Re: Switching from trunk(4) to aggr(4)
Hey,
My setup at home is almost identical. APU with aggr interface and a couple of
VLANs:
https://github.com/liv-io/ansible-playbooks-example/blob/master/bsd/host_vars/fw01.example.com.yml
# cat /etc/hostname.em{1,2,3}
up
# cat /etc/hostname.aggr0
trunkport em1 trunkport em2 trunkport em3 lacpmode active lacptimeout slow
description "i_data"
up
# cat /etc/hostname.vlan11
inet 10.1.1.2 255.255.255.0 NONE vnetid 11 vlandev aggr0 description "v_base"
up
# cat /etc/hostname.carp11
inet 10.1.1.1 255.255.255.0 NONE vhid 1 carpdev vlan11 advskew 10 pass
"" description "v_base"
# ifconfig aggr0
aggr0: flags=8943 mtu 1500
lladdr
description: i_data
index 11 priority 0 llprio 7
trunk: trunkproto lacp
trunk id: [(8000,,000B,,),
(8000,,0006,,)]
em1 lacp actor system pri 0x8000 mac, key 0xb, port pri 0x8000
number 0x2
em1 lacp actor state
activity,aggregation,sync,collecting,distributing
em1 lacp partner system pri 0x8000 mac, key 0x6, port pri
0x8000 number 0x12e
em1 lacp partner state
activity,aggregation,sync,collecting,distributing
em1 port active,collecting,distributing
em2 lacp actor system pri 0x8000 mac, key 0xb, port pri 0x8000
number 0x3
em2 lacp actor state
activity,aggregation,sync,collecting,distributing
em2 lacp partner system pri 0x8000 mac, key 0x6, port pri
0x8000 number 0x130
em2 lacp partner state
activity,aggregation,sync,collecting,distributing
em2 port active,collecting,distributing
em3 lacp actor system pri 0x8000 mac, key 0xb, port pri 0x8000
number 0x4
em3 lacp actor state
activity,aggregation,sync,collecting,distributing
em3 lacp partner system pri 0x8000 mac, key 0x6, port pri
0x8000 number 0x12f
em3 lacp partner state
activity,aggregation,sync,collecting,distributing
em3 port active,collecting,distributing
groups: aggr
media: Ethernet autoselect
status: active
It works well for me and I never had issues. I currently use a HP switch, but it
also works with Cisco.
Maybe some leftovers from the LACP config? I never encountered the "no carrier
status" issue though.
Let me know if I can extract any config for you.
HTH,
Livio
On 2020-12-12 16:44, Daniel Jakots wrote:
Hi,
I've been using a LACP trunk on my apu (with the three em(4)). On
top of which I have some vlans. I've been doing that for years and it's
working fine.
I thought about using aggr(4) instead (for no real reason). But the
aggr interface stays in "status: no carrier".
What I did is, I replaced my hostname.trunk0
trunkproto lacp trunkport em0 trunkport em1 trunkport em2
up
with a hostname.aggr0
trunkport em0 trunkport em1 trunkport em2
up
(and changing the parent in my hostname.vlan*). To apply the new
configuration, I just reboot.
