Bandwidth with P4T533

[chainbolt]

I have 2 samples of the board running ok, I have sold 2 othjer, they are also fine. Other ppl are reporting problems. I suspect that the new RIMM 4200 from Samsung has some quality fluctuations. There will be soon other RIMM 3200 and 4200 around.

Nevertheless, we are always looking for soemthing new, and it is a fact that board and this new RDRAM has HUGE potential: This is with my P4T533, aircooled, no voltage tricks, BIOS 1.03. No other 840 or 850E board let you run PC1060 at x4 up to 172 Mhz, not even remotely.


http://www2.gol.com/users/postbox/P4T533.jpg

[JCviggen]

dude stop it you're killing me with those benches :(

My P4T533 will run 154x4 somewhat stable.... boots at 160x4 and POSTS at 170 but thats about it. And imho its not the memory. The higher you put the CPU voltage, the lower the memory OC ... at least thats what the "bad" boards do. Some may be exceptional, most of them are very very mediocre.... damnit. I ran prime95 at 154x4 1.65V. At 1.78V same settings it errored out after 1 min and my comp blue screened about a min later. That was at 1°C CPU temp :rolleyes:

I wish i'd had a board come over from Tokyo m8, might have had more luck with that one

[chainbolt]

Hi: I think we all agree that there is something fishy here, be it the board or the RAM. What I wanted to demonstrate with this benchmark is more the fact, that the 850E platform with RIMM 4200 has a very big punch. RIMM 4200 is relatively cheap, I get it here for 120 USD the stick, you need only 1 stick, and it overclocks almost as good as DDR (not to talk about the MUCH better bandwidth).

They have to sort the hick-ups, that's for sure. Maybe the upcoming Epox for RIMM 4200 is doing better?

[JCviggen]

I dunno m8 ... somehow I've never been a fan of EPoX boards. I tried a lot of them, and some were "not bad" but others were just plain crap.

There is definately something wrong with the board... first there's the very flaky power on the board, which could be creating some noise to the memory power as well. Then there's the BIOS anomalies... 1003 final tends to give problems in rather low x4 frequencies, but is good at high 170+ x3 freqs. 1004 beta improves stability in the low frequenties, but high freq x3 modes have gone down the crapper. Then, to add to the embarrasment, if you switch to jumper mode you can get higher stable mem clocks than in jumperfree mode.

The board and the PC4200 does have a lot of potential.... although you have to keep in mind, there is DDR memory that does 200+ MHz easy, and it is no competition for PC4200 in the single stick now, but when dual DDR gets here, 200 FSB+ is going to prove unreachable for RDRAM...

*sigh*

too bad I can never test a board before I buy it :(

[Friar Tuck]

Originally posted by chainbolt [/B]




Nice bench mate... but as both of you know, there are BIG problems with this board..

Anyone thinking of buying it should first have a look at the thread on [H]ardOCP on the Strictly Intel Forum.

Asus have admitted that there is a problem with the board and are working to fix it.

This board will probably get sorted out eventually by Asus, but right now, it's no good.

FT.

[chainbolt]

Originally posted by Friar Tuck

Asus have admitted that there is a problem with the board and are working to fix it.



thanx, any link or details with information about this Asus admission?

[Friar Tuck]

On [H]ardOCP..the SI forum..Look for Zroc and talk to him.

He is working with Asus to try and get the boards sorted out....
Asus have openly admitted that there is a problem.

FT.

[OPPAINTER]

Well you can chalk up the one that I have as a POS. The highest I've been into windows at x4 was at 154 and that was only a few seconds. Also the highest it will do at x3 is 190, at 191 or above it won't even post at all. Makes me want to back to DDR:D

OPP

[Friar Tuck]

Originally posted by OPPAINTER
Well you can chalk up the one that I have as a POS.
OPP


There has been a solution to all the problems with this board found by the [H]ardOCP team ...specifically 'baWoofers'.. on SI.

The fix involves soldering tantallum caps onto the bottom of the motherboard to fix the v-core voltages.

This fix will be relayed to Asus by Zroc who is in touch with the Asus engineers.

Hopefully, Asus will produce a new revision of the board and recall all the old ones ..

FT.

[OPPAINTER]

Well give me the link:D
I don't see a V-core problem with this mobo, how does v-core affect memory speed problems??

OPP

[JCviggen]

OPP,

mine will boot (unstable) at 160x4, but in 3x I cant get much past 170 at 3x :(
This is my second board, exactly the same as the first one... past 170 is a no go

also, the higher the Vcore, the worse it gets. Try it yourself. Run 1.50 vcore and the highest memory setting you can do stable (prime95) at that speed. for example 150x4. change nothing except the voltage (like bump it to 1.7-1.8V) and start prime. You'll soon get what the problem is all about...

[OPPAINTER]

Thats pretty bad JC, Try a v-mem mod. I'm running mine at 2.72volts, thats the highest I can get, I think I need to hook somthing else up.
Anyway, this is the highest it will go, 190 and it's not stable. After that I cant even boot.

[OPPAINTER]

Mem bench at 190x3

[OPPAINTER]

I just turned down the memvoltage to 2.62 and went for 88/44, then I was able to accually get into windows at 193. Looks like I have more things to play with here before I give up:D

[OPPAINTER]

Got it into windows at 196x3 Still cant run 3D much, I'm stuck at about 190 right now with 3D. Didn't have any luck with 4x mode, did get up to 162 in windows though.

196*3

[RichBa5tard]

Looks like the v-mod worked out well OPP. :)

Only 2.7v-rimm max?

[JCviggen]

I dunno why my boards are not handeling 170+ well. I notice that when I dont add a fan to the northbridge cooling, my overclock is considerably worse. Would be my luck to find a crappy batch of I850E chips :rolleyes:

[JCviggen]

small update, with a lot of fiddling and tring different settings, I could run Super Pi and 3Dmark at 162x4. This is back with the shipping 1002 BIOS.

http://service.madonion.com/compare?2k1=4156989 just everything stock didnt bother to even tweak anything.

