Are PCI NIC always better than integrated ones? I'm on Gigabit

>>56042094
Without a doubt your router is your bottleneck, not your NIC.

Your plastic chinkshit router probably caps way under 1GB/s. There's very few that are capable of true wirespeed transfer, and those are more of an "enthusiast" class like the ubiquiti edgerouter.

>>56042143
I know it's capable of routing 1Gb/s I know because it's the ISP's router (chipset BCM43570), so many customers have it and they report speetest of 1Gb/s

>>56042175
That model is a wireless radio. That chip has nothing to do with hard ethernet.

>>56042175
That router caps at 866 MBits if you look at the homepage.

>>56042216
My bad, the chip is https://www.broadcom.com/press/release.php?id=s862249

>>56042175
bruh are you dumb?

>>56042280
This doesnt really say much. Unless you have a wired ethernet connection on at least cat6a, dont expect gigabit speeds.

And even then, unless you have a dedicated machine in a datacenter at your ISP's hub, dont expect exact gigabit speeds.

>>56042319
Torrenting from BTN, PTP and doing speedtests on a 40Gb/s ISP server.
Did I mention I did test on my LAN as well?

Unless you have some shitty GAYMEN mobo or cheap ass chink shit your NIC isn't the bottleneck.

Guess what? I can OC my CPU to 4.2 GHz but it doesn't mean shit because my mobo isn't stable over 3.0 because it's a peice of shit.

Buy this:
https://www.amazon.com/Ubiquiti-EdgeRouter-Advanced-Gigabit-Ethernet/dp/B00YFJT29C

and this:

https://www.amazon.com/Ubiquiti-Networks-Enterprise-System-UAP-AC-LITE-US/dp/B015PR20GY

Congratulations, you've got yourself a great router and AP that BTFOs Netgear/Linksys/Asus consumer shit

File: lan speed test.png (16KB, 412x437px) Image search: [Google]

16KB, 412x437px

I can get up to 979Mbps throughput on my two integrated NICs through a TP-Link router.

Fair to say that for Gigabit, at least, PCI-e NICs are a meme.

>>56042094
the bottleneck is 100% your shitty ISP provided router and not your NIC

>>56042725
I shall use this to determine what are my max integrated NIC speeds are.

>>56042094
No, not all Intel NICs support ToE, or any other offload. Most of the popular ones do, however. That said, the difference between something like a Realtek NIC and an Intel won't likely be seen as throughput but stability and features like ToE. The bottleneck in your situation is probably going to the gateway appliance given to you. Try connecting directly to your ONT and see what happens.

>>56042725
NIC cards are not a meme, but they won't help the average person if they have decent integrated NICs.

File: LAN Test.png (17KB, 412x437px) Image search: [Google]

17KB, 412x437px

>>56042913
>>56042725
Ok, here is the LAN test

>>56042094
There's so many other factors that come into play. ISPs plans are rated theoretical speeds under ideal circumstances. You may be on a 1gb/s plan but you may live further away from the exchange so you experience slightly slower speeds.

>>56042319
>CAT6a
>Is required for gigabit
Its required 10GbE

>>56042725
>testing GbE
>with a 20MB file
wut

>testing network bandwidth
>with SMB on a dekstop
wut

>>56045513
>You may be on a 1gb/s plan but you may live further away from the exchange
This nigga ain't on DSL, retard.

>>56042094
Nope, most on-board NICs are connected via PCI-E too

>>56045630
>Its required 10GbE
It's required for 10GbE at 100M. You can use Cat 6 for short 10GbE runs.


Use iperf, monkeys.

>>56045639
Possibly incorrect wording involving exchange, but they do only offer theoretical speeds under ideal circumstances. I'm on a 100/40 fibre plan, i get 96. When it rains i get 89. Other factors come into play

>>56045630
>>with SMB on a dekstop
Why? I have done a speedtest with the same pogram on two laptops directly connected to each other with a cat 5e cable. One of the laptops had a small shared ramdrive and the other one did the speedtest on that. I have also done another speedtest with iperf3 on the same two laptops and the results were nearly identical.

By the way a direct connection should theoretically be faster than a connection over a switch, right?

>>56045683
So what i'm getting at is it's most likely his cheap shit ISP provided router, OR infrastructure.

>>56045639
My buddy lives in an apartment complex with a 100/40 fibre plan and only experiences 60! Peak times as low as 30 he's told me.

>>56045683
>>56045797
That's because there is contention on the node. The uplink is probably 40Gbps at best, so if everyone has a 1Gbps plan and there's more than 40 people on the node...you're going to have problems when everyone wants to use it.

File: Capture.png (138KB, 562x516px) Image search: [Google]

138KB, 562x516px

OP here, funny thing is since the Windows 10 anniversary update I have +100Mbps on speedtests

>>56045683
>When it rains i get 89. Other factors come into play
True fiber internet (FTTH) is not affected by rains and shit

>>56045700
Bls resbond

>>56045683
You must be using FTTLA with a shitty coax end...
Real fiber does not give any fuck about rain.

>>56046039
>Erreur de division par zéro

Tu utilises IPoT, anon?

>>56047900
J'utilise https://lafibre.info/tester-son-debit/test-neutralite/

>>56047811
>>56045700
>Why?
because SMB can use compression

>By the way a direct connection should theoretically be faster than a connection over a switch, right?
latency? yes. throughput? no. if your switch isnt a piece of shit, it shouldnt matter