My trunk0 which works is
trunk0: flags=8843 mtu 1500
lladdr 00:0d:b9:43:9f:fc
index 7 priority 0 llprio 3
trunk: trunkproto lacp
trunk id: [(8000,00:0d:b9:43:9f:fc,403C,,),
(0080,00:00:00:00:00:00,,,)]
em2 lacp actor system pri 0x8000 mac 00:0d:b9:43:9f:fc, key
0x403c, port pri 0x8000 number 0x3
em2 lacp actor state
activity,aggregation,sync,collecting,distributing,defaulted
em2 lacp partner system pri 0x80 mac 00:00:00:00:00:00, key
0x0, port pri 0x80 number 0x0
em2 lacp partner state aggregation,sync,collecting,distributing
em2 port active,collecting,distributing
em1 lacp actor system pri 0x8000 mac 00:0d:b9:43:9f:fc, key
0x403c, port pri 0x8000 number 0x2
em1 lacp actor state
activity,aggregation,sync,collecting,distributing,defaulted
em1 lacp partner system pri 0x80 mac 00:00:00:00:00:00, key
0x0, port pri 0x80 number 0x0
em1 lacp partner state aggregation,sync,collecting,distributing
em1 port active,collecting,distributing
em0 lacp actor system pri 0x8000 mac 00:0d:b9:43:9f:fc, key
0x403c, port pri 0x8000 number 0x1
em0 lacp actor state
activity,aggregation,sync,collecting,distributing,defaulted
em0 lacp partner system pri 0x80 mac 00:00:00:00:00:00, key
0x0, port pri 0x80 number 0x0
em0 lacp partner state aggregation,sync,collecting,distributing
em0 port active,collecting,distributing
groups: trunk
media: Ethernet autoselect
status: active
And the aggr0 which doesn't come up is:
aggr0: flags=8843 mtu 1500
lladdr 00:0d:b9:43:9f:fc
index 6 priority 0 llprio 7
t
Re: Low throughput with 1 GigE interface
Thank you @Noth. You are right. The OpenBSD PF FAQ also says: > PF will only use one processor, so multiple processors (or multiple cores) WILL NOT improve PF performance. For PC Engines APU users, I can highly recommend to update the BIOS. It improved my networking performance quite a bit: https://pcengines.github.io/ When testing with iperf3 I now get 450Mbit/s instead of 200Mbit/s (with pf enabled). With pf disabled, I get ~750Mbit/s. Copying files from one to another machine (same physical NIC on APU, with VLANs and PF enabled) gave me around 930Mbit/s. Again, many thanks for all your help and inputs! On 2/5/2020 12:56 AM, Noth wrote: > According to the manufacturer of the APU2, the problem is with OpenBSD not > using all cores for network traffic management: > https://teklager.se/en/knowledge-base/apu2c0-ipfire-throughput-test-much-faster-pfsense/
Re: Low throughput with 1 GigE interface
Thank you for your inputs - @Jordan, @Tom, @Christian On 1/30/2020 9:07 PM, Tom Smyth wrote: > Livio are you running iperf on the apu ? > The apu doesnt have much cpu to generate packets from iperf... > Forwarding perf should about 450m on an apu c2 with pf enabled and about > 850m-900m with pf disabled > That issting with iperf through the apu2c2 with decent professional laptops > with iperf generated traffic and measured on these laptops @Tom: Yes, I am running iperf (server) on the APU. When I run iperf (server) on the notebook I get the following results: apu# iperf3 -c 192.168.20.40 Connecting to host 192.168.20.40, port 5201 [ 5] local 192.168.20.28 port 17990 connected to 192.168.20.40 port 5201 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 81.5 MBytes 680 Mbits/sec [ 5] 1.00-2.00 sec 81.1 MBytes 681 Mbits/sec [ 5] 2.00-3.01 sec 82.5 MBytes 685 Mbits/sec On 1/30/2020 11:29 PM, Christian Weisgerber wrote: > I vaguely remember a thread somewhere that concluded that one of > these network benchmark tools degenerated into a benchmark of > gettimeofday(2), which apparently is very cheap on Linux and not > cheap on OpenBSD. So you end up measuring the performance of this > system call. > > I don't remember whether it was iperf... You are probably right. I will now have to test the throughput with tcpbench as suggested by Jordan. On 1/31/2020 2:04 AM, Jordan Geoghegan wrote: > That sounds about right. I vaguely remember reading a thread about iperf on > misc some time in the past year mentioning that. > While OpenBSD obviously doesn't have the same network performance as Linux or > FreeBSD, as work continues on unlocking more of the kernel, things will > continue to get better. I think bluhm@ regularly runs some automated > benchmarks that show that OpenBSD maxes out at around 4-5 Gbit / second > throughput. Thank you, I will try to run some tests using tcpbench and send an update. The forwarding performance is the relevant part.
