-
Notifications
You must be signed in to change notification settings - Fork 371
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Polaris "failed to send message" errors during boot. #146
Comments
Your sitting at base clock, there is power play library failure due to VBIOS update.
Greg
On Jul 3, 2017, at 5:09 PM, jstefanop <notifications@github.com<mailto:notifications@github.com>> wrote:
On a fresh install of ROCm 1.6 on top of Ubuntu 16.04 we are getting the messages below during boot after ring tests. Each GPU hangs for about 10 seconds and boot continues normally.
Card seems to be initialized and working properly post boot regardless (full openCL performance on par with AMDGPU-PRO).
This issue is not present during Vega FE boot
[ 3.058782] amdgpu 0000:01:00.0: fb0: amdgpudrmfb frame buffer device [ 3.485773] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 4.310558] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 4.723086] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 5.547875] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 5.960395] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 6.785188] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 7.197712] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 8.022497] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 8.435022] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 9.259809] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 9.672330] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 10.502892] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 10.924276] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 10.938191] kfd kfd: Allocated 3969056 bytes on gart for device 1002:67df [ 11.346839] amdgpu: [powerplay] failed to send pre message 15b ret is 0 [ 11.767822] amdgpu: [powerplay] failed to send message 15b ret is 0 [ 12.196446] amdgpu: [powerplay] failed to send pre message 155 ret is 0 [ 12.618252] amdgpu: [powerplay] failed to send message 155 ret is 0 [ 12.628554] Virtual CRAT table created for GPU [ 12.630970] Parsing CRAT table with 1 nodes [ 12.632391] Creating topology SYSFS entries [ 12.633893] Topology: Add dGPU node [0x67df:0x1002] [ 12.635339] kfd kfd: Reserved 2 pages for cwsr. [ 12.636769] kfd kfd: added device 1002:67df [ 12.638159] [drm] Initialized amdgpu 3.16.0 20150101 for 0000:01:00.0 on minor 0 [ 12.641308] ACPI: Video Device [GFX0] (multi-head: yes rom: no post: no) [ 12.643153] acpi device:16: registered as cooling_device9 [ 12.644718] input: Video Bus as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/LNXVIDEO:00/input/input9 [ 12.646381] [drm] Initialized i915 1.6.0 20160919 for 0000:00:02.0 on minor 1
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub<#146>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD8DuYMVe1RePLJhNz5RNFgSP-3IiGaUks5sKWargaJpZM4OMuBf>.
|
If PPlib fails we drop the clock to minimum frequency. around 900 mhz or less. on vega10, which why your seeing low performance. we have no way for the Driver to drive the clock up with the driver since PPLIB interacts with System mangment unit which controls the clock. We will get a patch out. |
@gstoner this issue is only present on Polaris based cards. Even so, there is no performance impact on the polaris cards. The only real impact is slow boot since, if there are several cards on the system each card takes 10+ seconds to boot. This issue is NOT present on Vega FE, Vega FE boots properly and is running at 1600mhz, verified by both smi utility and power consumption. This issue is not the cause of very low Vega performance. |
Looking at the slow boot time is there more than 10 GPU in the system. |
Our systems are designed for 16+ but this issue is present even on normal boards with a single gpu running ROCm kernel. Every Polaris based gpu hangs for ~ 10 seconds as amdgpu drm posts them. |
How many PLX, and how deep, I mean are you cascading PLX switch like tree 2 to 3 layer deep.
Also
Greg
On Jul 4, 2017, at 3:20 PM, jstefanop <notifications@github.com<mailto:notifications@github.com>> wrote:
Our systems are designed for 16+ but this issue is present even on normal boards with a single gpu running ROCm kernel. Every Polaris based gpu hangs for ~ 10 seconds as amdgpu drm posts them.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#146 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD8DubdBknyYVqSZgwbPmkREG23KwNIQks5sKp56gaJpZM4OMuBf>.
|
Only two switches off the root complex. Each switch feeds 8 GPUs. Either way this issue is present with the GPU plugged directly into the root 16x lanes coming from CPU. I will try the ryzen setup later to rule out any issues with kabylake i3 we are using on our test bench. Are you able to reproduce the issue with your systems? |
What are the exact plx you are using? We only have plx8747 based systems.
Get Outlook for iOS<https://aka.ms/o0ukef>
On Tue, Jul 4, 2017 at 3:39 PM -0500, "jstefanop" <notifications@github.com<mailto:notifications@github.com>> wrote:
Only two switches off the root complex. Each switch feeds 8 GPUs. Either way this issue is present with the GPU plugged directly into the root 16x lanes coming from CPU.
