Bug 50834

Summary:	E100 driver exhibits poor performance in IA64 systems
Product:	[Retired] Red Hat Linux	Reporter:	Clay Cooper <clay_cooper>
Component:	kernel	Assignee:	Arjan van de Ven <arjanv>
Status:	CLOSED RAWHIDE	QA Contact:	Brock Organ <borgan>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	7.3	CC:	dale_kaisner, john_hull, matt_domsch, michael_e_brown, notting
Target Milestone:	---
Target Release:	---
Hardware:	ia64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2001-11-13 13:58:17 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Clay Cooper 2001-08-03 19:31:33 UTC

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.4.2-2 i686; en-US; m18)
Gecko/20010131 Netscape6/6.01

Description of problem:
On a pe7150 w/ fairfax beta3 installed, bios X15, onboard nic controlled by
e100 driver.

Nttcp-1.46 run between client and e100 controlled onboard nic on IA64
platform consistently reports speeds of 25-45MB/s on a 100MB network
connection.

Eepro100 on IA64 fails completely (see bug 50829)

Both the e100 and eepro100 driver when run on an I386 platform against the
same client produce speeds of 90-95MB/s on a 100MB network.

So:

                 IA64                    I386
e100             25-45MB/s               90-95MB/s
eepro100         Dies                    90-95MB/s

How reproducible:
Always

Steps to Reproduce:
1.Install fairfax beta3 on pe7150 w/ e100 controlling onboard nic
2.Start nttcp-1.46 between client and onboard on a 100MB network
3.Observe speed.
	

Actual Results:  Speeds of 25-45MB/s		

Expected Results:  Speeds of 90-95MB/s

Additional info:

Comment 1 Matt Domsch 2001-08-10 18:04:39 UTC

Bug reported to Intel (Walker, Timothy E [timothy.e.walker]) on 10 
August 2001.  Tim said he'd pass this on for investigation by the appropriate 
driver team.

Comment 2 Need Real Name 2001-08-16 10:50:09 UTC

What is the e100 driver version?

Comment 3 Clay Cooper 2001-08-16 12:42:26 UTC

E100 version 1.6.5, in the beta3 IA64 kernel (2.4.6-3.1)

Comment 4 Michael K. Johnson 2001-09-26 23:29:43 UTC

1) Would be interesting to know if eepro100 still dies on these systems
2) RC1 has a new version of the e100 driver that you can test

Comment 5 Clay Cooper 2001-10-08 14:53:56 UTC

E100 performance has not improved with kernel 2.4.9-0.18smp.

Comment 6 Matt Domsch 2001-10-17 19:31:30 UTC

Which I believe has e100 driver v1.6.22.  I've asked that this be included in 
the Dell/RH/Intel 3-way call (which isn't on for this week after all).  Does 
anyone know if Intel is actively working this issue?

Comment 7 Michael K. Johnson 2001-11-08 16:11:56 UTC

The problem here is swiommu performance -- the swiommu is a hack which
doesn't actually perform very well at all.  We are working on a patch
which gives the same functionality as swiommu with much higher performance
and our initial tests look very good, with some bugs still to be worked
out.

Comment 8 Clay Cooper 2001-11-13 13:58:11 UTC

w/ qa1108 (2.4.9-13.3smp) e100 driven nics reporting speeds of 95MB/s sustained
running nttcp.

Comment 9 Arjan van de Ven 2001-12-03 14:10:17 UTC

95Mbit sustained sounds like fixed to me...