P4 2.4C SL6WF @ 3.48ghz

[ThugsRook]

Test System
P4 2.4C SL6WF-L320
Abit IC7 (i875)
Kingston HyperX 3000 2x512mb c2622
BFG GF FX5900 @ 450x950
Swiftech MCX4000 | 92mm M1A | ASC
SBLive5.1 | WD800jb | Enermax 550w



Test Specs
2400mhz 200fsb 400ddr = 1:1 | 1.525v
2520mhz 210fsb 420ddr = 1:1 | 1.525v

3000mhz 250fsb 400ddr = 5:4 | 1.525v
3120mhz 260fsb 416ddr = 5:4 | 1.525v

3240mhz 270fsb 360ddr = 3:2 | 1.525v
3360mhz 280fsb 374ddr = 3:2 | 1.525v
3480mhz 290fsb 387ddr = 3:2 | 1.625v
3492mhz 291fsb 388ddr = failed



Benchmarks


http://www.bleedinedge.com/images/thugs/ico-3dm2k1.gif 3DMark2001
2400:200 = 16330
2520:210 = 16993

3000:250 = 18617
3120:260 = 19037

3240:270 = 19261
3360:280 = 19704
3480:290 = 20037



http://www.bleedinedge.com/images/thugs/ico-a32.gif Aida32 read / write
2400:200 = 5110 / 2382 mb/s
2520:210 = 5374 / 2518 mb/s

3000:250 = 5360 / 1981 mb/s
3120:260 = 5566 / 2058 mb/s

3240:270 = 5159 / 1869 mb/s
3360:280 = 5346 / 1940 mb/s
3480:290 = 5532 / 2006 mb/s



http://www.bleedinedge.com/images/thugs/ico-m86.gif Memtest86
2400:200 = 2936 mb/s
2520:210 = 3085 mb/s

3000:250 = 2733 mb/s
3120:260 = 2837 mb/s

3240:270 = 2484 mb/s
3360:280 = 2578 mb/s
3480:290 = 2671 mb/s



http://www.bleedinedge.com/images/thugs/ico-q3a.gif Quake III Extreme 1280x1024
2400:200 = 191.5 fps
2520:210 = 192.2 fps

3000:250 = 192.7 fps
3120:260 = 192.9 fps

3240:270 = 192.8 fps
3360:280 = 192.9 fps
3480:290 = 193.2 fps



http://www.bleedinedge.com/images/thugs/ico-ut2k3.gif UT2003 Demo 1280x960 flyby / botmatch
2400:200 = 193.6 / 71.4 fps
2520:210 = 199.8 / 75.2 fps

3000:250 = 213.2 / 86.0 fps
3120:260 = 216.1 / 89.3 fps

3240:270 = 217.6 / 90.7 fps
3360:280 = 220.1 / 94.1 fps
3480:290 = 221.8 / 96.5 fps



http://www.bleedinedge.com/images/thugs/ico-x2.gif X2 Threat Demo
2400:200 = 74.7 fps
2520:210 = 76.9 fps

3000:250 = 83.6 fps
3120:260 = 85.6 fps

3240:270 = 87.7 fps
3360:280 = 89.6 fps
3480:290 = 90.6 fps



http://www.bleedinedge.com/images/thugs/ico-xisle.gif X-Isle Tech Demo 1280x1024
2400:200 = 122.0 fps
2520:210 = 124.1 fps

3000:250 = 128.4 fps
3120:260 = 129.2 fps

3240:270 = 129.4 fps
3360:280 = 130.1 fps
3480:290 = 130.7 fps




:wave: -thugsrook

[ThugsRook]

just got this chip installed ~ lets see what she's got :chit:

* HT is disabled due to OS

[Thor]

Finally with the P4C. Go TR Go!

[ThugsRook]

thx Thor :-)

stock 200fsb results are posted.
210fsb is next...

[pointreyes]

Gee, I'm so disappointed in you. You are only going to go for 3500Mhz? :p :D

BTW: Is this retail or OEM and from what store?

[ThugsRook]

its an outpost.com OEM chip.


210fsb results have been posted.
220fsb is next...

[dbwillis]

Damn you!, who else is left on 533 chips now?

How much headroom yor 3500 got?

I assume youll test both 5:4 and 1:1 at best mem and CPU speeds..right?

[ThugsRook]

im not "offically" on a C chip til you see my sig say C chip ;)

actually ill be testing 1:1 5:4 then 3:2 cause my mem aint gonna go too far, and im only actually interested in c222 results for the moment.

[ThugsRook]

BAH!
this whole mem ratio thing is a waste of time.
im gonna skip over 220, 230, and 240fsb and go straight to 250fsb 5:4 (hopefully) :rolleyes:

[dbwillis]

There you go!

[pointreyes]

Originally posted by dbwillis
Damn you!, who else is left on 533 chips now?

How much headroom yor 3500 got?

I assume youll test both 5:4 and 1:1 at best mem and CPU speeds..right?

I have a 400fsb chip and check this out about the new Celerons: http://www.abxzone.com/forums/showthread.php?s=&postid=568637#post568637

TR, is that correct? You are using HyperX 3000?

[dbwillis]

Yeh, I read that, wondering if its the link I found on Intels site a few weeks back...

[ThugsRook]

Originally posted by pointreyes
TR, is that correct? You are using HyperX 3000?

yes im using my KHX3000, its some of the best stuff there is for c222 timings :-)

250fsb 5:4 is up and running :chit:
lost GAT tho :(
benchmarks coming up in a bit.....

[Ralf Hutter]

Well, well, well... :)

Welcome to the 21st century!

That Kingston is Winbond BH-5, right?

[ThugsRook]

i donno, i think theyre BH5, never looked :scratch:


250fsb results have been posted.
260fsb is next...

[ThugsRook]

260fsb results have been posted.
still on default voltage :-)
270fsb is next...

[ThugsRook]

:cry:

270fsb failed with a 5:4 mem ratio ~ dropping down to 3:2....

[ThugsRook]

....3:2 requires CPC to be disabled :rolleyes:

[Jim]

Originally posted by ThugsRook
....3:2 requires CPC to be disabled :rolleyes:


What was the max DDR MHZ/timings you got on the B chip??

[ThugsRook]

438ddr cas2622 ~ but that was while using the +4:5 ratio to boost the mem speed.

on this chip i couldnt do 1:1 440ddr
and i couldnt do 5:4 432ddr

so obviously mem ratios make a big difference in stability.
maybe a different bios version will yield slightly better results.

[ThugsRook]

270fsb results are posted.

had to run 3:2 mem ratio :(
lost CPC due to 3:2 mem ratio :(
...still on default voltage tho :-)

at least i know the memory cant get any worse from this point on ;)
hopefully ill make it to 300fsb so ill get my 400ddr back :chit:

280fsb is next...

[Jim]

I can here the CPU and DDR talking.....


"give me voltage Thugs." :wave:

[ThugsRook]

hehe~
the ddr has been @ 2.8v from the start
cas has been @ 2622 for all tests
CPC was enabled til i hit 3:2 ratio

i dont wanna add vcore! nooooooo! not yet! ;)

[ThugsRook]

280fsb results are posted ~ still on default voltage :chit:

290fsb is next...

[ThugsRook]

290fsb results are posted ~ had to boost the vcore to 1.625v tho.

300fsb is next??? :scratch:

[Jim]

300fsb is next??? :scratch:


Heheh....not too shabby on air.........:-)

[ThugsRook]

thx :-)


300fsb wouldnt even bootup on 1.65v :(
im just gonna take baby steps from here ~ 291fsb is next....

[Duvie]

Thugs...I thought it was nly the i865 chips you lost PAT or GAT when you don't use 1:1....

I am getting pc3500 cas 2 stuff for my hopefully new 2.6c rig but if the 2.6c goes to where I think it can 270fsb will rewuire a 5:4 and run my memory at 440mhz ddr...

[ThugsRook]

Duvie ~ i didnt lose PAT, i lost GAT (Gaming Accellerator)
running this much FSB and cas222 seems to make up pretty well for it tho.

:wave:

[Duvie]

What Bios are you running????

Maybe you need to get some hyper x pc3500 like me!!! At least I am hoping to get 1gb of it here shortly....

[ThugsRook]

im using bios v14 at the moment.
...a new bios might let me hang on to 1:1 and 5:4 ratios a little longer, donno.

back to the age old questions eh?
3:2 fast cas or 5:4 slow?
5:4 fast cas or 1:1 slow?

1gig of 500+ddr aint cheap ;)

[Mikki]

dbwillis, I'm still on a B-chip also with my main rig...current test rig is AMD...:p

Thugs, nice looking chip so far, and pretty good results also. I'm curious as to how you're going to make out with this. The ratios will help make use of your memory, which is cool, but would there be a significant difference in using faster sticks?

Looking great! :D

[Duvie]

Originally posted by ThugsRook
im using bios v14 at the moment.
...a new bios might let me hang on to 1:1 and 5:4 ratios a little longer, donno.

back to the age old questions eh?
3:2 fast cas or 5:4 slow?
5:4 fast cas or 1:1 slow?

1gig of 500+ddr aint cheap ;)

Sell that p4 2.4b (great overclocker chip for anyone) and the 1gb of that over achiever Hyper x stuff and you could get some!! Though you still couldn't do 1:1, but the 5:4 at 291fsb would be in the 464mhz range so maybe pc3700 stuff would suffice....All that ddr 500 stuff seems to be running cas 3 anyways...

[ThugsRook]

im still not sure about all that expensive ram making enough difference to actually be worth all that $$$. (remember id want 1gig)

it would probably be alot cheaper for me to get a cpu that OCs into the 250-265fsb range wouldnt you say?

but im sure you guys will still bring up that same question anyways. (5:4 or 1:1?)

;)

[Mikki]

Yeah, that's true...:p Were you planning on posting any B/C comparison results? :)

[ThugsRook]

yea ill post a B vs C comparison for yas.

the C chip is a little bit faster then a B chip mhz for mhz due to the FSB advantage.

the results above show FSB is just as important if not more important then memory. it also shows that CPU MHZ really doesnt mean very much to gamers with the latest vid cards ~ overall bandwidth means more ;)

the 3:2 ratio seems to really choak off any serious performance gains above 260fsb for me.

[ThugsRook]

291fsb failed @ 1.65v ~ no matter what i tried :(

could be a wall, or just all this chip has.

[Mikki]

I'd have to agree with your conclusions so far after looking over your results again. Also, along with the ratio question you have the timings question also...tight timings @ low fsb or loose timings @ high fsb? :p

Still a nice chip...:D

[Jim]

Use the ratio, go for the MHZ. (http://www.bleedinedge.com/forum/showthread.php3?s=&threadid=919) :D

Timings aren't that important. ;)

[ThugsRook]

my memory only does c222 ~ it cant run lame settings.

im not spending $400+ for new ram that may not be any faster at all ~ thats rediculous.


[Moderated by PiLsY.]

[ThugsRook]

Originally posted by ProfessorFate
Use the ratio, go for the MHZ. (http://www.bleedinedge.com/forum/showthread.php3?s=&threadid=919) :D

Timings aren't that important. ;)

you did very good work in that thread ~ but you didnt test c222.
backing off just 1 of my cas severly hurt gaming benchmarks ~ thats why i skipped 220-240fsb completely, the results were horid.

[Duvie]

actually you could get cas 2,3,3,7 4ns hand picked Geil 512mb pc3700 sticks for 170 or 340 total....

If you have to stay in the 250-265 range and still be in the 3.5ghz range you need a 2.8c...wait for intel drops in prices....A 2.6c will need in the 270's to run 3.5ghz and with a 5:4 you need to run 440-2mhz, but with some good pc3500 stuff you could/ should be able to run at cas2.

What would be sweet is a 2.8c that will do about 250-260 with some pc4000 stuff like pastorjays and run 1:1....

Thugs...should I look at the Geil pc3500 golden dragons instead of the hyper x pc3500??? I can get them for about 30 more bucks total 1gb... 1GB for 225.00 on ebay buy it now.

[Thor]

It's true, low latency spd modules sometimes can't run loose timings with CW/SD boards.

[Jim]

Originally posted by ThugsRook
you did very good work in that thread ~ but you didnt test c222.
backing off just 1 of my cas severly hurt gaming benchmarks ~ thats why i skipped 220-240fsb completely, the results were horid.

Heheh, the OCZ wouldn't do 222....wish it would.
It'll be interesting to see what the new line of DDR will do coming out that is a little more flexible on the timings...... Corsair Pro and OCZ's 4200EL when we get some retail box benchies. I don't necassarily see the need to spend more $ on a gig of ram than I have tied up in both my mobo and CPU for perhpas a marginal performance gain at the moment either Thugs . Hence 2X256 of OCZ3700 in my box. :wave:

[oldfart]

Originally posted by ThugsRook

of course you could be opening a pandora's box if that "killer ram" aint worth the difference, cause ill never shutup about it :-r It's not, and I haven't. ;) Once people stop wasting their time with SiSoft and test with actual applications, it is very apparent.
Look on the memory forum @ ABXZone. Lots of tests have been done there. A few on Anands as well. That ram is not all it is cracked up to be. Until decent timings can be had, stay away. Even if it were a little faster, the tiny performance difference is not worth the $$.

[ThugsRook]

i only ran a few tests while playing with the cas settings / mem ratios.....
may as well post them....


3DMark2001
3.2g 267fsb 5:4 c2733 = 18867
3.2g 267fsb 3:2 c2622 = 19103



3DMark2001
3.4g 283fsb 3:2 c2622 = 19707
3.4g 283fsb 3:2 c2722 = 19400 :!: 1 cas setting difference


:)

[oldfart]

And where are those guys who say ram timings dont make any difference?? :scratch:

[dbwillis]

Any idea if your gonna throw XP and HT into the mix Thugs?

[ThugsRook]

im not planning on it dbw ;)

[pointreyes]

Originally posted by oldfart
And where are those guys who say ram timings dont make any difference?? :scratch:

These guys: http://www.lostcircuits.com/memory/ddr400/
BTW: I shelled out quite a bit to have my dual Opteron to have tight memory timings. It did make a huge difference in preformance for me. Since I have a Tyan Thunder, I have to get memory that by SPD is tight because a Thunder board does not allow you to change the memory timings.

dbwillis, sorry but next week I will be on a 800fsb-a P4 3.0 on a P4C800-E Deluxe :cool:

[ThugsRook]

Originally posted by oldfart
And where are those guys who say ram timings dont make any difference?? :scratch:

yea thats why i gave up on testing 220fsb ~ i was wasting my time playing with cas settings in 1:1, when 5:4 2622 was faster after all.

[oldfart]

Originally posted by pointreyes
These guys: http://www.lostcircuits.com/memory/ddr400/
BTW: I shelled out quite a bit to have my dual Opteron to have tight memory timings. It did make a huge difference in preformance for me. Since I have a Tyan Thunder, I have to get memory that by SPD is tight because a Thunder board does not allow you to change the memory timings.

dbwillis, sorry but next week I will be on a 800fsb-a P4 3.0 on a P4C800-E Deluxe :cool:
Great article. Not a single bench was run :rolleyes:
They did make a valid comment:


For the average user, the low latency solution is the way to go but there are the speed-obsessed overclockers for whom nothing matters but frequency and a few numbers to show and brag
This would be the "must spend a bunch of $$ so I can run 1:1 and have a big SiSoft score crowd.

[ThugsRook]

i just noticed this oddity... :scratch:

Aida32
2.4C @ 2.4g 200fsb = 5110 / 2382 mb/s
2.4B @ 3.6g 200fsb = 5545 / 2422 mb/s

[PiLsY]

Oddity?

1200mhz cpu difference (and 1200mhz L1 and L2 cache difference remember) is obviously going to bump the benches up.

PiLsY.

[ThugsRook]

bios settings were identicle for both chips.
L1 L2 should not be part of this test, and im pretty sure it isnt after comparing my other results.

C chips are getting detuned by bios? :scratch:
after reading some comments on P4m chips @ 200-250fsb, this makes sense to me.