At 04:11 PM 10/17/2011, Henry Goffin wrote:
We rely on a stable OS timer, but we don't rely on CPU cycle counts
or directly query the hardware clock. I don't know if that answers
your question; I'm not familiar with these setups. We trust that
when the OS gives us a high res timer value, that
At 10:48 AM 7/5/2011, TRISTAN MARLER wrote:
Content-language: en
Content-type: multipart/alternative; boundary=--c07762eb3cd45620f9af
If you run fps_max 66 your hit detection will suffer. I've competed
on servers which were 200ish compared to a 1000fps server and
noticed a measurable
At 03:13 PM 3/12/2011, Munra wrote:
What's resource usage on say an 600 fps server with HPET enable compared to
it disabled. Or will resource usage be simply based on the configuration
of the game server itself?
You don't need high FPS. You only need 66fps with 66 tickrate (hz) on
the
At 10:07 AM 11/10/2010, ohn...@maxpowergc.com wrote:
OK, so I'm wondering if some people could share their experiences with me
on this issue. I'm running a TF2 srcds server... and I've tried it on
Ubuntu 10.10, Gentoo, Windows Server 2008, and Windows Server 2003, yet I
get the same results
At 06:30 PM 11/10/2010, ohn...@maxpowergc.com wrote:
Are you saying its going to always drop below the tick rate on a VDS,
direct hardware, or both?
If the FPS drops below the tickrate, then there will problems. If
you've switched out OS's and it's all doing the same on the hardware,
then
At 01:37 AM 5/12/2009, Yaakov Smith wrote:
Does SrcDS make any use of a graphics card, or does it just use the
processor?
Interesting idea though. Using the video card's processor to do fast
math calculations.
___
To unsubscribe, edit your list
At 10:26 AM 3/2/2009, CLAN RCR wrote:
Hey Midnight, I looked up HPET and this is what Microsoft says about it.
The High Precision Event Timer (HPET) was developed jointly by Intel and
Microsoft to meet the timing requirements of multimedia and other
time-sensitive applications. Originally, the
At 05:11 PM 3/2/2009, Blood Letter wrote:
I see you have experience with Nvidia's 690G.
LOL.
Yup. Nvidia chipsets are pretty much junk.
Turning on HPET breaks APIC interrupt routing.
Date: Mon, 2 Mar 2009 16:10:50 -0500
To: hlds@list.valvesoftware.com; hlds@list.valvesoftware.com
At 06:32 PM 2/7/2009, Chad Austin wrote:
Does anyone have experience using VMware with source engine game
servers? I have heard it has timing issues and would not be acceptable
for scrim servers due to lag spikes, is this true?
-Chad
Don't. Clocks inside of a VM don't sync up, and you'll see
At 06:21 AM 2/8/2009, Steven Hartland wrote:
The timing under VM's is not reliable enough for games. Basically time
can and DOES go backwards as the hardware timer is also virtualised.
This is much less apparent in single core VM's but in multi core VM's
its a major problem.
The newer VM's are
At 05:14 PM 11/21/2008, [EMAIL PROTECTED] wrote:
Hey Guys I am renting a Ded box and choose to have 2008 installed instead of
2003
I have noticed that the games run great however the OS seems to be very
resource hungry. Does any one have any good infomation or resources about
how to trim down
At 04:53 AM 10/25/2008, [ЯÑR] The-/iller wrote:
I use and have been using metmod plugins and windows exes for years and
they do theyre job, but what do you need to get higher than 512?
Decompile the fpsbooster hacks and increase the hack?
You can't on windows 2003. On windows 2008 you can,
At 06:44 AM 10/3/2008, Saint K. wrote:
Thanks.
The servers they referred too were prolly shared servers with overbooking
(from what I understand). The machine I want to run it off is dedicated our
machine. I wonder if these lag spikes are a result of running it in ESX, or
if a diff virtual
At 07:37 PM 9/11/2008, Nephyrin Zey wrote:
I have a Xeon 3210 Server. Running linux x64, I can run one 32x TF2
server per core with SourceTV enabled and get decent FPS. It uses
nearly 100% of a core though :(. Turning off SourceTV and some of the
'boosting' hax I use, the same 32x server uses like
At 06:52 PM 8/7/2008, Jake E wrote:
I've tried everything I can find on Google and I can't cap it... it is
currently running at over 250fps. How to I set a limit to avoid a crash?
+fps_max 125 +alias fps_max echo fps_max is currently disabled
Gary Stanley - [EMAIL PROTECTED] - [EMAIL
an
ioport read, which is very expensive.. I'm not sure on this; but that
would explain why I see so much jitter from sleep() calls
Windows 2008 uses HPET to service interrupts and timing, and it's
faster than the former.
