LanParty 875P-T issues

[Guest LightSpeed]

Where to start...buggy BIOS sounds like a good place.

First off, clock generator always adds 1 FSB to whatever you choose; this is downright annoying! Second of all, the board seems to have a huge attitude when attempting to turn on; namely the AGP/DIMM overvoltage protection. If I set my PSU to anything higher than 3.32v (set with DMM), the board has real problems turning on (PSU flashes on/off repeatedly).

5:4 ratio is just useless, and I would love to know why the mainboard temp is always showing 24-25C, even if the room temp is 10C (can you say BS reporting temp)

There is also the issue of losing the AGP/PCI lock which seems to happen ~275FSB.

It is sad to see that I am not the only one with these issues.

[Angry_Games]

the board does have overvoltage protection. This is built into the board and cannot be changed without some sort of modification (voiding warranty)

 

5:4 is still being looked at by bios engineers

 

as for the lock...nothing is guaranteed when overclocking. 275FSB might not be good for you, but its a lot better than 200FSB during overclocking...I've not clocked mine beyond 260x14 at the moment so can't comment other than there are no guarantees while overclocking.

 

No mobo mfg can guarantee any kind of overclock nor that any 'locks' will hold up beyond default speeds.

 

temps...mine reports temps correctly as almost everyone else's are reporting correctly.

 

Biggest issue I've seen is 5:4 doesn't work worth a darn unless you have specific RAM modules (and i dont have a clue which ones they are, OCZ Tony would have more info than me) and that the bios is very tight and overclocks are 'not as high as other motherboards'

 

that again is part of overclocking and no guarantees are made.

[Guest LightSpeed]

Overvoltage protection is one thing, but the PSU is not even giving an extra 1%, never mind the 10% tolerance for power (I don't like more than 3% personally but that is another story).

If I set the FSB to 200, I should see 200 in windows, NOT 201. Same if I set to 240, it should be 240 in windows, NOT 241!!! That has to be the lamest way to 'beat' other boards in default benchmarks, but besides that it is simply stupid.

As for RAM modules, I have several here and none of them seem to like 5:4, TCCD, Hynix, you name it. If you could mention 1 that actually works it would be something :nod:

I don't buy the 'BIOS is tight' as I can hit 275 effortlessly, but then I am not a first-time overclocker either. The board is very aggressive with the memory, especially the TCCD, but that doesn't keep the FSB down for me

Regarding temperature, could you explain how it is possible for a mainboard to be @25C when the CPU is @16C and the room is @10C? Even better, can you show me a screenshot of a mobo temp with anything other than 24 or 25C?

One can see from the latest BIOS release the LACK of testing, when the fixes were supposed to be 'enhanced D0 stepping support', which in reality castrates the L1 and L2 Cache on my D0 prescott... WTG

One last thing, the IRQ assignment for the PCI slots needs some work as well, and can I ask why the IDE controllers use IRQ 11/14 instead of 14/15?

[BigStan]

Overvoltage protection is one thing, but the PSU is not even giving an extra 1%, never mind the 10% tolerance for power (I don't like more than 3% personally but that is another story).

Just for comparison, I have my Powerstream rails set at +5% and I am not having any problems booting.

 

I'm guessing you are trying to reach +275fsb with the 14 multi of that 3.4E proc. That 14 multi gave me fits also. With the proc I have now, I am stable at 280fsb. It ight be able to go higher, but without a working 5:4 divider, I am unable to test it. But the 15 multi is the default. I tried the same proc you have for a while. But I had problems with the 14 multi similar to what you are seeing, apparent lack of AGP/PCI lock, unable to go much over 275fsb. I wonder if it is a problem of the a buggy BIOS trying to use the 14 multi.

[Guest LightSpeed]

Problem is not so prominent using Powerstream 520W, but with 600W it is very picky. When it happens, the light on the back of the PSU on the 3.3v rail flashes red, and sometimes I get a message saying my 9800 power connector is not plugged in. There is also a second or 2 delay from when I hit the power button to when the system actually turns on; not sure if it has to do with the 4 phase power or not.

I will pick up a E0 stepping one of these days; just haven't decided if I want 3.0 or 3.2 yet, but wanted the lower multi for testing high FSB RAM. Even using the 17x multi, the L1 and L2 cache scores are cut by more than half; something is definitely wrong with the latest BIOS.

Thanks for any other input.

What did you have to do in particular to hit 280? I had to use 1.8AGP to hit 275, 2.9v on the RAM, disable spread spectrum, and the 'auto detect pci clock' is enabled. Other than that I didn't have to do any voodoo

I did find I could go 4FSB higher using the orange slots, rather than the yellow ones.