I will try the ryzen setup later to rule out any issues with kabylake i3 we are using on our test bench.
Are you able to reproduce the issue with your systems?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#146 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD8DuY3qa0TtWKHcSdh5mJdQ_sVOt3SWks5sKqMTgaJpZM4OMuBf>.
|
Lets forget our custom systems for now, since we are testing on the simplest system possible to reduce variables. We can confirm this issue is present on your Ryzen platform as well, so its definitely a driver bug on the ROCm stack. Here is the exact test bench rig Ryzen 5 1400 |
We use Ryzen, EPYC,, Ryzen ThreadRipper, Intel Xeon E5, Intel Core I7, I5 for testing. We double check your finding to see if there is an issue is an issue or is a motherboard system bios issue. ROCm was Enterprise Server focused, the primary server system under test is SuperMicro SYS-1028 GQ, SYS-7048, SYS-4028 with Xeon E5 v3 & v4 processors. We also have a number Inventec P45 EPYC server and K888 Intel Xeon E5 systems under test. We also test Lenovo, Dell and HP servers. |
@gstoner yea like i mentioned its not that huge of an issue, since Polaris has its own consumer stack that works fine on all our systems, and this issue is not present on vega with this driver stack. Only reason I mention it is because Polaris cards are officially supported by the ROCm stack. |
We are going to dig in on Polaris since MI6 is Polaris 10 based. One thing we I need to evaluate is there is the issue in the VBIOS on the RX580 or this device ID issue for that card. We not in Markham so we do not alway get the newest consumer hardware |
Also I am going to check the performance Vega OpenCL - Windows vs ROCm, plus Polaris and Fiji Windows vs ROCm vs 17.20
On Jul 5, 2017, at 8:03 PM, jstefanop <notifications@github.com<mailto:notifications@github.com>> wrote:
@gstoner<https://github.com/gstoner> yea like i mentioned its not that huge of an issue, since Polaris has its own consumer stack that works fine on all our systems, and this issue is not present on vega with this driver stack. Only reason I mention it is because Polaris cards are officially supported by the ROCm stack.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#146 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD8DuWLWXyJJO6ySb5ZPoDXHop-FCBdNks5sLDJPgaJpZM4OMuBf>.
|
Was this resolved? |
We found the issue it on MSI motherboard there is APCI issue in the Linux kernel, we have back patch in 1.6.1 |
The SBIOS issue and Base Linux kernel issue not ROCm |
I have same issue. Do you have some way to resolve this? |
@maxsolyaris. We updated the 4.11 Linux kernel which fixed this issue, on MSI motherboard you also need to update the SBIOS. |
FWIW - I'm on latest bios on my MSI X370 A4 Gaming Plus (Sept 21 bios), and I still see these errors on bootup w/ RX470 and RX570 against the 4.11 rocm kernel. |
+1 i have this Issue on Msi z270-A Pro only on modded Bios on 570/580 Kernel needs to be updated to 4.14 at least |
Updated both kernel 4.14 and BIOS for H110 BTC+. Still having this crash when restarting. |
4.15 will be part of our next release. Note 4.14 Linux kernel does not have all the feature need for DGPU yet to run ROCm. Linux Kernel 4.16 is the first upstream release that should have everything you need build the driver for DGPU for ROCm Greg |
dear all. i stompled in googling this error. Iam on MSI Z170A Gaming M5 with 7x rx580 and 2~3 of them sitting at base clock(stock bios). A modified bios can not boot at all with linux.(arch) is there a way to handle this issue? |
I'm using Kernel 4.15.0-rc7 (!) and still getting error messages from my RX 460:
|
I'm seeing lots of these following errors on my ASUS B250 MINING EXPERT LGA 1151 Intel B250 HDMI SATA 6Gb/s USB 3.1 ATX Intel Motherboard: Mar 02 18:37:41 u-miner-1 kernel: ACPI Warning: _SB_.PCI0.RP08.PXSX.DSM: Argum I'm using Ubuntu 16.04 LTS Ben |
Use the 4.13 Linux kernel
Get Outlook for iOS<https://aka.ms/o0ukef>
…________________________________
From: Ben-and-Ellen <notifications@github.com>
Sent: Friday, March 2, 2018 8:44:20 PM
To: RadeonOpenCompute/ROCm
Cc: Gregory Stoner; Mention
Subject: Re: [RadeonOpenCompute/ROCm] Polaris "failed to send message" errors during boot. (#146)
I'm seeing lots of these following errors on my ASUS B250 MINING EXPERT LGA 1151 Intel B250 HDMI SATA 6Gb/s USB 3.1 ATX Intel Motherboard:
Mar 02 18:37:41 u-miner-1 kernel: ACPI Warning: _SB_.PCI0.RP08.PXSX.DSM: Argum
Mar 02 18:37:41 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP08.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 18:37:42 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX.DSM: Argum
Mar 02 17:28:17 u-miner-1 kernel: ACPI Warning: _SB.PCI0.RP12.PXSX._DSM: Argum
failed to send message 260 ret is 0
Mar 02 10:35:52 u-miner-1 kernel: amdgpu: [powerplay]
Mar 02 10:35:52 u-miner-1 kernel: amdgpu: [powerplay]
failed to send message 15b ret is 0
Mar 02 10:35:52 u-miner-1 kernel: amdgpu: [powerplay]
failed to send pre message 155 ret is 0
Mar 02 10:35:52 u-miner-1 kernel: amdgpu: [powerplay]
failed to send message 155 ret is 0
Mar 02 10:35:52 u-miner-1 kernel: amdgpu: [powerplay]
failed to send pre message 260 ret is 0
I'm using Ubuntu 16.04 LTS
4.4.0-112-generic
Ben
