| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11
Comments:
<0> all I'm trying to say is that memory controller must run at a fraction of the cpu speed
<0> not a direct fraction of the fsb
<1> thats true on any machine
<0> no its not
<0> although the numbers WILL be fractions of each other...
<1> so there are systems that have faster ram, than the cpu?
<0> the ram speed on athlon is derived from the cpu speed
<1> i give up, sorry
<1> you win
<0> dlion, if you change the cpu multiplier on an intel, the ram speed will be unaffected. If you change the cpu multiplier on an athlon 64 the ram speed will change
<0> try it
<0> that's as simple as I can explain what I'm trying to say
<1> you CANT change the multiplier on a p4
<0> unles...the bios changes the divider with the multiplier
<0> I'm gonna check
<0> oh
<0> you can't lower it?
<1> no, you cant
<0> wow
<0> did not know that
<0> ok brb
<2> ah yeah.. with another video card in its at 46 celcius now, far better
<2> i got the fan off thanks again.
<0> hrmm, it changed the divider to 8 along with the multiplier
<0> not terribly unexpected though
<1> dude, i think what you are talking about is not in proper terminolgy...
<1> is this what you are trying to say:
<1> like a p4 has an 800fsb, but the memory only runs at 200mhz?
<0> no
<0> I'm saying pfs derive their memory speed from the fsb, because the memory controller is seperate from the cpu, but athlons get their ram speed from the cpu speed
<0> p4s
<0> not pfs
<1> yes, the a64s have an integrated memory controller
<0> cpu-z has a box labeled fsb:dram, and on p4s it'll report 3:4, 4:5 etc, and on athlon 64s it'll report cpu/9, cpu/11 etc
<1> did you read what i pasted above?
<0> on athlons the ram is never set at 1:1 with fsb, even though it may be running is 1:1 the memory controller doesn't compute it that way
<0> ok...a p4 with an 800mhz fsb does not compute the ram speed as 1:4
<0> it doesn't count the fsb multiplier or whatever its called
<1> p4 controller alternates between 4 200mhz pipes, hence giving you 800mhz fsb
<1> but that is not the external clock
<0> a p4 with a 166mhz base fsb that supports 533 ddr2...
<1> when people say 1:1, they are talking about external clock
<0> external clock meaning 200mhz fsb or w/e
<0> right
<1> yea
<0> and like I've said 1000 times, I know that athlon 64s fsb:ram is usually 1:1, but that's not how it gets its ram speed
<1> ncrease the front s
<1> oops
<1> http://math.gogar.com/athlon64.cgi
<1> go there, and just next
<1> you will see what i mean
<1> err hit go
<1> do you see now?
<0> they give that ratio sure...but that's not how your cpu is computing the speed
<0> lets take the top one
<0> it says 11:12
<1> yeah
<0> BUT, your cpus memory controller is going every 11th cycle of the cpus clock, not 12 cycles for every 11 cycles of the fsb
<1> so what? this whole thing started because:
<1> [19:16] <4> i wanted to be able to just set a 250 fsb without worrying about my memory crapping out or resorting to cas 3 2t timings just to get there
<1> [19:16] <4> or running dividers
<0> yes
<0> because the athlon 64 memory controller is always running on a divider
<1> so him, me and everyone else knew what he ment except you
<0> and he didn't agree
<0> why is changing the cpu:ram ratio bad?
<0> he made it out to be bad
<0> and on athlon 64s it has no effect
<1> it all depends
<0> 200mhz ram at 200fsb or 244fsb is just as good
<1> on what timings you can run the ram at
<3> there is a defult setting for a reason
<1> no, its not b***goon
<1> 244fsb at cas 2.5 is ont as good at 244fsb at cas2
<1> ont=not
<0> how so?
<1> what is this? the newb hour?
<3> lol
<0> no...I mean...the ram is running at 200mhz either way
<1> dude, try this
<0> cause at 200mhz the ram will be cpu/9, but at 244mhz I set it to cpu/11
<3> even in not sure as to the anwser, but i can ***ume it has to do with the speeds at what its going at
<0> I'm not talking about running the ram at 244
<1> set your timings to 3-3-3-8 and run a cpu or memeory benchmark
<1> you think its gonna bee the same as 2-2-2-5?
<0> I'm talking about running the fsb at 244 and setting the ram to run at 200
<3> man thats slow
<3> junker computer?
<0> on athlon xps setting the ram to async mode could cause problems, make you run in 2t mode
<0> Chez, what are you talking about?
<1> i cant take this anymore... :(
<0> what is slow
<3> a fsb of 244mhz
<0> that isn't slow
<3> o.O
<0> you don't know what we're talking about
<3> i know my fsb is 800mhz
<0> not exactly
<0> it starts as 200mhz
<0> and that's the speed I'm referring to
<1> anyone else torn their hair out yet?
<0> well there is the possibility that I'm an idiot, but i believe you're just stubborn and aren't really listening to me
<1> dude, what nerp says was correct, whatever you were trying to point out had nothing to do with the conversation
<0> there's little point to running your ram above 200mhz on athlon 64, I've seen becnchmarks that show performance boosts only from cpu oc, and he made it sound like running a 'divider' on your ram would impact system performance seriously
<1> it isn't seriouosly
<0> its none
<0> running the ram at 250 instead of 200 won't give much of a boost
<1> yes it will
<0> they tested this by keeping the cpu speed the same and increasing the ram speed
<0> and there was very little benefit to ram + fsb oc with no cpu oc
<0> but when they oced just the cpu and not the fsb or ram there was a HUGE boost
<0> and when they oc
<0> ocd just the fsb there was very little boosyt
<1> where is this "they tested" you speak of?
<0> grr, I found it like a year ago
<0> I will look
<1> the reason why is proabbly because they had to relax the timings
<0> nope
<0> same timings
<0> they used pc-4000 ram
<0> kept same timings throughout
<0> well I'm playing around with some oc
<0> sorry to be a moron
<0> see ya later
<4> well this has been interesting.
<1> why did you just sit back nerp?
<4> well because I don't know the answer myself and I was trying to find some kind of explanation out there
<1> lol
<4> i can see what he means about the info in cpuz etc.
<4> but like, this sure is a concept that is quite well hidden by manufacturers, if true
<4> he's right that the real benefit from an OC is the increased mhz on the cpu, not the fsb
<4> but the reason people oc the fsb is to get the cpu faster.. :)
<5> if an oscilloscope takes analogue measurements will it still be able to test digital signals?
<6> From what I've herd, yes/no.
<6> Yes but not as convienent, if I recall properly.
<5> y
<5> dont remember?
<0> drat, this thing doesn't seem to be oc'ing as well as it was a month ago
<0> and I put on a better nb cooler too
<0> the one that was on there was p***ive
<1> Farcry 1600x1200 in FPS
<1> PC2300 - 3-3-3 1T- 62.25
<1> PC3000 - 3-3-3 1T- 68.14
<1> PC3000 - 2-2-2 1T- 71.31
<1> PC4700 - 3-4-3 1T- 73.52
<1> PC3800 - 2.5-2-2-1T - 74.42
<1> On farcry, we see ~18% faster performance from the old ram. But only ~8% faster than value ram. hint: spend the xtra $100 on a better video card. Notice async low latency Ballistix owning all even 1:1 TCCD.
<1> still dont think timings mean anything?
<0> I didn't say timings didn't matter
<4> he's saying that memory speed is actually determined by the memory controller and not the manufactuer's specs, bios settings and generally accepted # for what memory timing is.. basically
<4> memeory frequency
Return to
#hardware
or
Go to some related
logs:
#unixhelp
clm down etf
#freebsd
#gamedev
#computers
#openbsd
#3dsmax
#stocks
gf chicago efnet
#nhl
|
|