Gary Stanley - [EMAIL PROTECTED] - [EMAIL PROTECTED]
http
about this?
probably synflood. Put up a sniffer and see.
Gary Stanley - [EMAIL PROTECTED] - [EMAIL PROTECTED]
http://leaf.dragonflybsd.org/~gary
Put your hand on a hot stove for a minute, and it seems like an hour.
Sit with a pretty girl for an hour, and it seems like a minute.
THAT'S
has
useless 'tweaks' that do absolutely nothing.
Windows 2008 services timing with HPET and provides serverside FPS up to 1000.
-- Gary Stanley ([EMAIL PROTECTED] | [EMAIL PROTECTED])
Q: Because it reverses the logical flow of conversation.
A: Why is top-posting (putting a reply at the top
./hltv +connect localhost:27035
-
#!/bin/bash
LD_LIBARARY_PATH=$PWD:$LD_LIBRARY_PATH
screen -A -m -d -S cs16tv ./hltv +connect localhost:27035
-- Gary Stanley ([EMAIL PROTECTED] | [EMAIL PROTECTED])
Q: Because it reverses the logical flow of conversation.
A: Why is top
950fps.
-- Gary Stanley ([EMAIL PROTECTED] | [EMAIL PROTECTED])
Q: Because it reverses the logical flow of conversation.
A: Why is top-posting (putting a reply at the top of the message) frowned upon?
Let's not play Jeopardy-style quoting
source updates fixed it, and I don't think that fix for source
made it in to TF2 :)
-- Gary Stanley ([EMAIL PROTECTED] | [EMAIL PROTECTED])
Q: Because it reverses the logical flow of conversation.
A: Why is top-posting (putting a reply at the top of the message) frowned upon?
Let's not play Jeopardy
At 06:51 PM 7/12/2007, HLDS User wrote:
The onboard network card failed on my CS dedicated server. It has been
temporarily replaced by a 3com EtherLink XL (3C905B-TX).Is this card
suitable for supporting a high bandwidth server, or do others have
better recommendations?In the past,
At 05:56 PM 7/5/2007, HLDS User wrote:
This has probably been hit up several times, but I'll rehash it, flame
me later ;)
I just upgraded the processor on our dedicated cs server from a single
core AMD 3700+ to a Opteron 165 with dual cores, the OS is Windows
Server 2003, SP1.
Anyone that is
At 05:30 PM 6/17/2007, Steven Hartland wrote:
How about you read the contents which clearly states how to
unsubscribe from the list you subscribed to in the first place
instead of shouting at others for your own ineptitude.
Regards
Steve
Nothing like coming back from a vacation to read
04:27 AM 5/3/2007, Steven Hartland wrote:
You might be wise to raise to post this as a new thread so
it doesn't get lost in the noise of this one. If its Linux
including a thread dump of the crash would also be helpful
to the dev's in fixing the crash.
Steve
I've always found it helpful to
)
Velocity Servers, Inc
[EMAIL PROTECTED]
http://leaf.dragonflybsd.org/~gary
___
To unsubscribe, edit your list preferences, or view the list archives, please
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds
At 04:20 AM 3/19/2007, Steven Hartland wrote:
Gigabit Nick wrote:
but (going back to the original start of this thread fork) if there
was a solution to limit the manipulation of the high precision timer
to srcds and not make it system wide it would be good for us.
No and there never will be
At 12:35 PM 3/15/2007, Whisper wrote:
--
[ Picked text/plain from multipart/alternative ]
See the quoted post below:
Is there a way to do this for Windows 2000 or 2003 that is not a huge drama?
It would be fantastic to be able to run our Windows boxes at the exact FPS
we want to run them at,
quirks to get things to work properly.
G. Stanley
Engineering (AS36352)
Velocity Servers, INC
[EMAIL PROTECTED]
http://leaf.dragonflybsd.org/~gary
___
To unsubscribe, edit your list preferences, or view the list archives, please
visit:
http
At 12:30 AM 3/7/2007, Dan E wrote:
What should I be looking for in Process Explorer?
I didn't want to disable ACPI, but that's the only way that I've found to
achieve 1000FPS. There was another thing that I read about if you're
running a pure Intel based system (CPU and chipset and board), it
At 02:22 AM 1/24/2007, RMaioroff wrote:
Something also to consider is that many of the peers (routers) on that route
probably are configured to give BGP operations priority over ICMP traffic
(ping) traffic. So if it is indeed busy as Dan pointed out, then more BGP
overhead exists. If you get a
Looks like level3 is having a fit, might be return path issue.
traceroute to 74.130.80.1 (74.130.80.1), 64 hops max, 40 byte packets
1 ge0-1-1000M.ar1.ChicagoEquinix.velocity-servers.net
(66.225.194.1) 0.615 ms 0.577 ms 0.604 ms
2 unknown.ord.scnet.net (205.234.205.17) 0.480 ms 0.357 ms
That's nice. At least I have control over my network, do you?
At 11:25 PM 1/23/2007, J. Laws - Hi-Definition Gaming wrote:
Or you could stop using a Playskool network and have a return path along the
lines of this:
___
To unsubscribe, edit your
Make sure your duplex settings are correct; you could try turning off
flow control and disable IRQ mitigation (if your eth card supports it)
At 09:23 AM 1/19/2007, Andrius Pirus wrote:
--
[ Picked text/plain from multipart/alternative ]
What could be the problem if almost all (but 80%) of my
Netcom had a massive outage in '96 that lasted almost the same
duration as this valve outage.
At 11:06 AM 12/18/2006, -Mike- wrote:
There are far too many 90% empty datacenters practically sitting on
top of major exchanges down here in the SF Bay Area (and all over
the US) for Valve to have
I believe there are instructions on the mani site that explain
everything, and they (the instructions) are pretty simple and/or easy.
At 08:47 PM 11/22/2006, mike morache wrote:
--
[ Picked text/plain from multipart/alternative ]
what version and where do i get it? ilooked and there are several
Works good :)
What does hlds use for timing? queryperformancecounters or reading tsc/rdtsc?
At 01:53 PM 10/26/2006, Alfred Reynolds wrote:
All it does it stop the HLDS process from specifically picking Core 0,
now it leaves it to windows. If you still see your hlds processes
bunching up then
I am. I think it's a peering problem with one of their master
servers. After a few restarts everything came online.
At 12:03 PM 9/15/2006, Ryan Brady wrote:
This is a multi-part message in MIME format.
--
[ Picked text/plain from multipart/alternative ]
for some reason starting a few days ago
Well said. :)
Hard coding offsets is bad, generally, unless you are lazy.
At 01:55 AM 8/25/2006, Alfred Reynolds wrote:
The problem is that these plugins are hacking into hard coded offsets in
the game server, one tiny code change from us and the plugin blows up.
The solution is for them to
E.. Multiple timer enforcements are useless.. you only need 1, not 1
At 10:28 AM 8/13/2006, [xXx] con wrote:
1000 fps will still only deliver a constant 512 fps
running multiple timer enforcement helps when you have multiple servers in
one box.
High resolution timer support in windows is only supported with 1
processor, standard pc as Computer type..
At 06:23 AM 8/12/2006, Whisper wrote:
--
[ Picked text/plain from multipart/alternative ]
Was this on a single CPU or Dual CPU or greater server?
Can you install a non-acpi edition on
That has nothing to do with the original topic.. ACPI HAL manipulates
things like sleep() granularity, therefor you won't see good
performance.. only way to really get high precision sleep timers etc
in SMP is to use a real time OS, or don't use microsoft windows anymore.
It's an OS issue, not a
It's not 1ms, it's more like 1.95 (1000 / 1.95 = 512) .. only way to
increase it's granularity is to switch to a HAL that doesn't do ACPI
(which is only on single processor computers)
At 12:41 AM 8/11/2006, Frazer wrote:
I use the Multimedia SDK timeBeginPeriod and timeEndPeriod APIs. These
, for the application or device
driver.
No matter what, it seems pretty clear that the practical ceiling for srds is
512fps.
Frazer
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Gary
Sent: Friday, August 11, 2006 6:48 AM
To: hlds@list.valvesoftware.com; hlds
, it seems pretty clear that the practical ceiling for srds is
512fps.
Frazer
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Gary
Sent: Friday, August 11, 2006 6:48 AM
To: hlds@list.valvesoftware.com; hlds@list.valvesoftware.com
Subject: RE: [hlds] ogsWatcher
At 11:01 AM 5/17/2006, Kevin Ottalini wrote:
HLDS (HL1 servers) can easily and with little burden run at either ~500 fps
or ~1000 fps. There is no control over the actual maximum FPS since it is a
motherboard chipset related issue.
That isn't true. It's the software (OS). Windows is also not
It's just a game...
At 07:40 PM 4/14/2006, RaynServ wrote:
--
[ Picked text/plain from multipart/alternative ]
Now if I knew the reason why would I be making a support request asking why?
WTF kind of official response is this? Am I suppose to maintain maturity
in a world of children? Give me a
TSC has not ever really been SMP safe, but I suppose Microsoft hacked
a way to do it :)
I wish sleep(1) would actually sleep for 1, not to 1.95
At 06:07 PM 4/8/2006, Steven Hartland wrote:
Just trying to determine if using the external utils would have the
same effect
as removing the recently
I'm seeing this on all of my servers after I updated from version 24
to 29.. they don't seem to move off of cpu0 unless I deselect it.
