Can one use backreference as replacement value in sed?

Question

Source file:

Linux 2.6.32-754.18.2.el6.x86_64 (myhostname)  03/24/2020      _x86_64_        (64 CPU)
07:32:01 PM      57     25.79      0.00      4.99      0.00      0.00      0.00      0.02      0.00     69.20
07:32:01 PM      58     26.38      0.00      3.43      0.00      0.00      0.00      0.02      0.00     70.17
07:32:01 PM      59     24.49      0.00      8.73      0.00      0.00      0.00      0.03      0.00     66.75
07:32:01 PM      60     20.31      0.00      5.60      0.00      0.00      0.00      0.02      0.00     74.08
07:32:01 PM      61     20.08      0.00      4.09      0.00      0.00      0.00      0.02      0.00     75.81
07:32:01 PM      62     21.33      0.00      5.23      0.00      0.00      0.00      0.02      0.00     73.42
07:32:01 PM      63     18.49      0.00      4.04      0.00      0.00      0.00      0.02      0.00     77.45

Average:        CPU      %usr     %nice      %sys   %iowait    %steal      %irq     %soft    %guest     %idle
Average:        all     24.02      0.00      5.23      0.07      0.00      0.00      0.23      0.00     70.45

Desired output:

myhostname        64 CPU     24.02      0.00      5.23      0.07      0.00      0.00      0.23      0.00     70.45

I was able to extract required values from first string:

sed -n '1{s|^[^(]*(\([^)]*\))[^(]*(\([^)]*\)).*|\1 \2|p}' test.txt
myhostname 64 CPU

But I don't know how to use it for replacement.

I'm aware about other (perl/awk) ways to do the needfull but I'm looking for sed solution

Update: Alas both suggestions from comment haven't given me the desired output

[14]labuser@labhost:~> sed -n '1{s|^[^(]*(\([^)]*\))[^(]*(\([^)]*\)).*|\1 \2|;h;};${H;x;s/\n.*all//;p;}' test.txt
isgmwapp6n1 64 CPU

[14]user@labhost:~> sed -n '1h;${H;x;s/.*(\([^)]*\)).*(\([^)]*\))\n.*all/\1 \2/p;}' test.txt
[14]user@labhost:~/>
[14]user@labhost:~> sed --version |head -1
GNU sed version 4.2.1
[14]user@labhost:~> cat /etc/redhat-release
Red Hat Enterprise Linux Server release 6.10 (Santiago)

Update2 Third one also gives no output :( Attaching screenshot

Update3 I found out that first suggested solution don't work just because the last line of file is newline. Thus I was able to the needful via the following:

 sed -n '1{s|^[^(]*(\([^)]*\))[^(]*(\([^)]*\)).*|\1 \2|;h;};/Average.*all/{H;x;s/\n.*all//;p;}' test.txt

@mosvy, many thanks for you help. if you post the above as answer I gonna accept it.

Answer 1

sed -n '1h;/Average.*all/{H;x;s/.*(\([^)]*\)).*(\([^)]*\))\n.*all/\1 \2/p;}' /tmp/jeg

should work in any version of sed.

The h command on the first line copies its content to the "hold space".

Then on the last line (the one matching /Average.*all/, not $ because the file may contain extra empty lines at the end), H;x; (append to the hold space + exchange hold and pattern space) will prepend the first line to it, and then the s/../../p will trim it of unneeded stuff and print it.