they cranked the power to at least have some performance. right now AMD have more power headroom so they are less on edge, while all intel did for 3-4 gens was adding more juice to cpu. If they would increase TDP each time, marketing would cry.
TDP is the value that the cooler needs to be able to cool in order to guarantee that the CPU can run at its advertised base clocks. If this is a 84W TDP cooler running a 83W TDP CPU and it cannot do that, there is something bizarre going on. When has Intel "twisted" it to the point of worthlessness? Maybe you never understood it in the first place and thought it was supposed to mean "the power consumption of the chip when running my favorite game".
In light of the recent flaws we heard just a couple of hours ago (and how Intel knowingly shipped Cascade Lake with a flaws they were fully aware of, despite saying said SKUs being unaffected), people should THINK about some bits here, how it may end up being just another piece of the puzzle and eventually paint yourself the bigger picture already.
I'm not saying Intel made them EOL and tries to pulls those to be replaced by newer CPUs with different steppings which may be unaffected to prevent another law-suit, but this TDP-thing here looks like a flimsy excuse for the fact that Intel a) sets its boxed SKU as EOL and b) even asks given SKUs for being RETURNED for being replaced by tray-variants afterwards.
Whole thing tastes a bit sour, though it may be just me digging some Salt™ again …
Regardless of the reasoning, customers expect OEM boxed coolers to properly cool the CPU they are shipped with and if said coolers are insufficient then something needs to be done to address the problem. I do agree with Marlin though because Intel's rated TDP is not reflective of the CPU in a variety of workloads and that can present the company as dishonestly representing their wares.
I'm not sure this is the definition of tdp, for either Intel or amd. Definitely not for amd according to a really good gamers nexus video done in the past month. I think maybe this is Intel's definition of tdp, but I don't exactly remember.
Even if it is Intel's definition, the multi core enhancement of recent chips that violate their rules of time limits of power could potentially break these wimpy coolers and throttle the cpus. Maybe xeons are more tightly controlled than desktop parts however.
AMD historically has kept far closer to actual real world TDP they list on the box, sometimes go past, more often than not stay below (with exceptions of course)
Intel and dare say Nvidia have proven time and time again to not do such (exception being Nvidia 1xxx and newest "kind of") Intel wise, only at "base" clock, once turbo kicks in, not so much.
to each their own, IMO if you cannot trust what is written "on the box" how can you find proper cooler (very least, AMD with Ryzen if "auto" set, rarely passes by it's own rating, certainly not like Intel has done as of late to try and "one up" AMD.....outright dishonesty, something they (such as Intel / Ngreedia) lambast the crap out of AMD - taking well out of truthful context
yes Bulldozer era as well as Phenom 1 generation had issues, no doubts about that....
AMD general stay +/-5% TDP / ABP / TBP (whatever you want to call it) they have all the high end capacitors all that fancy stuff to ensure it
something dreadful going on over there at Intel seems like .. ehh sucks to be them, given they should @#$ know better these days.
"Intel has a long history of being much farther under top than AMD " oh really ?? explain how their cpus use more power the spec on the box. oh wait, Anandtech stated, and explained how here : https://www.anandtech.com/show/13544/why-intel-pro... spend less time waving the intel flag, and maybe you would see at least some of the lies from intel that you STILL believe.
So let me get this straight, I need to open my case, remove the CPU and cooler, go to them with both, and in return for the cooler I will just get the (same) CPU back??? Do they give me some money to buy another cooler?
Yup, literally the only thing which changes is the CPU, which is going to be replaced (for a reason, I might add).
… and in light of the recent flaws we heard just a couple of hours ago (and how Intel knowingly shipped Cascade Lake with a flaws they were fully aware of, despite saying said SKUs being unaffected), you shall THINK about some bits here, how it may end up being just another piece of the puzzle and eventually paint yourself the bigger picture already.
Before the conspiracy theories about fake TDPs get out of hand:
1) E-2274G = 4 GHz quad-core w/ HT, like 6700K or 7700K. 2) i3 8350K rarely hits 40 watts, torture test might do 50 - 60. 3) 22xx Xeons are 2 years more recent, typically lower voltage. 4) No recall for the 6 core E-2276G with the same cooler... 5) 65w 6+ core CPUs run 4+ GHz fine with Intel's stock coolers.
The thin aluminum stock cooler does still suck, though. AMD's equivalent is better. If you want to REEEE about something, there you go.
pretty bad when the maker in this case Intel effectively KNOWS it cannot properly cool the cpu in question, i.e TDP 84w with cooler designed for such, likely means it (the cpu) does NOT stay within that TDP envelope and/or the cooler certainly cannot.
For a "premium vendor" as everyone more often that not "waves the flag" saying Intel or nothing mantra....egg on their face huh (not once, not twice but many many times over)
good they are recalling I suppose, terribad for end user (consumer / big purchasers) to have to or not (lose warranty if not) send back?
Even far more odd considering it is Xeon class.
maybe they should reach out to OEM makers for something they know for sure will keep within operating spec instead of going about "all on their own"
I ran my trust old E8400 with stock cooler (even overclocked) for a LONG time before i bothered to go aftermarket wise..me thinks they probably should invest $$$$$ to assure this does NOT take place again...tis Intel, who knows
I have chucked every Intel cooler I've ever had, except maybe the Nehalem i7-920 cooler which was like 2lbs and had a huge copper slug in the center...only used it briefly though as I eventually overclocked that chip.
I think by now everyone knows Intel coolers are a joke. Almost every chip will eventually throttle with its bundled cooler.
Whole we have the usual suspects bleating over the inane conspiracy du-jour: This sounds far less like an IC issue, and far more likely HSF issue. DHA-A should be able to handle 83W just fine, but it's hardly a spring chicken in terms of design. Thermal compound degradation, malformed components (the heatsink is made up of friction-fit components, if one of these was out of tolerance then that affects thermal performance) etc. The bundled heatsink is rarely pushed to the rated performance, so batch issues can easily have slipped under the radar.
Whatever happened to engineering know-how at Intel? Testing a cooler to make sure it works properly for the CPU it's bundled with shouldn't be a problem for a company that use to pride itself as the technology leader in this space. And, why this is a particularly embarrassing problem here: people buy Xeon CPUs because they are expected to keep performing just fine at or near 100% their specified TDP year-in, year-out. That's why many XEONs and similar chips from AMD (EPYC, or other professional lines) are usually clocked a bit more conservative, and also priced higher than their consumer counterparts. So, Intel shipping garbage bundled coolers is really egg on their face, and was utterly avoidable. That company has problems, not just with getting 10 nm fabs going.
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.
26 Comments
Back to Article
A5 - Thursday, November 14, 2019 - link
Have to admit I didn't even know Intel sold boxed versions of the Xeons.deil - Thursday, November 14, 2019 - link
yeah and that one looks like Pentium cooler. small mistake propably a typo :) in specpsychobriggsy - Thursday, November 14, 2019 - link
Well I doubt many people bought boxed quad-core Xeons.But this does suggest that Intel might want to use more than just base clocks to decide TDP.
Marlin1975 - Thursday, November 14, 2019 - link
"The cooling system has been used by Intel for CPUs with an 84 W TDP since at least 2013, which makes the whole recall bizarre."No its not bizarre. Even anandtechs own reviews show Intels "TDP" ratinh is worthless now. They keep twisting what it means so much its next to worthless.
At least AMD admitted its CPUs were power hungry during its bulldozer days.
deil - Thursday, November 14, 2019 - link
everybody scales the CPU as far as they can so initial thing is always a bit different.bulldozer were bad so they cranked the power to at least have some performance.
right now AMD have more power headroom so they are less on edge, while all intel did for 3-4 gens was adding more juice to cpu. If they would increase TDP each time, marketing would cry.
Death666Angel - Thursday, November 14, 2019 - link
TDP is the value that the cooler needs to be able to cool in order to guarantee that the CPU can run at its advertised base clocks. If this is a 84W TDP cooler running a 83W TDP CPU and it cannot do that, there is something bizarre going on. When has Intel "twisted" it to the point of worthlessness? Maybe you never understood it in the first place and thought it was supposed to mean "the power consumption of the chip when running my favorite game".Smartcom5 - Thursday, November 14, 2019 - link
In light of the recent flaws we heard just a couple of hours ago (and how Intel knowingly shipped Cascade Lake with a flaws they were fully aware of, despite saying said SKUs being unaffected), people should THINK about some bits here, how it may end up being just another piece of the puzzle and eventually paint yourself the bigger picture already.I'm not saying Intel made them EOL and tries to pulls those to be replaced by newer CPUs with different steppings which may be unaffected to prevent another law-suit, but this TDP-thing here looks like a flimsy excuse for the fact that Intel a) sets its boxed SKU as EOL and b) even asks given SKUs for being RETURNED for being replaced by tray-variants afterwards.
Whole thing tastes a bit sour, though it may be just me digging some Salt™ again …
PeachNCream - Thursday, November 14, 2019 - link
Regardless of the reasoning, customers expect OEM boxed coolers to properly cool the CPU they are shipped with and if said coolers are insufficient then something needs to be done to address the problem. I do agree with Marlin though because Intel's rated TDP is not reflective of the CPU in a variety of workloads and that can present the company as dishonestly representing their wares.lipscomb88 - Thursday, November 14, 2019 - link
I'm not sure this is the definition of tdp, for either Intel or amd. Definitely not for amd according to a really good gamers nexus video done in the past month. I think maybe this is Intel's definition of tdp, but I don't exactly remember.Even if it is Intel's definition, the multi core enhancement of recent chips that violate their rules of time limits of power could potentially break these wimpy coolers and throttle the cpus. Maybe xeons are more tightly controlled than desktop parts however.
Ian Cutress - Thursday, November 14, 2019 - link
>Really good GN videoor
The extensive AnandTech articles on TDP and Turbo
Dragonstongue - Thursday, November 14, 2019 - link
AMD historically has kept far closer to actual real world TDP they list on the box, sometimes go past, more often than not stay below (with exceptions of course)Intel and dare say Nvidia have proven time and time again to not do such (exception being Nvidia 1xxx and newest "kind of") Intel wise, only at "base" clock, once turbo kicks in, not so much.
to each their own, IMO if you cannot trust what is written "on the box" how can you find proper cooler (very least, AMD with Ryzen if "auto" set, rarely passes by it's own rating, certainly not like Intel has done as of late to try and "one up" AMD.....outright dishonesty, something they (such as Intel / Ngreedia) lambast the crap out of AMD - taking well out of truthful context
yes Bulldozer era as well as Phenom 1 generation had issues, no doubts about that....
AMD general stay +/-5% TDP / ABP / TBP (whatever you want to call it) they have all the high end capacitors all that fancy stuff to ensure it
something dreadful going on over there at Intel seems like .. ehh sucks to be them, given they should @#$ know better these days.
Phynaz - Saturday, November 16, 2019 - link
Intel has a long history of being much farther under top than AMD. AMD even tried to create their own term so they wouldn’t look so bad.Spend less time waving the AMD flag and you’ll at least look smarter b
Qasar - Sunday, November 17, 2019 - link
"Intel has a long history of being much farther under top than AMD " oh really ?? explain how their cpus use more power the spec on the box. oh wait, Anandtech stated, and explained how here : https://www.anandtech.com/show/13544/why-intel-pro...spend less time waving the intel flag, and maybe you would see at least some of the lies from intel that you STILL believe.
Phynaz - Saturday, November 16, 2019 - link
Get off the AMD fanboy train and onto the clue trainQasar - Sunday, November 17, 2019 - link
right after you get off the intel fanboy train.TrinityAlex - Thursday, November 14, 2019 - link
So let me get this straight, I need to open my case, remove the CPU and cooler, go to them with both, and in return for the cooler I will just get the (same) CPU back??? Do they give me some money to buy another cooler?Smartcom5 - Thursday, November 14, 2019 - link
Yup, literally the only thing which changes is the CPU, which is going to be replaced (for a reason, I might add).… and in light of the recent flaws we heard just a couple of hours ago (and how Intel knowingly shipped Cascade Lake with a flaws they were fully aware of, despite saying said SKUs being unaffected), you shall THINK about some bits here, how it may end up being just another piece of the puzzle and eventually paint yourself the bigger picture already.
Ian Cutress - Thursday, November 14, 2019 - link
No, this is just for unsold boxed versions at retailers.yannigr2 - Thursday, November 14, 2019 - link
Probably the cooler was for the TDP Intel is advertising, not for the REAL TDP of the processor.Phynaz - Saturday, November 16, 2019 - link
Comments like this are proof people can’t stop proving they are absolute fucking idiots.Qasar - Sunday, November 17, 2019 - link
and EVERYTHING you post, proves you are just a troll, and just a child, point is ?brantron - Thursday, November 14, 2019 - link
Before the conspiracy theories about fake TDPs get out of hand:1) E-2274G = 4 GHz quad-core w/ HT, like 6700K or 7700K.
2) i3 8350K rarely hits 40 watts, torture test might do 50 - 60.
3) 22xx Xeons are 2 years more recent, typically lower voltage.
4) No recall for the 6 core E-2276G with the same cooler...
5) 65w 6+ core CPUs run 4+ GHz fine with Intel's stock coolers.
The thin aluminum stock cooler does still suck, though. AMD's equivalent is better. If you want to REEEE about something, there you go.
Dragonstongue - Thursday, November 14, 2019 - link
pretty bad when the maker in this case Intel effectively KNOWS it cannot properly cool the cpu in question, i.e TDP 84w with cooler designed for such, likely means it (the cpu) does NOT stay within that TDP envelope and/or the cooler certainly cannot.For a "premium vendor" as everyone more often that not "waves the flag" saying Intel or nothing mantra....egg on their face huh (not once, not twice but many many times over)
good they are recalling I suppose, terribad for end user (consumer / big purchasers) to have to or not (lose warranty if not) send back?
Even far more odd considering it is Xeon class.
maybe they should reach out to OEM makers for something they know for sure will keep within operating spec instead of going about "all on their own"
I ran my trust old E8400 with stock cooler (even overclocked) for a LONG time before i bothered to go aftermarket wise..me thinks they probably should invest $$$$$ to assure this does NOT take place again...tis Intel, who knows
ROFL
Samus - Friday, November 15, 2019 - link
I have chucked every Intel cooler I've ever had, except maybe the Nehalem i7-920 cooler which was like 2lbs and had a huge copper slug in the center...only used it briefly though as I eventually overclocked that chip.I think by now everyone knows Intel coolers are a joke. Almost every chip will eventually throttle with its bundled cooler.
edzieba - Friday, November 15, 2019 - link
Whole we have the usual suspects bleating over the inane conspiracy du-jour: This sounds far less like an IC issue, and far more likely HSF issue. DHA-A should be able to handle 83W just fine, but it's hardly a spring chicken in terms of design. Thermal compound degradation, malformed components (the heatsink is made up of friction-fit components, if one of these was out of tolerance then that affects thermal performance) etc. The bundled heatsink is rarely pushed to the rated performance, so batch issues can easily have slipped under the radar.eastcoast_pete - Saturday, November 16, 2019 - link
Whatever happened to engineering know-how at Intel? Testing a cooler to make sure it works properly for the CPU it's bundled with shouldn't be a problem for a company that use to pride itself as the technology leader in this space. And, why this is a particularly embarrassing problem here: people buy Xeon CPUs because they are expected to keep performing just fine at or near 100% their specified TDP year-in, year-out. That's why many XEONs and similar chips from AMD (EPYC, or other professional lines) are usually clocked a bit more conservative, and also priced higher than their consumer counterparts. So, Intel shipping garbage bundled coolers is really egg on their face, and was utterly avoidable. That company has problems, not just with getting 10 nm fabs going.