Re: Low throughput with 1 GigE interface
I ran fw_update and syspatch, which made the machine crash twice after booting, but now it is up and running again. The iperf results are still the same though: apu# iperf -s Server listening on TCP port 5001 TCP window size: 16.0 KByte (default) [ 4] local 192.168.20.28 port 5001 connected with 192.168.20.40 port 50064 [ ID] Interval Transfer Bandwidth [ 4] 0.0-10.0 sec 238 MBytes 199 Mbits/sec apu# iperf3 -s --- Server listening on 5201 --- Accepted connection from 192.168.20.40, port 50066 [ 5] local 192.168.20.28 port 5201 connected to 192.168.20.40 port 50067 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 22.2 MBytes 186 Mbits/sec [ 5] 1.00-2.00 sec 23.2 MBytes 195 Mbits/sec Should I try a -current or -stable Kernel? The update and crash output: apu# fw_update apu# syspatch Get/Verify syspatch66-001_bpf.tgz 100% || 102 KB 00:00 Installing patch 001_bpf Get/Verify syspatch66-002_ber.tgz 100% || 660 KB 00:03 Installing patch 002_ber Get/Verify syspatch66-003_bgpd.tgz 100% |***| 181 KB 00:00 Installing patch 003_bgpd Get/Verify syspatch66-004_net8021... 100% |*| 64839 00:00 Installing patch 004_net80211 Get/Verify syspatch66-005_sysupgr... 100% |*| 3023 00:00 Installing patch 005_sysupgrade Get/Verify syspatch66-006_ifioctl... 100% |*| 381 KB 00:03 Installing patch 006_ifioctl Get/Verify syspatch66-007_inteldr... 100% |*| 21468 KB 00:23 Installing patch 007_inteldrm Get/Verify syspatch66-010_libcaut... 100% |*| 20185 KB 00:17 Installing patch 010_libcauth Get/Verify syspatch66-012_suauth.tgz 100% |*| 7997 00:00 Installing patch 012_suauth Get/Verify syspatch66-013_ldso.tgz 100% |***| 295 KB 00:02 Installing patch 013_ldso Get/Verify syspatch66-015_ftp.tgz 100% || 65164 00:00 Installing patch 015_ftp Get/Verify syspatch66-016_ripd.tgz 100% |***| 45685 00:00 Installing patch 016_ripd Get/Verify syspatch66-017_inteldr... 100% |*| 268 KB 00:02 Installing patch 017_inteldrmctx Get/Verify syspatch66-018_smtpd_t... 100% |*| 224 KB 00:03 Installing patch 018_smtpd_tls Get/Verify syspatch66-019_smtpd_e... 100% |*| 224 KB 00:02 Installing patch 019_smtpd_exec Relinking to create unique kernel... done; reboot to load the new kernel Errata can be reviewed under /var/syspatch apu# reboot syncing disks... done rebooting... PC Engines apu2 coreboot build 20193012 BIOS version v4.11.0.2 login: kernel: protection fault trap, code=0 Stopped at uvm_map_lookup_entry+0x40: cmpq %r14,0x40(%rax) On 1/30/2020 7:06 PM, livio wrote: > @KatolaZ and @remi > > Thank you for your inputs on iperf2 vs. iperf3. > > After all the tests I needed a clean setup again and reinstalled both > OpenBSD and Window 10. > > With the new notebook (Dell vs Lenovo) I have different results. > Dell: ~ 