Epox 4pca3+ rev.2 Performance? ? [Archive]

Clevor

12-04-2003, 10:50 PM

I would seriously consider the epox 4pda2+ rev2.0 over this board. it overclocks like mad. never has a single problem with it. it outperformed my 4pca3+. the canterwwod might of been better if i could get over the fsb wall that most peeps experience.

So Esodactyryl, are you saying you can run over 280 fsb and 5:4 works with BH-5 at 2-2-2-5??? I don't mean Prime or screenies. Can you bench all this: 3DMark2001, 3DMark2003, Serious Sam 2nd Encounter demo loop, IL-2 demo loop for 30', PCMark2002, VulpineGL, CodeCreatures, Aquamark, UT2003 bench. Then Prime or the 16M SuperPI run.

I can't do this with the 4PCA3+ except below 280 fsb.

Also do the 'Aggressive Settings' in the BIOS turn on PAT, even partially?

Soulburner

12-04-2003, 11:56 PM

Clevor are you using a chip that has done higher speeds in a different board?

Clevor

12-07-2003, 02:49 PM

Yes, that's why I'm asking. The same chip does 288 on the P4C800-E, DFI 875P Infinity, and IC7-MaxIII, but only at 3:2 on the latter two boards. I really doubt the Epox Springdale board will be any different. I have tested a dozen SD/CW boards and virtually all of them will not run BH-5 at 5:4 over 280 fsb or so. I'm talking 3D stable as Memtest is clean and they will boot and pass SuperPI. It's an Intel bug on the chipset. The P4C800-E and P4P800 is somehow immune to it, though. The DFI will run 5:4 using ADATAPC450, but not above 281, even though it runs 287 at 3:2.

Who said life is simple? :p:

-CTPAX-

12-08-2003, 01:08 AM

I can bench at 3.5G 292(5:4), not prime stable though. I found, that raising vdimm to 2.9v helped prime stability up to 285. Strange thing is, that when i tried a pair of sticks Vitesta DDR500, the mobo refused to boot at 300fsb. With my ddr450 i have no problem. Also the Vitesta doesnt work with CPC enabled. Tried 280 1:1, worked like a charm :-)

Soulburner

12-08-2003, 11:23 AM

CTPAX sounds just like my chip, I cant go over 292. I don't think its your board at all.

Clevor

12-08-2003, 03:11 PM

What revision boards do you have? I have a 1.1. Actually, Memtest locks up over 280, 5:4 with BH-5, but I've not tried booting. Maybe I can bench. Normally, that spells trouble.

I have Vitesta, regular ADATA PC4000, and ADATA PC450 too. They all seem to work OK in the board at all mem ratios, except 5:4 over 280 does not clear 3DMark2001. I have not been able to run CPC on any ram in the board; too many Memtest errors.

I am running a Tiger1 NB cooler so I don't think that's holding me back.

I just tested an early IC7 board last night. With the latest BIOS, it's a great board, better than the IC7-MaxIII. Again, 5:4 issues. I can run OCZ PC3700 Gold up to 281, 5:4, 2-3-3-6. At 282 the board won't complete the boot. One mhz lower and it benches all 3D. It's like a switch or something.

With XMS3500, I can run 5:4 up to 280 fsb, not bad, but at 2-3-2-5 due to the 2.8 VDIMM limit. At 3:2, I can run 288 on the CPU with both ram. The board is very stable, only negative is the 5:4 limit. Sandra unbuffered bandwidth however is like the P4C800-E: about 100 points below the fast DFI and 4PCA3+ boards.

Soulburner

12-08-2003, 03:15 PM

Sounds exactly like mine....280 5:4 2-3-2-5 due to vdimm limit. Runs the same RAM speed 1:1 so I know I don't have anything else stopping me, just the volts. It will do 286 3:2 in 3DMark also where it becomes a CPU limitation it seems.

-CTPAX-

12-08-2003, 11:56 PM

Mine is rev 1.2. I tried Memtest once at 285, got no errors at all, so i dont think theres a 280 limit. I think the cpu can do above 3.5 with the Zalman, but 1.6v vcore is not enough. Also the onboard gigabit lan causes freezing under winxp and thats killing me :-((

Soulburner

12-09-2003, 12:09 AM

No there isn't a 280 limit but there is for me because my RAM won't go past 225 with the meager 2.85v option and undervolting.

Clevor

12-09-2003, 02:55 PM

No there isn't a 280 limit but there is for me because my RAM won't go past 225 with the meager 2.85v option and undervolting.

Hmmm, do you mean the IC7? The 4PCA3+ goes to 3.3 VDIMM, while the IC7 is limited to 2.8 (more like 2.75). My XMS3500 is capable of DR500, 2-2-2-5 at 3.2 volts, but what's the point if the 5:4 won't work over 280 fsb!

The P4C800-E is the only board I've seen that runs 5:4 over 280 with BH-5. I can do 287, 5:4, 2-3-2-5 with the same CPU, at the 2.85 VDIMM limit of the board. If only it had up to 3.3 like the Epox. :mad: I can do the same thing on the P4P800-D.

The low VCORE on the 4PCA3+ is annoying. I got a lousy 2.80C that maxes at 243 or 3.4 gig. Normally I can run 243, 1:1, 2-2-2-5 with the XMS3500 at maybe 3.0-3.1 volts, but the CPU needs 1.68 volts. Unfortunately, my 1.1 board doesn't have the bug where 1.45 nets you 1.70 VCORE.

Soulburner

12-09-2003, 03:14 PM

No, I have a P4P800. My ram runs out of juice at 220 2-2-2-5 and 225 2-3-2-5. Same thing at 280 5:4.

And the Epox has 1.85vcore so I don't consider that low.

Clevor

12-09-2003, 03:31 PM

Same with me on the Asus boards, XMS3500 good for DDR440, 2-2-2-5 and DDR463, 2-3-2-5. That is as good as it gets for 2.85 VDIMM. I do plan the VDIMM mod but you still can't get over 3.15 there. Also the P4P800-D has fluctuating VDIMM; doesn't seem to be as good as the P4C800-E.

And the Epox has 1.85vcore so I don't consider that low.

My 1.1 board only goes to 1.60 VCORE which is more like 1.55. I've tried the old BIOs that allowed up to 1.85 but nothing over 1.6 seemed to work.

Frank E. BoNeS

12-09-2003, 08:17 PM

Originally posted by Clevor
My 1.1 board only goes to 1.60 VCORE which is more like 1.55. I've tried the old BIOs that allowed up to 1.85 but nothing over 1.6 seemed to work.

They fixed it with Rev. 2....I just got one...the vcore does go up to 1.85 and works.

Soulburner

12-09-2003, 08:30 PM

Let us know how it does, I want that board.

Frank E. BoNeS

12-09-2003, 09:12 PM

Originally posted by Soulburner
Let us know how it does, I want that board.

So far its been great...the 3.3 vdimm soooo nice. I've had my 2.4 at 300fsb but it wasn't 100% stable....still need to do some tweaking....Voltages are pretty good....this board undervolts slightly on the vcore by about .02...same with the vdimm. It was well worth the $130...I like it better than my P4C800DLX (even with the voltmod)....

Clevor

12-09-2003, 10:46 PM

I like it better than my P4C800DLX (even with the voltmod)....

You should see 100 more unbuffered Sandra points with the Epox.

What ram are you running up near 300 fsb? Can you do 5:4 over 280?

SPQQKY

12-10-2003, 06:51 AM

I've been watching this thread as I was thinking of going to the dark side. I believe the 4pca3+ is the board I will be getting along with a 2.4CG. I will keep you updated, will probably be a week or better before I get the components and get rolling.

BTW: Do you think my Kingston HyperX pc3200 will work well with this mobo? I simply cannot afford any other RAM. I have TwinMos pc3700 that will go sky high, but only with 7-3-3-2.5 timings. The HyperX will do 228 on my 8rda+ at 6-2-2-2.

Frank E. BoNeS

12-10-2003, 01:34 PM

Originally posted by Clevor
What ram are you running up near 300 fsb? Can you do 5:4 over 280?

Yeah totally....At 300fsb, I'm running my HyperX 3200(BH-5) on the 5:4 divider at 3.2 volts, 2,2,2,6...

Craig

12-11-2003, 07:35 AM

Pretty good O/C you have going there for having just got the board. Congrats! !:thumbsup:

Could you spec your cooling set up please? And could you also post the settings you are making use of?

Thanks:toast:

Frank E. BoNeS

12-11-2003, 02:24 PM

Originally posted by Craig
Pretty good O/C you have going there for having just got the board. Congrats! !:thumbsup:

Could you spec your cooling set up please? And could you also post the settings you are making use of?

Thanks:toast:

Thanks :)

I'm using a SLK-900 and a vantec tornado for cooling...I also put a TT Tiger on the NB....

Here's my voltage settings at that speed....
Vcore: +0.2 - 1.7v
Vdimm: +0.6 - 3.2v
Vagp: +0.3 - 1.8v

I never had to raise the Vagp on my P4C800 but with this board I needed to increase it to run 3D at high bus speeds....it was Prime stable and all but 3d would lock the system up....raising the agp voltage solved that.

Clevor

12-11-2003, 03:14 PM

Maybe the Rev. 2.0 board has fixed things; I presumed you were on a Prommie. 300 fsb on air is exceptional for 3D stable.

Try 2.0 VAGP, you may get even higher :) .

With my XMS3500, you could do 2-2-2-5 there at maybe 3.0 volts or so.

My 1.1 board has a 280 fsb limit, but unless you have a 2.40C and/or Prommie, that makes a very fast rig for any other CPU, especially with the VDIMM which can drive BH-5/BH6 to DDR480+ at 2-2-2-5. 5:4 works fine here.

Actually, I got a Rev. 1.2 and 2.0 board also, just been too lazy to switch over :).

Soulburner

12-11-2003, 08:04 PM

Originally posted by Clevor
Try 2.0 VAGP, you may get even higher :)
I would be a little hesitant to try that, it actually does send that voltage to your AGP slot. At 1.7v my card starts to freak out and tear in Nature. I can't imagine my card even surviving 2 volts.

Clevor

12-11-2003, 08:23 PM

Nah, I think it bleeds off to the NB or something. I've run a 9700 Pro, 9800 Pro, and Geforce 5900FX and seen no problems. It might help get 2 mhz more on stubborn chips.

I've tried upping AGP on a dozen other boards. Never seen it help overclocks on the video card, or CPU. I think it helped on the P4P800-D to max out the VAGP.