For running at 1.55 Vcore, 2600 MHz isnt bad for a 1.6A I would think.

I can run Sandra at pretty high speeds but 3D fails quickly.

http://users.pandora.be/JCviggen/sandra167x4.JPG

This is still without any mods done to the board, I hope to get that done within a few days... although this board is unlikely to ever see 170x4 , I hope it can improve just a little bit...

keep you posted

[OPPAINTER]

Originally posted by RichBa5tard
Looks like the v-mod worked out well OPP. :)

Only 2.7v-rimm max?

I accually had to lower my voltage down to stock 2.5 RB to get past a 190 roadblock. I guess the mem doesn't hanle the heat of the higher voltage or somthing. Yes and all it gives me is 2.72 max, I can lower the resistence to 0-Ohms and still only get 2.72V.


JC,

I am able to get into windows at 198 now, the funny thing is I can't even complete a 3D bench at 194. I'll need at least that to beat my score. This card I have now is a POS 380 max core clock.

198*3

[JCviggen]

OPP,

My TH7-II used to be the same way. I could boot in the low 190s, but 187-188 was the MAX to run 3Dmark without trouble. Well actually it was enough trouble, but possible to complete.

the "problem" is that right now there are no CPU's yet with a high enough multiplier (and not too high) that overclock good enough to get the most out of the 4x memory. At 3x memory, I could have kept the TH7.
If I can do 165x4 for 3Dmark with a matching CPU, that could be pretty fast.

[Charles Wirth]

I have a P4T533 w/ Samsung PC4200 now, I will rip the IT7 apart this weekend and take the P4T533R for a spin @ 2c in the case. The PC4200 should like that.

[OPPAINTER]

JC, try the V-Rimm mod, It doesn't work out for me but supposedly this mem at 2.9 volts is supose to do 176*4. You would then have chips that could take advantage, a 2.63 would be great at 176, if it has a 20 multiplier, it should.

OPP

[OPPAINTER]

Originally posted by FUGGER
I have a P4T533 w/ Samsung PC4200 now, I will rip the IT7 apart this weekend and take the P4T533R for a spin @ 2c in the case. The PC4200 should like that.

2c in the case, thats exacxtlly what this mem needs:D

OPP

[JCviggen]

I'm not a soldering ace myself, I'm gonna let my friend take care of that and he needs to have time. I dont believe the memory is limiting me now, but the mainboard is. This Samsung stuff will do 170-180 I think in the right board... by putting 2 sticks into one the overclockability really got better for rambus, and since its a new variety, they should have a good production line. But ASUS is putting out a very crappy board and only a few ppl have gotten lucky with it. I'm not putting my trust in EPoX either :rolleyes:

After some mods we'll see how my board does. I would be happy with 170x4 but thats a high target for this board. The 2.66 chip should have a good multiplier yes... but we'll have to wait and see

also, there's the small problem that i'm absolutely broke right now :D

[OPPAINTER]

JC,

Heres your 170x4 :D

[JCviggen]

Cool stuff m8 :)

But I think without too much trouble I could do a 170x4 screenshot in sandra as well, the trouble is getting it to run 3Dmark :( And when it can just pass 3Dmark, super pi and Prime95 are still bound to fail... *sigh*
So far I havent gotten much above 162/163x4 through 3D2K1.
However, this is still without any mods. Also, I have heard it might actually work better with 2 sticks installed, I havent really got any money to buy another stick, but some friends have the same board so I'll check that out.

What BIOS are you running now ? Any particular settings that work well for you or its just an unstable screenie? :)

cheers !

JC

[OPPAINTER]

Unstable screenie:D
But I haven't even tested for stability at x4 yet so I'm not sure how far it will go stable. I can't really do anything with a stable 160 or 170 even, gunna need more for a good 3D score I think. I'm more interested in the 190s:D.
I'm using the 1004 bios.

OPP

[JCviggen]

the 1004 worked badly for me at high FSB but better at lower ones... you sure you cant get more stable clocks with the 1002 or 1003 at 190+ ?

With the multiplier on the 2.26 or 2.40/2.53 you need high x3 yea...

cheerio

[OPPAINTER]

Well I'll have try the other Bioses out now:D
I'm thinking that either this 2.26 is just getting old or this mobo's v-core flucuation problem is the problem. Then again it could be the low 12v:D Who knows, I'll try the other Bioses. I think like you said earlier the memory seams to be good it's just the mobo.

OPP

[OPPAINTER]

No help with the stability. I'm just going to find my highest Stable OC for now and leave it alone for a week. Looks like I should have a Prometeia by next end, this should get me stable at thoughs higher numbers:D

JC do you think since this P4T533 has a weird socket layout that the Prometeia will have problems with mounting? Nice review BTW:)

OPP

[JCviggen]

OPP,

the prometeia fits the P4T533 (i already tried), but it does require bending the hose with some force and you will not be able to close the left side panel. Although I think that, like me, you dont mind running one side open.... saves case cooling ;)

One thing though : havent tried this board yet, but if its like the first board, negative temps could be a problem. The temp reading I mean. both mbm and CPU cool give me a value in the 120-125C range when the temp drops below 0. when I boot at sub zero temp, CPU Cool even doesnt find the temp sensor anymore. When it goes above 0 it is normal again. Strangely, the BIOS does report the temp correctly. very weird

[OPPAINTER]

JC,

The temps on the MBM only do that with the 1002 and 1003 bios, they are good to go with the 1004 bios. If you look back at the thread I think posted some temps at -10 or something.

I'll be using my Swiftech QP case with it anyway, this should solve the side panel I hope. Some times I like the case closed and Quite there JC:D When I can, which isn't very often:D

OPP

[JCviggen]

I put the whole §§§§ back in the prometeia today with the 1.6A.

