Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Which chipset for NIC ? Which switching?

Status
Not open for further replies.

CMir

Programmer
Mar 6, 2003
26
0
0
US
I've been using NIC with Realtek cards from several companies ( Edimax, NPG, Longshine ). Someone told me that those from Lucent or Broadcom were better and that the NIC from 3COM, Level1 or Micronet were better than the NIC's from Edimax, Longshine,... .

The same doubt applies to the switching hubs or hubs in general.

I'd like to have an opinion concerning if it is worth expending more money in 3Com or Longshine taking into account the differences in network performance. What are really those differences ? Do we really notice them, e.g., in a P2P network? Or only on a server-based network.

Thanks.
 
Hi,

I have found that there are definite speed benefits when the make of card matches the make of the switch at 100Mbps or faster, but 10Mbit only cards it doesn't make much of a difference.

A switch will be faster than a hub because the major difference between switches and hubs is that switches only broadcast data on the appropriate ports, whereas hubs broadcast it on all ports, regardless of whether it is needed.

John
 
As I understood it's important to have the same company providing the switching hub and the NIC. My doubt is about the chipset or even the differences between,e.g., NGS and 3Com in terms of performance. Do you consider it important or better to use 3Com instead of NGS or Lonsgshine? Do you notice any difference ?

Thanks.
 
I found that when the make of the switch and NIC matched, there was a noticeable performance improvement.
My experience is:
3Com embedded & PCI NIC's to a 3Com switch, when I connected some PC's with Intel Pro 100 series NIC's to the same switch, there was a noticeable slowdown over the 3Com equipped machines, despite the Intel equipped machines having a superior specification, yet they both connected at 100Mbps according to Win2K Pro.
It may have something to do with full duplex operation, as the 3Com cards when switched to 100Mbps operation automatically enabled the full duplex, whereas when they were hard set to 10Mbps this was switched off and couldn't be enabled.
I couldn't find a way of checking whether the Intel NICs had this on or not, or a way of changing this setting either.

Hope that this helps.

John
 
You might look for a networking or server forum for more detailed info on NIC and chipset information on ethernet cards.

Must be fun trying to benchmark a network card. Seems like the cable would be important too.

If you do not like my post feel free to point out your opinion or my errors.
 
I almost forgot: in the same LAN we had a couple of machines with el-cheapo D-Link 10/100 cards, and they outperformed the Intels connecting to the 3Com switch - but unfortunately the number of machines wasn't good enough to be statisically significant (5 D-Link machines vs 200 3Coms and 100 Intels or something like that).

John
 
For ceh4702

Which forums would you advise me to go to ?

For jrbarnett

Thanks for your information. The main issue of my initial post was to get the real experience from those like you, Mainly it was intended to evaluate really the importance of the manufacturer concerning the network performance - in some hardware issues this is definitely important ( motherboards, processors, memories ) on others there may be given too much relevance to the manufacturer if we think of the real differences concerning all the other manufacturers. You helped me developing a point of view about this.

Thanks.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top