Soulburner

12-11-2003, 08:30 PM

Originally posted by Clevor
Nah, I think it bleeds off to the NB or something.
Are you telling me my results are just made up? Give me a break. I know what it does, but it does go to the slot as well. Otherwise my card wouldn't do that. I'm not the only one either.

Clevor

12-11-2003, 09:14 PM

Remember there are differences in hardware implemention between boards. I mean that on the 4PCA3+, upping the VAGP seems to stabilize the NB.

For yourself, you're talking about the P4P800, right? Well something is strange with that board. It's the one that shows graphical artifacts with Hynix PC4000 ICs, above around 233-240, 1:1. I'd swear it's a bad AGP slot, but it goes away with other ram. It's a documented problem with PC4000.

But last time I had the board installed I had no problems with a 9700 Pro and max VAGP with BH-5 ram. It did seem to help a bit with the CPU overclock.

Frank E. BoNeS

12-11-2003, 09:59 PM

I had to up the vagp to get this board 3d stable...like I said, with my P4C800, raising the vagp did nothing for my OC. I heard this from a lot of others with this board as well....I'm not sure why this board would be different but I think Clevor made the most sense.....

Remember there are differences in hardware implemention between boards. I mean that on the 4PCA3+, upping the VAGP seems to stabilize the NB.

OC_Newbee

12-12-2003, 09:25 AM

Hi,

Been reading in here alot but haven't really post anything yet. People in here have great skill and knowledge when it come to Xtreme OCing and very helpful to other who needs help(nice place to be at).

Back to the topic, this 4pca3+ board does seem like it need a little bit more of a raise in the vagp voltage to get everything more stable. I have mine set from 1.8 all the way up to 2.0 and anything lower than that make my system not be able to pass certain test. Been running this board with Vagp of 1.8-2.0v for over 2-3 months now with no problem to the board or my video card. This board is great IMO compare to my previous IC7-G.

dioSRL

12-12-2003, 09:37 AM

but it is not clear to me, raising the vagp voltage has anything to do with the vid card? Or is for nb only?

Anyway, I think its safe to upper a bit(ok, maybe a bit more) of any voltages :), this board undervolts enough.

OC_Newbee

12-12-2003, 09:45 AM

On my 4pca3+ and from what I've read, it seem that the vagp voltage help stablize the nb more than just feeding it to the vid card( I'm not really sure though).

althes

12-13-2003, 02:23 PM

Yup upping the VAGP stabilizes the nb. I have a rev1.2 and I use 2.2vagp to do 295-300fsb.
Using HyperX at 5:4 with 3.2V.
Just using 1.55v on cpu 2.4C.
I may get a rev2.0 to test. But have a Ai7 I am going to test now see what it can do with some adata pc4000.

SPQQKY

12-14-2003, 11:27 AM

Okay, I'm a bit confused. EPoX tech over at AOA recommends the 4pda2+ over the 4pca3+ for OCing, but he won't answer my question (or hasn't bothered looking back at the thread) about the vcore on the 4pda2+. Does the rev2.0 allow higher vcore than 1.55v like the rev1.x does? Reading this thread is giving me a headache as you guys are jumping from what board has what.....HELP!

Frank E. BoNeS

12-14-2003, 11:33 AM

I don't know about the vcore on the 4pda2+ but the rev. 2 4pca3's vcore goes up to 1.85...

Grace

12-14-2003, 01:28 PM

The Vcore with both Rev2 boards goes to 1.85v, solid stable where you'll set them.

The goods of the Epox 4PDA2+ and 4PCA3+ boards are in my opinion that they are very solid and stable, extremely well build, fully featured and literally undying. Also the voltages that you can set are very wide, up to 3.4Vdimm is more than enough. Another very good thing is that their 5:4 divider can work very very high without issues. I run semi-stable over 315FSB with the RAM at 5:4 mode and I suspect the stability issues where the motherboard blacking out.

The bad parts of it : The VAGP is NOT joined with the VDD, as with other 865/875 that people don't seem to check, so you'll need to mod it in order to get some real help there. Also the Epox safety part set the clock generator to very safe levels, as result you will lose about 50-200MHz of maximum speed at the very same settings you'd use on any other motherboard. Reliability does have drawbacks.

I hope this helps.

Soulburner

12-14-2003, 04:49 PM

Originally posted by Grace
Also the Epox safety part set the clock generator to very safe levels, as result you will lose about 50-200MHz of maximum speed at the very same settings you'd use on any other motherboard.
I don't understand what this is supposed to mean? Set the clock generator to safe levels? :confused:

Grace

12-14-2003, 05:09 PM

Originally posted by Soulburner
I don't understand what this is supposed to mean? Set the clock generator to safe levels? :confused:

I didn't exactly put that right, I were trying to be simplistic.

I don't know the tech terms either to write them down. What I want to say is that it will not use some commands that under extreme stress can kill the motherboard. That is what makes the motherboard unable to run at the same top speed with many others. Most motherboards if overclocked to very high FSB and give a lot power to the CPU, they just die one (close) day, you turn them off and they never come back.

Soulburner

12-14-2003, 09:06 PM

Well hopefully I will be able to try out this board soon and see how much faster it is than my P4P800.

benonmsn

12-15-2003, 03:54 PM

i didnt liek my 4pca3+ rev 2

dioSRL

12-15-2003, 03:57 PM

and why's that? i had to choose betwen and p4c800 non deluxe(same price). I chose the epox.. and I dont feel sorry (yet :P)

Soulburner

12-15-2003, 04:16 PM

Yeah you didn't even say what you didn't like about it.

It should beat my current board in every aspect.

benonmsn

12-15-2003, 05:42 PM

Originally posted by Soulburner
Yeah you didn't even say what you didn't like about it.

It should beat my current board in every aspect.

it would not let me overclock as far as my p4c800, or max 3.. for that matter. therfore i did not like it

Frank E. BoNeS

12-15-2003, 06:45 PM

Originally posted by benonmsn
it would not let me overclock as far as my p4c800, or max 3.. for that matter. therfore i did not like it

Well, mine OC's far better than my P4C800...I don't know....maybe I just got a lemon :confused: . I really like the 4pca3 a lot and would recommend it to anyone.....

Here's (http://service.futuremark.com/compare?2k1=7334847) my 2k1 compare...with crappy card clocks...:rolleyes:

Grace

12-15-2003, 07:10 PM

Originally posted by Frank E. BoNeS
Well, mine OC's far better than my P4C800...I don't know....maybe I just got a lemon :confused: . I really like the 4pca3 a lot and would recommend it to anyone.....

You probably didn't have a good P4C800 yourself. Or did something wrong.

I said it before many times, I'll say it again since it seems necessary as people don't read all the posts before they post themselves. The 865/875 Epox motherboard will not overclock as far as the rest of the competition due to design. They will lose 50-200MHz top speed at the exact same settings.

...all of them.

Craig

12-23-2003, 04:38 PM

I'll be water cooling the CPU with a Cascade & the GPU with a ?? block. NB will be a good air cooler like the Swifty or Micro cool that The Dude just tested.

I like the fact this board has high voltage settings, I won't need to risk a vmod of it. I also like the fact that it's a hard board to kill. Those two features are enough to offset the 100+mzh or so I may lose at the max O/C levels. Those few extra mhz are not worth killing a board for. I frankly can't afford to kill to much stuff.

Many thanks to all taking part in this thread, you've shown me what I needed to know about this board.:toast: :thumbsup:

althes

12-24-2003, 11:14 AM

good luck with the board.

OC_Newbee

12-24-2003, 05:33 PM

Originally posted by Craig
I'll be water cooling the CPU with a Cascade & the GPU with a ?? block. NB will be a good air cooler like the Swifty or Micro cool that The Dude just tested.

I like the fact this board has high voltage settings, I won't need to risk a vmod of it. I also like the fact that it's a hard board to kill. Those two features are enough to offset the 100+mzh or so I may lose at the max O/C levels. Those few extra mhz are not worth killing a board for. I frankly can't afford to kill to much stuff.

Many thanks to all taking part in this thread, you've shown me what I needed to know about this board.:toast: :thumbsup:

Craig I feel exactly the same way as you are. I wasn't so sure about doing the vdimm mod on my IC7-G(scare I might break some part), So I end up selling that board and grab me this rev 2.0 instead. Beside this board treated me so well.

btv94

12-25-2003, 09:54 AM

OC_Newbee,

Can you bench and post the score from your setting/sig?

Just wonder how fast can 4pca3+ Rev 2.0 with 2-2-2-5 mem setting will be?

Thanks,

The Runner

12-25-2003, 10:08 AM

are these available in the UK? by the looks of it, they're rock solid stable and good for loads of HDD's. My kind of board as I'm not trying to be the next Fugger :)

Maranello

12-25-2003, 09:19 PM

Sweet thread. :thumbsup:

Answered alot of my questions. One thing though, does anyone know if newegg is shipping the v2.0 4pda3+? They list the 2+ board as being v2.

Frank E. BoNeS

12-25-2003, 10:25 PM

Originally posted by Maranello
One thing though, does anyone know if newegg is shipping the v2.0 4pda3+? They list the 2+ board as being v2.

I'm not sure about Newegg....you could contact them about it....I got mine Here (http://www.censuspc.com/Merchant2/merchant.mvc?Screen=PROD&Product_Code=4PCA3P&Category_Code=MOT0021)...pretty decent place.

Maranello

12-25-2003, 10:37 PM

Thanks for the link. :D

SPQQKY

12-25-2003, 10:41 PM

I got my rev 2.2 4pca3+ from the Egg.

dioSRL

12-26-2003, 05:00 AM

Originally posted by SPQQKY
I got my rev 2.2 4pca3+ from the Egg.

Could you please post a picture of the board? I am curious what was done in this revision.

Thanks

Maranello

12-26-2003, 10:37 AM

Originally posted by SPQQKY
I got my rev 2.2 4pca3+ from the Egg.

Ordered mine last night from newegg. I checked the reviews to see if anyone posted what version they got... saw v2.2. Got a 2.4C also.

