Thursday, June 11th 2020

AMD Responds to Allegations of Ryzen Power Reporting Deviation Reducing CPU Life

AMD on Thursday posted its first official response to reports that the deviation in power reporting by the CPU VRM controller to a Ryzen processor against its actual power delivery may be shortening processor lifespan, sparked by HWInfo introducing the "Power Reporting Deviation" sensor. "We are aware of the reports claiming that select motherboards may be under-reporting certain power telemetry data that could alter the performance and/or behavior of AMD Ryzen processors under certain conditions. We are looking into the accuracy of these reports," the response begins.

AMD also clarified that Ryzen processors aren't dictated entirely by the power telemetry from VRM controllers. "We want to be clear with our customers: AMD Ryzen processors contain a diverse array of internal safeguards that operate independently of external data sources. These safeguards enforce the safety and reliability of the processor during stock operation. Based on our initial assessment, we do not believe that altering external telemetry in the manner described by those public reports would have a material impact on the longevity or safety of a user's processor." AMD's response also suggests that the company isn't aware of power reporting deviation or at least the extent to which motherboard manufacturers rely on tricking the external power telemetry setup to send more power to processors, in a bid to notch ahead in stock performance or overclocking headroom.
Source: Tom's Hardware
Add your own comment

44 Comments on AMD Responds to Allegations of Ryzen Power Reporting Deviation Reducing CPU Life

#26
TheoneandonlyMrK
Vayra86Still waiting on my first CPU demise ever...but I did see degradation a few times.

Then again I don't hoard old parts. It will be interesting to see when the 3570k says thxbai, running a mild OC all its life.
Ten years probably minimum, you have to push quite hard to degrade them both in clock's and just as important load's.
Posted on Reply
#27
Turmania
genuine question, what is the longetivty expectance if you use them at stovk with no overclocking in terms of voltagesd applied for both Intel and AMD?
Posted on Reply
#28
AsRock
TPU addict
Pinktulips7AMD are always Lying on their Products!!! Now AMD got caught again with their Pants down!!! intel mean Big Dog where AMD and their Fanboys are beating around the Bush to make excuse, Fuc* them
They all do it, as long as it gets passed the warranty it's a none issue.
Posted on Reply
#29
TheoneandonlyMrK
Turmaniagenuine question, what is the longetivty expectance if you use them at stovk with no overclocking in terms of voltagesd applied for both Intel and AMD?
Over ten years I would say, long before the CPU dies the motherboard will, and would be hard to replace.
Posted on Reply
#30
Vya Domus
silentbogoHad hard time diagnosing one of my customer's PCs, where no OC was in play and dude actually knows what he's doing
"No OC"

Riiiiiight. Who would ever show up with a dead CPU and admit it was because of overclocking ? Come on.
Posted on Reply
#31
DemonicRyzen666
GoldenXSay that to the over 10 Sempron LE-1150 I have.
Only AMD CPU that died on me so far is a K6/2.
I 'm using a sempron le 1150 too on windows 10 no less!. I would be using a phenom 9850 on this board but I can't get it boot with out clearing CMOS evertime. It even refuses to restart with out CMOs clear. ( yes I changed the battery too) I suspect the power supply is the problem. As it's getting warm with just this single core and two 3470 in crossfire lol.
Posted on Reply
#32
1d10t
silentbogoI can't tell you how many dead APUs I had to change on HP Pavilion/Envy laptops(especially A10-4800), or how many desktop 1155 Celerons/Pentiums died in pain and suffering due to internal shortage, or most importantly, how many sFM2+ Athlons got incinerated by mismanaging auto-voltage on cheap "updated" FM2 motherboards.
I am going to the office right now, I can snap my collection of deaddies that weren't made into keychains or nerdy pendants yet. I even have weird ones, like a 2500K that died on H61 board, and 2500S that "burned-out" over time due to sudden TIM separation from heatspreader (owner wasn't watching temps and only noticed when his PC started to BSOD and died).
What do you expect from OEM and ODM? Basic power delivery, crappy capacitor, minimum cooling and unjustified PSU ? :D
Many low end up to mainstream notebook are using Compal motherboard, they are the cheapest and also the worst.I see many notebook vendor now reserving Quanta for their high end, heck I never saw Pegatron plugged into $2500+ notebook.
Posted on Reply
#33
silentbogo
1d10tMany low end up to mainstream notebook are using Compal motherboard, they are the cheapest and also the worst.I see many notebook vendor now reserving Quanta for their high end, heck I never saw Pegatron plugged into $2500+ notebook.
I think you are stuck in early 2000s or sumptin'. Nowadays every manufacturer is capable of making both cheap crap and semi-decent designs(but mostly crap), and you can easily find Compal in high-end $1000+ laptops: off the top of my head I can name Dell Latitude E6540, or Precision M4800/M6800, or Lenovo Y50-70 (first and last are currently on repair in my office). Or an abominable Quanta designs in bottom-of-the-barrel Acer laptops/netbooks.
Posted on Reply
#34
AxilomX
I personally still have a working Athlon X2 4200+, an X4 640 and a A8-7650 that still runs. So I can attest to how long they can last. I guess we all have to bet on the silicon lottery.
Posted on Reply
#35
r9
Ryzen I wasn't that fast definitely lagging behind Inter . So AMD had an option to sell it as a slow chip with average overclocking or they could overclock the f-k out of it and sell it as "stock" with poor overclocking.
Few Ryzen generations later seems they made the right call.
They days where you can get a E5200 2.5GHz and OC it to 4.7GHz or get Athlon II X3 2.8GHz and unlock it it to Phenom II x4 @3.8GHz are gone RIP.
Posted on Reply
#36
Jism
People need to let go of the idea that CPU's need a manual input these days while it has build in hardware to check constantly the silicon's performance. That is what makes this boost functionality of AMD so great. With light-weight workloads you get the highest boost clock (duh, and higher clocks need a bit more voltage) and with heavy workloads you'll get something on avg on what your cooling allows (less voltage, with no more then 1.39V or so).

