1:1 VS 5:4...How much difference does it really make??? ***added test***

[Duvie]

This test was done for2 reasons...1) To disspell the myth of a handful of knuckleheads who think cause they run 1:1 though the cas timings may be weak means it equal to a much higher clocked cpu...2) To prove to myself I have the optimal setup of memory and speed for this Abit mobo which I clearly think shows abit has problems....


Test systems:

p4 2.6@3.033ghz w/ 1:1 for 466mhz ddr cas 2.5,7,3,3
GAT stop functioning at 446mhz ddr, and cas 2,7,3,3 stopped at 446mhz ddr.


p4 2.6@3.2ghz w/ 3:2 for 38mhz ddr cas 2,5,2,2 spd
NO Gat cause Abit version 18 wont boot Gat settings no matter the cas speed or how much you underclock.
I tried CPC enabled/disabled...cas 2.5 to 3,8,4,4...NB trap 800 to By CPU.....NO POST


p4 2.6@3.2ghz w/ 5:4 for 394mhz ddr cas 2,7,3,3
NO Gat cause Abit version 18 wont boot Gat settings no matter the cas speed or how much you underclock.
I tried CPC enabled/disabled...cas 2.5 to 3,8,4,4...NB trap 800 to By CPU.....NO POST

p4 2.6@3.2ghz w/ 5:4 for 394mhz ddr cas 2,7,3,3 CPC enabled (Bios version 14)
Gat works only at turbo but CPC needs to be disabled to run and the GAT results are worse then not having them on.


________________________________________________________________________________________________

-------------------------3.033ghz w/ 466ddr-------3.2ghz w/ 338ddr------3.2ghz w/ 394ddr------3.2ghz w/ 394ddr (CPC)

Sissoft MEM---------------5650's------------------------4800------------------------5380's---------------------5420's

Tmpgenc-------------------2:24---------------------------2:18------------------------2:18------------------------2:17

SuperPi 1M----------------0:45---------------------------0:44-------------------------0:43-----------------------0:43
superPI 2M---------------1:45----------------------------1:43-------------------------1:43-----------------------1:42
Prime95 (ver22)
1792k----------------------76.643------------------------73.320----------------------73.309--------------------73.137

Besweet Wav-AC3------4:25----------------------------4:11------------------------4:08------------------------4:04
DVDshrink 3.0------------17:58--------------------------17:28----------------------17:26-----------------------17:32

Cinebench2003
single cp/multi------------307/363-----------------------323/384-------------------323/381-------------------329/388
AutoCAD 2002------------0:48---------------------------0:46-------------------------0:45-----------------------0:45

WinRAR---------------------3:08---------------------------3:08------------------------3:11-------------------------3:11

3dmark2k1SE-------------9085---------------------------9076-----------------------9077-------------------------Didn't Do

__________________________________________________________________________________________________


3.2ghz to 3.033ghz represents 5.7%

Tmpgenc = 3.033ghz cpu is 5.1% behind fastest 5:4 setting despite 70mhz more and 200+ more mb/s from sissoft...

SuperPI 1M = 3.033ghz cpu is 4.6% behind fastest 5:4 3.2ghz...guess this would to be around 3.15ghz to equal.
SuperPI 2M = 3.033ghz cpu is 1.9% behind fastest 5:4 3.2ghz...guess this would need to be around a 3.08ghz cpu...

Prime95 = 3.033ghz cpu is 4.8% behind fastest 5:4 3.2ghz....would need a 3.16ghz cpu to equal it...

Besweet Wav-Ac3 = 3.033ghz cpu is 8.6% behind fastest 5:4 3.2ghz...Go figure!! Timings mean more I guess..Ran multiple times.

DVDshrink3.0 = 3.033ghz cpu is 3.1% behind fastest 5:4 3.2ghz....would need a 3.12ghz cpu to equal it...

Cinebench 2003 = 3.033ghz cpu is 7.2% behind fastest 5:4 3.2ghz....again it must be the slower timings at 1:1...

AutoCAD 2002 = 3.033ghz cpu is 6.7% behind fastest 5:4 3.2ghz....again this render program cares less about speed vs timings

WinRar = 3.033ghz cpu is 1.6% faster and the only program I tested that was faster then a higher speed cpu with lower speed mem.

3dmark2k1SE = 3.033ghz cpu is equa to higher speed cpu with lower speed memory but I dont think this is a surprise. Many know the 3dmark2k1se as ery non cpu dependent. Memory bandwidth wins in these programs. Sorry it is my only game test but I should have UT demo to test here in a few days to see if this is indictative of gamers.