I had a 4PDA2+ v1 that I loved... regret ever selling it.

ps24eva

12-26-2003, 10:41 AM

Originally posted by Grace
The bad parts of it : The VAGP is NOT joined with the VDD, as with other 865/875 that people don't seem to check, so you'll need to mod it in order to get some real help there.

what does this mean?

vicious

12-26-2003, 12:30 PM

how do i tell what revision my 4pca3+ is, i just got one from newegg

Maranello

12-26-2003, 02:05 PM

Originally posted by vicious
how do i tell what revision my 4pca3+ is, i just got one from newegg

Bottom of the last PCI slot there should be a sticker with the version #.

vicious

12-26-2003, 04:08 PM

turns out its in the bottom left hand corner not on the sticker on the last pci slot or at least thats how it is on mine and mine is a rev 2.2

Melkizedek

12-26-2003, 04:31 PM

my rev 2.0 simply sux... i dunno why but after a week of tests I have lost 600 (six hundreds) points in cpumark... really strange:smileysex

only 8047 in 3dmark03 before the lost...
http://service.futuremark.com/compare?2k3=1725191

now simply i can't score even with VGA at 530/430mhz...
:confused:

SPQQKY

12-26-2003, 05:23 PM

Must have been a bug in the first place as I have never seen 1500 points for the cpu score. My 4PCA3+ seems to be moving along just fine, however, I have a problem with my vdimm. No matter what I set it to, it will not go above 2.96 volts. Multimeter readings indicate 3.01v, but setting it to 3.1. 3.2 or 3.3 will not yield any extra voltages. This has been reported by a couple of people with rev 2.2's over at AOA. Rev 2.0 owners do not seem to have this problem. I am able t orun stable at 245MHz 1:1 with my TwinMos pc3700. I have some Geil Platinum pc4000 on the way that I hope will allow 260 1:1, that seems to be my cpu's limit (260 x 13 = 3380MHz).

Soulburner

12-26-2003, 07:09 PM

Yeah 1500 is not possible, a P4EE at 4Ghz scores 1000 points and you have no EE.

Your score is correct.

Soulburner

12-26-2003, 07:10 PM

Originally posted by SPQQKY
I have a problem with my vdimm. No matter what I set it to, it will not go above 2.96 volts. Multimeter readings indicate 3.01v, but setting it to 3.1. 3.2 or 3.3 will not yield any extra voltages.
How's your PSU?

SPQQKY

12-26-2003, 10:19 PM

PSU is fine, and would not be the reason why voltages would not change and I do mean ZERO change. This has to be a bug with the regulator. Perhaps they used a different one that actually does not allow the higher voltage but did not change the BIOS due to the other revisions.

Soulburner

12-26-2003, 10:41 PM

Actually it would be a reason for zero change, if your 3.3v rail was not supplying enough juice.

But if yours is fine then I don't know what the problem is.

Melkizedek

12-27-2003, 12:34 AM

Originally posted by Soulburner
Yeah 1500 is not possible, a P4EE at 4Ghz scores 1000 points and you have no EE.

Your score is correct.

Correct... but now my best score (even after a format) in CPUMark is 305 with 2-3 fps in cputest 2... how is it possible?
all the hardware is tested and working great:eek:
My Scsi Raid Controller is broken so I have attached only 1 SCSI HD (FUJITSU MAP3184 18gb 15000rpm ) at an old Adaptec U2W scsi controller. With this setup i scored 907-910 (4180 mhz, ram 5:4 2,2,2,5) and now (with the same setup) the best score i 305... I have also lost about 400 points in 3dmark03... Maybe this controller is dying also? Fujitsu speedest drive kill the controller?
AGP/PCI frequency is standrd...:smileysex

Soulburner

12-27-2003, 01:20 AM

Are other programs scoring correctly?

3DMark2001?

Melkizedek

12-27-2003, 01:56 AM

superPi (1mega) = 34 sec

http://melkizedek.altervista.org/immagini/shot_279.jpg

Soulburner

12-27-2003, 02:22 AM

Looks like just a 3DM03 bug.

Melkizedek

12-27-2003, 02:31 AM

Originally posted by Soulburner
Looks like just a 3DM03 bug.

So i have downloaded at least 15 bugged 3dmark03 set... 5 with flashget, 3 with DAP, 2 with Getright, 5 with Iexplorer ?
:mad: :mad:
formatted the system 2 times
enabled/disabled all system service (one at time)
reinstall all the whql or tweaked Radeon Drivers i know... every time uninstalled with drivecleaner...

i'm gettin mad

Soulburner

12-27-2003, 02:44 AM

Try 3DMark2001SE v330 and PCMark2002.

Maranello

12-29-2003, 09:39 AM

Just an FYI, ordered the 4PCA3+ and a retail 2.4C from newegg....

4PCA3+ is v2.2

2.4C is SL6Z3, pack date 11/14/03, FPO# 3332A583

Just incase anyone was curious what newegg was shipping.

vicious

12-30-2003, 12:51 PM

thats exactly what i got too

Clevor

12-30-2003, 03:26 PM

And I bet you will see 3.1-3.3 VDIMM only yield 3.07 or so. Got the same problem with a Rev. 2.2 I got from another vendor. :mad:

SPQQKY

12-30-2003, 03:50 PM

Dito on the vdimm, but it is better than only getting 2.85v like we see with other mainboards. I know it isn't yielding what is advertised (well, they don't really advertise it at all), but with the right memory, it's still a kicking board. Now I just wish my 2.6 was a better clocker. I will stick with the 4PCA3+ despite the vdimm issue. It's still a great board.

btv94

12-31-2003, 11:07 AM

Originally posted by Maranello
Just an FYI, ordered the 4PCA3+ and a retail 2.4C from newegg....

4PCA3+ is v2.2

2.4C is SL6Z3, pack date 11/14/03, FPO# 3332A583

Just incase anyone was curious what newegg was shipping.

2.4c sl6z3 or 2.8c from newegg . which one will be better for my rig?

Thermaltake 6000a case
Thermaltake Power supply w0011+PFC 480W
Epox 4PCA3+ V2.2
Thermalright SLK-947-U
and I have on the way 2x 512MB OCZ PC3500 Plat Limited Editions.

Thanks,

dioSRL

01-01-2004, 08:38 AM

I have reasons to believe that this mobo has less mem bandwith than others..

For example, I have 4300 MB/sec as reported by Sandra. Generic DDR400(single side) 2,5-3-3-7(SPD); 2.6C. Latest bios: 12/25/2003. I know sandra is not accurate, but still it seems a bit low. Other PCA3+ users, please post feedback at stock speed/spd. I think it's the only way to compare, since CPU's/Ram may differ a lot. Thanks.

As soon as possible I will test it with some OCZ Platinum. But I think even with double side and 2-2-2-5, it wount be more than ~4600MB/s...

SPQQKY

01-01-2004, 09:44 AM

I am getting pretty low bandwidth myself even at 252MHz 1:1 with my Geil Platinum pc4000....only seeing about 5890/5860. I would think I would get much more than that. I got 5820/5810 with the TwinMos pc3700 at 245MHz 1:1....I have a 2.6C and I will test out stock and spd sandra mem.

SPQQKY

01-01-2004, 09:53 AM

Here is my Sanra memory bandwidth scores at 200MHz 1:1 7-3-3-2.5
Higher than yours, dioSLR, so I think you have a problem somewhere, though mine is not as high as I would like.

dioSRL

01-01-2004, 01:10 PM

strange, your stiks are double sided? tomorrow I will test with some good memory, I'll post some screenies. Anyone else? Rev. 2.0 would be good. Latest bios also, while not required..

rlc

01-01-2004, 01:57 PM

althes

01-01-2004, 02:34 PM

I have to try a rev2.2 I have a rev1.2 and I like it alot. Takes alot of voltage to get the high bandwidth but I like it.:D

vicious

01-01-2004, 06:32 PM

I am getting 4815/4890 with everything at stock for sisoft memory bench

dioSRL

01-01-2004, 07:34 PM

what board revision/memory type/bios date/ etc etc etc?

vicious

01-01-2004, 08:32 PM

rev 2.2/mushkin lvl 2 black 3500

vicious

01-01-2004, 08:45 PM

i just flashed my bios to 12/25/2003 and now i get 4816/4800

Soulburner

01-01-2004, 09:02 PM

Originally posted by vicious
i just flashed my bios to 12/25/2003 and now i get 4816/4800
That's the same as before.

SPQQKY

01-01-2004, 10:54 PM

Should I add some heatsinks to the mosfets and southbridge? Do you think it would help with stability? I can run at 252MHz 1:1 but after about 10 minutes it will lock up. I know it helped my old 8RDA+, but will it help these Intel boards? I'm just a greenhorn with Intel.

Slickthellama

01-02-2004, 12:06 AM

I was thinking of upgrading my AOpen AX4SPE-N (865PE), Im not sure whether I should anymore. Rite now I have 280FSB totally stable on my 2.6C. The AOpen undervolts like .05volts! So i set it too 1.65 but it reads 1,6 avg in windows. So that's totally stable. I have a maze4-1. Should I get a 4PCAb+ rev 2.2? Or an Abit AI7? I have GeiL PC4300 memory with slow timings. But it doesnt seem to affect benchmark scores. I run 3:2 at 280 to make it stable. What should I upgrade too? And is anyone interested in buying this board? I'll sell it cheap. like 20$ and we split shipping.

Grace

01-02-2004, 12:14 AM

I don't think that upgrading this motherboard to another 865 is a wise idea, actually you won't gain anything. If you get an 875 motherboard, that would be something and I'm still against it since the performance increase is next to invisible.

If you want to run 1:1, consider the Abit motherboards.

Slickthellama

01-02-2004, 01:05 AM

the problem is my board restricts me to 2.85 voltage when the memory is rated at 3Volts for 24/7 usage. This is totally a waste if I cannot get a board with better voltage supply.

EDIT:plus there isnt a single person on the entire face of the earth with my motherboard so if i have a problem noone can help. I dont even know of a person living that owned a Pentium 4 AOpen motherboard.

btv94

01-02-2004, 01:17 PM

vicious

01-02-2004, 01:20 PM

Mine was retail and i got an SL6Z3 from newegg

SPQQKY

01-02-2004, 01:49 PM

I thought I was having data corruption due to overclocking, turns out my HDD took a dump on me. Dropped in a 40 Gig outa my sons system and it's running good again. It was a Maxtor 80 Gig ata133 w/8MB cache and if I am going to have to replace it, I may go SATA. Problem is, I am unable to get access to the backup partition and get to all my files......:(

btv94

01-02-2004, 01:50 PM

Thanks vicious,

Just ordered one retail box. Hopefully 2.4C SL6Z3 is a good one to overclock.

dioSRL

01-02-2004, 06:10 PM

2.6C
2x512 OCZ PC3500 Platinum Dual Ch.
Epox PCA3+ REV 2.0 / 1.2

There is a difference of ~150MB/s, exact same settings/hardware/bios.

Anyone else with rev 2.0?

SPQQKY

01-02-2004, 11:48 PM

Best I can muster outa my system. I really need a better NB cooler and no one answered if putting sinks on the southbridge and mosfets and regulators would help stability. I can't see it hurting I guess.

charlie

01-03-2004, 01:37 AM

what's the highest 5:4 bench 3D stable on the epox??
Curious :D And v2.0 vcore 1.85V? and Vdimm?? V-mods out there?