200Mbit/s > Lenovo: ~ 145Mbit/s > > iperf2 vs. iperf3 (I also ran the corresponding version on Windows): > > apu# iperf -s > > Server listening on TCP port 5001 > TCP window size: 16.0 KByte (default) > > [ 4] local 192.168.20.28 port 5001 connected with 192.168.20.40 port 50052 > [ ID] Interval Transfer Bandwidth > [ 4] 0.0-10.0 sec 241 MBytes 202 Mbits/sec > > apu# iperf3 -s > --- > Server listening on 5201 > --- > Accepted connection from 192.168.20.40, port 50054 > [ 5] local 192.168.20.28 port 5201 connected to 192.168.20.40 port 50055 > [ ID] Interval Transfer Bitrate > [ 5] 0.00-1.00 sec 22.2 MBytes 186 Mbits/sec > [ 5] 1.00-2.00 sec 23.5 MBytes 197 Mbits/sec > [ 5] 2.00-3.00 sec 23.4 MBytes 196 Mbits/sec > [ 5] 3.00-4.00 sec 23.3 MBytes 196 Mbits/sec > [ 5] 4.00-5.00 sec 23.2 MBytes 195 Mbits/sec > [ 5] 5.00-6.00 sec 23.4 MBytes 196 Mbits/sec > [ 5] 6.00-7.00 sec 23.4 MBytes 196 Mbits/sec > [ 5] 7.00-8.00 sec 23.4 MBytes 197 Mbits/sec > [ 5] 8.00-9.00 sec 23.0 MBytes 193 Mbits/sec > [ 5] 9.00-10.00 sec 23.5 MBytes 197 Mbits/sec > [ 5] 10.00-10.05 sec 1.05 MBytes 197 Mbits/sec > - - - - - - - - - - - - - - - - - - - - - - - - - > [ ID] Interval Transfer Bitrate > [ 5] 0.00-10.05 sec 233 MBytes 195 Mbits/sec receiver
Re: Low throughput with 1 GigE interface
@KatolaZ and @remi Thank you for your inputs on iperf2 vs. iperf3. After all the tests I needed a clean setup again and reinstalled both OpenBSD and Window 10. With the new notebook (Dell vs Lenovo) I have different results. Dell: ~ 200Mbit/s Lenovo: ~ 145Mbit/s iperf2 vs. iperf3 (I also ran the corresponding version on Windows): apu# iperf -s Server listening on TCP port 5001 TCP window size: 16.0 KByte (default) [ 4] local 192.168.20.28 port 5001 connected with 192.168.20.40 port 50052 [ ID] Interval Transfer Bandwidth [ 4] 0.0-10.0 sec 241 MBytes 202 Mbits/sec apu# iperf3 -s --- Server listening on 5201 --- Accepted connection from 192.168.20.40, port 50054 [ 5] local 192.168.20.28 port 5201 connected to 192.168.20.40 port 50055 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 22.2 MBytes 186 Mbits/sec [ 5] 1.00-2.00 sec 23.5 MBytes 197 Mbits/sec [ 5] 2.00-3.00 sec 23.4 MBytes 196 Mbits/sec [ 5] 3.00-4.00 sec 23.3 MBytes 196 Mbits/sec [ 5] 4.00-5.00 sec 23.2 MBytes 195 Mbits/sec [ 5] 5.00-6.00 sec 23.4 MBytes 196 Mbits/sec [ 5] 6.00-7.00 sec 23.4 MBytes 196 Mbits/sec [ 5] 7.00-8.00 sec 23.4 MBytes 197 Mbits/sec [ 5] 8.00-9.00 sec 23.0 MBytes 193 Mbits/sec [ 5] 9.00-10.00 sec 23.5 MBytes 197 Mbits/sec [ 5] 10.00-10.05 sec 1.05 MBytes 197 Mbits/sec - - - - - - - - - - - - - - - - - - - - - - - - - [ ID] Interval Transfer Bitrate [ 5] 0.00-10.05 sec 233 MBytes 195 Mbits/sec receiver On 1/30/2020 6:37 PM, Remi Locherer wrote: > > iperf vs. iperf3?