Conclusion it appears in the apps I run which tend to lean at Mulitmedia, rendering, archiving, Transcoding, and scientific that memory speed does not equate to raw speed and many who sacrifice larger overclocs to run at 1:1 with lower timings are not running their machine optimally. Now I do conclude this is not proven for gaming yet, but it is nowhere near the fantasy numbers I ave heard ppl pull out of their arses around here....It is likely higher speed memory like 466mhz ddr could run at 3.06ghz and equal a 3.2ghz 400mhz 5:4 system as long as the timings are no that much different.


__________________________________________________________________________________________________
***ADDED***

Test systems:

#1) p4 2.6@3.033ghz w/ 1:1 for 466mhz ddr cas 2.5,7,3,3
GAT stop functioning at 452mhz ddr, and cas 2,7,3,3 stopped at 456mhz ddr.

#2) P4 2.6@3.033ghz w/ 5:4 for 373mhz ddr cas 2.5,7,3,3

#3) p4 2.6@3.033ghz w/ 5:4 for 373mhz ddr cas 2,5,3,3 CPC enabled



-------------------------(#1)3.033ghz w/ 466ddr-------(#2)3.033ghz w/ 373ddr-----(#3)3.033ghz w/ 373ddr

Sissoft MEM---------------------5650's---------------------------5030's----------------------------5100

Tmpgenc--------------------------2:24-----------------------------2:26------------------------------2:24

SuperPi 1M-----------------------0:45-----------------------------0:46------------------------------0:45
superPI 2M-----------------------1:45-----------------------------1:51------------------------------1:49
Prime95 (ver22)
1792k----------------------------76.643--------------------------77.340----------------------------77.142

Besweet Wav-AC3--------------4:25-----------------------------4:28-----------------------------4:22

DVDshrink 3.0-------------------17:58----------------------------18:05----------------------------17:57

Cinebench2003
single cp/multi------------------307/363------------------------305/361--------------------------306/362

AutoCAD 2002--------------------0:48-----------------------------0:49-----------------------------0:48

WinRAR-----------------------------3:08-----------------------------3:25-----------------------------3:19

3dmark2k1SE---------------------9085----------------------------9008-----------------------------9037

UT2003demo(flyby)------------186.82--------------------------185.79---------------------------186.32
UT2003demo(botmatch)--------83.65---------------------------81.30-----------------------------82.46


1:1 vs 5:4 (same timings)/1:1 vs 5:4 w/ cas 2,5,3,3 cpc

Sandra MEM---- 1:1(+12.3%) / 1:1(+10.8)

Tmpgenc-------- 1:1(+1.3%) / 1:1(+0.0%)
SuperPi 1M----- 1:1(+2.2%) / 1:1(+0.0%)
SuperPi 2M----- 1:1(+5.7%) / 1:1(+3.8%)
Prime95--------- 1:1(+0.9%) / 1:1(+0.6%)
Besweet-------- 1:1(+1.1%) / 5:4(+1.1%)
DVDshrink3.0-- 1:1(+0.6%) / 5:4(+0.1%)
Cinebench03--- 1:1(+0.6%) / 1:1(+0.3%)
AutoCAD2002-- 1:1(+2.1%) / 1:1(+0.0%)
WinRAR---------- 1:1(+9.0%) / 1:1(+5.9%)
3dMark2k1SE--- 1:1(+0.9%) / 1:1(+0.5%)

Interpolate as you wish...The cas 2 timings almost negated the memory speed....

 

[orion7144]

I have been saying this since the P4C's came out and alot of people still swore that it is not possible. I kept saying that you can not compare an AMD rig to the P4C's. The performace you gain by uping the FSB far outweighs running a slower FSB to be able to run 1:1

My OCZ PC3700 Gold will do 275 1:1 2.5-3-3-7 (this is what I run everyday) However I have some OCZ PC3500 EL that will only do 250 at the same CAS setting. Running the PC3500 at 275 5:4 netted very similar, non-synthetic, scores as the PC3700 1:1.

 

[oldfart]

Nice to have someone else backing me up on this for a change! As you know, I've been fighting this myth for a long time.

I've been posting about this here, ABXzone, and BE forums. All you have to do is actually test it and the results are right there in front of you. The guys who are out there pimping overpriced "high speed"

memory test with SiSoft mem bench, Aida and the like since these synthetic apps show huge gains but have no relevance to real world apps.

You should throw in a SiSoft mem bench just for fun.

 

[orion7144]

Originally posted by: oldfart
Nice to have someone else backing me up on this for a change! As you know, I've been fighting this myth for a long time.

I've been posting about this here, ABXzone, and BE forums. All you have to do is actually test it and the results are right there in front of you. The guys who are out there pimping overpriced "high speed"

memory test with SiSoft mem bench, Aida and the like since these synthetic apps show huge gains but have no relevance to real world apps.

You should throw in a SiSoft mem bench just for fun.

I think I put those numbers from my tests in the IS7 thread. I will go through the pages and see if I can't find them.

 

[orion7144]

O.k hear are some of my older numbers with the IS7

225 1:1 3-4-4-7

Sisoft mem 9829 Combined performance 2335/5200

250 5:4 same timings So mem at 200mhz 50mhz slower but still faster

Mem 10500 Combined 2477/5691

200 1:1 Gat set to Street racer

Mem 9651 combined 2079/4639

As a refference I also did my old 2.26 @ 2.7 5:4

Mem 4779 combined 1519/3583

 

[jiffylube1024]

But we like mindlessly plugging $300 RAM dual channel setups . Nice work as usual, Duvie. I'm running some corsair XMS3200 @ 209 MHz (really pushing the BH-6 chips here too!) but still get good results, even though I'm not running at l33t memory speeds .

 

[Duvie]

I added the "synthetic" sissoft scores (averaged for 3 times) for Oldfart.....I almost did not like using 3dmark but think it is at least relevant to changes within a closed system such as mine....


I have a disk but together by Orion coming that will give me a popular gaming test and some more rndering test that is more down my alley of work and fun....I know I am bucking the trend since there are so many gamers here, but this is what I use ppl and as far as uses go this is what these power machines should be used since the gpus do most of the work now....

 

[Duvie]

Olfart did you notice that I gain posting ability with GAT again my using bios version 14...Bios version 17 gave it to me but not as effective for some reason as bios 14.....In the end it really did nothing for me as I don't even use it since I get better results without it but CPC enabled.

I need to do some prime95 to see if the older bios gave me a touch more stability!!! You never Know....I was able to post at a whopping 236mhz 1:1 with cas 2.5,7,3,3 run benches and run 3 passes of memtest before I stopped it and a 2 hour mpeg2 encode as I went and played B-ball. 237 froze in memtest in the 2nd pass....238 just gives a beep tone....

If I had 2.9 and 3.0v options I know I could take this memory a bit more cause I haven't even had to run cas3 yet.....

1:1 is worthless in my setup but nice to know I have the headroom for the future and possibly a 2.4c that does 3.5 could come my way (wink...wink Shim!!!)

 

[orion7144]

Originally posted by: Duvie
Olfart did you notice that I gain posting ability with GAT again my using bios version 14...Bios version 17 gave it to me but not as effective for some reason as bios 14.....In the end it really did nothing for me as I don't even use it since I get better results without it but CPC enabled.

I need to do some prime95 to see if the older bios gave me a touch more stability!!! You never Know....I was able to post at a whopping 236mhz 1:1 with cas 2.5,7,3,3 run benches and run 3 passes of memtest before I stopped it and a 2 hour mpeg2 encode as I went and played B-ball. 237 froze in memtest in the 2nd pass....238 just gives a beep tone....

If I had 2.9 and 3.0v options I know I could take this memory a bit more cause I haven't even had to run cas3 yet.....

1:1 is worthless in my setup but nice to know I have the headroom for the future and possibly a 2.4c that does 3.5 could come my way (wink...wink Shim!!!)

I may be selling one of my 2.4C's that will do 3.5 since I may be getting a EE from a "friend" at Intel.

 

[virtuamike]

My 1:1 tops out around 265FSB, 5:4 around 280FSB. I'll be back later with some benchies.

 

[Corporate Thug]

http://www.pcstats.com/articleview.cfm?articleID=873

 

[virtuamike]

PAT enabled, OCZ EL PC4000 512MBx2. Stable running Prime95 for 2+ hours case fans off. 1.8 bios helped with my 1:1 FSB, before I couldn't get that high because board kept setting my ram to 2.5-8-4-4.

** 263 1:1 3-8-4-4 GAT: A-A-A-D-D
Sandra buffered: 6030/6107
Sandra unbuffered: 3101/3200
3DMark2001: 14229
DVDShrink: 17:46

** 280 5:4 3-8-4-4 GAT: A-A-A-D-D
Sandra buffered: 5901/5891
Sandra unbuffered: 2769/2869
3DMark2001: 14582
DVDShrink: 16:22

I'm curious though, wondering what my system would do with a 2.6C@3.4 260FSB 1:1.

BTW all air-cooled. Zalman 7000 AlCu on CPU, new IC7 heatsink lapped and epoxied on the NB w/ fan removed. Trying to make my system silent.

 

[oldfart]

Originally posted by: shady06
www.pcstats.com/articleview.cfm?articleID=873

Linkified for you

While bandwidth is still very important to the Intel Pentium 4, it's not as important as it once was in the i845PE days of single channel memory controllers. Thanks to the i865PE/i875P's dual channel memory controller things are much brighter. On average, the system with the memory running at 400 MHz (5:4 memory divider enabled) with aggressive memory timings performed 2-3% faster than the system using high speed memory with loose timings.

Gee, what a surprise. Again, I've been telling people this since just before the P4 "C" chips came out. Where are the "5:4 ratio will cripple your system" people? Oh yeah, they are busy running SiSoft mem benches.

 

[Henny]

Thanks. It's refreshing to see someone make assertion backed up with real data.

Originally posted by: Duvie
This test was done for2 reasons...1) To disspell the myth of a handful of knuckleheads who think cause they run 1:1 though the cas timings may be weak means it equal to a much higher clocked cpu...2) To prove to myself I have the optimal setup of memory and speed for this Abit mobo which I clearly think shows abit has problems....


Test systems:

p4 2.6@3.033ghz w/ 1:1 for 466mhz ddr cas 2.5,7,3,3
GAT stop functioning at 452mhz ddr, and cas 2,7,3,3 stopped at 456mhz ddr.


p4 2.6@3.2ghz w/ 3:2 for 38mhz ddr cas 2,5,2,2 spd
NO Gat cause Abit version 18 wont boot Gat settings no matter the cas speed or how much you underclock.
I tried CPC enabled/disabled...cas 2.5 to 3,8,4,4...NB trap 800 to By CPU.....NO POST


p4 2.6@3.2ghz w/ 5:4 for 394mhz ddr cas 2,7,3,3
NO Gat cause Abit version 18 wont boot Gat settings no matter the cas speed or how much you underclock.
I tried CPC enabled/disabled...cas 2.5 to 3,8,4,4...NB trap 800 to By CPU.....NO POST

p4 2.6@3.2ghz w/ 5:4 for 394mhz ddr cas 2,7,3,3 CPC enabled (Bios version 14)
Gat works only at turbo but CPC needs to be disabled to run and the GAT results are worse then not having them on.


________________________________________________________________________________________________

-------------------------3.033ghz w/ 466ddr-------3.2ghz w/ 338ddr------3.2ghz w/ 394ddr------3.2ghz w/ 394ddr (CPC)

Sissoft MEM---------------5650's------------------------4800------------------------5380's---------------------5420's

Tmpgenc-------------------2:24---------------------------2:18------------------------2:18------------------------2:17

SuperPi 1M----------------0:45---------------------------0:44-------------------------0:43-----------------------0:43
superPI 2M---------------1:45----------------------------1:43-------------------------1:43-----------------------1:42
Prime95 (ver22)
1792k----------------------76.643------------------------73.320----------------------73.309--------------------73.137

Besweet Wav-AC3------4:25----------------------------4:11------------------------4:08------------------------4:04
DVDshrink 3.0------------17:58--------------------------17:28----------------------17:26-----------------------17:32

Cinebench2003
single cp/multi------------307/363-----------------------323/384-------------------323/381-------------------329/388
AutoCAD 2002------------0:48---------------------------0:46-------------------------0:45-----------------------0:45

WinRAR---------------------3:08---------------------------3:08------------------------3:11-------------------------3:11

3dmark2k1SE-------------9085---------------------------9076-----------------------9077-------------------------Didn't Do

__________________________________________________________________________________________________


3.2ghz to 3.033ghz represents 5.7%

Tmpgenc = 3.033ghz cpu is 5.1% behind fastest 5:4 setting despite 70mhz more and 200+ more mb/s from sissoft...

SuperPI 1M = 3.033ghz cpu is 4.6% behind fastest 5:4 3.2ghz...guess this would to be around 3.15ghz to equal.
SuperPI 2M = 3.033ghz cpu is 1.9% behind fastest 5:4 3.2ghz...guess this would need to be around a 3.08ghz cpu...

Prime95 = 3.033ghz cpu is 4.8% behind fastest 5:4 3.2ghz....would need a 3.16ghz cpu to equal it...

Besweet Wav-Ac3 = 3.033ghz cpu is 8.6% behind fastest 5:4 3.2ghz...Go figure!! Timings mean more I guess..Ran multiple times.

DVDshrink3.0 = 3.033ghz cpu is 3.1% behind fastest 5:4 3.2ghz....would need a 3.12ghz cpu to equal it...

Cinebench 2003 = 3.033ghz cpu is 7.2% behind fastest 5:4 3.2ghz....again it must be the slower timings at 1:1...

AutoCAD 2002 = 3.033ghz cpu is 6.7% behind fastest 5:4 3.2ghz....again this render program cares less about speed vs timings

WinRar = 3.033ghz cpu is 1.6% faster and the only program I tested that was faster then a higher speed cpu with lower speed mem.

3dmark2k1SE = 3.033ghz cpu is equa to higher speed cpu with lower speed memory but I dont think this is a surprise. Many know the 3dmark2k1se as ery non cpu dependent. Memory bandwidth wins in these programs. Sorry it is my only game test but I should have UT demo to test here in a few days to see if this is indictative of gamers.


Conclusion it appears in the apps I run which tend to lean at Mulitmedia, rendering, archiving, Transcoding, and scientific that memory speed does not equate to raw speed and many who sacrifice larger overclocs to run at 1:1 with lower timings are not running their machine optimally. Now I do conclude this is not proven for gaming yet, but it is nowhere near the fantasy numbers I ave heard ppl pull out of their arses around here....It is likely higher speed memory like 466mhz ddr could run at 3.06ghz and equal a 3.2ghz 400mhz 5:4 system as long as the timings are no that much different.

 

[Thor86]

Hmm, well I beg to differ on this, as I run my setup 1:1 at high fsb, with not so tight memory timings, and I find it is what the
"BEST" thing there is out available for me at this time. I prefer the high fsb settings as multi-tasking just seems so much snappier in everyday usage, over highest cpu mhz.

I have looked into 5:4, and yes, with certain setups utilizing high cpu speeds and tight memory timings, the performance is equal if not 2-3% better (as Oldfart stated) as per benches, but tell me which ram sticks can do DDR480+ with less than 2.9 vdimm?

Most memory modules that are rated at tight timings require too much vdimm for high FSB IMHO or cannot come anywhere close to what high-end, expensive ram can do at 2.85 or less vdimm.

 

[WA261]

Great info. I guess it is just me, but the only time I see people arguing over this is when fsb is "=".... "=" fsb 1:1 being superiour to 4:5

 

[oldfart]

Originally posted by: Thor86
Hmm, well I beg to differ on this, as I run my setup 1:1 at high fsb, with not so tight memory timings, and I find it is what the
"BEST" thing there is out available for me at this time. I prefer the high fsb settings as multi-tasking just seems so much snappier in everyday usage, over highest cpu mhz.

I have looked into 5:4, and yes, with certain setups utilizing high cpu speeds and tight memory timings, the performance is equal if not 2-3% better (as Oldfart stated) as per benches, but tell me which ram sticks can do DDR480+ with less than 2.9 vdimm?

Most memory modules that are rated at tight timings require too much vdimm for high FSB IMHO or cannot come anywhere close to what high-end, expensive ram can do at 2.85 or less vdimm.

When you run a 5:4 ratio, you dont need to run DDR480. That would be a 600 MHz FSB! Its the opposite. When you run 1:1 you need the high DDR speeds and high VDimm to get there. When you run a ratio, you lower the DDR speed and dont need the high Vdimm.

 

[virtuamike]

Actually I'm hitting 263 1:1 with 2.78-2.80V on my RAM. Voltage fluctuates and IC7 undervolts slightly, still able to get up to DDR525 which is pretty damn good in my opinion. If I can find a 2.6C that'll run at 260FSB then that'd be the perfect CPU/mem combo. Well, either that or I spring for an Asus P4C800 and try for 280 1:1 with my 2.4C, but I think it'd be cheaper to do CPU swap than mobo swap.

 

[Thor86]

Originally posted by: oldfart

When you run a 5:4 ratio, you dont need to run DDR480. That would be a 600 MHz FSB! Its the opposite. When you run 1:1 you need the high DDR speeds and high VDimm to get there. When you run a ratio, you lower the DDR speed and dont need the high Vdimm.

You mean 300fsb. And we know some 2.4Cs are hitting them with ease. Hmm, depends on what cpu/fsb you plan on using. Anything over 270fsb would require overly-high vdimm while using slower DDR sticks.

 

[oldfart]

Originally posted by: Thor86

Originally posted by: oldfart

When you run a 5:4 ratio, you dont need to run DDR480. That would be a 600 MHz FSB! Its the opposite. When you run 1:1 you need the high DDR speeds and high VDimm to get there. When you run a ratio, you lower the DDR speed and dont need the high Vdimm.

You mean 300fsb. And we know some 2.4Cs are hitting them with ease. Hmm, depends on what cpu/fsb you plan on using. Anything over 270fsb would require overly-high vdimm while using slower DDR sticks.

And 1:1 would be better? In your example of 300 MHz FSB, you would need to run DDR600 @ 1:1 How much Vdimm would that need? I think DDR480 is possible while DDR600 is not.
And yes, it does depend on CPU/FSB speed.

 

[Duvie]

Originally posted by: WA261
Great info. I guess it is just me, but the only time I see people arguing over this is when fsb is "=".... "=" fsb 1:1 being superiour to 4:5

Actually I have been seeing many ppl state their 2.4@3.0 or 2.4@2.8ghz with a 1:1 is as fast as a 2.4@3.2c with a 5:4....I never see these ppl compare same speed fsb....If that was the case that would be a no brainer and there would have been no point to doing this study. However 1:1 with cas3,4,4,8 may not even be all thatit is cracked up to be if with the 5:4 or even the 3:3 the same fsb could yield cas 2,2,2,5 and Gat Streetr racer settings.

Again I ran more of a non gaming suite of test but it clearly shows that the proceesor would stll need t be 3.1+ghz to even compare and by then the ram I have would not even cut it or be the worst possible cas settings and likely would stil fall short.

I ran the 1:1 the 5:4 and 3:2 and not one setup ovr the other felt any snappier...heck look at the differences most were in the5% or less which is questionable territory of noticeable in the first place.


***Same fsb and timings....no argument 1:1 is better****

It is the ppl who hold off what could be a much higher fsb overclock cause they can't get 1:1 to go higher or the fools here. I have seen many by dropping to a 5:4 get much more of an overclock...look at mine....I could only do 236fsb with 1:1 but can get 250 at 5:4..14more fsb at 13x and that is a lot....Now you have to be a fool or illiterate to look at these numbers and say you would have ran a 1:1 setup at this point.


Had I had say some ocz gold pc4000 I very well may have been able to do 250 with my 2.8v on this abit and then the cpu limitation would have reared its ugly head first menaing no matter if I shifted to5:4 I wasn't going to get anymore out of this oc then 1:1 is GOD!!!! Ofcourse unless at that point I was having to run cas 3,4,4,8 whereas I could be running cas 2 and then in some apps as pictured above that would have made more of a difference then 500mhz versus 400mhz...

 

[Thor86]

My point is if I wanted to run 5:4, at 300fsb, you'd need ddr480 at least. Show me which memory can do this besides the "expensive" kind.

 

[Duvie]

I paid 215 for 2x512mb sticks oc Geil PC3500 Ultra Platinums and they do 472ddr at cas 2.5,8,3,3 and I am sure the reason they went no further is cause the abit mobo not only has 2.8v as its maximum vdimm but that it also undervolts it to 2.78v....If I had 2.9v this does 480 I guarantee!!!!

By no means expensive when you figure it was for 1gb...

 

[virtuamike]

I don't think 300FSB 5:4 is all that common. Most people top out around 280FSB. There's a very good reason why 3.6GHz P4s aren't on market yet.

 

[oldfart]

Thor86
I'm not sure what your point is. Unless I misunderstood, you were saying 5:4 is a problem because of the high DDR speeds and high Vdimm to do it. No matter what FSB you run, 5:4 will ALWAYS require a lower DDR speed. Its basic math. If you run a crazy high FSB speed like 300 MHz you MUST run a ratio. You have no other option.

It you run 300 MHz FSB, you have 3 choices

DDR600 1:1 (good luck)
DDR480 5:4 (expensive, but possible)
DDR400 3:2 (piece of cake)