Fairly nice CPU, a malay one... dunno the packdate, I swapped it for something... dont remember what that was also ;)

it seems running slightly different settings does enable me very high FSB's in x3 mode

The CPU was limiting me or I could have put it at the roof of 200 MHz (would have been a 100% OC)

pitty

[JCviggen]

sandra

[JCviggen]

I think with a wire trick and some unhealthy voltage, I could get a 100% overclock out of this one

[Gwar]

JCviggen, will your rdram 4200 run 2x600=1200MHz ? Im geting this Kingston stuff pretty high..:) why are my images comming out as links?

http://home.earthlink.net/~xgwar/150fsbmemory1.jpg

[JCviggen]

yeah I dunno what is up with the pics as links, I get the same thing.

this is the PC4200 that came with the board...

[Charles Wirth]

My IBM Deathstar took a dump on me, but the new mobo seems to kick some ass.

Will be reloaded and benching again in a few hours.

[JCviggen]

Well, after all the capacitor adding, Vrimm modding, I'm starting to see some encouraging results now...

http://users.pandora.be/JCviggen/WCPUID175x4.JPG

These are not stable speeds, but at least impressive screenshots I would think. 3Dmark will run at around 170x4 now.
The memory voltage has been bumped up to a (possably slightly unhealthy) 2.97V, and it had no problem getting into windows and running sandra at 177x16 and 1.80Vcore. I had to bump the vcore that high because my 1.6A was being cooled by the intel heatsink and didnt feel like going much further. At lower core voltages, 177-178 may be possible.

Oh, this is actually a different stick of memory than I had first. Went over to a friend who had the same board and compared things... my board ran better with his memory, and his one ran better with mine ... go figure. Since a trade was mutually beneficial to us, we made it a deal. So its pretty clear there are better or worse memory sticks out there, but I guess you have to be a bit lucky with the combo board/memory.

Its still not a very stable board, even after all the mods I did to it, but I can live with it. It can bench great with sufficient stability to run 3Dmark, PCmark and Super pi, and for every-day use I just knock it down to still very respectable speeds.

Still running BIOS 1002 btw, might try the latest one also though.

next target for me is 200x16 in 3x mode ;)

JC

[chainbolt]

Originally posted by JCviggen
Well, after all the capacitor adding, Vrimm modding, I'm starting to see some encouraging results now...

http://users.pandora.be/JCviggen/WCPUID175x4.JPG

These are not stable speeds, but at least impressive screenshots I would think. 3Dmark will run at around 170x4 now.
The memory voltage has been bumped up to a (possably slightly unhealthy) 2.97V, and it had no problem getting into windows and running sandra at 177x16 and 1.80Vcore. I had to bump the vcore that high because my 1.6A was being cooled by the intel heatsink and didnt feel like going much further. At lower core voltages, 177-178 may be possible.

Oh, this is actually a different stick of memory than I had first. Went over to a friend who had the same board and compared things... my board ran better with his memory, and his one ran better with mine ... go figure. Since a trade was mutually beneficial to us, we made it a deal. So its pretty clear there are better or worse memory sticks out there, but I guess you have to be a bit lucky with the combo board/memory.

Its still not a very stable board, even after all the mods I did to it, but I can live with it. It can bench great with sufficient stability to run 3Dmark, PCmark and Super pi, and for every-day use I just knock it down to still very respectable speeds.

Still running BIOS 1002 btw, might try the latest one also though.

next target for me is 200x16 in 3x mode ;)

JC

the funny thing is that I get this with my board without any mods, just so. I hope the new RAM is soon availbale, EPIPA (?) announced weeks ago, they would come with RIMM 4200-232 pin, but nothing so far.

http://www2.gol.com/users/postbox/P4T533.jpghttp://www2.gol.com/users/postbox/P4T533.jpg

hmmm, this is a strange message board, why is the jpg is not showing up?

[Reddog]

I just have one question... Will this board give me any problems running my 2.26 (533fsb) at stock speeds?

I'm not planning on running this board at O/Ced speeds.

Thanks

Reddog

[Charles Wirth]

I have no problem at stock speeds, BSOD after 150FSB for me

Here is a unique problem that maybe someone else is having.

I get "no CPU installed, no CPU installed" at boot up, The system promptly boots. Just the voice reporter reporting incorrect info.

Ripping my P4T533 out and putting back in the IT7, sending it back for exchange and hopefully the 3rd board works correctly.

[chainbolt]

Originally posted by Reddog
I just have one question... Will this board give me any problems running my 2.26 (533fsb) at stock speeds?

I'm not planning on running this board at O/Ced speeds.

Thanks

Reddog

For me and few other ppl. here the P4T533 runs without problem and modifications at stock speed (Prime95). Other users report problems. It's a mystery. Epox has just announced that retail will commence for the EP 4T4A+, this board has the new ICH4 (the P4T533 has the older ICH2) and has the VDIMM voltage adjustment. Maybe better to wait a few days until you get this board.

[Charles Wirth]

Ok, straight up 3.4Ghz -+.01%

P4T533 RIM4200 (188FSB) vs IT7 PC3200C2 (170FSB)

What would be faster?

http://fugger.netfirms.com/8393.jpg

Must be the new core :eek:

[chainbolt]

Originally posted by FUGGER
Ok, straight up 3.4Ghz -+.01%

P4T533 RIM4200 (188FSB) vs IT7 PC3200C2 (170FSB)

What would be faster?

http://fugger.netfirms.com/8393.jpg

Must be the new core :eek:


Fugger: this is very good, this entry in the ORB "Digital Jesus on Anus P4T5333" :D


But probably this here 2800@4010 with a P4T533 is faster, and at the same same time the world record in Super Phi (that is what they claim)

http://ja0hxv.calico.jp/pai/ep4p4tesp.html

http://ja0hxv.calico.jp/pai/sp10419021.gif

[Spystar]

I reckon that Pifast is a better indicator of memory bandwidth than PCMark. I use it for my reviews at Hexus.net.

You can download it
here (http://img.hexus.net/hexus_pifast.rar) . Just unrar and run the test.

I asked BMG to run this test at 3350MHz / 159FSB on his Asus P4T533 with RAMBUS @ 4x. He got a time of 62.20 seconds.

I tried it with my ABIT IT7 and Mushkin PC3200 @ 2-6-2-2 timings and got 65.08s at the same speed and FSB with the Mushkin running at 212MHz.

RAMBUS is faster when run at full speed or at very high FSBs with 3x.

[kiko]

and with extensions disabled?

my shot with a small 1.6A@2.4.

http://kikoguru.supereva.it/img/150.png

3dmark w/my 4200

http://kikoguru.supereva.it/img/13850.png

http://service.madonion.com/compare?2k1=4378851

i am still testing and benching @150.

Ciao

[kiko]

http://kikoguru.supereva.it/img/160.png

http://kikoguru.supereva.it/img/14175.png

http://service.madonion.com/compare?2k1=4413129

http://kikoguru.supereva.it/img/162.png

http://kikoguru.supereva.it/img/2800.png

(not stable 100% coz Intel standard HSF)

http://kikoguru.supereva.it/img/104.png

:)

Ciao

[macci]

Here's some numbers I got w/ my good old Samsung PC800 stuff (2 x 128/8) :)
Sandra 170 4x Turbo => 4242/4241 (http://www.solidhardware.com/macci/mem/170_4x_full.gif)

Running on a Turbopll+vcore+Vmem modded P4T533-C. No extra cooling for ram. Had to bump the voltage up to ~3V thou :D
This stuff wouldn't run higher than ~150 4x unmodded.

Was able to run some 3DMark tests at 169 4x mode too so this ram should do over 220FSB 3X mode. Lets hope that this mobo can run up there (I'm sure that this 1.6A can do it easily) :D

[JCviggen]

Originally posted by macci
Here's some numbers I got w/ my good old Samsung PC800 stuff (2 x 128/8) :)
Sandra 170 4x Turbo => 4242/4241 (http://www.solidhardware.com/macci/mem/170_4x_full.gif)

Running on a Turbopll+vcore+Vmem modded P4T533-C. No extra cooling for ram. Had to bump the voltage up to ~3V thou :D
This stuff wouldn't run higher than ~150 4x unmodded.

Was able to run some 3DMark tests at 169 4x mode too so this ram should do over 220FSB 3X mode. Lets hope that this mobo can run up there (I'm sure that this 1.6A can do it easily) :D

That'd be nice :D
However simple math usually doesnt tell the whole story.. :(
for example, I can to 175x4 for screenshots and 168-170x4 for 3Dmark...(2.98V) but it will run 3Dmark no higher than 195x3 or so, something in that neighbourhood. I imagine with some kick ass chipset cooling it'd improve, but 220 is a pretty long way from that. Still, 220x3 would be awesome...

[macci]

Yeah I guess I need to add some extra chipset cooling there.
Anyways 228 3x (http://www.clio.ne.jp/home/tatumiya/Pikiroku/sl668spi46.jpg) has been done (by Anpanman) on a ln2 cooled th7-2 so it should be possible w/ this mobo too.

[JCviggen]

Thats some damn nice FSB.... makes me wonder if I should decently cool my north bridge :D
Lets see, with a good 2.53, put some LN2 on it ... 19x210 = 4 GHz... with RDRAM at 630... ouch that'd be nice. There is the minor inconvenience that this board only goes up to 200 FSB lol.
Too bad reality is never as easy as dreaming about overclocks ;)
Gonna give it a shot soon though :)

First thing i'm going to do is get a different PSU... this P4T533 isnt too good with voltages... at 1.80V 3360 MHz my 12V is between 11.1 and 11.35. Not too good. Cant adjust this PSU... maybe i'll try using 2 PSU's for this board... it'll need juice for 4 GHz...

[kiko]

not bad :)

http://kikoguru.supereva.it/img/17k.png

http://service.madonion.com/compare?2k1=4425040

P.S. no MOD on asuz no MOD on suma. Only stock cooling. :)

Ciao

[Magnum_]

http://users.pandora.be/crush/HW

Here ya find my results with the board. Everything is pretty standard, the memory I used = deliverd with the asus P4T533.

I can go up to 166*4 (1.7v), boot into windows, and run the mem bandwith bench in sandra. Also everything else in sandra works FINE. Though, if I start really using the machine, I get random lockups, bluescreens, reboots, U name it :(. Another strange thingy is, that I cant get any higher with the memory @ 3* 1xxmhz. At 170 it wont even boot anymore :(

I also have a pcmark benchmark (http://service.madonion.com/compare?pcm=579447), but I didnt run it in optimal circumstances ;) I'll get a newer one soon :)

For normal use, 160x4 is the maximum, I've been running for a couple of weeks now at this speed, without any big problems :)

I was also doubting about getting a new PSU (currently I've got a 340w chieftec, wich is getting pretty hot actually), because I can see some fluctuations in the 5v rail, also vcore voltage isnt as stable as I wished. The tool provided by asus doesnt have an exact scaling, so I guess that with 3rd party soft, those fluctuations will become even more obvious...
Maybe I'll have to get an extra psu, superflower 420watt is looking pretty nice and cheap :) Together with the enermax, that should DEFINATELY be enough !


mot : http://service.madonion.com/compare?2k1=4220501
this is the 3dmark result, pretty crappy due to GF3 that just wont OC decently... People with TI200 cards are getting WAY better results :( Not usual for a gainward. But in the meanwhile I sold that card, I was looking out for R9700 :)

[kiko]

http://kikoguru.supereva.it/img/rimmina.png

http://kikoguru.supereva.it/img/prob.png

those are my values electrical workers w/cpu @2560 vcore default. :)

Ciao

[Klosterbruder]

My highest booting FSB so far was 196 same as JC, I don't know if the ram or the Intel stock fan as the limit for my 1,6 GHz P4. But I'll have a water cooling system soon (waiting 2 month for it and then they told me they forget it :( ). Unfortunately my 1,6 Ghz P4 has almost reached the limit of the FSB, perhaps I'll buy a new 2,4 Ghz P4. For pictures click on the link in my signature.

[spaceboy]

Originally posted by kiko
http://kikoguru.supereva.it/img/rimmina.png

http://kikoguru.supereva.it/img/prob.png

those are my values electrical workers w/cpu @2560 vcore default. :)

Ciao


nice work man!!












ando sei sparito?:D

[sysfailur]

Originally posted by JCviggen
Thats some damn nice FSB.... makes me wonder if I should decently cool my north bridge :D
Lets see, with a good 2.53, put some LN2 on it ... 19x210 = 4 GHz... with RDRAM at 630... ouch that'd be nice. There is the minor inconvenience that this board only goes up to 200 FSB lol.
Too bad reality is never as easy as dreaming about overclocks ;)
Gonna give it a shot soon though :)

First thing i'm going to do is get a different PSU... this P4T533 isnt too good with voltages... at 1.80V 3360 MHz my 12V is between 11.1 and 11.35. Not too good. Cant adjust this PSU... maybe i'll try using 2 PSU's for this board... it'll need juice for 4 GHz...

Some even nicer FSB... =) I'll say it again... crazy japs! :P

http://members.jcom.home.ne.jp/aguri-no.3/index.html/bd7II_wcpuid_s256.49.gif

[kiko]

Originally posted by spaceboy



nice work man!!












ando sei sparito?:D

a donne :D

http://kikoguru.supereva.it/img/58sec.png

with turbo enabled

http://kikoguru.supereva.it/img/4xmonco.png

with turbo disabled

Intel HSF vcore 1.7v

Ciao

[kiko]

http://service.madonion.com/compare?2k1=4542033

http://kikoguru.supereva.it/img/mhm.png

:)

Ciao

[kiko]

http://service.madonion.com/compare?2k1=4548867

:)

Ciao

[kiko]

turbo on

http://kikoguru.supereva.it/img/viaggia.png

turbo off

http://kikoguru.supereva.it/img/sgrung.png

Intel HSF 1.65v @2.91 1.7v @3.04 T ambient 23°

Ciao

[kiko]

http://kikoguru.supereva.it/img/78xx.png

http://service.madonion.com/compare?pcm=692574

:)

Ciao

[kiko]

http://kikoguru.supereva.it/img/7931.png

http://service.madonion.com/compare?pcm=697363

Ciao

[spaceboy]

'un te se filano..
posta con [img]...
;)

randi :D

[kiko]

Originally posted by spaceboy
'un te se filano..
posta con [img]...
;)

randi :D

[IMG] code is OFF only 4 me? :)
non me lo fanno modificare dal profile

P.S. http://kikoguru.supereva.it/img/porelli.png

Ciao

[kiko]

163 4x turbo @2.5v. :)

http://kikoguru.supereva.it/img/163.png

Ciao

[Klosterbruder]

go with higher fsb and *3 setup and you get crazy points.

[kiko]

Originally posted by Klosterbruder
go with higher fsb and *3 setup and you get crazy points.

it's a 2.53 and is a crap. :/ (164 max w/3x also @1.8v)
I have sold my 1.6A. :/

Ciao

[Klosterbruder]

hmm that's a pity. I think the resurrection of 1,6a Ghz will come with granite bay, imagine 1.6@3.2 and 6.4 GB/s bandwith, yeah that is crazy.

[JBELL]

Originally posted by kiko
[IMG] code is OFF only 4 me? :)
non me lo fanno modificare dal profile

P.S. http://kikoguru.supereva.it/img/porelli.png

Ciao

png is not a supported file format in php or mysql for image creation... sorry.

use gif or jpg for images.

[macci]

Originally posted by macci
Here's some numbers I got w/ my good old Samsung PC800 stuff (2 x 128/8) :)
Sandra 170 4x Turbo => 4242/4241 (http://www.solidhardware.com/macci/mem/170_4x_full.gif)

Running on a Turbopll+vcore+Vmem modded P4T533-C. No extra cooling for ram. Had to bump the voltage up to ~3V thou :D
This stuff wouldn't run higher than ~150 4x unmodded.
Same RAM, same mobo but more voltage (3.5V :D) and colder ambient temp (+6C or so)
Sandra MEM @177.62MHz 4X TURBO => 4409/4394 (http://www.solidhardware.com/macci/mem/177.6_4xmem.gif) :)

PCMark @176 4X TURBO (http://service.madonion.com/compare?pcm=897618)

PiFast @175 4x turbo (http://www.solidhardware.com/macci/mem/175_pi.gif)

[Raptor]

so, is the P4T533 problems all solved?

[Klosterbruder]

I'll send my to Asus soon, the new version should be better, but i haven't heard a lot about it. So i don't know, if the problems are fixed. There are no reports about, cause the most people have changed to DDR Ram, after the problems with the P4T533 . But i'll stay @rambus for a while.

[chainbolt]

Originally posted by Klosterbruder
I'll send my to Asus soon, the new version should be better, but i haven't heard a lot about it. So i don't know, if the problems are fixed. There are no reports about, cause the most people have changed to DDR Ram, after the problems with the P4T533 . But i'll stay @rambus for a while.

I think I remember some posts that after a certain serial number z....XX or something, the problems are solved. It was mentioned in the huge {H} thread. But I am out there and don't even read this forum anymore.

Fact is: despite dual channel, the P4T533 is the ruling bandwidth champion.

[Klosterbruder]

Originally posted by chainbolt
I think I remember some posts that after a certain serial number z....XX or something, the problems are solved. It was mentioned in the huge {H} thread. But I am out there and don't even read this forum anymore.