C

SPQQKY

01-03-2004, 08:12 AM

Versions 2.0 and 2.2 have vcore of 1.85v and v-dimm of 3.3v.......I'm sure a vmod is out there, but why? :P

btv94

01-03-2004, 02:12 PM

Originally posted by Clevor
And I bet you will see 3.1-3.3 VDIMM only yield 3.07 or so. Got the same problem with a Rev. 2.2 I got from another vendor. :mad:

Owner of rev 2.2. what vdimm are you running?

Clevor

01-04-2004, 09:33 PM

what's the highest 5:4 bench 3D stable on the epox??

I've tried all the revisions, they all overclock the same: 5:4 max is around 278-279 or so with good BH-5. 3:2 over 280 fsb is unstable, unlike the Abit and DFI boards. I've gotten to 282, 5:4 with the 4PDA2+, Rev. 2.0.

I am getting pretty low bandwidth myself even at 252MHz 1:1

The Epox boards don't get high buffered scores, but their unbuffered is second to none. More on that later.

I have to try a rev2.2 I have a rev1.2 and I like it alot.

You might want to hold on to that 1.2, particularly if it's one that overvolts the VDIMM and has the bug where 1.475 nets you 1.58-1.60 VCORE and 1.45 nets you 1.70. That's really all you need for a reasonable overclock, besides:

There is a difference of ~150MB/s, exact same settings/hardware/bios.

Yes, I notice that with the same BIOS, the Rev. 2.0 is shy about 100 Sandra unbuffered over the 1.1/1.2 boards, particularly at the first memory mark. However I found a way to gain this back, and then some: flash to the beta BIOS that has the aggressive memory settings like the 4PDA2+. Run Expert (Auto-Enabled-Auto). This will turn on PAT at all mem ratios. I get the following unbuffered marks at 248, 1:1, 2-2-2-5 at 3.2 volts with XMS3500. This is higher than at 250, 1:1 with the old BIOS on the fast 1.1/1.2 boards. I threw everything in 3D at this rig last night and it passed, including 30' into the IL-2 Sturmovik demo loop.

SPQQKY

01-04-2004, 10:34 PM

Which Sisoft Sandra are you using? I am using the Sandra 2004 standard from OCCafe. How do I get the unbuffered mem benches to run?

Soulburner

01-04-2004, 10:42 PM

Uncheck the following 9 boxes:

SPQQKY

01-05-2004, 10:43 PM

Thanks for the help, Soulburner. I ran some benches tonight (don't know why I kept running buffered Sandra :p: ), but I am pretty happy with the results. This doesn't rank up with the Xtremers, but it's enough to satisfy me.
Using the 12/16 BIOS that Clevor suggested, I was able to run 275MHz 1:1 with the Geil Platinum pc4000 @ 2.5, 7,4,4 and got some good buffered Sandra runs, knocked a couple of seconds off of my SuperPi and ran PCMark2002 successfully.

:toast:

Maranello

01-06-2004, 04:50 AM

Originally posted by SPQQKY
I thought I was having data corruption due to overclocking, turns out my HDD took a dump on me. Dropped in a 40 Gig outa my sons system and it's running good again. It was a Maxtor 80 Gig ata133 w/8MB cache and if I am going to have to replace it, I may go SATA. Problem is, I am unable to get access to the backup partition and get to all my files......:(

Had the same thing happen. Using an ATA133 40GB Maxtor and thought I was having stability issues... for some reason the drive kept corrupting. I use Norton Ghost so I was able to restore it and get some info off another partition before installing a Western Digital. Been running smooth ever since I installed the WD..... guess Maxtor just doesnt play well with the Epox.

drunkenmaster

01-06-2004, 07:19 AM

hmm, would the 865 version of this board actually be a better option than 875? As in, i'd guess they've got pat enabled but same as asus, at 1:1 and stock? or not, have they pulled more outta it?
I'm gonna do some thread searching, looking at reviews and seeing if i can get one semi cheap somewhere, but if you can help out quicker, so be it ;)

If it doesn't do PAt properly, then, it can't be better than the 875 can it? WHy would epox tech recommend it as the better option.

Maranello

01-06-2004, 07:37 AM

The only problem (other then the Maxtor) that Ive had with the board was my ATI at or above 300fsb... 299 it will run all day but 300+ the 9800 gives me errors or windows will dump my desktop to safe mode color and tell me the agp crapped out. From what Ive been told its the loss of the agp lock at 300+ that causes this and I tend to believe it since my TI4200 gives no errors..... but Ive only run my 4200 with the WD hdd so I dont know how much was the 9800 and how much was the Maxtor. Im running 3:4 since my Geil PC3200 wont do more then 220fsb 1:1 or 5:4.

Voltage readings are are another area that Ive noticed... doesnt bother me really tho. I can run the vcore at 1.60v and no matter what app I use it always shows 1.50v. Strange but like I said... doesnt annoy me.

Other then those Im very happy with my 3+.

Soulburner

01-06-2004, 09:37 AM

Originally posted by Maranello
Im running 3:4 since my Geil PC3200 wont do more then 220fsb 1:1 or 5:4.
You mean 5:4?

Voltage readings are are another area that Ive noticed... doesnt bother me really tho. I can run the vcore at 1.60v and no matter what app I use it always shows 1.50v. Strange but like I said... doesnt annoy me.
That's because its not reading vcore....up the VDDQ and check it again.

Maranello

01-06-2004, 11:29 AM

Meant 6:4 (CPUz shows 3:2)... sorry... early morning typo :D.

I ran 6:4 to get over 300fsb (200fsb memory at 300fsb CPU).

VDDQ is AGP... correct?

SPQQKY

01-06-2004, 04:25 PM

I had no problems running the RAM benches at 277 but I started getting file corruption and PCMark would halt at file copy anything above that. 277MHz locked me up in SuperPi.

Clevor

01-06-2004, 04:39 PM

I was able to run 275MHz 1:1 with the Geil Platinum pc4000 @ 2.5, 7,4,4 and got some good buffered Sandra runs, knocked a couple of seconds off of my SuperPi and ran PCMark2002 successfully.

Spooky, you can get even higher bandwidth by running Expert (Enabled-Enabled-Disabled), which is what I got 3700+ unbuffered (functional) on the 4PDA2+, however the 4PCA3+ was not stable with those tweaks. I only tried my ADATA PC450, not Vitesta sticks. The latter are faster sticks and might be able to handle it (doubt it though). You will get awesome unbuffered on PC4000 if you can run that tweak on either board.

Try checking 3D too. Most the Hynix 4.0 ns stuff can bench Sandra, Memtest, and 2D fine at 2.5-4-4-7 but won’t clear 3D until you drop down to 3-4-4-7 (my Soltek 875P board can run it at 3D though). The Vitesta sticks can usually do the faster timings 3D stable.

hmm, would the 865 version of this board actually be a better option than 875?

As I mentioned above, the 4PDA2+ gets even better unbuffered at 1:1 due to the ability to run the faster ram tweak. However at 5:4 and 3:2 PAT is partially enabled, even with the tweak. I also did not try running over 250, 1:1 so maybe full PAT is disabled there too (doubt it). So the 4PCA3+ will be the faster board if you have to run other than 1:1 (as long as ‘Intel FAST’ is enabled). You can see it in the unbuffered bandwidth. I see no diff between partially enabled and disabled.

Not sure there is an overclock limiter on the Epox boards - with the Beta BIOS or latest BIOS anyway. I recently tried a 2.60C and 2.80C and both overclocked virtually the same on the P4P800 as on the Epox boards. A limiter might be in effect over 280 fsb.

When you can run 250, 1:1 at 2-2-2-5, 5:4 won’t get you faster :D. I upgraded from a 2.40C at 250, 1:1 to a 2.80C at same settings (3.5 gig) and gained 1400 3Dmark2001 points! 1 GB XMS3500 here. Next step up is a 3.0C I got that can do 248 fsb or 3.72 gig 3D stable on air. And that one will be even faster on the 4PDA2+.

DDR500, 2-2-2-5 at 5:4 would be nice, with the CPU at 315 fsb or whatever, but let's be realistic . . . I'm on air after all. If you haven't figured out by now, these SD/CW boards really feel happier at 1:1.

Clevor

01-07-2004, 03:46 PM

I tried some Vitesta sticks while I had the board up. It would not run the aggressive memory settings that netted me 3700+ with BH-5 on the 4PDA2+, but I got there anyway :).

Highest I could bench the 3DMarks on the board is 283, 1:1, 2.5-4-4-7. Needs around 2.95 VDIMM. Got a 12,192 PCMark2002 score and a 39 sec SuperPI 1M time, which is significant, since I'm at 3.4 gig and I'm usually stuck there at 40 sec. Funny thing is if I relax timings to 3-4-4-7 the board won't boot!

This is also the first time I've been able to bench 3D over 279 fsb before, because of the 5:4 issue with BH-5, so the PC4000 helps with that.

For me, this is the highest (functional) unbuffered I've gotten with PC4000 stuff:

drunkenmaster

01-08-2004, 11:28 AM

i've had a little look around, is the v2 supposed to adhere to fmb1.5 standards, or even 2.0 like gigabyte? IE is it worth me trying this when i'm pretty much planning on getting a prescott in 18 days?

i couldn't see stuff plastered all over epox sites like adverts as you can see on other manu's sites.

S_T

01-09-2004, 06:44 AM

Just got my new 4pca3+, cause my IC7 died after bios flashing. I lost 11mhz fsb. With my Ic7 I could run two instances of prime at 252 now the bord will lock up even with 245 in under a minute no matter what CPU voltage I use.

Is there any new bios or beta bios out, which fixes this or upgrades the Sata Raid since the last bio is from August?

dioSRL I also have rev2, with timings 2,2,2,5 fast cs disable, I get 48xx/47xx. Nearly the same as with my IC7.

SPQQKY

01-09-2004, 06:52 AM

The 12/16 BIOS works best for me. I also had stability problems with the 8/19 BIOS, tried the 12/25 BIOS but my system would not boot wtihout keeping optimized defaults, but some have good luck with them.
Here you go.

4PC33C16 (http://www.aoaforums.com/forum/attachment.php?attachmentid=7711)

ps24eva

01-09-2004, 03:43 PM

Originally posted by drunkenmaster
i've had a little look around, is the v2 supposed to adhere to fmb1.5 standards, or even 2.0 like gigabyte? IE is it worth me trying this when i'm pretty much planning on getting a prescott in 18 days?

i couldn't see stuff plastered all over epox sites like adverts as you can see on other manu's sites.

anyone?

i'm curious too

S_T

01-10-2004, 05:01 AM

Thanks for the link SPQQKY, will try it out.

althes

01-15-2004, 08:11 PM

How is the new bios working out.

Clevor

01-15-2004, 09:04 PM

i've had a little look around, is the v2 supposed to adhere to fmb1.5 standards, or even 2.0 like gigabyte? IE is it worth me trying this when i'm pretty much planning on getting a prescott in 18 days?

Just in from an Epox tech on the AOA forums: yes, all revisions support Prescott, even 1.x.

Craig

01-18-2004, 01:05 PM

Well I've ordered 2X 256 Mushkin PC3500 Lvl 2 black RAm and sent a e-mail to Axion tech which is still listing the revision 2 board as being in stock, to please confirm they still do have the revison 2 boards of the EP 4PCA3+.

If they do I'm going to order one.

Then wait to order the CPU & vidio card when the Prescotts come out. I want a 3.2, question is though, which 3.2? On a good stable board with good BH- 5 RAM and plenty of vdimm for the RAM it would be fun to be one of the first to push the Prescott.

btv94

01-18-2004, 03:46 PM

I talked to axion last week, Rev 2 means 2.0 and higher. They don't guaranty just for Rev 2.0

S_T

01-19-2004, 04:28 AM

Just tested 12/16 bios, performance is clock by clock the same than on my IC7 and they finally updated the sata rom so raid 1 is now available.

Stability is for me a bit better. Got it now stable up to 244fsb but this is still 8mhz fsb less than on my IC7.

Craig

01-19-2004, 03:03 PM

Does anyone know of a site that has the rev. 2 for sure?

If I can't get the rev. 2, would I be better set with a IC7 Max3 for running 1:1?

Thanks

Grace

01-19-2004, 05:11 PM

Originally posted by Craig
Does anyone know of a site that has the rev. 2 for sure?

If I can't get the rev. 2, would I be better set with a IC7 Max3 for running 1:1?

Thanks

Yes, the IC7-MAX3 is a monster for running 1:1. Just make sure that you RAM can follow, you don't want to mess with dividers.

dioSRL

01-21-2004, 12:56 PM

What NB temps are u running with PCA3+ rev. 2.0? I have around 34C full load@12x285 .. It seems a bit low, does anybody know where the temp probe is? Default voltages, SLK900+Antec 92mm@5v.. 3 antec case fans in a 1080AMG. Room temp ~26C, cpu temp ~61C full load.

OC_Newbee

01-21-2004, 01:26 PM

Mine at 32-33C full load with 2 instance prime95. I have the tiger1 nb heasink/fan on mine with some As5. Also SLK900u +92mm Delta fan@ 7v and 6x 80mm Vantec Stealth case fans. Cpu temp is at 52-53C full load with case close at 1.7v in bios(1.65v in window), reading of the temp probe in my thermaltake hardcano 10. Hardcano and USDM show the same load temp, but idle USDM is way of buy like 10C more.

dioSRL

01-21-2004, 01:32 PM

I have stock heatsink on the mobo and it seems too low.. The cpu is priming @ 63C full load :D reading Epox USDM.. hope to hit 300FSB with this CPU (L319)

OC_Newbee

01-21-2004, 01:34 PM

Lap the stock heatsink and put As5 on it and see if you can lower the temp down a little. What is your vcore in the bios? 63C seem pretty high.

Grace

01-21-2004, 01:59 PM

The reading of the 'case' sensor is completely irrelevant with the northbridge temperature. If my memory serves me well, this sensor is located to the right of the CPU socket, near the memory dimms. It is meant to give an idea of the air temperature in the case, nothing else.

dioSRL

01-21-2004, 02:01 PM

also clockgen does not work with this new bios anymore.. I cannot set AGP...

dioSRL

01-21-2004, 02:08 PM

Originally posted by OC_Newbee
Lap the stock heatsink and put As5 on it and see if you can lower the temp down a little. What is your vcore in the bios? 63C seem pretty high.

1,488v. It has AS5, but with 5v fan I cannot ask for more.

The reading of the 'case' sensor is completely irrelevant with the northbridge temperature. If my memory serves me well, this sensor is located to the right of the CPU socket, near the memory dimms. It is meant to give an idea of the air temperature in the case, nothing else.

Yeah, I thought so.. But for other revisions? I have a friend with rev. 1.2 and it has for sure much higher temps, so it must be reading the NB..

A good thing for me is the 92mm fan witch I think it gives a bit of airflow to the NB also.. Take a look at the picture, the NB heatsink has the fins just under the airflow. Hoping to break 300+FSB on air.. with the same fan, but @12v. CPU&NB lapped. Grace, maybe this board will not fail to attend higher fsb, as you said earlier..

Grace

01-21-2004, 02:29 PM

Originally posted by dioSRL

Yeah, I thought so.. But for other revisions? I have a friend with rev. 1.2 and it has for sure much higher temps, so it must be reading the NB..

Grace, maybe this board will not fail to attend higher fsb, as you said earlier..

Other revisions may have a different winbond sensor or if I remember well it is the older revisions that had the sensor to the right of the CPU, so this one could have the sensor anywhere else, ex between the PCI slots as Asus usually likes to do.

And let me repeat myself again and again and again, so much that it will become boring one day I guess. The motherboard has NO problem reaching the highest of FSB frequencies, even with dividers it will not create memory based problems up to 320FSB+. It is one of the BEST motherboards for going at high FSB speeds. Of course over 300FSB it won't be 100% stable but no motherboard is, you'll have to use clockgen in order to go further. As I said earlier, the motherboard RESTRICTS THE D@MN CPU by 100-200MHz. You will be losing CPU TOP SPEED at the exact same settings set on most other motherboards.

EXAMPLE :

Got it now stable up to 244fsb but this is still 8mhz fsb less than on my IC7.

At 240-255MHz no motherboard should have any problem and no motherboard has a problem. Respectful motherboard manufacturers at least. The IC7 pushes the CPU 100-200MHz further than the 4PCA3+ just because it is another motherboard.

I hope I made myself clear enough this (8-9th?) time.

dioSRL

01-21-2004, 06:48 PM

sorry, but for me it's the same thing. If it restricts the CPU or if she cannot handle high FSB, it has the same result. BTW, just broke FSB300.. not prime95 stable, but anything else...

Grace

01-21-2004, 08:52 PM

Originally posted by dioSRL
sorry, but for me it's the same thing. If it restricts the CPU or if she cannot handle high FSB, it has the same result. BTW, just broke FSB300.. not prime95 stable, but anything else...

Sorry, but these are 2 completely different things.

If it couldn't handle high FSB speeds, with the higher CPU models it wouldn't have any problem following the rest while the 2.4C would be terrible for it for example. But since it restricts the CPU, no matter WHICH CPU you'll use, you'll be always 100-200MHz handicapped compared using the same CPU on any of the rest.

OC_Newbee

01-22-2004, 11:02 AM

Originally posted by dioSRL
sorry, but for me it's the same thing. If it restricts the CPU or if she cannot handle high FSB, it has the same result. BTW, just broke FSB300.. not prime95 stable, but anything else...

Nice OC, that's one special 2.4C you got there dioSRL. You can do all of that with stock voltage that is awesome. Mine needed 1.7v in the bios to do 285fsb and 1.725v for 288fsb, so I just leave it at 285 with 1.7v in the bios

Eru

01-23-2004, 03:32 AM

Where do u folks have your ram 1 and 3 or 2 and 4 (1 being the closest to the cpu)? My cpu does only 235 with 1.725 volts :mad: Looks like its a bad cpu..

dioSRL

01-23-2004, 11:44 AM

no matter WHICH CPU you'll use, you'll be always 100-200MHz

I understand perfectly. BUT now I have a 2.4 and the possible problem is NOW. Hoping for 4GHz on air with some proper mosfets&NB cooling + some Delta 80CFM on my SLK900. It passed prime95(for less than 15min-too hot for current setup, but still) at 310MHz so a big hope is for 334MHz wich I'm gonna need for 4GHz.... But I am almost certain I could never do this on air.. Wish me luck! I dont need it to be stable, just to load windows and.. a pifast/sandra/super pi would be nice also :D

dioSRL

01-24-2004, 11:50 AM

hm, I need memory(mine is 133MHz and doing 214MHz@2.9v) ... so I can improve my record. For stability... that's other thing, didn't work yet on it. So no additional NB cooling.. no mosfets, I dont even lapped the stock NB heatsink...

dioSRL

01-25-2004, 04:08 PM

Also, another problem. I tested two kinds of memory: Twinmos PC3200 2.5-3-3(1x512) and OCZ PC3500 Platinum non LE(2x512). With 3:2@ FSB300 MHz nothing happened. There's a "bug" with this board or what? the memory is running at 200MHz... it has twinmos chips/ CH-5 chips... none worked. Grace, any ideas?:P what is limiting my OC, CPU or Epox board? BTW I achieved this by BIOS, clockgen gave me some strange results and I did not use it anymore.

I am in trouble now, what memory should I buy(BH-5 its not an option) in order to keep this high FSB? I need 1GB, not very expensive..

With my sycron PC2100 ... watch the results: - sorry no pifast or anything more relevant.. It's not prime95 stable, maybe it's my memory failing at such high speeds.. and it's @1,6v cpu-z is reading wrong.

Grace

01-25-2004, 04:26 PM

Originally posted by dioSRL
Also, another problem. I tested two kinds of memory: Twinmos PC3200 2.5-3-3(1x512) and OCZ PC3500 Platinum non LE(2x512). With 3:2@ FSB300 MHz nothing happened. There's a "bug" with this board or what? the memory is running at 200MHz... it has twinmos chips/ CH-5 chips... none worked. Grace, any ideas?:P what is limiting my OC, CPU or Epox board? BTW I achieved this by BIOS, clockgen gave me some strange results and I did not use it anymore.

I am in trouble now, what memory should I buy(BH-5 its not an option) in order to keep this high FSB? I need 1GB, not very expensive..

CH-5 chips tend to create problems over 255MHz FSB. Twinmos are just re-labeled chips of other companies, you probably hit another CH-5. If it is a memory issue.

Now, I can't really tell you what is wrong unless I do throughout testing myself! Raise the Vagp high, to 1.8-1.9v if your video card can handle it (or install a PCI card if handy). That will help the chipset, granted that its cooling is at least average.

...you know that you won't be able to run over 299FSB at any time stable on that motherboard without unlocking the PCI lock, right?

......plus you've pushed the CPU a whole lot already.

BH-5 not being an option...hard to imagine any good RAM without CH-5 that is cheap as well AND decent! A-Data Pc4000 that is good in order to run 1:1 may (well, most probably will) not reach 300MHz. Most black out at 290-295.

dioSRL

01-25-2004, 04:29 PM

thanks, but I don't need 1:1, just 3:2.. to hell with that CH-5 .. but ofcourse if it can hit 300MHz.. it would be a good option for 1:1.. i remember now(becouse u noticed it) the vitesta a-data pc4000 running @300MHz in a friend computer.. yeah, I probably would give it a try.

...you know that you won't be able to run over 299FSB at any time stable on that motherboard without unlocking the PCI lock, right?
I am trully aware of this, but is running rock stable(yeah.. no prime yet) @ 321FSB, directly from bios. Dont ask me how, it does. I have to find some other memory to fully test it.

Also, how unlocking can be done, without clockgen? I really dont like it.

Grace

01-25-2004, 04:59 PM

Originally posted by dioSRL
thanks, but I don't need 1:1, just 3:2.. to hell with that CH-5 ..

I am trully aware of this, but is running rock stable(yeah.. no prime yet) @ 321FSB, directly from bios. Dont ask me how, it does. I have to find some other memory to fully test it.

If you don't need 1:1, merely 3:2, then........

Samsung Original Pc3200 perhaps?

321FSB stable is literally impossible, especially directly from the BIOS. Your AGP/PCI is unlocked for sure at that speed as well, the devices can just handle whatever it is set at. Also, stable would mean dual Prime95 stable for 24 hours straight for example, not a boot and benchmarks.

...plus you should slow it down, over 300FSB for long won't make your motherboard live long enough for you to enjoy it.

dioSRL

01-25-2004, 05:06 PM

plus you should slow it down, over 300FSB for long won't make your motherboard live long enough for you to enjoy it.

Define "long" please..Thats for sure, I dont keep it 24/7, I just need it for a couple of testing with a Radeon 9800XT and some Corsair BH-5 witch I will borrow from a friend in 2-3 days.. But why should the mobo die? It is proper cooled, under a big window and outside there are like -5C .. And the 80CFM Delta is throwing cold air all over NB and mosfets. I have like 40C full load(2xprime~5-10min)

Also, stable would mean dual Prime95 stable for 24 hours straight for example, not a boot and benchmarks.
You are perfectly right, I will do that as soon as I get proper memory .. hoping not to be snowing that night :D

what tool should I use, beside clockgen? Thanks.

Grace

01-25-2004, 05:15 PM

Originally posted by dioSRL
But why should the mobo die? It is proper cooled, under a big window and outside there are like -5C ..

what tool should I use, beside clockgen? Thanks.

Ask it yourself when it does. Usually they don't give an answer...:p:

Cooling is NOT everything. Processors die in Prometeia's, VGAs under TECs. Because it is not hot it doesn't mean it cannot die! There are a whole lot more factors than heat. Over 280FSB and/or 3.8GHz is way too much for a motherboard to provide. Defining long..from 2 weeks to 12 months, noone can tell.

And the clockgen is the only effective one, as far as I can tell....:(

dioSRL

01-25-2004, 05:21 PM

hah, lol, sure cooling isnt everything.. but at stock voltages for the mobo, 2.8v for the memory and 1.6v for the cpu..thats strange.. cpu dies of high voltage(SDNS) memory.. the same. That's why I am curious.. still she's under warranty :P but for sure I dont want to loose a good board.

Grace

01-25-2004, 05:27 PM

Originally posted by dioSRL
hah, lol, sure cooling isnt everything.. but at stock voltages for the mobo, 2.8v for the memory and 1.6v for the cpu..thats strange.. cpu dies of high voltage(SDNS) memory.. the same. That's why I am curious.. still she's under warranty :P but for sure I dont want to loose a good board.

They die from overloading, EMI, pipeline puncture (which happens usually due to the first 2) and other things as well.

Alexandrus

01-26-2004, 12:28 PM

Originally posted by Grace
Now, I can't really tell you what is wrong unless I do throughout testing myself! Raise the Vagp high, to 1.8-1.9v if your video card can handle it (or install a PCI card if handy). That will help the chipset, granted that its cooling is at least average.

I thought the VDD and VAGP were not joined on this mobo.

Originally posted by Grace
...you know that you won't be able to run over 299FSB at any time stable on that motherboard without unlocking the PCI lock, right?
And you'd be amazed how stable the rig really is at 321MHz FSB, I just can't believe it, this guy has tons of good luck, or maybe it's my luck, I was with him when he got the mobo, and I found that L319 2.4C for him :)

Originally posted by Grace
321FSB stable is literally impossible, especially directly from the BIOS. Your AGP/PCI is unlocked for sure at that speed as well, the devices can just handle whatever it is set at. Also, stable would mean dual Prime95 stable for 24 hours straight for example, not a boot and benchmarks.
You're right, but as long as 3DMark and Sandra run just fine, I think the system is pretty stable for everyday use, although Prime will not run. It just prooves that AGP/PCI lock still works, somehow.

dioSRL

01-26-2004, 12:32 PM

yeah, but I did not use any pci device, only my hdd. No onboard lan, no onboard sound, nothin. Even the mouse was on ps2, not USB. Next time I will disable all from bios.

Alexandrus

01-26-2004, 12:35 PM

Still, the AGP card should be affected. And we should really try a PCI sound card, to see if the PCI lock is still on.

Grace

01-26-2004, 12:39 PM

Originally posted by Alexandrus
I thought the VDD and VAGP were not joined on this mobo.

And you'd be amazed how stable the rig really is at 321MHz FSB, I just can't believe it, this guy has tons of good luck, or maybe it's my luck, I was with him when he got the mobo, and I found that L319 2.4C for him :)

You're right, but as long as 3DMark and Sandra run just fine, I think the system is pretty stable for everyday use, although Prime will not run. It just prooves that AGP/PCI lock still works, somehow.

They aren't joined, the higher Vagp helps the unlocked AGP over 300FSB.

I can't be amazed anymore, I've done far more impossible things.:p:

And if Prime95 can't run stable, the system is NOT stable. Sandra and 3dMark are not applications that actually test the system stability, merely performance. They don't have any kind of error recognition as well. And the AGP/PCI lock won't work at 303FSB and over on any motherboard, it just proves that his devices handle the flying AGP/PCI, whatever it is set to now. You better manually set it to 72/36 or so.

PCI sound cards will not test the PCI lock, most can handle 40+MHz. Many AGP cards can handle 76+MHz easily as well. If you want to test the lock, set a RAID on the ICH5-R SATA ports. These die if the PCI goes even to 34.

Edit : Don't try it, I did that before you, on the very same motherboard.

dioSRL

01-26-2004, 12:42 PM

thats why for instance I dont want any kind of raid in my system. When I'll have such, I'll buy some 2.8(SL6Z5 if I can find). And if i manually set to 72/36 it wount get unlocked?

Grace

01-26-2004, 12:44 PM

Originally posted by dioSRL
And if i manually set to 72/36 it wount get unlocked?

No, it stays within +/- 1MHz from the setting.

Alexandrus

01-26-2004, 12:50 PM

Well, strangely enough, at 334MHz FSB on a IC7 mobo, with SATA RAID on ICH5R, without AGP/PCI set to 72/36, the system is stable enough for a screenshot :) And for moving the mouse a few centimeters :)

Grace

01-26-2004, 12:59 PM

Originally posted by Alexandrus
Well, strangely enough, at 334MHz FSB on a IC7 mobo, with SATA RAID on ICH5R, without AGP/PCI set to 72/36, the system is stable enough for a screenshot :) And for moving the mouse a few centimeters :)

Been there done that....more than that actually...I'm trying, oh well.

Do THAT (http://www.xtremesystems.org/forums/showthread.php?s=&threadid=24520&perpage=25&pagenumber=3) and then I'll praise you.:D

dioSRL

01-26-2004, 01:04 PM

we are all on air here..

Grace

01-26-2004, 01:08 PM

Originally posted by dioSRL
we are all on air here..

Yep, HiPro5 where on air as well. The processor cooling doesn't matter for the northbridge. Anyway his were at a mere 3GHz.:D

dioSRL

01-26-2004, 01:14 PM

ok, gimme some ES(that's why I thought he was cooling with something powerfull... 371FSB its not "possible" on air). and some

.......and my new pi-fast on this Mobo........

ABIT IC7-G
Modded Prommie with R507 gas
2 x 256MB Kingston Value Rams PC2700(DDR333) with BH6 chips on board

with my current cpu&cooling I think I cannot go more than 4GHz... with some luck.

Alexandrus

01-26-2004, 01:42 PM

Heh, with an ES CPU, some BH-5 RAM such as OCZ Platinum LE, and of course, some other hardware(fans and stuff, volt mods), nothing is out of reach, not even Hipro5's record(I'm just saying, but probably that's the best FSB we'll ever see on i875P).

Grace

01-26-2004, 01:56 PM

Originally posted by Alexandrus
Heh, with an ES CPU, some BH-5 RAM such as OCZ Platinum LE, and of course, some other hardware(fans and stuff, volt mods), nothing is out of reach, not even Hipro5's record(I'm just saying, but probably that's the best FSB we'll ever see on i875P).

No matter what you have, it is all useless until you know how to fully use them I'm afraid....For example, you mention BH-5 RAM when it has nothing to do with the max FSB, plus BH-6 is better for top end overclocking!

Even if you had that system and an ES, you wouldn't get anywhere close to that. Not specifically you, but any person and I include myself, except very very few people. But of course, my opinion and only. HiPro5's knowledge and skills are not anything that can be underestimated.

Alexandrus

01-26-2004, 02:40 PM

Why would BH-6 be better than BH-5, I have no idea. For high FSB any PC4000 or higher RAM will do, provided the 3:2 divider and loose timmings are used.
Plus, Hipro5's rig is not Prime stable either, if I am not mistaken, right ?

dioSRL

01-26-2004, 02:46 PM

Even if you had that system and an ES, you wouldn't get anywhere close to that. Not specifically you, but any person and I include myself, except very very few people. But of course, my opinion and only. HiPro5's knowledge and skills are not anything that can be underestimated.
I totally agree, but at least I can hope at it. If you play long enough with good stuff, there's almost nothing you cannot do, sometimes. When you change cpu/mem like socks, it is almost impossible you could not manage to find some good ones. I am limited at what I buy, with limited money (500USD/month, including my wife :) and all the house) And I have let's say good results. If I would have at most 3k and lot of places to buy stuff... eh it will be completely different.

Grace

01-26-2004, 03:07 PM

I totally agree, but at least I can hope at it. If you play long enough with good stuff, there's almost nothing you cannot do, sometimes. When you change cpu/mem like socks, it is almost impossible you could not manage to find some good ones. I am limited at what I buy, with limited money (500USD/month, including my wife and all the house) And I have let's say good results. If I would have at most 3k and lot of places to buy stuff... eh it will be completely different.

You got the point. Enjoy what you have at its maximum and don't overworry yourself with that.

Originally posted by Alexandrus
Why would BH-6 be better than BH-5, I have no idea. For high FSB any PC4000 or higher RAM will do, provided the 3:2 divider and loose timmings are used.
Plus, Hipro5's rig is not Prime stable either, if I am not mistaken, right ?

BH-6 stretch more than BH-5 when high voltages are used. PC4000 RAM is not ideal at all, the Hynix chips that those carry usually become crazy when running at 3:2.

Also I don't think his is Prime stable, but his is the world record. No system over 300FSB can become Prime stable that easily, certainly not anywhere near that. I merely mentioned it since you boasted 330FSB, I wanted to tell you that this is not something that would make me feel wondered, neither consider it strange as you said.

Here, that screenshot (http://www.vr-zone.com/guides/Intel/Northwood/page2.htm), third from the top of mine is 9 months old...

dioSRL

01-26-2004, 03:16 PM

congrats, nice cpu :toast: too bad it wasnt on a Epox :P And that fsb alexandrus mentionned it is on a IC7 rev 1.0 .. 334MHz L321@1.7v

Grace

01-26-2004, 04:02 PM

Originally posted by dioSRL
congrats, nice cpu :toast: too bad it wasnt on a Epox :P And that fsb alexandrus mentionned it is on a IC7 rev 1.0 .. 334MHz L321@1.7v

It was very good but not stable there at all, plus I burned it to oblivion months ago.

Now THAT'S (http://www.xtremesystems.org/forums/showthread.php?s=&threadid=27420) what I consider a good CPU.:D

dioSRL

01-26-2004, 04:10 PM

yeah :) but if I hit 4GHz with air will be better :P All I need to try it is one damn 256MB stick. A good one. With the diference ofcourse, your rig is allday, mine.. not! I simply cannot stay with 80CFM fans in my system.. I ran 24/7 a quiet 92mm@5v.

Alexandrus

01-27-2004, 12:29 AM

Grace, not all PC4000 have Hynix chips, some have some Samsung chips. Anyway, BH-6 are good, I admit, but for a Sandra bench and a screenshot ... don't really need tight timmings ;)

As for the records, well, everything is air cooled on the two rigs I mentioned, dio's and the other one. Water would be an improvement, phase change would be even better.

PS. Nice overclocks you managed there.

Grace

01-27-2004, 03:46 AM

Originally posted by Alexandrus
Grace, not all PC4000 have Hynix chips, some have some Samsung chips.

Yep, Kingston HyperX Pc4000 for example, which switched to Hynix as soon as they got :owned: by the competition. :D
Hynix chips can take higher speeds easily.

dioSRL

01-28-2004, 01:20 PM

beah, I hit a wall at 322MHz .. No matter how much voltage I throw at it it will not pass this FSB speed. I really hope it's the memory.. But at 1:1 my "crappy" 133MHz jumps to 230MHz.. At FSB 334MHz the memory should be OK ~216MHz... .. maybe the memory controller is under stress and the memory cant handle the situation anymore :) I found some Corsair PC3500 with BH-5, unfortunately only one stick.. I will do more testing in the weekend.

dioSRL

01-30-2004, 03:49 PM

ok with some BH-5 corsair pc3500 (3:2)... and some 1.7v CPU .. that's the end of line. Maybe I will make it stable to bench something(If I am very very lucky..).. But I am tired already ;) Me wants some good 2.8C and some good wc.

OC_Newbee

01-30-2004, 04:25 PM

Originally posted by dioSRL
ok with some BH-5 corsair pc3500 (3:2)... and some 1.7v CPU .. that's the end of line. Maybe I will make it stable to bench something(If I am very very lucky..).. But I am tired already ;) Me wants some good 2.8C and some good wc.

This is still on air right and the 4pca3+ Rev 2.0, WOW!
That's one heck of a chip. Nice OC mann!

dioSRL

02-04-2004, 04:07 PM

thanks, man. But I am rather ANGRY! I cannot manage to find some ram to help me(that BH-5 above was 1x256...). Now I am testing 1GB of Corsair PC4400PT and I am not happy AT ALL. Beside the fact it wount do 275MHz 1:1(2.9v), even @ 5:4 FSB 300MHz it's very very unstable. Since it's the first ram that I ran @ 5:4 and 1:1 on this board, may I ask, there is a BOARD issue or Corsair is just crappy? Mann its the dummiest memory I've ever seen, it will not POST at SPD, FSB200, 1:1, because at this speed the SPD is 3-4-4-6 and should be 3-4-4-8. I have to throw in my "magic stick" of PC2100, and set the timings manually. Crappy EXPENSIVE focking ram! Or is just me playing stupid?

Mann, I WOULD KILL for 1GB of BH-5 or BH-6!:frag: :slash:

OC_Newbee

02-04-2004, 05:49 PM

Yeah bh-5/bh-6 is the way to go with these 865/875p board. I wish I have your chip right now, boy I say boy......my system would be bangin......................300+fsb at 5:4 2-2-2-5. My two stick of corsair can do 247fsb 2-2-2-5 at 3.3v. Maybe one day I will be lucky like you and find a golden chip like that. You should try and get some hyperX pc3000, those might be still around somewhere, or the Muskin level 2. Well anyway good luck in finding the right memory to match your cpu.

Clevor

02-04-2004, 06:15 PM

After four tries, I finally got an MO step 2.40C that does 301 fsb, on air too. As you know, I got three 4PCA3+ boards, including a Rev. 2.0. And XMS3500 that do 250, 1:1, 2-2-2-5 at 3.2 volts. :banana:

But nah, none of my Epox boards run 5:4 or 3:2 over 279-282 or so. What a waste of 3.3 volt VDIMM. I'll give it a shot of course. And since I just got a Corsair HydroCool, I'll try that to see if it improves the 5:4.

These MO steps are funny. I have a 3321 and 3326 CR, both of which run 300 fsb totally clean in Memtest at about the same VCORE. But the 3321 won't boot WinXP till 293 or so, and has to drop down to 288 to be 3D stable, whereas the 3326 boots fine and is 2D stable at 301. Benches all 3D at 298. It's currently on an IS7 and has that 299+fsb-unstable thing. Will be trying that 72/36 bus and/or Clockgen trick.

dioSRL

02-04-2004, 06:19 PM

But nah, none of my Epox boards run 5:4 or 3:2 over 279-282 or so. I am running 3:2 with crappy ram @310FSB with no problems. Even higher, but I did not test it enough to say. thats really strange and annoying.

Soulburner

02-04-2004, 06:23 PM

Um my crappy Costa Rica chip from last June will run 3D01 at 286 on air....1.7v...

dioSRL

02-04-2004, 06:27 PM

well.. clevor, did you tried to run that xms single channel?

Clevor

02-04-2004, 06:40 PM

I am running 3:2 with crappy ram @310FSB with no problems. Even higher, but I did not test it enough to say. thats really strange and annoying.

It's possible the higher strung the ram, the worse the 5:4 problem is. I really need to get some Infineon CAS 2.5 sticks to test this out. :)

I have seven Epox boards, and I cannot run 5:4 or 3:2 over 279-282 on any of them. I've tried Corsair, HyperX, and SanMax BH-5, as well as SanMax BH-6, and OCZ PC3700 Golds. 2D/Memtest/SuperPI is fine, but 3D kicks out or the system randomly reboots above 279 on the 4PCA3+ and 282 on the 4PDA2+.

I simply run 250, 1:1 with a 3.0C; I get a bigger CPU anyway over 300 fsb on a 2.40C :D. From what I've seen so far, 250, 1:1 gives around 3600 unbuffered, and I won't touch that at 300, 5:4, DDR480 at 2-2-2-5.

Moreover, if you can get 250, 1:1 with full PAT on an 865P board, you will get even better bandwidth than any 875P board, but it's harder to do. So far I've managed 245, 1:1, 3D benchable.

Clevor

02-04-2004, 06:45 PM

well.. clevor, did you tried to run that xms single channel?

Heh heh, you must not have seen my other posts. Yes, seven XMS3500 sticks do DDR500, 504, 506, 510, 510, 512, and 518. At maybe 3.28 volts. But up there, it's harder to do DC.

You know my two best sticks I thought were my worse and I was flogging them the last six months, really abusing them. They don't do the best mid-range but they top out at DDR512 and 518. So maybe burning in these XMS3500 helps. So I am flogging the 504/506 sticks now. The two 510 sticks have the best mid-range, and only have 3 hours on them.

OC_Newbee

02-04-2004, 06:56 PM

Originally posted by Clevor
After four tries, I finally got an MO step 2.40C that does 301 fsb, on air too. As you know, I got three 4PCA3+ boards, including a Rev. 2.0. And XMS3500 that do 250, 1:1, 2-2-2-5 at 3.2 volts. :banana:

But nah, none of my Epox boards run 5:4 or 3:2 over 279-282 or so. What a waste of 3.3 volt VDIMM. I'll give it a shot of course. And since I just got a Corsair HydroCool, I'll try that to see if it improves the 5:4.

These MO steps are funny. I have a 3321 and 3326 CR, both of which run 300 fsb totally clean in Memtest at about the same VCORE. But the 3321 won't boot WinXP till 293 or so, and has to drop down to 288 to be 3D stable, whereas the 3326 boots fine and is 2D stable at 301. Benches all 3D at 298. It's currently on an IS7 and has that 299+fsb-unstable thing. Will be trying that 72/36 bus and/or Clockgen trick.

Three 4PCA3+ and none is good over 282 with both 5:4 and 3:2. Man I'm Sorry to hear that you got bad luck with these 4PCA3+ board bro. I would have give up and grab a different board already after the second fail to go higher.

Clevor

02-04-2004, 07:08 PM

I would have give up and grab a different board already after the second fail to go higher.

But nothing runs BH-5 finer. Even at 277-282, 5:4, 2-2-2-5 the benchmarks are excellent.

The 4PCA3+s I have are different revs, and with Epox boards, most times the early revs are keepers. Reason I keep the Rev. 1.1 is it has the best VDIMM (overvolts). Reason I keep the 1.2 is it has the bug where I can get 1.58-1.6 and 1.7 VCORE with the 7-03 BIOS. Reason I keep the 2.0 is it has the VCORE up to 1.85.

Moreover, I still have to check this out, but the Rev. 1.x boards seem to give better unbuffered bandwidth over the 2.xs, same BIOS.

It's easier finding a good 3.0C. I only tried two, both are late LxxAxx Malays, and both do 3.6-3.7 gig at around 1.58 volts, on air too. I think the lousy cores they put in the 2.6-2.80Cs. I've had no luck with 2.60-2.80Cs. They barely do 3.4 on air, with lots of volts.

dioSRL

02-05-2004, 02:24 AM

Heh heh, you must not have seen my other posts. Yes, seven XMS3500 sticks do DDR500, 504, 506, 510, 510, 512, and 518. At maybe 3.28 volts. But up there, it's harder to do DC

Yeah, I am aware of this :) So, even in single channel it wount make it 3:2/5:4?

Clevor

02-05-2004, 03:41 PM

So, even in single channel it wount make it 3:2/5:4?

Actually I thought about this last night. I do recall booting at DDR518 and 3D was OK, but this is SC at 1:1. I took a look at the bandwidth and it was around 2450 unbuffered. But even if 3:2 and 5:4 would work, too much of a performance loss in SC.

Another possibility is all my sticks are 512 MB, but the SanMax BH-6 I tried were 256.

As you may know, this problem is not isolated to Epox boards. The Abits will run 3:2 fine but I've never gotten much beyond 280 at 5:4. The DFI will only run 3:2 with BH-5, but with PC4000 it runs 5:4 OK too. The Soltek boards, both the 865/875P, won't run 5:4 or 3:2 with BH-5. The A4XC-MaxII won't run over 270 with ANY ram, either at 5:4, 3:2, or even 1:1 with PC4000! It's the only board I know that can't even run PC4000 stable over 270 at 1:1. As usual Memtest is clean so the ram and CPU looks good, but 3D won't clear and the board exhibits random rebooting - typical 5:4 bug.

But there is more than one way to skin a cat. I plan to put the 300 fsb chip in the P4C800-E and do the VDIMM mod to try to get maybe 3.0-3.1 volts. I should be able to run 5:4, 2-2-2-5 in the 300 area then. It will probably beat a 2.80 C (3.5 gig) on the 4PCA3+ at 250, 1:1, 2-2-2-5, but not a 3.0C (3.75 gig). The 2.80C might give it a run for it's money though.

I already ran 298, 3:2, 2-2-2-5 on the IS7 and was not impressed with the 3DMark scores. BTW, you can run BH-5 at SR-Enh-A-D-D and it partially enables PAT. Unlike the 4PDA2+, it works and give extra unbuffered. Heck, I got 3100 unbuffered at DDR384! At 277, 5:4, 2-3-2-5, I got 3300+, a bit more than the 4PCA3+, same settings at 2-2-2-5! But I got a 41 sec SuperPI time compared to 40 on the Epox 875P, so either the unbuffered is bogus or the 865P chip is holding it back.

OC_Newbee

02-05-2004, 05:51 PM

Originally posted by Clevor
Heck, I got 3100 unbuffered at DDR384! At 277, 5:4, 2-3-2-5, I got 3300+, a bit more than the 4PCA3+, same settings at 2-2-2-5! But I got a 41 sec SuperPI time compared to 40 on the Epox 875P, so either the unbuffered is bogus or the 865P chip is holding it back.

3100 unbuffered at 384mhz 277fsb ,5:4 2-3-2-5, is this from you 4pca3+ or 4pda2+. My 4pca3+ unbuffered at 287fsb 5:4 459mhz 2-2-2-5, I got around 3100. The is weird, I've try every bios that Epox released from 7/03(4PC33703), 8/19(4PC33819), 12/16(4PC33C16) And 12/25(4PC33C25) and they all got about the same numbers. On the lastest bios I even have my MAM setting like this- expert: enable, enable, disable.

Clevor

02-05-2004, 06:12 PM

OC-Newbie, hate to tell you this, but it's a fact of life: it's the 256 MB sticks. :(

There is a significant difference in unbuffered bandwidth between single-sided and double-sided sticks with the SD/CW boards. That's why I got rid of all my SS sticks, particularly with PC4000, and got 512 MB sticks. Fortunately, my XMS3500 were all 512 MB leftover from the 845PE days of single channel boards.

For example, I can get 3550 unbuffered with PC4000 at 273, 1:1, 3-4-4-7 using 512 MB sticks. With SS sticks that would be 3000-3100. I've tested 285+, 5:4, DDR470 or so at 2.5-3-3-7 and got 2400-2700 unbuffered on bad boards with SS sticks. Man, an RDRAM board running 156/4x can match that! So if you have to beg, borrow, or steal . . . get DS sticks. From my results in DC with my XMS3500, the denser sticks don't seem to affect the overclock that much with quality ram.

I got the 3100 unbuffered on the IS7 board at 298, 3:2, 2-2-2-5 or DDR384 running SR-Enh-A-D-D. Also got the 3300+ at 277, 5:4, 2-3-2-5. As I mentioned, with the 4PCA3+, I get around 3250 with 512 MB BH-5 at the same settings but at 2-2-2-5, so that is good marks on the IS7, an 865P board. But doesn't necessarily translate into performance. Perhaps Abit is rigging the BIOS.

OC_Newbee

02-05-2004, 06:45 PM

O I C, so sandra unbuffered score favored the DS stick more huh. Does that also affect real life stuff? I mean do SS performance worse than DS stick at anything(game, app, etc) or just Sandra.
That explain why you got 3100 unbuffered on the IS7 since your fsb is at 298 and mine is at 287, even at 3:2 your fsb is much higher. That's a nice soore though on your 4PCA3+, 3300+ at only 277 5:4 2-3-2-5. I have one stick of 512mb of bh-5 and to bad I can't find another matching stick(Imma try harder to find that sucka).

Thanks mann.

Clevor

02-05-2004, 07:19 PM

O I C, so sandra unbuffered score favored the DS stick more huh. Does that also affect real life stuff? I mean do SS performance worse than DS stick at anything(game, app, etc) or just Sandra.

Yes, because where unbuffered bandwidth goes, so does performance. If you compare 3DMark scores at 273, 1:1, 3-4-4-7, DS versus SS, 3500 versus 3100 unbuffered, the scores go up with the DS sticks.

CAS settings do interact with unbuffered bandwidth so you want the highest unbuffered at the fastest CAS settings you can muster.

That explain why you got 3100 unbuffered on the IS7 since your fsb is at 298 and mine is at 287, even at 3:2 your fsb is much higher.

Sandra unbuffered is not affected by fsb so much as DDR and CAS settings. You should get the same unbuffered at DDR500, 2-2-2-5 whether by 315 fsb, 5:4 or 250 fsb, 1:1.

298, 3:2 is DDR384 whereas 287, 5:4 is DDR459. I get the same unbuffered as you because of my DS sticks, even though you are at DDR459-DDR384=75 mhz higher!

That's a nice soore though on your 4PCA3+, 3300+ at only 277 5:4 2-3-2-5.

Nope, that is on the IS7 with PAT partially enabled, 512 MB sticks of course. I get 3250 on the 4PCA3+ at 277, 5:4, 2-2-2-5, same sticks. However the 4PCA3+ is faster by 1 sec at SuperPI 1M, which is fairly significant.

OC_Newbee

02-05-2004, 08:21 PM

Do me a favor, what kind of video card you have? If you have something similar to my 9500np softmod to a 9700 or maybe a 9700 at stock would be good. Run 3dmark01se and see what your score for a stock video card. My 9500np to 9700 at stock(276/270) score, 16600-16690(somewhere between there). I just want to see how much of a lost for my Single sided stick compare to a double sided stick.

I run 3dmark at the speed in my sig right now only the video card at stock. Please run your at same speed you got that unbuffered score at 277fsb 5:4 2-2-2-5.

Clevor

02-05-2004, 08:51 PM

I don't have a 9700 Pro. Mine died and Newegg replaced it with a 9800 Pro. :banana: Only got two 9800 Pros and a Geforce 5900FX (I have five rigs after all). My 9800 Pros overclock like crap, maybe 428 max, still with artifacts.

But I can tell you going from 3300 unbuffed (DS) to 3000 unbuffed (SS), at 277, 5:4, 2-2-2-5 you lose maybe 175-250 3DMark2001 points.

GRiM

02-13-2004, 04:48 PM

Had to post as i now have information about how these epox boards overclock.
I can overclock my 2.8C just as far on 4pda2+ as i could with AI7. Same volts and everything else.

gouda96

02-13-2004, 08:15 PM

My 4pca3+ can do 5:4 up to 310 just fine. I am using 2x256 khx 3200 bh-5 @ 3.2v 1:1 my ram could do upto 520 all day long, so there is a pretty sizeable loss in stability for my ram 5:4 v 1:1. I am gonan do a vdim mod on this board because I don't wanna relax timings, and nor do i want to drop to 3:2. I am thinking 3.5v will remove my ram issue.

althes

02-14-2004, 10:57 AM

Using an prescott on my epox running nicely so far.:D

dioSRL

03-06-2004, 02:22 AM

does anyone tried the latest bios? 2/09/2004? I have a max vdimm of 2.9v!!! WTF?

rev2 board here.

later edit: yeah this is correct, I restored the 12/25/2003 bios and back to 3.3v. This is bad news, or epox screwed this up. I dont like playing like this with vdimm.

Alexandrus

03-06-2004, 10:14 AM

The new BIOS is very weird when it comes to VDIMM. I think it has the VDIMM register corrupted or something, because on my 4PCA3+ rev. 1.2, no matter what VDIMM is select in BIOS, when I save and restart it's back to 2.70V(default+0.10V). Very weird.

gouda96

03-06-2004, 02:35 PM

I was told that the vdim is also limited to 2.9v, so no more 3.3v if you upgrade to that bios, which is rediculous.

Alexandrus

03-06-2004, 02:46 PM

Nope, I can select up to 3.30V, but no matter what I select it's going back to 2.70V :(

dioSRL

03-06-2004, 03:05 PM

mine rev.2 only 2.9v with that... its really §§§§ed up :D

SPQQKY

03-06-2004, 08:00 PM

EPoX got pi$$ed at the rev2.2 owners whining over at AOA, so they threatened to remove the option completely from all BIOS updates. I guess they were serious.

gouda96

03-06-2004, 08:33 PM

Originally posted by SPQQKY
EPoX got pi$$ed at the rev2.2 owners whining over at AOA, so they threatened to remove the option completely from all BIOS updates. I guess they were serious.

That is a realy terrible response to that kind of complaint. Instead of making another rev, or changing something in the bios for the .2 owners, they screw everyone over.

That will deffinately dissuade almost any future customer from buying that mobo.