LinuxQuestions.org
Review your favorite Linux distribution.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Distributions > Slackware
User Name
Password
Slackware This Forum is for the discussion of Slackware Linux.

Notices


Reply
  Search this Thread
Old 09-04-2017, 11:34 AM   #76
55020
Senior Member
 
Registered: Sep 2009
Location: Yorks. W.R. 167397
Distribution: Slackware
Posts: 1,129
Blog Entries: 4

Rep: Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549

Rob Enderle?
Rob The World's Worst Shill Rentaquote Enderle?
Rob SCO Group Apologist Enderle?
Rob "Free Software And The Idiots Who Buy It" Enderle?
Rob Versus Groklaw Enderle?
Rob "The Enderle Group" == Just Him And His Gerbil Enderle?

NO

NO NO NO NO NO NO NO

Just

NO
 
Old 09-04-2017, 11:44 AM   #77
onebuck
Moderator
 
Registered: Jan 2005
Location: Midwest USA, Central Illinois
Distribution: Slackware®
Posts: 12,892
Blog Entries: 27

Rep: Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183
Member response

Hi,

You can read the information! His rep has nothing to do with past or future issues for hardware. Or reading it will be to hard to make a good informative opinion from presented information?
Quote:
Good manners is the art of making those people easy with whom we converse. Whoever makes the fewest persons uneasy, is the best bred in the company.”- Swift
Have fun & enjoy!
 
Old 09-04-2017, 11:53 AM   #78
55020
Senior Member
 
Registered: Sep 2009
Location: Yorks. W.R. 167397
Distribution: Slackware
Posts: 1,129
Blog Entries: 4

Rep: Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549Reputation: 1549
He has a fifteen year track record of saying whatever the client wants him to say, so long as it's not too technical for Rob. Today's client, apparently, is AMD. Meh. There are plenty of non-tainted non-superficial sources of info about Ryzen out there, and that's where I'll invest my time and attention, cheers.

Edit: Nevertheless, it's worth saying this properly: thanks Gary for your ongoing work finding interesting stuff to read, even if some of your readers are occasionally horrible and bad-tempered

Last edited by 55020; 09-04-2017 at 12:20 PM.
 
Old 09-04-2017, 02:56 PM   #79
onebuck
Moderator
 
Registered: Jan 2005
Location: Midwest USA, Central Illinois
Distribution: Slackware®
Posts: 12,892
Blog Entries: 27

Rep: Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183Reputation: 2183
Member response

Hi,

I do not have a problem with that. I leave things mostly up to the reader to discern for judgement and understanding.

Hope this helps.
Have fun & enjoy!
 
Old 11-01-2017, 12:50 PM   #80
BrZ
Member
 
Registered: Apr 2009
Distribution: Slackware
Posts: 541

Rep: Reputation: 120Reputation: 120
Quote:
Originally Posted by MarcT View Post
As I understand it, it seems Ryzen CPUs manufactured from week 25 of 2017 are probably OK. I say probably because I've been away and haven't been following the latest discussions on the AMD forum.

The date code is engraved on the CPU itself, under the RYZEN logo. Mine says "UA 1711PGS" => Week 11 of 2017, so is affected (and does segfault if stressed).
It seems stable if I disable the uOP cache in the UEFI, so I'm pondering the RMA option.

Let us know how you get on with yours!
Sorry man. I know I'm late.
Initially the script failed after a couple of hours but after AGESA 1.0.0.6b it is failing consistently with ~150 seconds of execution. I opened a ticket with AMD and the RMA was approved and scheduled for the next days. They asked for the whole package and a copy of the bill of sale. No debugging procedure or pictures. Got the 1700 on March so I'm curious to see the batch on my actual sample. Again, sorry for not being able to reply early. Damn, it looks like a necropost =]
 
1 members found this post helpful.
Old 11-01-2017, 01:56 PM   #81
bassmadrigal
Senior Member
 
Registered: Nov 2003
Location: Newport News, VA
Distribution: Slackware
Posts: 4,423

Rep: Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365
So, after my UPS decided to kill my old motherboard, I was forced to replace it, and I figured this was the perfect time to upgrade to Ryzen (I intended on waiting until our move occurred next year, but I had the money and figured it was better than putting some into a system I didn't want to keep). I ended up with the 1800x and the ASRock Taichi motherboard (same as MarcT) with 32GB (16GBx2) of Corsair Vengeance LPX 2666MHz RAM, a Corsair HX1000i PSU, Samsung 960 EVO NVMe drive, Thermaltake Contac Silent 12 HSF (I hate that AMD doesn't include a stock cooler on these), and a Thermaltake Chaser MK-1 case. I'm carrying over many harddrives and a video card (and the PSU and RAM were carry overs too). I'll mention of a few of the issues I ran into in case someone else has them.

