http://hexus.net/tech/news/cpu/89636-intel-skylake-bug-seizes-pcs-running-complex-workloads/
Maybe I'll wait for the fix before upgrade to Skylake PC.
And Intel forums thread:
https://communities.intel.com/mobile/mobile-access.jspa#jive-content?content=%2Fapi%2Fcore%2Fv3%2Fcontents%2F524553

---

I'm hearing that a BIOS update should fix this issue, but I can't say for sure that it will work.

"Hello All,
Intel has identified an issue that potentially affects the 6th Gen Intel® Core™ family of products. This issue only occurs under certain complex workload conditions, like those that may be encountered when running applications like Prime95. In those cases, the processor may hang or cause unpredictable system behavior. Intel has identified and released a fix and is working with external business partners to get the fix deployed through BIOS."

Official response is BIOS fix. We'll see if it's enough...

---

Worth reposting this from archae86 at Einstein, then.

For some years now Intel microprocessors have carried a provision that allows microcode patches to be uploaded to the CPU on every boot of the machine. This is why it is absolutely crucial to make sure that the BIOS version you run on a given motherboard is specifically stated to support the specific CPU revision you plug into it.

I think people get careless on this point because so many of the bugs are rather uncommon in their manifestation, so plenty of people "get away" with using BIOS versions not claimed to support their specific CPU stepping.

This episode suggests why that is a really bad idea.

(that's in response to this particular issue today, which I reposted over there)

Microcode patch are dating back the P6 Intel microarchitecture i.e Pentium Pro that also received exhaustive fpu verification (FDIV bug docet).

Marco.

---

pcgameshardware.de

(...)
Intel selbst hat auch nicht genauer erklärt, wo denn nun das Problem bei den Skylake-Prozessoren liegt. Mangels Specification Update lässt sich nur raten. In einem Forum wird darüber diskutiert, dass der Fehler bei Prime 95 und Skylake genau dann auftritt, wenn eine Fast Fourier Transformation (FFT) mit der Länge 768K mit AVX-Befehlen ausgeführt werden soll. Bei AVX2 soll der Fehler nicht vorhanden sein.
(...)

Google Translator say:

Intel itself has also not explained precisely where for now the problem is with the Skylake processors. Lack Specification Update can only be guessing. On a forum discussed the fact that the errors in Prime95 and Skylake exactly occurs when a fast Fourier transform (FFT) is to be executed with the length 768K with AVX commands. In AVX2 the fault should not be present.

We use here FFTW (app) and AVX (CPU).
So SETI members could be affected?

I remember the AMD Phenom (1st generation) bug, in the BIOS was an option to enable a workaround (disabled something in the CPU).
The CPU was limited.

Intels solution will be the same way?
That the Skylake will not reach the whole performance?

---

pcgameshardware.de

(...)
Intel selbst hat auch nicht genauer erklärt, wo denn nun das Problem bei den Skylake-Prozessoren liegt. Mangels Specification Update lässt sich nur raten. In einem Forum wird darüber diskutiert, dass der Fehler bei Prime 95 und Skylake genau dann auftritt, wenn eine Fast Fourier Transformation (FFT) mit der Länge 768K mit AVX-Befehlen ausgeführt werden soll. Bei AVX2 soll der Fehler nicht vorhanden sein.
(...)

Google Translator say:

Intel itself has also not explained precisely where for now the problem is with the Skylake processors. Lack Specification Update can only be guessing. On a forum discussed the fact that the errors in Prime95 and Skylake exactly occurs when a fast Fourier transform (FFT) is to be executed with the length 768K with AVX commands. In AVX2 the fault should not be present.

We use here FFTW (app) and AVX (CPU).
So SETI members could be affected?

I remember the AMD Phenom (1st generation) bug, in the BIOS was an option to enable a workaround (disabled something in the CPU).
The CPU was limited.

Intels solution will be the same way?
That the Skylake will not reach the whole performance?

I don't think I have seen an option on an Intel based board to enable/disable microcode updates. If you want the fix you load the version of BIOS/UEFI with the fix. If you do not want the fix then you keep the old BIOS/UEFI version.
It is possible the next stepping of Skylake will not require a patched BIOS.

Perhaps the app devs can tell us if the 768K length issue effects the SETI@home AVX app. If it does. Then it might be good for those with Skylake hardware to use SSE3 or SSSE3 app until they get a patched BIOS/UEFI for their system.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Perhaps the app devs can tell us if the 768K length issue effects the SETI@home AVX app. If it does. Then it might be good for those with Skylake hardware to use SSE3 or SSSE3 app until they get a patched BIOS/UEFI for their system.

I don't think it's this general, as the bug very probably requires some very special combination of instructions to appear. This applies to pretty much all CPU bugs which make it out into the wild. If any 768k FFT would have triggered the issue, I'm convinced it would have been discovered much earlier.

MrS

---

Scanning for our furry friends since Jan 2002

How will Intel fix this CPU bug?
In the BIOS is then an option to disable AVX?

It would be better to use instead of...
i5-6600K and i7-6700K
the older...
i5-4690K and i7-4790K ?

How I could identify newer Skylake versions (later steppings) which haven't this bug?

---

How will Intel fix this CPU bug?
In the BIOS is then an option to disable AVX?

No, the BIOS loads new Microcode in the CPU at boot time (this new Microcode will change how the CPU do AVX instructions internally, probably no speed change but only "correctness" of AVX instructions)
https://en.wikipedia.org/wiki/Microcode

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

How will Intel fix this CPU bug?
In the BIOS is then an option to disable AVX?

No, the BIOS loads new Microcode in the CPU at boot time (this new Microcode will change how the CPU do AVX instructions internally, probably no speed change but only "correctness" of AVX instructions)
https://en.wikipedia.org/wiki/Microcode

probably no speed change ?

You mean a SAH or AP app which use AVX could be then slower than a SSE3 or SSSE3 app - because of the AVX bug fix?

Maybe I'll go with an i7-4790K instead of an i7-6700K (?) - short time and I'll order the parts of/for my new build.

The i7-4790K hava a TSX bug - maybe one day we'll have a SAH or AP TSX app?

Oh, maybe I'll go better with an unbugged AMD CPU? ;-)

---

How will Intel fix this CPU bug?
In the BIOS is then an option to disable AVX?

No, the BIOS loads new Microcode in the CPU at boot time (this new Microcode will change how the CPU do AVX instructions internally, probably no speed change but only "correctness" of AVX instructions)
https://en.wikipedia.org/wiki/Microcode

probably no speed change ?

You mean a SAH or AP app which use AVX could be then slower than a SSE3 or SSSE3 app - because of the AVX bug fix?

Maybe I'll go with an i7-4790K instead of an i7-6700K (?) - short time and I'll order the parts of/for my new build.

The i7-4790K hava a TSX bug - maybe one day we'll have a SAH or AP TSX app?

Oh, maybe I'll go better with an unbugged AMD CPU? ;-)

I think it is still unknown if the AVX issue even effects computers running SETI@home AVX apps. So far the only i7-6700K hosts I have found in the top 500 computers list have only ever used stock apps. So they would never see the AVX issue from SETI@home apps.
probably no speed change = No one is expecting there is be any speed change in how AVX instructions are processed after the fix. However until the fix is published we will now know 100%.

The TSX feature has been removed from Haswell, & Broadwell, CPUs made since 2014-08. Also microcode updates remove TSX from systems that have a CPU with the issue.

Also some AMD CPUa have an error with 256 bit AVX instructions, but AMD has never publicly admitted to there being a problem.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[