AGESA 1.0.0.6b Might Fix The Ryzen Linux Performance Marginality Problem

[simi_id]

Hello,

I have ASRock X370 Gaming Professional with P3.10 bios which has Agesa 1.0.0.6b
My CPU is a rma one from week 33, the previous one was from week 14.

https://www.dropbox.com/s/x78yfs9ykbeynh5/2017-09-18%2012.43.06.jpg?dl=0

And when I ran this test from here:

https://www.reddit.com/r/Amd/comments/6rwggi/ryzen_build_loop_compile_failures_under_linux/

I still get crash error with that test. Is that a segfault ? Is the test flawed ? How can I test kill-ryzen script step by step ?

[simi_id]

Update: I ran the same test from there on my laptop which has an Intel 7700hq and I get same error in 3 seconds.
"Test Failed: loopCountToFailure=[0] elapsedTimeInSeconds=[30]"
I think the test is flawed.
How can I run killRyzen test which is everyone testing ? I'm not a linux guy and I would like a step by step tutorial as the test above.

[BillBroadley]

simi_id could try downloading the newest gcc source code (google building gcc from source). Then use "make -j 8" instead of make. Takes a good while to finish and if it crashes/hangs during the build that's a very bad sign.

[simi_id]

BillBroadley Hello, as I said I am not a linux guy or programmer ... I don't have any idea how to build from source from scratch. Where can i find a step by step tutorial ? Also what is the latest 7.2 or 6.4, I saw them both released recently.

Thanks

[scorpio810]

BillBroadley Hello, as I said I am not a linux guy or programmer ... I don't have any idea how to build from source from scratch. Where can i find a step by step tutorial ? Also what is the latest 7.2 or 6.4, I saw them both released recently.

Thanks

Don't panic ! if you not a Linux software programmer or a Gentoo guy this problem not concern you. ;-)

Edit : your CPU is good 1733SUS !!

[mat69]

Update: False alarm!

My week week 28 ("UA 1728PGT" is on the chip) seems to be fine.
One RAM module was broken and I never thought of running memtest. Silly me.

[BillBroadley]

BillBroadley Hello, as I said I am not a linux guy or programmer ... I don't have any idea how to build from source from scratch. Where can i find a step by step tutorial ? Also what is the latest 7.2 or 6.4, I saw them both released recently.

Thanks

sudo apt install build-essential libgmp-dev libmpfr-dev libmpc-dev
wget [URL]https://ftp.gnu.org/gnu/gcc/gcc-7.2.0/gcc-7.2.0.tar.gz[/URL]
tar xvf gcc-7.2.0.tar.gz
mkdir build
cd build
../gcc-7.2.0/configure --disable-multilib --enable-languages=c,c++,fortran,go
time make -j 16

Sure, not many people do parallel compiles often. But if you are paying extra for additional cores you should be able to use them. Some games, video encoders/decoders (like handbrake), or batch processing of photos/audio/video can make use of extra cores.

It should look like this when done:

/libgo/go/cmd/cgo/gcc.go ../../gcc-7.2.0/gotools/../libgo/go/cmd/cgo/godefs.go ../../gcc-7.2.0/gotools/../libgo/go/cmd/cgo/main.go ../../gcc-7.2.0/gotools/../libgo/go/cmd/cgo/out.go ../../gcc-7.2.0/gotools/../libgo/go/cmd/cgo/util.go zdefaultcc.go
make[2]: Leaving directory '/home/bill/build/gotools'
make[1]: Leaving directory '/home/bill/build'
real    18m15.689s
user    228m21.438s
sys     8m36.111s
bill@naples:~/build$

[soulsource]

Sure, not many people do parallel compiles often. But if you are paying extra for additional cores you should be able to use them. Some games, video encoders/decoders (like handbrake), or batch processing of photos/audio/video can make use of extra cores.

The Linux compilation segfault issue has afaik not been observed in multithreaded programs, even if they push all logical cores to full load, but only while running several compiler processes at the same time.

[soulsource]

Reports around the internet indicate that even some of the replacement CPUs still exhibit the problem.

The first few RMA requests indeed seem to have been a game of luck. Meanwhile one can be more or less certain to get a week 25 or newer part though.

[Brisse]

Just updated BIOS for my PRIME X370-PRO this morning. According to release notes, it has the new agesa.

"PRIME X370-PRO BIOS 0902
Update AGESA 1.0.0.6B"

Went ahead and ran some kill-ryzen, but it didn't take long...

[loop-5] TIME TO FAIL: 53 s

[loop-6] TIME TO FAIL: 119 s

[loop-11] TIME TO FAIL: 132 s

So no, it's definitely not fixed.