Rereading through this thread did provide some help, especially in getting sensors working by loading the nct6775 module. I already had 14.2 installed onto a Samsung 960 EVO NVMe drive using UEFI and I had kernel 4.9.38 on there (grabbed from a previous -current update) and 55020's 4.13.7 kernel and installed that.

I did run into an issue that I didn't realize was even going to be an issue. Having a drive set up as UEFI booting doesn't transfer between machines. I initially thought something was screwed up since I was able to boot that NVMe drive on my old system, but it kept refusing to do so on the new system. After researching, I found you need to make the UEFI on the motherboard aware of your EFI partition, which can be done using a convoluted command with efibootmgr (see below -- I extracted it from the eliloconfig script) after booting the Slackware installer. If you only have one kernel and initrd, you can simply use eliloconfig, which I did the first time, but that erased my elilo.config, and I didn't think to back it up, so I had to repopulate it with my extra kernels. While I definitely see some benefits with UEFI, the lack of being able to boot off any bootable harddrive by just selecting it seems like a major step backwards... but I will say that being able to update the firmware using a network connection right in the UEFI is pretty sweet. It's quite a bit easier than doing it through DOS as many BIOSes required. As a side note, I may rework the eliloconfig script to allow it to retain an existing elilo.conf. It would've saved me a bit of time.

Code:
efibootmgr -q -c -d /dev/nvme0n1 -p 1 -l "\\EFI\\Slackware\\elilo.efi" -L "Slackware"
Once I got the machine booted, I was obviously impressed with the speed. I finished getting everything configured and the case set up. I then moved the computer back to my office and hooked everything back up. Turned on the computer and I got that horrible burning electronics smell. This is when I figured out that my UPS was bad and it wasn't just my last motherboard crapping out on me. While the "Master" plug reads 119V and 60Hz with no load using my multimeter, something is horribly off and it likes to fry motherboards (luckily, it's been just motherboards and not the PSU, harddrives, or CPU/RAM -- counting my lucky stars there). So, I got a replacement motherboard and now have my PSU plugged directly into the wall as I research what UPS I want to get to replace it (trying to decide if I want to stick with APC since they are considered one of the best, but this left a sour taste in my mouth and the UPS was only 3.5 years old).

Once I got the new motherboard installed, I continued with getting the relatively new install set up. I was noticing a lot of unknown devices in lspci, which I thought was weird considering I had a 4.13 kernel. What I found out is that the pciutils package is what provides that information, and you can update the database using update-pciids as root. Now most of the items in lspci display proper names.

Things were great until about 6 hours later... I went to wake up the computer and it had froze. The really weird thing is I could not ssh in and the screens wouldn't wake up and I couldn't turn on/off numlock, but my NFS shares were still working fine. Still unsure how that happened. When I rebooted, I checked the logs, but nothing showed any problems at the time of the freezing. I was thinking it might be due to an issue in the 4.13 kernel, so I grabbed 4.14-rc6 and compiled that after running through make menuconfig. Holy crap! This thing is fast. With -j17, I compiled the kernel in under 5 minutes. Unfortunately, after installing that kernel and rebooting, I ran into another freeze after like 12 hours. Again, everything by NFS stopped working. Consider me perplexed. I then build the newly released 4.14-rc7 (and I forgot to add the -j17 and it ended up taking about an hour to compile it). I added the new kernel to my elilo.conf and figured when it froze next, the new kernel would be used... but it hasn't frozen yet. It's now been up for 44 hours without an issue, so I'm thinking it might've just been the parts "settling"? Hopefully the freezing is behind me and I can finally move it back to its normal IP to take over as my server (which another machine has been temporarily fulfilling that position).

Quote:
Originally Posted by MarcT View Post
The date code is engraved on the CPU itself, under the RYZEN logo. Mine says "UA 1711PGS" => Week 11 of 2017, so is affected (and does segfault if stressed).
I wish I remembered this while I had my HSF off. I'd love to know when mine was built. I assume it is not visible on the packaging anywhere or as some parameter in /sys/?

But I did compile qt5 in under an hour (if I remember right) with -j17 without issue, so I'm hoping mine is a late enough batch that doesn't have that issue. Guess I'll see if any segfaults start happening while compiling.

Sorry for the novel.
 
2 members found this post helpful.
Old 11-02-2017, 12:36 AM   #82
Richard Cranium
Senior Member
 
Registered: Apr 2009
Location: Carrollton, Texas
Distribution: Slackware64 14.2
Posts: 2,740

Rep: Reputation: 1161Reputation: 1161Reputation: 1161Reputation: 1161Reputation: 1161Reputation: 1161Reputation: 1161Reputation: 1161Reputation: 1161
Quote:
But I did compile qt5 in under an hour (if I remember right) with -j17 without issue [...]
Dayum!
 
Old 11-02-2017, 07:15 AM   #83
MarcT
Member
 
Registered: Jan 2009
Location: UK
Distribution: Slackware 14.2
Posts: 107

Original Poster
Rep: Reputation: 45
Quote:
Originally Posted by bassmadrigal View Post
I did run into an issue that I didn't realize was even going to be an issue. Having a drive set up as UEFI booting doesn't transfer between machines. I initially thought something was screwed up since I was able to boot that NVMe drive on my old system, but it kept refusing to do so on the new system. After researching, I found you need to make the UEFI on the motherboard aware of your EFI partition, which can be done using a convoluted command with efibootmgr (see below -- I extracted it from the eliloconfig script) after booting the Slackware installer. If you only have one kernel and initrd, you can simply use eliloconfig, which I did the first time, but that erased my elilo.config, and I didn't think to back it up, so I had to repopulate it with my extra kernels. While I definitely see some benefits with UEFI, the lack of being able to boot off any bootable harddrive by just selecting it seems like a major step backwards...
There is a UEFI shell which allows for booting from arbitrary devices, but the syntax is really awkward.
Quote:
Originally Posted by bassmadrigal View Post
Once I got the machine booted, I was obviously impressed with the speed.
Oh yes, they are fast!
Quote:
Originally Posted by bassmadrigal View Post
Once I got the new motherboard installed, I continued with getting the relatively new install set up. I was noticing a lot of unknown devices in lspci, which I thought was weird considering I had a 4.13 kernel. What I found out is that the pciutils package is what provides that information, and you can update the database using update-pciids as root. Now most of the items in lspci display proper names.
Yep - less of an issue in -current which has a more up to date pciutils package.
Quote:
Originally Posted by bassmadrigal View Post
Things were great until about 6 hours later... I went to wake up the computer and it had froze. The really weird thing is I could not ssh in and the screens wouldn't wake up and I couldn't turn on/off numlock, but my NFS shares were still working fine. Still unsure how that happened. When I rebooted, I checked the logs, but nothing showed any problems at the time of the freezing. I was thinking it might be due to an issue in the 4.13 kernel, so I grabbed 4.14-rc6 and compiled that after running through make menuconfig. Holy crap! This thing is fast. With -j17, I compiled the kernel in under 5 minutes. Unfortunately, after installing that kernel and rebooting, I ran into another freeze after like 12 hours. Again, everything by NFS stopped working. Consider me perplexed. I then build the newly released 4.14-rc7 (and I forgot to add the -j17 and it ended up taking about an hour to compile it). I added the new kernel to my elilo.conf and figured when it froze next, the new kernel would be used... but it hasn't frozen yet. It's now been up for 44 hours without an issue, so I'm thinking it might've just been the parts "settling"? Hopefully the freezing is behind me and I can finally move it back to its normal IP to take over as my server (which another machine has been temporarily fulfilling that position).
I have had random freezes at idle, although not in a long while. I'm running 4.13.10 here. I think I've also disabled C6 powersave state in the UEFI.
However, I haven't yet sent my CPU off for RMA (although it was approved), so I know it still has "issues".
Anyhow, I think the problem is one or more of the cores crashes/stalls, so NFS, the NIC driver & HDD/NVMe were presumably running on a different core.

Quote:
Originally Posted by bassmadrigal View Post
I wish I remembered this while I had my HSF off. I'd love to know when mine was built. I assume it is not visible on the packaging anywhere or as some parameter in /sys/?
Not that I know - AFAICT the only place you can see the CPU date code is engraved on the CPU lid itself. I was lucky in that I'd photographed the CPU with my phone before building.


Finally - JFYI, there was a preliminary patch for the on-die CPU temp sensor on LKML. I don't think it was accepted in it's current form, but at least there's been some progress.

In other news, I'll hopefully be progressing my CPU RMA next week.

Last edited by MarcT; 11-02-2017 at 07:18 AM.
 
1 members found this post helpful.
Old 11-02-2017, 07:56 AM   #84
bassmadrigal
Senior Member
 
Registered: Nov 2003
Location: Newport News, VA
Distribution: Slackware
Posts: 4,423

Rep: Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365Reputation: 2365
Quote:
Originally Posted by MarcT View Post
There is a UEFI shell which allows for booting from arbitrary devices, but the syntax is really awkward.
Oh, I didn't notice that in the UEFI screen. But yeah, that doesn't sound much better. Looking at the dump from efibootmgr, I can certainly believe the syntax is awkward.

Quote:
Originally Posted by MarcT View Post
Yep - less of an issue in -current which has a more up to date pciutils package.
I actually went and grabbed the source from -current and built an updated package, but it was still using an out-of-date pciid and there were still unknown devices. So once I ran the command, it populated a lot more entries.

Quote:
Originally Posted by MarcT View Post
I have had random freezes at idle, although not in a long while. I'm running 4.13.10 here. I think I've also disabled C6 powersave state in the UEFI.
However, I haven't yet sent my CPU off for RMA (although it was approved), so I know it still has "issues".
Anyhow, I think the problem is one or more of the cores crashes/stalls, so NFS, the NIC driver & HDD/NVMe were presumably running on a different core.
Interesting. I wouldn't have thought it was tied to a specific core.

Quote:
Originally Posted by MarcT View Post
Finally - JFYI, there was a preliminary patch for the on-die CPU temp sensor on LKML. I don't think it was accepted in it's current form, but at least there's been some progress.
That's good to know. It looks like it is this 3-part patch. Although, I don't know if I want to be patching any kernel I may build until this gets added into mainline.

I was actually quite surprised that ASRock still doesn't have a firmware update to deal with the temperature offset of the CPU in the Health section of the UEFI. It threw me for a loop when I put the computer together that it stated my CPU was running at 60C. I thought I had somehow missed some peel-off coating on the CPU or HSF, so I took it off and double checked everything. Then I read up and found out that AMD, for whatever reason, decided to make a few of their processors seem 20C hotter than they actually were (I'm still not believing it's running at 40C, but that's at least within relatively normal operating temps).
 
  


Reply

Tags
ryzen, slackware


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
Ryzen processors quantumbb Linux - Hardware 4 03-06-2017 05:39 PM
LXer: AMD Ryzen 7 1700 Specs Confirmed, A $320 Rival To Intel’s $1099 i7 6900K – 8 Cores, 65W, 3.7Ghz With Unlimited Boost & Unlocked Multip LXer Syndicated Linux News 0 02-12-2017 11:43 AM
upgrading slackware64 13.1 multilib to slackware64 -current multilib Cultist Slackware 4 03-12-2011 10:04 AM
Updating from Slackware64-current to Slackware64 13. glore2002 Slackware 4 08-28-2009 07:50 PM
[SOLVED] Slackware64-C client wont detect Slackware64-C Samba server slackass Slackware 5 08-08-2009 12:23 AM

LinuxQuestions.org > Forums > Linux Forums > Linux - Distributions > Slackware

All times are GMT -5. The time now is 03:39 PM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration