AMD Confirms It's Issuing a Fix To Stop New Ryzen Processors From Crashing Desktops (digitaltrends.com) 113
AMD says the company has been able to figure out why FMA3 code is causing system hangs on PCs using a new Ryzen desktop processor. From a report: Although AMD didn't provide a detailed report on the problem's root cause, the company said that BIOS changes will be distributed to motherboard manufacturers to resolve the issue. Customers are encouraged to keep an eye on their motherboard vendor's website for an update. "We are aware of select instances where FMA code can result in a system hang," the company said. "We have identified the root cause." AMD released three Ryzen-branded desktop processors at the beginning of March that plug into motherboards based on AMD's new AM4 socket. The trio of processors include the Ryzen 7 1800X, the Ryzen 7 1700X, and the Ryzen 7 1700. However, all three reportedly cause a hard system lock when running certain FMA3 workloads. The problem was replicated across all three processors and a variety of motherboards.
w00t - the K6 bug all over again! (Score:4, Interesting)
Kidding, but I'm old enough to remember when running Windows 95 on the old AMD K6 boxen was a no-go...
That said, does this fix affect performance any (no matter the OS)?
Re:w00t - the K6 bug all over again! (Score:5, Insightful)
K5 had a bug like that too. And let's not forget Athlon/Athlon XP era and AGP issues. The Piledriver Opterons required a patch to fix a bug in the hypervisor system which allowed for escape from a VM. AMD has had just as many horrible bugs as Intel, which can be summed up like this: Making flawless high-performance chips is difficult.
Re: w00t - the K6 bug all over again! (Score:1)
Re: (Score:2)
AFAIK Athlon/Athlon XP had no such bugs, the issues were bugs in the VIA chipsets. AMD's chipsets worked fine.
I don't remember any bugs on Piledriver. I do remember that Bulldozer had a TLB bug though. The workaround was to disable the TLBs and it significantly hampered performance.
Re: (Score:2)
The AMD 761 held fixes for the bug, but VIA and nVidia had problems due to AMD withholding information of the bug. Ugh, I can still remember all that annoying AGPGART messing about...
As for Piledriver: https://lists.debian.org/debia... [debian.org]
Re:w00t - the K6 bug all over again! (Score:5, Insightful)
Look at an intel errata list some time. There are huge numbers of bugs in all CPU's in recent times. Bios patches trap the errant instructions and use a work around. Nothing really to see here. I've had several intel instabilities get resolved with a bios flash. It is yet another reminder to always wait a few months after major revisions for the dust to settle unless your goal is to actually be an early adopter for the hell of it.
Re: (Score:1)
Just a microcode update
Re: (Score:2)
Re: (Score:2)
I've had multiple in my hands at my local Frys Electronics here in las vegas. I would have to say their available. The Asus x370 Prime, and the white MSI board with the dual m.2 are in stock as we speak.
Re: (Score:3)
AMD's issue here isn't necessarily that they have more of these problems or that the problems are more serious. It's that they have a reputation for having more of these problems and for them being more serious when they happen.
A lot of people, self included to a degree (though I do try to counter it) have picked up trust issues around AMD products over the years. In many cases, including mine, that may well be because we tried running AMD CPUs which just run a bit hotter than the Intel equivalents with coo
Re: w00t - the K6 bug all over again! (Score:1)
It doesn't matter how hoy the CPUs get. AMD has damn critical bugs in their chip microcode. They have a long history of failing to catch microcode issues in emulation before chip tape out. It must be horribly embarrassing for the engineers that work there.
Re: (Score:2)
So you forgot the Intel TSX bug, the Intel F00F bug, the Intel FDIV bug, etc.
Re: (Score:2)
Re: (Score:1)
Remember intel inside can't divide ....
Re: (Score:2)
Really? How about using the damned stock cooler? That's what I've done and had no issues.
Re: (Score:2)
Re: (Score:2)
That's been the case on the AMD side too, like the hypervisor bug in Piledriver Opterons for example.
Re: (Score:2)
Sage advice for nearly every industry.
Never buy the first model year of a new car design. You're bound to get engine or transmission issues at a higher rate than after they've worked things out a bit.
Never buy the first release of a new game console. Just look at the XB360 RROD, or the PS1's CD-ROM issues.
Never pre-order or buy day-one titles. Look at Ubisoft's Assassin's Creed (Brotherhood?) issues. Or Batman Arkham Knight on PC - with 2 stop sales and recalls/refunds. Or really any
Re: (Score:1)
I remember it working but needing some weird patch of config change on Win95.
On Win98 it worked without any special hacking needed.
Re: (Score:2)
Kidding, but I'm old enough to remember when running Windows 95 on the old AMD K6 boxen was a no-go...
That said, does this fix affect performance any (no matter the OS)?
I never had any issue or quirks with Win95 on my K6. Though that might have been second stepping.
Re: (Score:1)
Re: (Score:2)
A+BxC in one instruction is the fix. Apparently this instruction fails when the Ryzen is at some power saving stage.
I believe it is a bug in the Ryzen Microcode, that can be changed by modifying the motherboard. In otherwords, fix the MB for now and fix the instruction in the Ryzen ASAP. I believe there will be a microcode patch for Windows and Linux and whatever to fix the CPU at some near distant future.
By the way, my CPU has an INTEL patch to correct faulty instructions. It should not happen
What the fuck is FMA? (Score:1)
C'mon Slashdot, give us at least some clue instead of simply copying and pasting someone else's article.
Re: (Score:2)
Re: (Score:1)
Re:What the fuck is FMA? (Score:4, Informative)
Apparently, I had to look into the forum posts that the FA referenced, and FMA instructions are Fused Multiply Add, whatever the fuck that is.
After looking at Wikipedia for 5 seconds, FMA instructions perform round(a+b*c) in a single operation, so you can a) speed up and b) get more accurate results whenever you need such a mathematical operation (which is actually reasonably frequently, in numerical computing).
Re: (Score:2)
> FMA instructions are Fused Multiply Add, whatever the fuck that is.
Really? You're posting on /. and you can't google it?
* Multiply-Accumulate [wikipedia.org]
This commonly shows up when you are lerping (linear interpolating) between two values, a and b, you have a interpolation parameter usually called t:
Compilers will see this pattern and generate a FMA instruction for it, or you can write your own.
Copying the code snippet from this StackOverflow Question: [stackoverflow.com] How to use Fused Multiply-Add (FMA) instruct
Re: (Score:1)
Re: What the fuck is FMA? (Score:1)
Re:What the fuck is FMA? (Score:5, Informative)
FMA instructions are Fused Multiply Add instruction. Usually their are on SIMD registers and allow you to do "a += b *c". Modern cores can do that on a vector in a single cycle. Actually, they may be able to do more than one FMA on a vector register in a single cycle.
FMA are most commonly used to compute dot product, and are therefore very helpful in linear algebra. (And so they are useful in a ton of data mining algorithms.)
Re:What the fuck is FMA? (Score:4, Informative)
Also known as the Multiply-Accumulate (MAC) instruction in DSPs. MAC is an extremely common instruction in signal processing kernels (the inner loop that does the calculations). It is vital to be able to do a lot of them per clock cycle. In fact, it's often why DSPs have special looping registers so you can do zero-overhead loops and thus doing a sequence of MACs without incurring branch (and branch prediction) times thus being able to do nothing but this instruction for very little overhead
Re:No Big Deal (Score:5, Informative)
Re: (Score:2)
no he was just being a fanboy asshole, and would prefer AMD die so he could pay $20,000 for intels next generation CPU.
Really? (Score:2)
Re: (Score:2)
Yes that was exactly what I was wondering too. It seems to be clear evidence of AMD conducting only superficial testing prior to release.
Re: (Score:2)
...haha.
if by a complex problem that occurs in combination of different states you mean BENCHMARKING said feature, then sure. yeah. very complex. just hammering that instruction seems to be the cause. doesnt matter which core inside either. ..basically, benchmarking one of the new features of their new cpu makes the bug appear.
Why I wait before buying.. (Score:2)
Re: (Score:2)
Why bother? Just do what time and again I've found works best for a trouble-free life: totally avoid any/all AMD CPUs or GPUs.
Re: (Score:3)
That's only trouble-free for people for whom the price of Intel compared to AMD is already not a concern.
So in other words, what you really mean is that what works best for a trouble-free life is to just be rich enough that you can buy your way past any problems.
And who can argue with that?
Re: (Score:2)
I've personally found time and again that buying cheap is nearly always a false economy, since you need to service it far more often and/or replace it far sooner (Sometimes even immediately) compared to a quality/reliable product. Espcially one with some headroom to meet future needs.
It reminds me of a (Discworld) story of a poor Policeman (Sam Grimes) who could only ever afford $1 boots on his salary, and he got through 3 pairs a year. He married into a rich family so could finally afford a pair of $5 boot
Re: (Score:2)
Actually, I have. A 308 GTS to be exact. And it was one of the most reliable cars I've ever owned.
Re: (Score:2)
"Reliable" is relative. The cost of ownership of a Ferrari, and even a 308 requires dedication and planning. http://www.birdman308.com/tuto... [birdman308.com]
The 308 is one of my all-time favorite cars but if I don't have the time to perform the maintenance myself I would not own one. Especially if it's a steel body prior to 1984. Another favorite is the F355 but that has maintenance headaches, too. https://www.youtube.com/watch?... [youtube.com]
Re: (Score:2)
I can only talk about mine. It was the quad carb (ie. not even injected) version and always started almost immediately even after weeks of not starting it, and it literally never had any mechanical problems in maybe 5 years of owning it. That said you don't normally use a car like that as an everyday driver either (although i did for a while), so it only had like 70k miles on it when I sold it.
Re: (Score:2)
> "Reliable" is relative.
True. Compared to several German brand cars I've also had, it was much more reliable.
Re: (Score:2)
That theory works great until Really Good Quality Boot company decides to juice up its quarterly profits by outsourcing production to dumpy boot company, effectively selling their reputation for short term profit. And then after you spend $100 on two pairs of Not actually Really Good Quality Boot Company boots before figuring it out you move to Really Good Boot Comapny's closest competitor. Only they instituted the same changes to compete and suck just as bad.
End Result is you've spent $150 for 3 pairs of $
Re: (Score:2)
Re: Why I wait before buying.. (Score:1)
Re: (Score:2)
i upgrade often, i choose AMD because im not rich and its my hobby. Intel until now has has a better product. Ryzen has caught AMD up, Hopefully VEGA does the same. I honestly dont have the money right now for a full Ryzen upgrade, so im not worried. But its a new architecture so we already knew there would be problems on the motherboard and cpu side. so im kind of glad that i have to wait. Its stupid to be biased. You make stupid decisions.
Re: (Score:2)
The Sam Vimes theory of economy has a lot of basis in reality.
Look at real quality furniture for example. I inherited a kitchen work table that's been in the family for over 120 years. It's been sanded down and re-polished a number of times, but it's still a large thick slab of wood on REALLY sturdy legs. Great for baking or cooking in large batches etc(like say 10 plates of various rolls, bisquits and cookies. At the time it was made, it would have cost a working class family more than a year and a half's
Re: (Score:3)
>> Sure, I could buy something that would last us maybe 20-25 years at best
It seems to me that even expensive furniture (at least in the US) is made with well-disguised cheapest possible materials such as composites and laminates, so actually only good for about 10 years at most.
Re:Why I wait before buying.. (Score:4, Informative)
Why bother? Just do what time and again I've found works best for a trouble-free life: totally avoid any/all AMD CPUs or GPUs.
Heh, if you want to avoid bugs, better stay away from Intel also. They have a 1000 page errata list with every single processor.
Re:Why I wait before buying.. (Score:5, Informative)
But I can use FMA3 instructions on multiple Intel microarchitectures without it ever causing my system to freeze up. That's the kind of "quality" only AMD can provide.
Try TSX instruction in a Haswell or Broadwell.. Well if your bios haven't been updated by Intel to disabled the instructions completely.
Errata is unfortunately pretty common in CPUs these days. I hope this one is fixed without having to disable the entire extensions like Intel does.
Re: (Score:1)
Here's an interesting quote from someone who apparently worked at intel:
Re: Why I wait before buying.. (Score:1)
Re: (Score:1)
90% of your systems are Pentium 2s and 3s? Bullshit...
Re: (Score:3)
I guess having a patch within a week is pretty good though.
Assuming that the patch actually fixed the problem and just doesn't disable the instruction (or feature set) on the CPU. Disabling CPU functionality would "solve" the problem, but then you would be effectively getting a reduced capacity CPU and not what you were paying for.
Re: (Score:2)
platform security processor (Score:1)
Hey AMD, it's because of your Platform Security Processor (PSP) backdoor, aka Intel's equivalent of Intel Management Engine. You know, you should really get rid of it. You really should.
Re: (Score:2)
I knew it would be a bad idea to install that laptop logged into Slashdot at collapsed-drunkard level in the pub restroom.
Re: (Score:2)
s'okay; at least it's still out of the reach of the old GNAA trolls...