If people are so affraid of it turn off PBO/XFR then. You'll get the base clocks 24/7 with no automatic overclocks at all. I have my 2700x for a year, no problems and its constantly hanging all core 4.2Ghz. This is due to the big cooling thats attached and the slight undervolt. Over time when the water warms up it drops back to 4150Mhz, 4100Mhz and so on. It's normal.
Posted on Reply
#37
SKD007
SteevoMy 1100T begs to differ. Also all these products are aged before they would die. So, poor troll.....poor poor troll.
I second that. My old gaming 6 core 1100T CPU is still used by my mom and it’s on a H100i LC running at 3.7ghz all the time...
Posted on Reply
#38
ValenOne
Tom's Hardware FUD is BS when Ryzen XTs made minor OC as official configs. My Ryzen 9 3900X at 4.35 GHz for "all 12 cores" are attached to AIO water cooling with minimal voltage increase. "All core" clock speeds are under 4.6 Ghz boost mode.
Posted on Reply
#39
TheGuruStud
Just more quality journalism and totally not a hit piece from the reputable site of Tom's
Posted on Reply
#40
Dicfylac
Let's see what's the next intel move when amd will launch the 3000 xt ryzen series.
Afterall the gap seems to be greater than it looks like.

The master cpu killer is the software who turns it obsolete, "my two cents".
Posted on Reply
#41
Melvis
ZoneDymothough I do love the "do your own research" remark, like they cant even link a single article to back up their claim lol.
This is such a troll move, just shows how much knowledge the person actually knows, probably linked some none English websites and google.com and then couldn't answer multiple questions when given proof that proved them wrong even when given plenty of chances and time to do so and then repeats them selves again and again making themselves look even more foolish. Trolls will be trolls.
Posted on Reply
#42
R0H1T
Just an update on this saga if you will ~
Update 6/20/20 7:25am PT: We tested several motherboards and firmware revisions, finding that one motherboard vastly misreported its power telemetry data, resulting in higher performance. However, newer BIOS revisions corrected the issue. We also found that other motherboards send power data that can result in slightly reduced performance. According to our testing, while the HWinfo tool does shine a light on some issues, it's readings aren't entirely accurate. You can see the full testing here.
Also a major issue I've had with the hwinfo's latest parameter is this ~
According to the HWinfo utility, the Taichi motherboard is still feeding incorrect power telemetry data to the SMU—the utility lists the deviation at ~7%. However, our measurements align more with our expectations of VRM losses, so the HWinfo data could be a misreport. (It's still unclear exactly how HWinfo determines deviation.)

HWinfo hasn't shared information with us to clarify how it measures deviation, so the tool is a bit of a black box. The HWinfo tool reports a variance of 12% with the auto VDD settings above, implying that the tool makes its decisions based on reference full scale current values, and not those optimized by vendors.

Modern chips rely upon accurate telemetry data, and HWinfo's new deviation feature helps shine a light on how some motherboard vendors have found a way to misreport power telemetry. Unfortunately, the inner workings of the tool aren't entirely clear, and HWinfo doesn't specify how it assigns the deviation value. From our testing, it appears the tool doesn't take what we would consider legitimate adjustments to the full scale current into account, which causes inflated deviation readings.
www.tomshardware.com/features/amd-ryzen-motherboard-power-cheat-testing
Basically the stats reported themselves can be inaccurate to the tune of 12% ~ which for a fully loaded 3900x or 3950x is quite a lot of power deviation!
Posted on Reply
Add your own comment
Apr 24th, 2024 12:55 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts