Moving backwards....

[PiLsY]

It would appear my conclusions about the current crop of intel mobos running out of power at over 3.5ghz or so would be correct. I have noticed my stability moving slowly backwards with my IT7. From my original 3.6ghz stable / 3.66ghz for benching over the last few days its dropped off to 3.43ghz stable / 3.5ghz for benching. Its definitely not the cpu as ive been able to drop voltage along the way - infact its burned in a little and im now running below default voltage at 3.433ghz. Extra vcore makes no difference to stability and neither does dropping the fsb:mem ratio to 1:1. Ive pulled everything apart down to constituent parts to no avail.

My belief is the extra current the 2.8 is drawing at these speeds is actually wearing out the board. I dont expect it to last more than another fortnight. Power regulation is getting steadily worse with larger and larger variation on vcore under load. Its getting bad enough that im now concerned about actually damaging the chip.

To this end ive ordered 2 new boards today in preparation for the demise of this IT7 - a soltek P4X400 board (has nice large mosfets and some rather powerful looking caps) and the via reference P4PB400 board (some extended delivery time on it - should arrive in a week or so). This IT7 will be moved down to a secondary rig with the 2.26b in it and ill see what happens there. Its a shame as i'll be losing Raid, firewire and all those USB ports, but theyre no use to me if it cant overclock.

I must admit when I originally purchased the IT7 the small size of the mosfets and low power ratings on the caps bothered me a little, but it seemed to have no negative effect on the boards clocking capability. Now that im using more and more power I can see that my original fears were founded. In all honesty I cant blame Abit as they probably never envisaged people starting to approach 4ghz on the IT7. However, the via boards are designed to support the future hyperthreaded P4s at speeds approaching 4ghz anyway, so the whole power regulation circuitry looks a lot more capable.

I cant comment on how my original (made by abit) board would have done, but as a rough guess id suggest it wouldve been a lot more capable than my current (made by ECS - aka PC Chips) board. These ECS boards seem lower quality - they run a lot hotter and clock a lot lower.

Anyway, ill post back on this thread once I have some solid results for the Via boards, but im expecting a lot more stable voltage at higher speeds. I hope im not disappointed.

PiLsY.

[RAMMER]

Pilsy-

I'm having similar issues, although on a much smaller scale. My IT7\2.53 combo ran solid @ 160, 3:4 mem timings 2,6,2,2 for weeks.I'm slowly having to back down to 157 fsb, 3:4, 2.5,6,3,3 to avoid BSOD's. I initally thought it was due to the new 9700 Pro, but I tried the old Ti 4600 with the same results. I've reinstalled everything , no go. Tried a nice XMS 3200 256 stick that will do 223 fsb on my KX7 rig.My old Samsung will still do 212 fsb @2,6,2,2, 2T on the KX7 . I noticed vcore makes no difference as well.
My question is: does the VIA P4X400 board support 3:4 strap ??
I'd lose serious bandwidth if not-due to this 2.53.

[PiLsY]

The P4x400 supports 1:1, 3:4, 4:5 and 4:6 ratios .

From what ive been told theres a dipswitch on the board that lets you set the ram speed to either 133, 166 or 200 and this automatically changes the ram divider depending on what FSB range youre in. For example, at 133 fsb with ddr set to 200mhz youd be running a 4:6 divider. At 133 fsb with it set to 166 ddr youd be running a 4:5 divider.

Im hoping the soltek has the ability to set these in bios at any fsb to whatever ratio you like, but in all honesty im more concerned at the moment with getting a board that will cope with the larger current draw more effectively than the IT7.

Id suggest you test to see if its your board (or cpu) flaking out or the ram. Set the 1:1 divider in bios and see if you can then run 160fsb again. If you can then your ram is starting to die. If not its either the board or cpu. Unless you have a spare cpu lying around thatll do 160fsb you wont be able to tell if its the board or cpu doing it. From what youre saying though with the vcore making no difference id suggest its either the board or ram. With my board lowering ram timings makes no difference, it just wont stay stable.

For example, trying to run at 171fsb gives me exactly the same random instability using 1.6vcore (actual volts), 3:4 divider with semi tight timings as at 1.8vcore (actual), 1:1 divider, minimum timings and 3v through the ram. Its therefore definitely the board. Further proof can be had in that the higher I run the vcore the more flaky it gets - if I run 1.85vcore over 3.5ghz it locks up on the desktop. If I drop this to 1.75vcore its stable, but will randomly lock up.

PiLsY.

[QuadDamage]

have you guys tried IT7 Max2 yet? excellent mobo, but has some issues to be fixed though.

[PiLsY]

Yup I have.

Wasnt impressed at all. All voltages were all over the place and that was with my 2.26b. It got returned, and when I took it back I was asked before I even mentioned the problem whether it was the voltages, so its obviously not an isolated incident.

I think its a case now of avoiding intel chipsets until they bump up the voltage regulation to cope with the new P4s. The upcoming 845PE chipset may be the key as it is designed to run the hperthreaded 3ghz plus P4s as well as DDR333 memory. Unless they improve the bandwidth quite a lot over the original 845E though I cant see it being able to compete with the SIS 648 and P4X400 chipsets though.

PiLsY.

[Marci]

Knew there was a reason why I was happy with my AMD rig... lol

[PiLsY]

Smartarse :p.

[JCviggen]

Pilsy,

I'm slightly tempted by that Via P4X400 board.... to be honest I very much doubt I will be able to reach very high mem speeds on it, but on the other hand, with as many possible dividers as that thing has, its easier to make something from a bad situation than it is with this IT7. Also I'm worried about possible low CPU OC, but again, cant really know before I try. Price is very nice, boards are hard to come by tho... I just know of that one place in germany...

last time I checked, the VIA P4X's were pretty good at 3Dmark too compared to I845E

[toofast]

Hi PiLsY
Do you get PCI lock on those boards ? got to quite like it on the Intel chipsets

The only thing thats putting me off is the max 2.8v-dimm

[NWEng]

A quick BIOS mod will take care of the 2.8v vdimm max to 3.2v. Not a problem.

[PiLsY]

Just got the soltek board today - will be stripping everything down in a couple of hours and installing. Look for some results around midnight here (5 hours or so).

Having looked at the soltek ive already sussed out vmods for it - its a piece of proverbial. Its a VERY well laid out board - no components over 3mm high in the cpu area, tons of room and basically a very nice piece of kit. The silver looks a bit cack (a bit like AS2 colour but not shiny), but its sposed to do a very good job of EMI shielding and heat dissipation. The manual is EXCELLENT. Ive never seen as good a manual as this. It goes through every detail of the bios and board, even telling you how to overclock via both jumpers and bios and going through the merits of each. it describes voltage adjustment and full performance tuning too. The cd is smack up to date on drivers and you get tons of useful free software with it. Full license software too, none of this time demo crap.

The part I like best though is that the mosfets and all related voltage regulation stuff is rated up to 45 amps. I mean jesus thats a huge amount. Theyve included a 4 pin molex and stated in the manual "This should be used for non atx 2.03 compliant PSUs or in addition to the standard P$ connector under situations where the cpu is highly overclocked and additional voltage stability over the 4 pin P4 connector is required". It boosts not only the 12v, but also supplies extra ground lines for a cleaner signal and an extra 5v feed from looking at the traces.

All in all this board oozes quality and looks like a true winner.

As i mentioned above its a shame to see only 2.8v via bios, but im confident ill have sussed the vmem mod by the end of tonight.

PiLsY.

[JCviggen]

Pilsy,

sounds promising, but I always need to see FSB's timings, bandwidth and some marks to be convinced

give her hell and report back

[majormav]

will wait up to see the results pilsy m8 ... hope its as good as you initially think ...waiting

[PiLsY]

Voltages are nice and steady. First power up just now but only on air. Will slap it into the prometia now .

Hopefully on target for at least an indication by midnight .

PiLsY.

[zakelwe]

How is it going ?

Regards

Andy

[chainbolt]

I cannot confirm these findings. I am running my IT7-II since 2 weeks day by day with a CPU voltage of 1.75 (BIOS 1.80) and the VDIMM at 2.8. The CPU is running between 3600 an 3700 Mhz and I don't have experienced a degrading effect (so far). This is now the 2nd 2.66 I have in this rig and both did not show a "degrading" effect. But who knows ...

[Ragnarok]

pilsy wat are the temps on ya mosfets like?

Recently I've been having a series of problems with my BD7II's OC capability.. only to find the problem attributed to overheating mosfets. It got to the point that I couldnt OC at all above stock speed. Miraculosly after I chucked a 120 fan over the mosfets all teh instabilities are gone. I suspect one of the mosfet may even be dead (perhaps drawing too much power, as the BD7II only has 3 phase power)

Dunno , just a thought... may be cooling for ur mosfets will help too?

[chainbolt]

Originally posted by Ragnarok
pilsy wat are the temps on ya mosfets like?

Recently I've been having a series of problems with my BD7II's OC capability.. only to find the problem attributed to overheating mosfets. It got to the point that I couldnt OC at all above stock speed. Miraculosly after I chucked a 120 fan over the mosfets all teh instabilities are gone. I suspect one of the mosfet may even be dead (perhaps drawing too much power, as the BD7II only has 3 phase power)

Dunno , just a thought... may be cooling for ur mosfets will help too?

I am by the way also using a 120 mm fan for exactly the same purpose, cooling the RAM, the capacitors and the fets..

[PiLsY]

Mosfet temps were VERY high. The laser thermometer I was using is for sub zero readings and tops out at 75c. Mosfet temps were above this. As I said in a different thread I put tweakmonster copper ramsinks on the Fets and an 80mm fan blowing over it. this took temps on the fets down to around 50 - 60c, but thats still way too high.

Chainbolt, how is it whenever anyone else reports a problem you can never "confirm" their findings? Is everything you buy hand carved in gold or something?

Just a thought....

PiLsY.

[PiLsY]

Am currently waiting for a response from soltek regarding a problem with the current Bios's. There is no TRAS, TRCD or TRP adjustment at all. TRCD and TRP are set to 3 no matter what, and TRAS is set to 6. I need to be able to run TRAS 7 to achieve DDR400, and tweaking the TRP down to 2 would be nice as well.

Until I hear back from soltek im not going to comment on the board as a whole.

PiLsY.

[JCviggen]

Pilsy -


even WPCREDIT doesnt help? Well I can see the problem there being that you cant boot high enough at those timings

On a side topic, I saw on the [H] frontpage that ASUS have now added HyperTransp to their P4T533 .... this board has a truckload of problems, 2-phase power supply and in some cases doesnt even run 2.4/2.53 at default because of power needs..... how on earth do they expect to run a 3.06 P4 stable.?

These mobo manufacturers can be such a bunch of nitwits sometimes...

[Marci]

Could be worse... imagine if Intel & AMD decided to make their own boards in-house rather than just chips n chipsets... *scared*

[toofast]

Hi PiLsY did your SolteK mobo come with dipswitches ? just curious

[PiLsY]

Nope - theyre silk screened onto the PCB and have the solder pads in place, so I could add it np, but it didnt come with them.

I feel I should say now incase anyone goes and buys this board - DONT. Theres so many bugs its just not funny.

Setting memory to DDR400 gives no post, but overclocking to DDR400 using the 4:5 divider is no problem. For some reason soltek have disabled the 1/5 pci divider. Getting into bios with a usb keyboard is 50/50 at best. CPU overclocking is absolutely crap - I cant get over 3.4ghz stable. Memory overclocking isnt much better - because of the lack of TRAS, TRCD and TRP its not really possible to tune your memory in at all. Its an all or nothing scenario. The lack of the 1/5 divider combined with via onchip USB sees the return of the old KT266 problem - the USB craps out randomly at anything over 160fsb. I also just lost my windows install for the third time. It undervolts the same as the IT7, but voltages are much more stable, so the power regulation is good.

All in all I suppose its a good board for a day to day overclocker, but anyone who really wants to push things would be better off looking elsewhere.

Gonna have to kill it and send it back. Dont want this thing in my rig a minute longer than it has to be.

PiLsY.

[PiLsY]

Right a final note.

The soltek board is unstable at any FSB over 156mhz. Its not the cpu thats crapping out, its the memory. Running 1:1 with lowest timings it fails memtest. The strange thing is it boots all the way up to 166mhz. Can also boot at asynch mem speeds, always get the same errors though. Its no more unstable with mem @ 4:5 and 3:4 using 157fsb than it is at 1:1. Definitely the board. The chipset gets seriously hot - burning hot in fact. Its warm enough that you can smell the heat coming off the board. Id be tempted to try cooling it better, but in all honesty I really cant be bothered to mod a board just to get it to run at stock speeds (should be 166mhz capable - the chipset certainly is).

This board is a dog - avoid at all costs.

PiLsY.