AMD finally solves Ryzen linux bug

[Shiv78]

2 minutes ago, HisEvilness said:

Yeh I did read that part, where can I see what week my chip was produced?

probably on the box?

[HisEvilness]

Just now, Shiv78 said:

probably on the box?

Yeh i got the box here but i do not know what part of any code would reveal what week it was produced, 

[Humbug]

15 minutes ago, HisEvilness said:

I have to ask how can i check if my CPU is effected?

46 minutes ago, raphidy said:

Welp, time to RMA my 1700.

This is very specific bug that comes up under a very specific use case.

https://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response

Personally I wouldn't bother with the RMA process and related wait time etc unless I run those specific workloads...

 

But your decision to make of course, not mine...

[HisEvilness]

Just now, Humbug said:

This is very specific bug that comes up under a very specific use case.

https://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response

Personally I wouldn't bother with the RMA process and related wait time etc unless I run those specific workloads...

 

But your decision to make of course, not mine...

I sometimes run a security related Linux custom OS that does lots of works loads to some extent similar to that the above article.
Just want to make sure i do not run into problems down the road.

[Windows7ge]

2 hours ago, FratStar said:

Theadripper and Epyc have not been effected by the issue. Only normal Ryzen silicon have experienced the issues.

Then I wonder if the issue has to do with some sort of communication between the CPU and the chipset.

[Sauron]

It sounds like the perfect way of handling this, offering a free replacement for anyone interested. After all, they know very well very few people are actually affected by this and will care enough to ask for a replacement.

1 hour ago, HisEvilness said:

I have to ask how can i check if my CPU is effected?

Quote

users will have to find out if their cpu is the old version or new one, by looking at the label on their cpu box

either way if you needed this functionality you'd know by now.

[CommandMan7]

I wonder how many people who lost the silicon lottery will use this as an excuse to try again for a better overclocking chip.

[dexT]

1 hour ago, CommandMan7 said:

I wonder how many people who lost the silicon lottery will use this as an excuse to try again for a better overclocking chip.

They'll be in for a surprise when the older chips go higher. Its not like at the R7 launch where the chips clock the same anymore.

[NumLock21]

2 hours ago, HisEvilness said:

Yeh I did read that part, where can I see what week my chip was produced?

Looks like serial number on box won't help. Have to look at cpu itself luckily, these new ryzen have a cut out window. 

It's the batch number. 

E.g. 

UA 1706PGT

17 is 2017

06 is week 6 

so 1706 means week 6 of the year 2017

[raphidy]

2 hours ago, Humbug said:

This is very specific bug that comes up under a very specific use case.

https://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response

Personally I wouldn't bother with the RMA process and related wait time etc unless I run those specific workloads...

 

But your decision to make of course, not mine...

Yeah, I just personally don't want to keep something with a defect. They say heavy compilation on linux, but not on Windows? Rather be safe than sorry.

3 hours ago, HisEvilness said:

Yeh i got the box here but i do not know what part of any code would reveal what week it was produced, 

It's not on the box, it's on the HIS aka, gotta pull that cooler off.

 

I'll install linux and bench it. So much headache with ryzen, gosh.

[HisEvilness]

47 minutes ago, NumLock21 said:

Looks like serial number on box won't help. Have to look at cpu itself luckily, these new ryzen have a cut out window. 

It's the batch number. 

E.g. 

UA 1706PGT

17 is 2017

06 is week 6 

so 1706 means week 6 of the year 2017

Well will have a look when i put in the new mobo this month i guess and find out.

[NumLock21]

22 minutes ago, HisEvilness said:

Well will have a look when i put in the new mobo this month i guess and find out.

The number is directly on the cpu ihs. You can see it from the box cut out window.

[CommandMan7]

4 hours ago, dexT said:

They'll be in for a surprise when the older chips go higher. Its not like at the R7 launch where the chips clock the same anymore.

I can't find any news about this, is this a problem someone's tested or is it just word of mouth?

 

I would expect newer CPU's to clock better due to process and design adjustments.

[mr moose]

9 hours ago, Tenelia said:

Let us take a moment to recall that Intel has a track record of refusing such things...

I was wondering how long it would be before someone had an unnecessary dig at Intel,  less than 20 posts GG LTT.

 

It seems it doesn't only have to be a negative news story about AMD for people to bring Intel into it.

 

[dexT]

1 hour ago, CommandMan7 said:

I can't find any news about this, is this a problem someone's tested or is it just word of mouth?

 

I would expect newer CPU's to clock better due to process and design adjustments.

All of the best dies go to Threadrippah and EPYC chips instead of going to the R7 line now. At launch the chips weren't binned as tight as they are now either so 1700/X rarely do 4/4.1.

 

Someone has binned a lot of R7, I dont believe there's any chance of the details being in an article or youtube video though.. I dont even get the full details and testing procedure lol.

 

 

[Misanthrope]

"Hey everybody AMD has a bug in Linux!" 

-"buuuu!"

 

"Hey everybody AMD finally fixes the issue"

-"Took them long enough the bastards"

 

You people don't deserve 6 core chips for cheap.