At 10:03 PM 4/4/2006, David Harrison wrote:
Bit of a weird problem - we have four CS (1.6) servers running on one
machine (dual Opteron, dual core). For some
the programatic binding to CPU0
and work around the windows timing problem in another (hackier) way. I
don't have an ETA for that update, in the meantime you can use taskman
to force the whole process to another CPU (does anyone have a command
line tool to do this?).
- Alfred
Gary wrote:
I'm seeing
are going to remove the programatic binding to CPU0
and work around the windows timing problem in another (hackier) way. I
don't have an ETA for that update, in the meantime you can use taskman
to force the whole process to another CPU (does anyone have a command
line tool to do this?).
- Alfred
Gary
sleep() returns about 1.95 on windows.. (1000 / 1.95ms = 512fps)
usleep() on *NIX returns closer to 1 depending on the OS.
It's a bug in HAL i believe on windows. If you don't use ACPI, it
doesn't use ACPI as a timer anymore and uses TSC..
Only way to get super high frames is to a.) not use
Windows is -not- an realtime OS... you will never get a deterministic
behavior to the way you want it. ACPI's default time quantum is crap,
TSC is better but it doesn't work for SMP. So you have to decide
which is better, ACPI with SMP and 512fps maximum, or TSC-like with
single processor support
You should use qstat and wrap it with awk.. simple..
At 12:32 AM 2/2/2006, MOFCLAN.NET :: Bush wrote:
This is a multi-part message in MIME format.
--
[ Picked text/plain from multipart/alternative ]
Does anyone know of a PHP script or be willing to write me a PHP
script that will grab JUST
I've noticed limelight is doing weird metrics to their network, so I
just adjust my routing and the servers appear to operate just fine now.
Before I had problems with servers not getting steam connected, vac2
init problems etc. :)
Maybe you should consider getting a carrier that doesn't use
Make the directory read only :)
I'd run prime95 to see if you get any errors. I don't use tyan
products anymore because of the high failure rate/issues (the i8254
was acting strange on the last tyan server I had)
Curious to know what type of SATA controller you have on that board,
though. I
I know you have filters setup to protect your infrastructure, but:
To: 207.173.177.11
9 gw-cust-VALVESOFTWARE-COM.bllv.eli.net (207.173.69.238) [AS
23352] 72 msec 60 msec 64 msec
10 gw-cust-VALVESOFTWARE-COM.bllv.eli.net (207.173.69.238) [AS
23352] !A !A !A
69.28.151.178
5
70.380 ms
11 ge-2-1-2.nyc33.ip.tiscali.net (12.118.94.46) 67.381 ms 66.700 ms
66.651 ms
Looks like tiscali.net might have a routing problem?
- Alfred
Gary wrote:
I know you have filters setup to protect your infrastructure, but:
To: 207.173.177.11
9 gw-cust-VALVESOFTWARE
- at some stage on December
1st, my s DMC server disappeared from the master server list
and I have no idea s why. I'm aware of at least one other
person who had the same problem on s the same date.
s Is anyone aware of any updates that happened then?
s leo bounds wrote:
Gary
Some of my CS servers aren't showing up on query lists since about 2
days ago. At first I thought it was one of my access-lists on my 4506
(or maybe a routing issue, checked), but it's only like a few
scattered servers.
/ges
At 06:27 PM 12/2/2005, Napier, Kevin wrote:
I belive 27010 udp I
-heapsize 128000
At 09:25 PM 11/13/2005, [EMAIL PROTECTED] wrote:
Re-readit works normally AFTER firing the GUI version of it but won't
fire if I DL the files and then immediately fire it up with a console.
Weird. We tried the suggested /3GB switch in boot.ini when it was suggested
but to
Your bandwidth has nothing to do with why it's lagging.
At 07:33 PM 11/2/2005, [xXx] [EMAIL PROTECTED] wrote:
This is a multi-part message in MIME format.
--
[ Picked text/plain from multipart/alternative ]
can someone plz help me with this issue
i have a dual xeon 3 gig processor with a 100mb
You won't.
At 02:01 PM 10/20/2005, [xXx] [EMAIL PROTECTED] wrote:
your right there too oz, i would like to see if i get even better latency
with diff ports
G. Stanley
Engineering
Velocity Servers, DBA
[EMAIL PROTECTED]
___
To unsubscribe, edit
I think they've already said that a broadband codec would be released
soon after the 1108 launch.
I'm only a lowly ISDNer (broadband take-up is slow here in the UK) but I
think the patch has been excellent so far. Our clan server is taking a
bit of a hit performance wise, but I think its mainly
64 matches
Mail list logo