Re: Low throughput with 1 GigE interface
To answer your second question, I did not change any sysctls or other settings on the OpenBSD. The only thing I ran was pfctl -d. My installation guide was: https://github.com/elad/openbsd-apu2 - amd/install66.fs - stty com0 115200 - set tty com0 On 1/30/2020 5:39 PM, livio wrote: > Yes, I tried yet another cable. I hope this gives some credibility: > https://ibb.co/m4mrWt3 > > I now tried with 3 different cables (and vendors). As you can see the > patch cable is brand new. I am also setting up a new Windows 10 notebook > on the right. > > But again, I achieve 940Mbit/s with the exact same setup and FreeBSD 10. > > On 1/30/2020 5:17 PM, Ian Darwin wrote: >> Peter wrote: >> >>> chi# iperf -c beta.internal.centroid.eu >>> >>> Client connecting to beta.internal.centroid.eu, TCP port 5001 >>> TCP window size: 17.0 KByte (default) >>> >>> [ 3] local 192.168.177.40 port 13242 connected with 192.168.177.2 port 5001 >>> [ ID] Interval Transfer Bandwidth >>> [ 3] 0.0-10.0 sec 536 MBytes 449 Mbits/sec >>> >>> ... on an APU1C4, could it be you have a slow switch or router? Any other >>> hardware that could slow yours down? >>> >>> I'm happy with this result, the APU1 is not really a powerhorse. >> That is pretty normal. From an older Intel-cpu laptop with a bge interface, >> to my APU2, both on a TP-Link gig switch, I get >> >> $ iperf -c gw-int >> >> Client connecting to gw-int, TCP port 5001 >> TCP window size: 32.5 KByte (default) >> >> [ 3] local 192.168.42.46 port 21653 connected with 192.168.42.254 port 5001 >> [ ID] Interval Transfer Bandwidth >> [ 3] 0.0-10.0 sec 502 MBytes 421 Mbits/sec >> $ >> >> Again, that's with no tuning. Did you try a different cable? >>
Re: Low throughput with 1 GigE interface
Yes, I tried yet another cable. I hope this gives some credibility: https://ibb.co/m4mrWt3 I now tried with 3 different cables (and vendors). As you can see the patch cable is brand new. I am also setting up a new Windows 10 notebook on the right. But again, I achieve 940Mbit/s with the exact same setup and FreeBSD 10. On 1/30/2020 5:17 PM, Ian Darwin wrote: > Peter wrote: > >> chi# iperf -c beta.internal.centroid.eu >> >> Client connecting to beta.internal.centroid.eu, TCP port 5001 >> TCP window size: 17.0 KByte (default) >> >> [ 3] local 192.168.177.40 port 13242 connected with 192.168.177.2 port 5001 >> [ ID] Interval Transfer Bandwidth >> [ 3] 0.0-10.0 sec 536 MBytes 449 Mbits/sec >> >> ... on an APU1C4, could it be you have a slow switch or router? Any other >> hardware that could slow yours down? >> >> I'm happy with this result, the APU1 is not really a powerhorse. > That is pretty normal. From an older Intel-cpu laptop with a bge interface, > to my APU2, both on a TP-Link gig switch, I get > > $ iperf -c gw-int > > Client connecting to gw-int, TCP port 5001 > TCP window size: 32.5 KByte (default) > > [ 3] local 192.168.42.46 port 21653 connected with 192.168.42.254 port 5001 > [ ID] Interval Transfer Bandwidth > [ 3] 0.0-10.0 sec 502 MBytes 421 Mbits/sec > $ > > Again, that's with no tuning. Did you try a different cable? >
Re: Low throughput with 1 GigE interface
I am happy run the tests with another cable (although the one I was using is brand new). I still receive 940Mbit/s with FreeBSD 12.1 with the exact same setup. The only(!) difference is the physical mSATA SSD (one for OpenBSD, the other for FreeBSD). They have the identical specs though. Results with a different cable: OpenBSD apu.liv.io 6.6 GENERIC.MP#4 amd64 apu# iperf3 -s --- Server listening on 5201 --- Accepted connection from 10.10.1.240, port 64453 [ 5] local 10.10.1.241 port 5201 connected to 10.10.1.240 port 64454 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 16.8 MBytes 141 Mbits/sec [ 5] 1.00-2.00 sec 17.7 MBytes 149 Mbits/sec [ 5] 2.00-3.00 sec 17.5 MBytes 147 Mbits/sec [ 5] 3.00-4.00 sec 17.7 MBytes 148 Mbits/sec [ 5] 4.00-5.00 sec 17.3 MBytes 146 Mbits/sec On 1/30/2020 5:04 PM, Peter J. Philipp wrote: > On Thu, Jan 30, 2020 at 04:50:59PM +0100, livio wrote: >> Hi Peter, >> Thanks for your reply. I would already be quite happy with ~500Mbit/s. >> My test do not involve a switch, just a notebook and the APU through >> a Cat.6a cable. I achieve 940Mbit/s with the exact same setup but >> FreeBSD 12.1 on the APU. >> >> I am happy to change parameters, provide additional logs and run any >> number of tests. I am currently out of ideas. > I go from my APU this way: > > cat5e-->switch (netgear)-->cat5e-->switch (netgear)-->cat6a-->Xeon workstation > > Where the path between the first switch and the Xeon is all 10 GbE but this > shouldn't matter. I'd bet on that you have a bad cable. It's happened to me > before. > > Regards, > -peter > >> $ uname -a >> OpenBSD apu.liv.io 6.6 GENERIC.MP#4 amd64 > I have the same version on my APU and -current on the Xeon. > > >> $ ifconfig em0 >> em0: flags=8843 mtu 1500 >> lladdr 00:0d:b9:41:70:20 >> index 1 priority 0 llprio 3 >> media: Ethernet 1000baseT full-duplex (1000baseT >> full-duplex,master,rxpause,txpause) >> status: active >> inet 10.10.1.241 netmask 0xff00 broadcast 10.10.1.255 >> >> Thank you, >> Livio >> >> On 1/30/2020 4:38 PM, Peter J. Philipp wrote: >>> On Thu, Jan 30, 2020 at 03:43:41PM +0100, livio wrote: >>>> Dear all, >>>> >>>> I am unable to achieve decent throughput with a 1 GigE interface >>>> (Intel I210) on OpenBSD 6.6. When running iperf3 I get around 145Mbit/s. >>>> >>>> The config/setup is: APU2c4, Win10 notebook, no switch, Cat.6a cable, >>>> MTU 1500, 1000baseT, full-duplex, pf disabled, BSD.mp, no custom Kernel >>>> parameters/optimizations. >>>> >>>> With an increased MTU of 9000 (on both devices) the throughput is around >>>> 230-250Mbit/s. >>>> >>>> When running the same test with a FreeBSD 12.1 on the APU I achieve >>>> around 940Mbit/s (MTU 1500). >>>> >>>> The BIOS has been updated to the latest version (v4.11.0.2). The >>>> hardware of the device is: https://pcengines.ch/apu2c0.htm >>>> >>>> dmesg output: >>>> https://paste.ee/p/OeRbI >>>> >>>> Any inputs and help is highly appreciated. >>>> >>>> Many thanks, >>>> Livio >>>> >>>> PS: I ran the same tests on an APU1c4 with Realtek RTL8111E interfaces. >>>> The results were lower - around 95Mbit/s. >>>> https://pcengines.ch/apu1c4.htm >>> Hi, >>> >>> Without any tuning arguments I get: >>> >>> chi# iperf -c beta.internal.centroid.eu >>> >>> Client connecting to beta.internal.centroid.eu, TCP port 5001 >>> TCP window size: 17.0 KByte (default) >>> >>> [ 3] local 192.168.177.40 port 13242 connected with 192.168.177.2 port 5001 >>> [ ID] Interval Transfer Bandwidth >>> [ 3] 0.0-10.0 sec 536 MBytes 449 Mbits/sec >>> >>> ... on an APU1C4, could it be you have a slow switch or router? Any other >>> hardware that could slow yours down? >>> >>> I'm happy with this result, the APU1 is not really a powerhorse. >>> >>> Regards, >>> >>> -peter >>> >>>> PPS: Others also seem to have low throughput. None of the tuning >>>> recommendations I found online substantially improved my results: >>>> https://www.reddit.com/r/openbsd/comments/cg9vhq/poor_network_performance_pcengines_apu4/ >>>>
Re: Low throughput with 1 GigE interface
Hi Peter, Thanks for your reply. I would already be quite happy with ~500Mbit/s. My test do not involve a switch, just a notebook and the APU through a Cat.6a cable. I achieve 940Mbit/s with the exact same setup but FreeBSD 12.1 on the APU. I am happy to change parameters, provide additional logs and run any number of tests. I am currently out of ideas. $ uname -a OpenBSD apu.liv.io 6.6 GENERIC.MP#4 amd64 $ ifconfig em0 em0: flags=8843 mtu 1500 lladdr 00:0d:b9:41:70:20 index 1 priority 0 llprio 3 media: Ethernet 1000baseT full-duplex (1000baseT full-duplex,master,rxpause,txpause) status: active inet 10.10.1.241 netmask 0xff00 broadcast 10.10.1.255 Thank you, Livio On 1/30/2020 4:38 PM, Peter J. Philipp wrote: > On Thu, Jan 30, 2020 at 03:43:41PM +0100, livio wrote: >> Dear all, >> >> I am unable to achieve decent throughput with a 1 GigE interface >> (Intel I210) on OpenBSD 6.6. When running iperf3 I get around 145Mbit/s. >> >> The config/setup is: APU2c4, Win10 notebook, no switch, Cat.6a cable, >> MTU 1500, 1000baseT, full-duplex, pf disabled, BSD.mp, no custom Kernel >> parameters/optimizations. >> >> With an increased MTU of 9000 (on both devices) the throughput is around >> 230-250Mbit/s. >> >> When running the same test with a FreeBSD 12.1 on the APU I achieve >> around 940Mbit/s (MTU 1500). >> >> The BIOS has been updated to the latest version (v4.11.0.2). The >> hardware of the device is: https://pcengines.ch/apu2c0.htm >> >> dmesg output: >> https://paste.ee/p/OeRbI >> >> Any inputs and help is highly appreciated. >> >> Many thanks, >> Livio >> >> PS: I ran the same tests on an APU1c4 with Realtek RTL8111E interfaces. >> The results were lower - around 95Mbit/s. >> https://pcengines.ch/apu1c4.htm > Hi, > > Without any tuning arguments I get: > > chi# iperf -c beta.internal.centroid.eu > > Client connecting to beta.internal.centroid.eu, TCP port 5001 > TCP window size: 17.0 KByte (default) > > [ 3] local 192.168.177.40 port 13242 connected with 192.168.177.2 port 5001 > [ ID] Interval Transfer Bandwidth > [ 3] 0.0-10.0 sec 536 MBytes 449 Mbits/sec > > ... on an APU1C4, could it be you have a slow switch or router? Any other > hardware that could slow yours down? > > I'm happy with this result, the APU1 is not really a powerhorse. > > Regards, > > -peter > >> PPS: Others also seem to have low throughput. None of the tuning >> recommendations I found online substantially improved my results: >> https://www.reddit.com/r/openbsd/comments/cg9vhq/poor_network_performance_pcengines_apu4/ >>
Low throughput with 1 GigE interface
Dear all, I am unable to achieve decent throughput with a 1 GigE interface (Intel I210) on OpenBSD 6.6. When running iperf3 I get around 145Mbit/s. The config/setup is: APU2c4, Win10 notebook, no switch, Cat.6a cable, MTU 1500, 1000baseT, full-duplex, pf disabled, BSD.mp, no custom Kernel parameters/optimizations. With an increased MTU of 9000 (on both devices) the throughput is around 230-250Mbit/s. When running the same test with a FreeBSD 12.1 on the APU I achieve around 940Mbit/s (MTU 1500). The BIOS has been updated to the latest version (v4.11.0.2). The hardware of the device is: https://pcengines.ch/apu2c0.htm dmesg output: https://paste.ee/p/OeRbI Any inputs and help is highly appreciated. Many thanks, Livio PS: I ran the same tests on an APU1c4 with Realtek RTL8111E interfaces. The results were lower - around 95Mbit/s. https://pcengines.ch/apu1c4.htm PPS: Others also seem to have low throughput. None of the tuning recommendations I found online substantially improved my results: https://www.reddit.com/r/openbsd/comments/cg9vhq/poor_network_performance_pcengines_apu4/
