however, Xeon series doesn't do too good on HLDS as expected while somebody talk about it in the past (btw, i have no budget to try xeon cpu & mb, can't confirm). it drive me to something interested like HLDS (or the compiler) *may* have problem on cache management with certain limit of L2/L3 cache (correct me if i m wrong). so that even the architecture is good enough, the performance won't boost as expected. anyone can comments on?
chris
Matt D wrote:
while the "EE" (extreme edition) P4 is certainly faster at pretty much everything, the "E" class (prescott) P4's may not be *any* faster at HLDS than the older "C" (northwood) P4's.
These new "E" pentium 4 are a totally new design on the core, and as such have been desgned to run at higher frequencies. Essentially they are - same speed as equivalent clock "C" P4 - much greater heat output 10-20% more (and way more than ANY athlon ;-) - longer pipeline to enable greater clockspeeds - larger L2 cache to improve performance
The larger cache should make the core faster, but this gets offset by the deeper pipeline which is a performance penalty, especially when cache preloads are incorrect (cache misses).
There are about a bazillion reviews on the various hardware sites that all say the same thing...at identical clock speeds the "E" prescott is a bad buy due to the heat penalty with no performance gain, but once the clock speed ramps up, the prescott architecture will be worthwhile
The Extreme Edition P4 (avaliable in both 3.2 and 3.4ghz variants) is essentially a gallatin xeon with 2 mb L2 cache shoehorned into the normal P4 socket 478, otherwise architecturally its identical to the northwood P4.** The extra cache seems to make a fairly appreciable difference to many things, including serving games, I've used mine to do 64 player WolfET with much success.
Matt
**note: the fact that intel can take a xeon and make it into a desktop chip is more evidence to suggest that P4 and xeon are pretty much identical cpus in different packages. The biggest performance difference between the two is the mobo chipset that connects them to the rest of the world.
http://indigo.intel.com/compare_cpu/showchart.aspx?mmID=857353yes, you will see a speed increase with the larger cache sizes.
kev
Christopher Luk wrote:
are 3.2G(EE) CPU >> with 1M L2 cahce. any one have triedUPDATE: should be 3.2G(E) with 1M L2 cache, not 3.2G(EE). Christopher Luk wrote:
i just came across intel's page and found that there
it before? >>
,857995,852351&catID=7,8,9,10,11,12,13,14,15
performance? >>further, for intel 3.2G, there are models with
1. (3.2GEE) 512k L2 + 2M L3,
2. (3.2GE) 1M L2 + no L3 and
3. (3.2G) 512k L2 + no L3.
any comments on performance with recent HLDS
just fine w/ >> >75% CPU loading but low enough ping forbtw, i got an 2.8C already running an dod 32ppl server,
local users (~40). i will >> purchase an 2.8E in the next
few days with the same intel D865PERL >> main board as my
old one. just wonder is there any performance gain >>
with larger L2 cache. the theory ppl taught a while for
internal cahce >> vs HLDS performance. any comments?
list archives, >> please visit:chris
_______________________________________________
To unsubscribe, edit your list preferences, or view the
http://list.valvesoftware.com/mailman/listinfo/hlds_linux
http://list.valvesoftware.com/mailman/listinfo/hlds_linux_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit:
_______________________________________________
To unsubscribe, edit your list preferences, or view the
list archives, please visit:
http://list.valvesoftware.com/mailman/listinfo/hlds_linux
_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
