ywkaras commented on issue #7546:
URL: https://github.com/apache/trafficserver/issues/7546#issuecomment-786971941
I've not been able to reproduce the problem on the AMD proxy. There are my
results:
```
-bash-4.2$ cat http2_benchmark.report
**http2load**
finished in 10.17s, 98320.46 req/s, 1.50GB/s
requests: 1000000 total, 1000000 started, 1000000 done, 1000000 succeeded, 0
failed, 0 errored, 0 timeout
status codes: 1000000 2xx, 0 3xx, 0 4xx, 0 5xx
traffic: 15.29GB (16419037931) total, 7.66MB (8031931) headers (space
savings 96.63%), 15.26GB (16384000000) data
min max mean sd +/- sd
time for request: 111us 130.27ms 1.51ms 4.90ms 92.15%
time for connect: 3.38ms 87.15ms 38.33ms 20.31ms 69.00%
time to 1st byte: 5.59ms 106.70ms 47.40ms 19.96ms 74.00%
req/s : 476.66 803.50 671.46 97.62 65.50%
**dstat**
You did not select any stats, using -cdngy by default.
----total-cpu-usage---- -dsk/total- ---net/lo-- -net/total- ---paging--
---system--
usr sys idl wai hiq siq| read writ| recv send: recv send| in out | int
csw
1 0 99 0 0 0|2597B 211k| 0 0 : 0 0 | 0 0 |
28k 84k
13 5 81 0 0 1| 0 19M|1590M 1590M:2169B 2064B| 0 0 |
340k 295k
**perf stat**
Performance counter stats for process id '52642':
167,926.60 msec task-clock # 16.268 CPUs utilized
644,529 context-switches # 0.004 M/sec
16,677 cpu-migrations # 0.099 K/sec
7,170 page-faults # 0.043 K/sec
292,679,346,215 cycles # 1.743 GHz
(67.12%)
41,677,240,403 stalled-cycles-frontend # 14.24% frontend cycles
idle (67.21%)
22,998,344,897 stalled-cycles-backend # 7.86% backend cycles
idle (67.20%)
212,494,852,375 instructions # 0.73 insn per cycle
# 0.20 stalled cycles
per insn (67.15%)
38,284,614,744 branches # 227.984 M/sec
(67.13%)
1,354,399,053 branch-misses # 3.54% of all branches
(67.09%)
10.322305247 seconds time elapsed
**perf report**
# Total Lost Samples: 0
#
# Samples: 1M of event 'cycles'
# Event count (approx.): 448142823289
#
# Overhead Shared Object Symbol
# .......... ....................
..................................................
#
7.17% libcrypto.so.1.1 [.] _aesni_ctr32_ghash_6x
5.81% libc-2.17.so [.] __memcpy_ssse3
3.51% [kernel.kallsyms] [k] copy_user_generic_string
2.65% libtscore.so.10.0.0 [.] freelist_new
1.94% [kernel.kallsyms] [k] acpi_processor_ffh_cstate_enter
1.60% libc-2.17.so [.] __memcmp_sse4_1
1.04% libtscpputil.so.10.0 [.] memcmp
0.91% traffic_server [.] HpackIndexingTable::lookup
0.80% [kernel.kallsyms] [k] _raw_spin_lock_irqsave
...
```
I'm seeing much lower req/s on the AMD than on the Intel Xeon.
Xeon:
```
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 64
On-line CPU(s) list: 0-63
Thread(s) per core: 2
Core(s) per socket: 16
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 79
Model name: Intel(R) Xeon(R) CPU E5-2683 v4 @ 2.10GHz
Stepping: 1
CPU MHz: 1200.091
CPU max MHz: 3000.0000
CPU min MHz: 1200.0000
BogoMIPS: 4199.63
Virtualization: VT-x
L1d cache: 32K
L1i cache: 32K
L2 cache: 256K
L3 cache: 40960K
```
AMD:
```
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 128
On-line CPU(s) list: 0-127
Thread(s) per core: 2
Core(s) per socket: 64
Socket(s): 1
NUMA node(s): 1
Vendor ID: AuthenticAMD
CPU family: 23
Model: 49
Model name: AMD EPYC 7742 64-Core Processor
Stepping: 0
CPU MHz: 2250.000
CPU max MHz: 2250.0000
CPU min MHz: 1500.0000
BogoMIPS: 4491.55
Virtualization: AMD-V
L1d cache: 32K
L1i cache: 32K
L2 cache: 512K
L3 cache: 16384K
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