Fact is: despite dual channel, the P4T533 is the ruling bandwidth champion.
I hope so.

[chainbolt]

I flashed the 1.05 BIOS, and now the speed selection for RAM is gone. I want my 1.04 back!! But I did not back it up before I flashed. Could somebody kindly mail me the BIOS 1.04?? Thanx

postbox@gol.com

[Zroc]

Originally posted by chainbolt
I flashed the 1.05 BIOS, and now the speed selection for RAM is gone. I want my 1.04 back!! But I did not back it up before I flashed. Could somebody kindly mail me the BIOS 1.04?? Thanx

postbox@gol.com
Done.

[mkleingeld]

I've just had this p4t533 for 2 weeks. I'm moderately pleased with it. it doesn't give too much ghz without getting unstable with bsod's but it does give the best speed for the ghz it's at probably. I do notice that the vcore is very unstable if set at 1.7 and above and is therefore better left alone at a lower level. fsb 154 is the best I can do with x 4 and turbo. I can go a little bit higher but that becomes very unstable and gives errors immediately with super pi. But at that setting it does Seti@home units between 1.45 and 2.15 hours. Pretty fast :-)
My board came with 1.05 bios.
Chainbolt, what do you mean with speed selection for the ram? i guess something else then 3x, 4x and turbo? I've downloaded 1.04a but can't install it without some dos flash utility that I don't have at this point.
Also I would like to get some very clear information on how to do a vrimm mod, since I'm nowhere near the limit of my cpu on this board. I've read something on thenakedreview.com but it's not clear. Thanks!

[chainbolt]

Originally posted by mkleingeld

Chainbolt, what do you mean with speed selection for the ram? i guess something else then 3x, 4x and turbo? I've downloaded 1.04a but can't install it without some dos flash utility that I don't have at this point.
Also I would like to get some very clear information on how to do a vrimm mod, since I'm nowhere near the limit of my cpu on this board. I've read something on thenakedreview.com but it's not clear. Thanks!

My mistake, I was sleepy and simply oversaw that the feature wa still present.

(a) I can run my P4T533 at 172 mhz in x4, and then the bandwidth goes up to more than 4300 MB/sec. of course at that speed not stable.
(b) over the last 5 months I have tested around 20 RIMM 4200 modules 256 MB and 512 MB non-ECC and ECC. result: unfortunately there are wide fluctuations: I have seen modules which made it just to 150 mhz in x3, fault-free, and other going up to 164 mhz in x4 fault-free. In order to see whether a module runs fault-ree you have to run a non-resident memory test from a floppy like memtest-80 or goldmem. That you can take Sandra screenshot for bandwidth has NO MEANING AT ALL.
(c) ACE at naked hardware had come with a Dimm mod, he is also member of xtreme. Ask him, if you want I can give his e-mail address. I found that it is important to set the core voltage not too high, if you do so, you obviously take away some voltage from the DIMMs. The P4T533 is OVERvolting, and I don't think you gain much with a volt mod. There is also an electronic company
here Japan "OJI" which professional modifying boards, clock generator changes, DIMM voltages, core voltage etc etc.http://www2.gol.com/users/postbox/P4T533.jpg

[TechTones]

What is the best overclocking bios with this board?

I'm not using the Raid yet but someone told me all the other bios but the 1005 had problems when using the Raid? Is this true?

[TechTones]

Also

I put a 2.8 in the P4T533 today. I put the overvolt vcore jumper back to normal. The "Auto" default of 1.525v reads 1.58v in the bios but reads HIGHER in windows XP using MBM at 1.62v!
WTF is that all about? I don't know if my chip is getting the 1.525v
from the bios or 1.62v in windows?

Maybe this is how Asus has helped fix the undervolt problem by overvolting it?

BUT, I put a 1.8 in and the default of 1.50v read correctly.

[chainbolt]

Originally posted by TechTones
Also

I put a 2.8 in the P4T533 today. I put the overvolt vcore jumper back to normal. The "Auto" default of 1.525v reads 1.58v in the bios but reads HIGHER in windows XP using MBM at 1.62v!
WTF is that all about? I don't know if my chip is getting the 1.525v
from the bios or 1.62v in windows?

Maybe this is how Asus has helped fix the undervolt problem by overvolting it?

BUT, I put a 1.8 in and the default of 1.50v read correctly.

MBM has no value at all, I wonder why ppl still using this. The only corect reading is from the BIOS.

I also get a higher voltage reading for the core in the health monitor than what I have set. The reason is simple: the P4T533 is overvolting, obviously a little different from CPU to CPU.

[TechTones]

I seem to be stuck at 157-158/4x.

I'm going to do the vdimm mod.
What is the best bios for the board?

[chainbolt]

Originally posted by TechTones
What is the best overclocking bios with this board?

I'm not using the Raid yet but someone told me all the other bios but the 1005 had problems when using the Raid? Is this true?

I have tried 1.03, 1.04 and 1.05 (ands several beta versions in between). I never found any difference in overclocking. There is however (and unfortunately) a HUGE difference between Samsung 32-bit RIMM 4200 modules. During the last 4 months I have tested around 20 modules and also a 512 MB module. The differences between modules are ranging from 140 ~ 165 for faultfree test runs with the memory testing programs GOLDEMEN or MEMTEST-80. The 165 Mhz moduel was the one I could run with Windows at 172 Mhz producing a bandwidth of more than 4300 MB/sec (see the screenshot from above). Unfortunately SANDRA screenhots do not tell anything about system stability, I think everybody here knows that you can take a Sandra bandwidth bench with a highly unstable system.

Most important when you are in overclocking the P4T533 is to find at least one module which is overcklocking above 160 and that seems not so easy. :(

I have the feeling that the problems reported with this board are entirely the fault of the quality fluctuations with the Samsung CRIMM 4200 memory modules. I never could install XP with the RAM speed set to x4. I always thought that this is the P4T533's fault. Then I got the new RDRAM Epox for the same memory type, and it was the same: even at 133 Mhz I could not install XP or W2K at x4. The is is clealry a memory fault.

[TechTones]

Hmmm

I used a 400 bus CPU to install winxp so I have no idea if it defaulted to 3x at install.

What serial number is your board? How long ago did you buy it. There has been some changes made to it.

[chainbolt]

Originally posted by TechTones
Hmmm

I used a 400 bus CPU to install winxp so I have no idea if it defaulted to 3x at install.

What serial number is your board? How long ago did you buy it. There has been some changes made to it.

the changes are just rumors. It's still revision 1.03 they are selling. It's amazing how much assumptions, speculations, and guessing have been done about this board (looking at the thread at Fake[H].

[TechTones]

Originally posted by chainbolt
the changes are just rumors. It's still revision 1.03 they are selling. It's amazing how much assumptions, speculations, and guessing have been done about this board (looking at the thread at Fake[H].

I never mentioned revision which is 1.03,
I meant serial number. Everyone knows the new Oct board started with Asus's dating system of 2aZ. Means it was made in
Oct with the new circuitry and chips. Mine was a 2BZ, made in Nov 2002

[chainbolt]

Originally posted by TechTones
I never mentioned revision which is 1.03,
I meant serial number. Everyone knows the new Oct board started with Asus's dating system of 2aZ. Means it was made in
Oct with the new circuitry and chips. Mine was a 2BZ, made in Nov 2002

:) who is everyone? I got my first P4T533 in spring this year, some time before this board was selling in US or EU countries. I never had a problem with this board. Now, after half a year or so, I got another one, which also works w/o the slightest hick-up (when using the selected RAM :D). If Asus made important changes to the board they would have changed the revision number, that is what I think. And this "October" productioin story sounds more like another urban myth to me. After testing so many RIMM 4200 modules which such a wide range of results, I am 100% sure the problems reported are related to quality fluctuations with the Samsung RIM 4200 and not the board. reason: using the same RAM I got the same problesm with the 32-bit RIMM 4200 EPOX board

But for the record: my SN is 27ZGOZ6248 and nowwhere on the lable is anything said about the production time. Or is this written somewhere else?

Both sample are good for 172~176 Mhz in x3. But this only with hand selected RAM. I am currently using a 512 MB stick ECC.

[TechTones]

You are wrong Chainey :eek:

I have had both the old and new board side by side and these are the changes. Ask Asus.

Here are the changes on the new boards:

1. The DC/DC controller (vCore) is changed.
On the old board it was a Intercil HIP6302, on the new it's a RichTek RT9241, both rated at 25A, vCore accuracy of +/- 1%

2. Two capacitors (the ones closest to the memorybanks are slightly bigger 560 microfarad vs. 510

3. The power transistors close to the CPU socket (on each side of the 7 capacitor group) is now a Philips PHB96NQ03LT and o the old board it's a Infinion IPB05N03LA

4. Two surface mount capacitors (I assume, it could be jumpers but it's most likely it's capacitors) on the back side of the board, close to the memory banks.

5. PCB is colored slightly differently.

They also come flashed with the latest 1005 bios that enables HT for the 3.06 and have enclosed a sheet of paper explaining it. If you're going to order one only do it from Newegg. They have the latest, newest stock.

[TechTones]

Originally posted by chainbolt
:) who is everyone? I got my first P4T533 in spring this year, some time before this board was selling in US or EU countries. I never had a problem with this board. Now, after half a year or so, I got another one, which also works w/o the slightest hick-up (when using the selected RAM :D). If Asus made important changes to the board they would have changed the revision number, that is what I think. And this "October" productioin story sounds more like another urban myth to me. After testing so many RIMM 4200 modules which such a wide range of results, I am 100% sure the problems reported are related to quality fluctuations with the Samsung RIM 4200 and not the board. reason: using the same RAM I got the same problesm with the 32-bit RIMM 4200 EPOX board

But for the record: my SN is 27ZGOZ6248 and nowwhere on the lable is anything said about the production time. Or is this written somewhere else?

Both sample are good for 172~176 Mhz in x3. But this only with hand selected RAM. I am currently using a 512 MB stick ECC.

27ZGOZ6248

Well 2 is for 2002, 7 is for the month. Yours is from July. the new boards started coming out in Oct. or A. Asus went from single numbers for the month, to letters once the month was two digits.
A is for Oct. B for Nov etc.

[chainbolt]

Originally posted by TechTones
27ZGOZ6248

Well 2 is for 2002, 7 is for the month. Yours is from July. the new boards started coming out in Oct. or A. Asus went from single numbers for the month, to letters once the month was two digits.
A is for Oct. B for Nov etc.

thanx, so this is again one of the very first samples. It runs without the slightest problem, just as mt previous one. I have
my 512 MB module running at 180 Mhz. Need another proof that this story that Asus has changed something after the October production and that the baords before are faulty is wrong? :)

This is aircooled, the vcore at 1.70 volt. The RAM is at 3 x 180 Mhz = 540 Mhz

http://www2.gol.com/users/postbox/2.4@3.2.JPG

[chainbolt]

Originally posted by TechTones
You are wrong Chainey :eek:

I have had both the old and new board side by side and these are the changes. Ask Asus.



I have asked Asus: if you want I can give you the the name and telephone number of the local distributor. This is Asus Japan. They say nothing has been changed.

Manufacturers are always changing components during the production process, that does not mean that the architecture of the board has been changed. If they do this, they normally change the revision number. This is now the 3rd P4T5333 (all are from June/July) which is running w/o any problem and overclock like hell.

[TechTones]

Nothing has changed?

That's why almost everyone so far is NOT having problems with the new boards where just about everyone had trouble with old production. I don't care what the reason was they changed the components, but now with the new ones most of the boards now work.

When you send an old one in for RMA, it comes back with 2 small
capacitors on the back of the board near the mem slots. When you buy a new board, they are there already. So nothing has changed huh?

On top of that the board now overvolts the CPU. Setting it at 1.82v gives you 1.92v in windows.

I have been following this board for months, and believe me, plenty has changed. Just because you have an older board that does work, doesn't mean nothing wasn't changed in the new ones.

[chainbolt]

Originally posted by TechTones
Just because you have an older board that does work, doesn't mean nothing wasn't changed in the new ones.

that is of course right :) but may I ask:

(a) Why the 3 different boards I have tested so far (all produced in June or July) function w/o problem when using selected RAM?
(b) why is ASUS telling me that nothing has been changed?
(c) why it's still the same revision 1.03 since the start of production until now?
(d) why is ASUS handing out a review (!!) board produced in July?
If they would have changed something after October to remedy a problem, they would certainly not give a board produced in June to a major review site. That would be incredible stupid.
(e) why is it so, that according to your comment from above still some ppl experience problems, although you say Asus has changed the board to solve the problem?

Please don't think, I am not aware about the problems. If you check the long P4T533 thread at {H}, you will find my results/comments among the first posts.

What I think is that the problems reported by many users are at least partially related to the Samsung RAM. Out of maybe 20 or so 256 MB and 512 MB modules I have tested so far, maybe 5 were working w/o a problem.

When using the moduls I had identified as problematic with the EPOX, I got EXACTLY the same problems: instability, zero overclocking, problems when installing Windows. The matter is not solved, either way.

[TechTones]

Well Chains

I have a Epox 4T4A+ running right now, revision 2.0 and one stick
of Samsung 32 bit week 0234. I think the boards sucks so far.

1] I can only run Prime95 stable at 140/4x where the same stick on the new Asus P4T533 runs at 145/4x with Prime.

2] I can run my 2.8 gig at 169/3x on the Asus. On the Epox it won't even boot up at 153/3x. I can't believe it. This Epox board is really bad for overclocking. For default it's OK, but I hear with 2 sticks of rdram it's BSOD/reboot city.

3] The Asus board OVERVOLTS the Vcore, where the Epox UNDERVOLTS. I set 1.85v in bios, I get 1.77v. Very bad. But curiously the Vrimm is right on. Not .01v off. Strange.

I think the new Asus is a quality stable board. Not bad for overclocking at all. The Epox, even revision 2.0, is not good. I do realize I have a crappy stick of memory. I need to try some new ones.

As far as the changes on the Asus, I had both boards side by side. An old one unfixed, and a new one. The old one was terrible, where the new one was completely the opposite.

[chainbolt]

Originally posted by TechTones
Well Chains

I have a Epox 4T4A+ running right now, revision 2.0 and one stick
of Samsung 32 bit week 0234. I think the boards sucks so far.

1] I can only run Prime95 stable at 140/4x where the same stick on the new Asus P4T533 runs at 145/4x with Prime.

2] I can run my 2.8 gig at 169/3x on the Asus. On the Epox it won't even boot up at 153/3x. I can't believe it. This Epox board is really bad for overclocking. For default it's OK, but I hear with 2 sticks of rdram it's BSOD/reboot city.

3] The Asus board OVERVOLTS the Vcore, where the Epox UNDERVOLTS. I set 1.85v in bios, I get 1.77v. Very bad. But curiously the Vrimm is right on. Not .01v off. Strange.

I think the new Asus is a quality stable board. Not bad for overclocking at all. The Epox, even revision 2.0, is not good. I do realize I have a crappy stick of memory. I need to try some new ones.

As far as the changes on the Asus, I had both boards side by side. An old one unfixed, and a new one. The old one was terrible, where the new one was completely the opposite.

I agree on the Epox, I also put great hope in it because it has the more advanced IHC and DIMM voltage. But nothing than problems. All of them came at the end came down to RAM. After trying I don't know how many modules I got my sample finally running at 155 Mhz in x4. My best P4T533 did 174 Mhz in x4 and my current does 180 Mhz in x3.

It's still a mystery: as you have a changed board, Asus did obviously change something, ok, but still some ppl obviously have problems. And after my experience with so many "bad" RIMM 4200 modules, I suspect, at least partially, the Samsung RAM at fault.

[chainbolt]

By the way, Tech, do you know which chipset stepping your P4T533 has?

[TechTones]

Sorry Chains,

I never took off the sink to see, just added a fan. I could use Sisoft Sandra, it tells you that but the Asus is not installed right now. I took it apart to set up the Epox.

Yes, I found it funny I couldn't find not ONE review on the Epox 32 bit. Not one. But, when I heard version 2.0 was out, and one guy in this forum had it running Prime 160/4x with Samsung no problems, I decided to bite. Now it looks like a RMA.

The real funny thing is increasing the Vrimm to even 2.92 didn't help at ALL. The two best Rambus boards I've owned were the TH7-II and the Iwill p4R533. I got that as a sample from Iwill.

With the 16 bit modules increased Vrimm helped a lot. I tried the Epox just for the adjustable vrimm and AGP. Funny how the Asus does better at default vrimm all around.

I think Epox quality is terrible. If you get a good board you're tripping, but get a bad board and it's problems galore. I just RMA'ed my 8RDA+. Unstable and bad hard drive performance.

[chainbolt]

Originally posted by TechTones
I think Epox quality is terrible. If you get a good board you're tripping, but get a bad board and it's problems galore. I just RMA'ed my 8RDA+. Unstable and bad hard drive performance.


LOL, this is my sermon since years. I remember one board from this company which I had to RMA 4 times in a row (four times). I think it was the 8KHA or something. 4 times! One RMA case was a short circuit on the board which also killed the CPU.

Regarding the DIMM voltage: it's true, I found the same, increasing the voltage at the Epox did not help at all. But again: this might be more related to the Samsung RAM than to the board itself.