23.09d - Is QAT Broken?
-
@RobbieTT said in 23.09d - Is QAT Broken?:
@jimp said in 23.09d - Is QAT Broken?:
You can't compare 23.05.1 directly because it did not have that tunable so I don't get where your assertion is coming from that it had both enabled there. Your shell output doesn't show that. There is nothing in the 23.05.1 output showing userspace.
Understood but the man pages lists ks;us as the default and the absence of an explicit command to demure from that usually equals that the default is set. Apologies if this is not the case.
The man page is not complete/accurate there. It's only the default on 4xxx devices as well, for all others it defaults to kernel only.
https://github.com/freebsd/freebsd-src/blob/main/sys/dev/qat/qat_common/adf_cfg.c#L37
-
@jimp said in 23.09d - Is QAT Broken?:
The man page is not complete/accurate there.
I didn't stand a chance. Back to the cup of tea.
๏ธ
-
@RobbieTT said in 23.09d - Is QAT Broken?:
@jimp said in 23.09d - Is QAT Broken?:
The man page is not complete/accurate there.
I didn't stand a chance. Back to the cup of tea.
๏ธ
There are many nuances indeed! It's a good discussion to have, and the civil approach is appreciated :)
-
@marcosm said in 23.09d - Is QAT Broken?:
@RobbieTT said in 23.09d - Is QAT Broken?:
@jimp said in 23.09d - Is QAT Broken?:
The man page is not complete/accurate there.
I didn't stand a chance. Back to the cup of tea.
๏ธ
There are many nuances indeed! It's a good discussion to have, and the civil approach is appreciated :)
Yes, thank you for a civil discussion @marcosm and @stephenw10. I don't know why these conversations often become aggressive with users trying to provide input to netgate. Thanks @marcosm and @stephenw10 .
-
@bcdouglas
Sounds like I am being admonished; if so I will take it on the chin. It was not my intent to cause waves but clearly something unintended was triggered.As users we don't always have the technical language for this kind of discourse but all I can say is that I did my best to read-into the topic to try and understand the apparent changes or limitations, only to find gaps in the documentation.
Threads such as this may put-off others from providing feedback but it shouldn't. Please set this aside and do comment when you think something does not make sense. Nothing can move forward without feedback.
Anyway, I'll take the thumping on this one.
-
I really don't think there was any issue here. Reading back I think there was a misunderstanding earlier on but in general this was a useful discussion. No admonishment required!
-
@RobbieTT Not from me. The conversations from the Netgate side often take a weird turn when users try to ask honest questions and help.
-
-
No observable change in functionality with the newly-enabled QAT 200xx devices (tested on an Xeon D-1536NT with 23.09.b.20231020.0600 installed) from that of the C3xxx series on the same beta load.
๏ธ
-
The change we made there should now recognise that device as QAT capable on the dashboard. It should also load the qat module if it's not already.
However the driver itself already worked with the hardware so if it was loaded then the kernel could already use it for kernel mode crypto. -
@stephenw10
It does show correctly and I have provided feedback on issue #14844.๏ธ
-
-
Interesting conversations, indeed. I hope that if I installed the QAT device that all need for it, whether kernel or user induced, would utilized the device...that's the reason for purchasing the device. I know in both 23.05.1 and 23.09 releases, the QAT driver takes almost 3 minutes to load on one of my boxes.
-
@RobbieTT said in 23.09d - Is QAT Broken?:
There is certainly little point leaving QAT idle when it could be put to use; well, in my view. QAT is one of things that attracted me to Netgate / pfSense+.
I agree and what also attracted me to plus as well and having it in two boxes.
-
Wendell has just posted this video on QAT and where it should be used:
Clearly I agree.
๏ธ
-
@RobbieTT said in 23.09d - Is QAT Broken?:
Wendell has just posted this video on QAT and where it should be used:
Clearly I agree.
๏ธ
Right on...wished he had also test on FreeBSD...