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#146 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD8DubSKx6P7HOyrrxtgy2sHRHf8QTZDks5tagOEgaJpZM4OMuBf>.
|
Still have this problem on ASUS B250 MINING EXPERT |
Same problem here |
Same problem
|
same problem... |
I might have figured this out for the B250 mining expert on ubuntu 16.04. If I had a card plugged into the primary A1 slot (the first x16/full size) while mining I would get these errors and rocm-smi would timeout on the card in that particular slot (while throwing more of these errors). But if I don't use it, everything is happy. Note: I am using an older AMDGPU Pro driver (17.40) and I'm still on the 4.10 kernel, but it's crunching away happily now and I can use rocm-smi without error. So not sure if it's related at all or what it doesn't like about that first x16 slot (did not try the second either) but might be related? Edit: Oops, actually I see this issue is for ROCm driver, so probably not helpful :/ |
* first pass of the release notes for 6.2.1 (#131) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * Spolifroni amd/release notes 621 (#135) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * added documentation highlights (#136) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * Added information for rocdbgapi (#138) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Updates to documentation section; changed "key" to "notable" (#139) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * Updated the release date and made changes to component details (#140) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * Updated the known issues intro (#141) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * test (#142) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * Spolifroni amd/release notes 621 (#143) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * Reworded some things (#146) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * Added info for rocal 2.0.0 (#147) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Some small changes to the release notes (#148) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * Updated with more components for RC3 (#149) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * Small changes to wording, punctuation; fixed a list (#150) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * Updated versions and removed previous release notes. (#151) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * Update to hightlights, SMI, small fixes (#152) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * Updated the known issues wording for rocAL (#153) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * small fixes (#155) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray "notable" (#156) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * Added offline installer highlight (#157) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added link to offline installer; aligned rn with other FBGEEM doc (#158) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component (#159) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed broken links (#160) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * updated the links again and removed rocAL optimization and known issues (#161) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date (#163) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes (#165) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * Moved known issue to omnitrace (#166) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked omnitrace wording (#167) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed rocdbgapi (#168) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Changed wording in offline installer changes (#169) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Updated wording for Offline Installer changes * Updated to show no new Known Issues. (#170) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Updated wording for Offline Installer changes * changed Known Issues to say that there are no known issues * updated the upcoming changes (#171) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Updated wording for Offline Installer changes * changed Known Issues to say that there are no known issues * added rccl plugin removal * added lack of mi300x support to hardware (#172) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Updated wording for Offline Installer changes * changed Known Issues to say that there are no known issues * added rccl plugin removal * added lack of MI300X supporort * removed a contraction (#173) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Updated wording for Offline Installer changes * changed Known Issues to say that there are no known issues * added rccl plugin removal * added lack of MI300X supporort * i don't like contractions. te irony * Changed the link in known issues (#174) * first pass of the release otnotes for 6.2.1 * something went wrong building the relnotes the first time; this should be OKer * Partially complete release notees for 6.2.1 * added a line about there being no OS changes in 6.2.1 relative 6.2.0 * Updated version and date * made wording changes and added documentation highlights * added information about rocdbgapi * Changed 'key' to 'notable'; clarified that changes are from 6.2.0 to 6.2.1; clarified the open-source nature of the documentation; brought a note back. * updated the release date in conf.py; removed added api calls for HIP; added fixed issues to rcodbgapi * changed the opening intro to Known Issues * fixed the major copy-pasta error with upcoming changes * removed a word just to see what happens * putting the "are" back * removed the HIP changes; they were in 6.2.0 * corrected some formatting errors * changed some wording * changed a word * reworded the known issues * added info for rocAL 2.0.0 * Updated the wording on the rocAL changes * made some small changes. * minor wording change * added more component changes * fixed a bad table; made some minor changes to punctuation and spelling. * The hipify version needs to be updated to reflect that its version reflects the rocm version. So it went from 6.2.0 to 6.2.1 * undid the hipify version change, but updated the version of amd smi * removed the previous release notes. * updated release date to Sept 12 * modified the ROCm SMI entry; workaround reworded and put into known issues; one line added to resolved issues * Added the FBGEEM support highlight * updated wording on rocAL known issues * made some small edits * removed a stray 'notable' * added offline installer highlight * added a link to the offline installer doc; removed the second uppercase E in FBGEEM long-form to align with the other documentation * fixed a link that had to go to latest rather than to 6.2.1 * trying to trigger a pr * undoing the last change * changed a link; fixed wording; added a 'removals' section for one component * fixed up the list for rocAL to make it more compact * fixed broken links to component documentation * Removed optimizations and known issues from rocal * updated doc links of 404ing components to their readthedocs documentation. Tensile won't be released until later so the link goes to github. Will need to double-check links after release to make sure they still work. * updated release date * small changes * moved known issue to omnitrace * tweeked the omnitrace workaround language to be more precise * fixed ROCdbgapi * Updated wording for Offline Installer changes * changed Known Issues to say that there are no known issues * added rccl plugin removal * added lack of MI300X supporort * i don't like contractions. te irony * fixed the label in known issues github link and also changed it from being a link to known issues to issues, since there are no verified known issues at this point * removed link to github and reference to the list of known issues * remove "6.2.1 does not support MI300X" and add MI300X GPU recovery failure KI * update words * removed info re. rocdbgapi known issues (#176) * Added point about version change to rocal * Put link to prerequisites in rocal --------- Co-authored-by: Peter Park <peter.park@amd.com>
On a fresh install of ROCm 1.6 on top of Ubuntu 16.04 we are getting the messages below during boot after ring tests. Each GPU hangs for about 10 seconds and boot continues normally.
Card seems to be initialized and working properly post boot regardless (full openCL performance on par with AMDGPU-PRO).
This issue is not present during Vega FE boot
[ 3.058782] amdgpu 0000:01:00.0: fb0: amdgpudrmfb frame buffer device [ 3.485773] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 4.310558] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 4.723086] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 5.547875] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 5.960395] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 6.785188] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 7.197712] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 8.022497] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 8.435022] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 9.259809] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 9.672330] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 10.502892] amdgpu: [powerplay] failed to send pre message 260 ret is 0 [ 10.924276] amdgpu: [powerplay] failed to send message 260 ret is 0 [ 10.938191] kfd kfd: Allocated 3969056 bytes on gart for device 1002:67df [ 11.346839] amdgpu: [powerplay] failed to send pre message 15b ret is 0 [ 11.767822] amdgpu: [powerplay] failed to send message 15b ret is 0 [ 12.196446] amdgpu: [powerplay] failed to send pre message 155 ret is 0 [ 12.618252] amdgpu: [powerplay] failed to send message 155 ret is 0 [ 12.628554] Virtual CRAT table created for GPU [ 12.630970] Parsing CRAT table with 1 nodes [ 12.632391] Creating topology SYSFS entries [ 12.633893] Topology: Add dGPU node [0x67df:0x1002] [ 12.635339] kfd kfd: Reserved 2 pages for cwsr. [ 12.636769] kfd kfd: added device 1002:67df [ 12.638159] [drm] Initialized amdgpu 3.16.0 20150101 for 0000:01:00.0 on minor 0 [ 12.641308] ACPI: Video Device [GFX0] (multi-head: yes rom: no post: no) [ 12.643153] acpi device:16: registered as cooling_device9 [ 12.644718] input: Video Bus as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/LNXVIDEO:00/input/input9 [ 12.646381] [drm] Initialized i915 1.6.0 20160919 for 0000:00:02.0 on minor 1
The text was updated successfully, but these errors were encountered: