difference between intel gigabit ethernet connection and a atheros qualcomm ethernet connection

[Xtreme Gamer]

What is the difference between intel gigabit and atheros qualcomm ethernet connections?

[dalekphalm]

In a most basic sense? Nothing. However, the Intel controller is much more stable, and has excellent, and WIDE driver support, meaning that it'll work on pretty much any OS, on any platform (Windows - pretty much any version of it, Linux, etc).

 

Most modern platforms use Intel controllers, except of course AMD.

 

Speed wise, probably very similar, though the Intel controller in my opinion is more reliable of the two.

[PBaines]

They are just two different controllers, as above the Intel one will be more reliable.

[Ssoele]

In a most basic sense? Nothing. However, the Intel controller is much more stable, and has excellent, and WIDE driver support, meaning that it'll work on pretty much any OS, on any platform (Windows - pretty much any version of it, Linux, etc).

 

Most modern platforms use Intel controllers, except of course AMD.

 

Speed wise, probably very similar, though the Intel controller in my opinion is more reliable of the two.

 

I can back this claim up completely, Intel NICs are by far my preferred choice.

[Kuzma]

Intel NICs are the best man :D

[looney]

Yes, always go intel if you want to play it safe.

[MG2R]

I've heard Logan from Teksyndicate talk about Intel NICs haveing a better implementation of the protocol stack and having a better buffering system than competing solutions. He backed this up by some numbers related to latency and throughput. I'll see if I can find the video.

 

Can't seem to find it in a timely  manner, sorry

[mark13568]

With Intel Lan, you also have the support to boot from PXE. Overall Intel is a lot more reliable, I have never seen a Intel NIC fail me as a systems admin (servers or stations)

[LAwLz]

I've heard Logan from Teksyndicate talk about Intel NICs haveing a better implementation of the protocol stack and having a better buffering system than competing solutions. He backed this up by some numbers related to latency and throughput. I'll see if I can find the video.

Don't listen to Logan, he barely knows anything about anything (seriously, he just makes stuff up or has ready like half an article on something and then completely misinterprets it, but still talks as if he understands it). If he's talking about for example the IP stack then he is wrong, because that's implemented in the OS (and the UEFI), not the NIC or NIC drivers.

He probably heard about Intel I/OAT which, among other things, decreases the overhead when moving packets between buffers but that requires specially designed hardware that's typically only available on high end server equipment, and not consumer grade motherboards.

 

I like Intel NICs because I've never had a problem with them. I have tried to find benchmarks with different NICs and the only thing I found was that Intel performs just the same as some low end Realtek NIC. What you get with a good Intel NIC compared to a bad low end NIC would be:

1) Higher stability and quality.

2) Probably wider hardware and software support (shouldn't really be an issue if you're just running Windows).

3) It might use a few less CPU cycles, but it wouldn't really matter with a modern CPU.

 

To answer your question OP. If you're just going to use your computer for general stuff most people use their computers for, then you will probably not notice any difference.

[MG2R]

Don't listen to Logan, he barely knows anything about anything (seriously, he just makes stuff up or has ready like half an article on something and then completely misinterprets it, but still talks as if he understands it)

I tend to disagree. Talk to him about tech laws and privacy concerns with them and I would be quite certain that he knows his sh*t better than you. Also, Wendell is a very, very knowledgable guy when it concerns hardware, so I wouldn't just ignore everything Teksyndicate has to say.

[Xtreme Gamer]

Intel NICs are the best man :D

 

Which one thats affordable would you recommend?

[Stomplar]

I agree, Logan knows his stuff. If you dont like him, thats cool too. i respect that.

[LAwLz]

I tend to disagree. Talk to him about tech laws and privacy concerns with them and I would be quite certain that he knows his sh*t better than you. Also, Wendell is a very, very knowledgable guy when it concerns hardware, so I wouldn't just ignore everything Teksyndicate has to say.

OK excuse me. Don't trust anything he says about hardware or software. He has a strong opinion about privacy laws and such though, but for hardware and software, he usually makes things up which sounds good and has no evidence to back his claims up with.

Wendell seems knowledgeable but Logan, no.

 

 

I agree, Logan knows his stuff. If you dont like him, thats cool too. i respect that.

It's not really about "my opinion", it's that he is flat out lying and spreading misinformation which even less knowledgeable people than parrots without doing research. He is making people dumber with his videos because people trust what he says. He is very charismatic, but his knowledge in most areas is next to zero and he tries to make up for lack of knowledge by saying whatever he thinks is true, no matter if it is or not.

 

 

Which one thats affordable would you recommend?

Why do you want a new one? What you are going to use it for and your budget determines which one that's most suitable for you.

[Skanic]

What about an Intel nic not working with a dlink dir615? I was only able to geht 10 Mbit instead Of my 110 mbit . In THW qualcom it worked at full speed.

[Skanic]

Anyone?

[Gob]

Don't listen to Logan, he barely knows anything about anything (seriously, he just makes stuff up or has ready like half an article on something and then completely misinterprets it, but still talks as if he understands it). If he's talking about for example the IP stack then he is wrong, because that's implemented in the OS (and the UEFI), not the NIC or NIC drivers.

 

I don't understand this comment. You put words into his mouth, and then call him wrong. He could be refering to the OSI protocol stack, in which layers, such as the physical or data link layer, are implemented by the NIC and/or NIC drivers.

 

OK excuse me. Don't trust anything he says about hardware or software. He has a strong opinion about privacy laws and such though, but for hardware and software, he usually makes things up which sounds good and has no evidence to back his claims up with.

Wendell seems knowledgeable but Logan, no.

 

Well Logan AND Wendell produce the Tek Syndicate's content. You think Wendell is knowledgeable, don't you think Logan runs his content by Wendell before it is released? I am willing to bet that Wendell would agree with most of what Logan says. So don't you think it is weird you consider Wendell is knowledgeable but Logan not? This sounds like a personal dislike for the guy. I agree that Logan does not appear to have a formal technical background, but I don't think that makes all of his reviews/opinions about hardware invalid (though I am sure he is wrong about some things, everyone is). Most hardware reviewers out there do not have technical backgrounds (including Linus).

[Blade of Grass]

I don't understand this comment. You put words into his mouth, and then call him wrong. He could be refering to the OSI protocol stack, in which layers, such as the physical or data link layer, are implemented by the NIC and/or NIC drivers.

 

 

Well Logan AND Wendell produce the Tek Syndicate's content. You think Wendell is knowledgeable, don't you think Logan runs his content by Wendell before it is released? I am willing to bet that Wendell would agree with most of what Logan says. So don't you think it is weird you consider Wendell is knowledgeable but Logan not? This sounds like a personal dislike for the guy. I agree that Logan does not appear to have a formal technical background, but I don't think that makes all of his reviews/opinions about hardware invalid (though I am sure he is wrong about some things, everyone is). Most hardware reviewers out there do not have technical backgrounds (including Linus).

The 7 layer OSI model is STANDARDISED. Therefor it's the EXACT same on EVERY SINGLE NETWORK DEVICE. IF it wasn't they wouldn't be able to communicate natively.

[LAwLz]

What about an Intel nic not working with a dlink dir615? I was only able to geht 10 Mbit instead Of my 110 mbit . In THW qualcom it worked at full speed.

Sounds strange. Are you using the latest Intel drivers and the latest firmware for the D-Link router? Are you using wireless? Is it Mb or MB? How do you measure the speed?

 

 

I don't understand this comment. You put words into his mouth, and then call him wrong. He could be refering to the OSI protocol stack, in which layers, such as the physical or data link layer, are implemented by the NIC and/or NIC drivers.

Like blade of grass said, the ISO layers are exactly the same no matter what device you use.

I "put words in his mouth" because that was one of the few stacks I could think of which he could be referring to. Yes I think Wendell sounds like he knows what he is talking about. If Wendell approves of everything before Logan says it, then I guess Wendell is more ignorant than I thought. Anyway my point was that please don't use Logan as a source to back your claims up with. He doesn't know what he is talking about half the time. That's all I wanted to say.

[Gob]

The 7 layer OSI model is STANDARDISED. Therefor it's the EXACT same on EVERY SINGLE NETWORK DEVICE. IF it wasn't they would be able to communicate natively.

 

I think you may be confused. Yes, the ISO protocol stack is standardized (that is the definition of a protocol: xmpp, http, IP, CDMA, OSI, etc), but the implementation (hardware used, software used, etc) is not. A protocol or model is just a specification of how devices/programs/etc should interface with each other, but the devices/programs will use different hardware/software to implement the same protocol.

[SteepGnomeKing]

Don't listen to Logan, he barely knows anything about anything (seriously, he just makes stuff up or has ready like half an article on something and then completely misinterprets it, but still talks as if he understands it). If he's talking about for example the IP stack then he is wrong, because that's implemented in the OS (and the UEFI), not the NIC or NIC drivers.

I see you say this a lot but never back it up. I get you don't like him but the bashing seems to go a bit far with no proof as to what your saying. He's a knowledgeable guy who know quite a bit. No one knows everything and can be wrong some times but that's no reason to bash him and say he make stuff up just to make himself sound smart. 

[Gob]

Like blade of grass said, the ISO layers are exactly the same no matter what device you use.

 

Lol, I never said the layers were different on any particular device, that doesn't make sense. I said that the physical and data link layers were implemented by the NIC. Which is correct. I don't know what your point is. 

 

 

I "put words in his mouth" because that was one of the few stacks I could think of which he could be referring to. 

 

I just think it is funny that you are bad mouthing Logan about how he doesn't back up anything he says, and then you don't back up anything you say, but say that he is wrong for something he didn't say.

[LAwLz]

I think you may be confused. Yes, the ISO protocol stack is standardized (that is the definition of a protocol: xmpp, http, IP, CDMA, OSI, etc), but the implementation (hardware used, software used, etc) is not. A protocol or model is just a specification of how devices/programs/etc should interface with each other, but the devices/programs will use different hardware/software to implement the same protocol.

The only part of the OSI stack that the NIC handles is the physical layer, and the MAC sublayer. In the physical layer the difference between NICs is basically what throughput and connects they can handle. Can NIC X and NIC Y both handle gigabit Ethernet? Then they are basically the same here. In the MAC sublayer they basically just add their MAC address and check for frame errors. Not really much they can differentiate on there either.

Conclusion: No matter what NIC you got the OSI stack is the same, and the implementation of it is almost identical. The things that do differ between cards are obvious things like is it FastEthernet or Gigabit Ethernet. I doubt that's what Logan was talking about when he said Intel has a better implementation of the protocol stack.

 

 

I see you say this a lot but never back it up. I get you don't like him but the bashing seems to go a bit far with no proof as to what your saying. He's a knowledgeable guy who know quite a bit. No one knows everything and can be wrong some times but that's no reason to bash him and say he make stuff up just to make himself sound smart.

But I do. I have only been asked to back my claim up once and that was in this thread, and I did back up my claim (and people replied with "I see your point. He really knows nothing."). Qain seems to do the same thing. Want an example? In "inbox 0007: Qain vs Networking" he says that

[...] The 5GHz might be able to penetrate some extra walls which is the only reason why I might recommend this one [...]

Their "networking expert", which claims to have worked in the networking industry for 13 years (when the video was recorded) doesn't know that 5GHz is worse at penetrating solid objects than 2.4GHz... He doesn't know what he is talking about yet he pretends like he does.

Inbox 0025

He doesn't understand what cas latency is yet he pretends like he does, and gives a flat out wrong explanation. CL is the delay (measured in clock cycles) between the memory controller issuing a command, to the RAM accessing that memory column. Since it is measured in clock cycles, 1 CL on 1600MHz RAM is longer than 1 CL on 2133MHz RAM. The whole "you want low CL for video editing and high MHz for gaming" is flat out wrong as well. If anything, the opposite would be true since you want very high bandwidth when working with video, and having to wait 1ns longer for the file which will take several hours to render isn't a big deal.

 

 

He does not understand what causes input lag. In the video he says that because the monitor only has DVI, it has no input lag (he repeats that it has no input lag at all several times) which is simply false. He does not understand how a monitor works if he think that having a single connector means no input lag.

 

He doesn't do any reason tests in the video even though it's suppose to be a review. Where are the gamut tests? Where are the color balance tests? Where are the black levels tests and where is the brightness test? What about contrast? What about gamma? He just reads what the box says which is something that's OK for an unboxing, but for a proper review that's disgraceful.

 

He says he wants to make a note that it's a backlit panel at 3:15, as if this is not something all monitors are. All LCD monitors are backlit.

 

At 4:48 he says that 2560x1440 makes the pixels so small that you can turn off AA. 2560x1440 on a 27" monitor has a PPI of ~108. Most desktop monitors are about 100 PPI (some are slightly below, some are slightly above) so the PPI is nothing special.

 

For his

vs

videos please see my blog because I've already written in detail all the things wrong with those videos. (Warning, lots and lots of text to explain everything wrong in those videos).

Blog post about video number 1

Blog post about video number 2

 

 

The Tek 0041

Where do I begin...

The biggest problem isn't that it takes forever to encode. The problem is that it takes forever to decode, since there isn't any hardware acceleration support (and probably won't be for quite a while). Not even that is the biggest problem. The biggest problem is that the standard isn't complete yet.

You don't measure video quality in 1080p and 4K. You can't just say that H.264 will be 1080p quality, and H.265 (from here on called HEVC) will be like 4K quality with the same file size. I can easily make a 4K file that looks worse than a 1080p file, and I can make a 4K file that's smaller than a 1080p file.

HEVC can be implemented as open source, just like VP9. X265 is an open source HEVC encoder, and OpenHEVC is an open source decoder. The difference between HEVC and VP9 (which is the codec Logan was thinking of but didn't know the name) is that VP9 is royalty free. There is a big difference between open source and royalty free.

 

 

That build is awful. It is in fact so awful that the parts aren't even compatible. If you bought what he recommended, you would end up with a build that wouldn't fit together. Things that are wrong with his build:

• Wastes money on high frequency RAM. It will make next to no difference if you get 1600MHz or 2133MHz RAM.
• He recommends a Kingston SSD, as well as OCZ and ADATA SSDs, but completely ignores Intel, Crucial and Samsung SSDs. He says that he recommends them because none of the ones he have used have failed but staitiscs such as this shows that OCZ is by far less reliable than Intel and Samsung. He doesn't really have that much personal experience with them either. Basically it's just "oh it haven't failed on me and I've used it for like 2 months, so it must be good!".
• EVGA fanboy. Their cards aren't any more "top notch" than any other rebranded Nvidia reference card. Their warranty is good, but saying that they are top notch when basically all other cards are exactly the same (except for the stickers) is fanboy-ish.
• The case and cooler he recommends does not work with the cooler he recommends. The NZXT Tempest 410 can not fit the Kraken X60.

 

He thinks that thermal mass == weight, which it isn't. He was just talking about the weight of the heatsink yet he throws in "thermal mass" to sound smarter than he really is (actually makes him sound dumb because he uses scientific terms in the wrong way).

He changed the thermal paste from the stock one which I am going to assume performs about the same as the Xigmatek Freezing Point, or Zigmatek PTI-G4512. If that is true, then the "good stuff", aka the Tuniq TX-2 Logan puts on is pretty much the same as stock thermal paste (PTI-G4512 is actually better than Tuniq TX-2 according to hardwaresecret's tests.

The lapping he did on the H100 looks absolutely horrible. I would not be surprised if that's the reason why the Xigmatek outperformed the H100.

I find it very, very, VERY hard to believe that it outperforms the H100 by the way. That would mean it also outperforms the Noctua D14, which is just flat out bullshit. It's simply not true.

I got more things written down if you want. I think this is enough for now though to prove that Logan often has no idea what he is talking about. Just says whatever sounds good.

Oh and yes, I do in fact dislike Logan (you might have been able to tell hehe). I think he is a pretentious douchebag who spreads misinfromation.

[Gob]

The only part of the OSI stack that the NIC handles is the physical layer, and the MAC sublayer. In the physical layer the difference between NICs is basically what throughput and connects they can handle. Can NIC X and NIC Y both handle gigabit Ethernet? Then they are basically the same here. In the MAC sublayer they basically just add their MAC address and check for frame errors. Not really much they can differentiate on there either.

Conclusion: No matter what NIC you got the OSI stack is the same, and the implementation of it is almost identical. The things that do differ between cards are obvious things like is it FastEthernet or Gigabit Ethernet. I doubt that's what Logan was talking about when he said Intel has a better implementation of the protocol stack.

 

First of all, the MAC sub layer is a sub layer of the Data Link Layer. So saying it also includes the data link layer is clearly correct.

Reference: http://msdn.microsoft.com/en-us/library/windows/hardware/ff571073%28v=vs.85%29.aspx

Second, stop saying the OSI stack is the same across NIC's, clearly it is, it is a protocol stack, you are being redundant.

Third, almost identical, is not identical.

Fourth, a realtek networking chipset is different from a marvell networking chipset which is different from an intel networking chipset (not to mention the different chipsets within each brand). These are all DIFFERENT implementations (ie different hardware) of NICs conforming to the OSI standard.

[LAwLz]

First of all, the MAC sub layer is a sub layer of the Data Link Layer. So saying it also includes the data link layer is clearly correct.

Reference: http://msdn.microsoft.com/en-us/library/windows/hardware/ff571073%28v=vs.85%29.aspx

Yes I know. Not sure why you're pointing this out because I never said you were wrong on that part. I am not sure if you're missing the point deliberately or if you didn't understand my point. In the MAC sublayer it adds the MAC address and checks for frame errors. In layer 1 it simply converts the frame into whatever signal is suitable for the medium (wireless, electrical pulses and so on). If both cards are gigabit cards, then the physical layer part of the OSI stack will be exactly the same (or at least 99% the same). The way to detect errors is also in the protocol so they can't differentiate themselves there.

 

 

Fourth, a realtek networking chipset is different from a marvell networking chipset which is different from an intel networking chipset (not to mention the different chipsets within each brand). These are all DIFFERENT implementations (ie different hardware) of NICs conforming to the OSI standard.

How? Please explain how they differ in a significant way. I am sorry, but I simply can't see how they could differ in any meaningful way. Different ways to detect errors? Nope. Different way of converting the frame to signals? Nope. Different way of adding the MAC address? Nope. I assume you got some solid evidence to back the ridiculous claim up with, right? You haven't been arguing in blind faith this whole time, right?

[Gob]

Yes I know. Not sure why you're pointing this out because I never said you were wrong on that part. I am not sure if you're missing the point deliberately or if you didn't understand my point. In the MAC sublayer it adds the MAC address and checks for frame errors. In layer 1 it simply converts the frame into whatever signal is suitable for the medium (wireless, electrical pulses and so on). If both cards are gigabit cards, then the physical layer part of the OSI stack will be exactly the same (or at least 99% the same). The way to detect errors is also in the protocol so they can't differentiate themselves there.

 

Because your last statement was phrased in an argumentative way. Yes, it is hard to understand what is written in a textbook. The reason I am ignoring this is because it is not relevant to what I am saying, nor was I ever arguing about any of these points.

 

How? Please explain how they differ in a significant way. I am sorry, but I simply can't see how they could differ in any meaningful way. Different ways to detect errors? Nope. Different way of converting the frame to signals? Nope. Different way of adding the MAC address? Nope. I assume you got some solid evidence to back the ridiculous claim up with, right? You haven't been arguing in blind faith this whole time, right?

 

No, I cannot give you specifics on networking chipset hardware architectures because I am not an electrical engineer. But you don't need a degree to know that they are different. You are thinking at a high level. That is like saying an intel cpu and an amd cpu are exactly the same, because they are cpu's and they both implement the x86 instruction set. But to end this ridiculous conversation, you're right all NIC's have the exact same hardware, they all use the same drivers, and they all have the same stability.