The strangest thing is *sometimes* when I load my confirmed stable BIOS settings, they seem to fail, but the funny thing is they are only saved AFTER I pass everything in my testing suite (spi, seti, all 3dmarks, memtest, etc...). For example, I can pass memtest for hours, but if I change a few things and then load up the same BIOS, memtest fails like no tomorrow, and the worst thing is that I cannot reproduce it on demand!

Which BIOS are you using Stan?

[BigStan]

I'll have to get back with you this evening on the BIOS version. I flopped around a bit, and I don't remember which on I settled on. When I get home this evening, I'll check for you.

 

Mem is at 2.8v. Which was odd, because on my Abit IC7 Max3, I needed 2.9v to hold 277. On this mobo I get mem errors at 2.9v, but error free at 2.8v

 

AGP volts at 1.7.

 

CPU at 1.5v

 

Spread Spec off, auto detect pci clock is enabled. All legacy ports, ie parallel, com, etc., that I don't use are diabled.

[BigStan]

I can pass memtest for hours, but if I change a few things and then load up the same BIOS, memtest fails like no tomorrow, and the worst thing is that I cannot reproduce it on demand!

I just saw this. When making several changes in the BIOS, it seems to lose it's mind. I have to reset the CMOS and start over.

[Guest LightSpeed]

Thanks for confirming the BIOS anomalies with multiple changes...thought I was losing it at first, but repeated saving of known stable config and reloading the same ones just is not as easy as it sounds :nod:

Thanks for checking Stan; I was ready to gut my case and RMA the board, but there is some hope still

I can't even get memtest to load without 1.8vagp >270FSB with 1/25 BIOS using the OCZ Plat 4200s and am debating just getting some 3200r2 instead, but my memtest errors in test7 only lead me to believe it is either the BIOS or the RAM

I can use 1.6vagp with the HyperX PC4300 :confused:

[BigStan]

Try more VDIMM, or maybe even less VDIMM on those OCZ 4200s. You sort of need to retry things that may/may not have worked on other mobos. This mobo pumps more bandwidth thru your mem than other mobos do. So you might need to relearn what you mem can/can't do. Like I mentioned, for some reason I had to back my voltage down from what I used to use. Which thru me for a loop for a week of testing them with what I knew previously were known good settings.

[Carnajo]

Sure, the BIOS is tight and gices "lower" overclocks, and no overclock is guaranteed. Fair enough.

 

But a 5:4 divider is advertised as one of the features/functions of the board, we expect it to work.

[Guest LightSpeed]

Well I had already typed a response but it disappeared saying I wasn't logged in

Here is the condensed version

I spent the last week testing 4 sticks from 2 dual channel kits, in each orange and yellow slot; enough to make me sick of it and take a break on my northwood/asus rig to regain my sanity.

The latest BIOS seems to be for the birds though; I was happy @275 for the time being using 1/11, which was a great improvement over 11/30 which would not even let me boot into windows with the HyperX.

5:4 has nothing to do with overclocking; if anything it is underclocking, but it is a feature and should work, and the clock generator +1FSB should be adjusted. After all, *every* utility sees the extra FSB, and we actually like to be able to use the setting we select, without a little extra 1FSB, or .9, or .8. Alot of the mobo makers boost the 200FSB setting, but they don't mess with the FSB settings when the user decides to do it themself, and none of the reputable ones add a cheesy FSB to every one from 200 to whatever the board loses its lock at

The reason I am happy with 275 is that I like to run my RAM at 1:1, not 5:4 or 3:2, but I know something is whacky when I can't get past 280FSB no matter what divider I use, or voltage for that matter, and temps are once again not an issue.

If I were to compare the amount of hoops one has to jump through on a DFI board to a circus dog, I would say the dog has the easier job of us :cool:

I think the reason you only need 2.8vdimm is because you are overvolting your 3.3v rail by 5% = ~.1v

I am giving my ram 2.9v, but letting the mobo do the overvolting rather than the PSU, which seems to be the logical way to me, seeing as DFI gave us vdimm to 3.2v, etc... Using a solid PSU like Powerstream, I don't see any need to overvolt the rails that much. I usually go 3.35v, 5.1v and 12.2v, but again have tested 3.30, 5.5, 12 up to the above mentioned settings and all kinds in between. The most ripple I have seen is ~2.5%, and that is usually on the 5v rail.

[Guest LightSpeed]

To explain my mysterious CMOS demonic possession or whatever it is, here is an example:

I run 14x 275 till I get bored, then go to 17x to do some testing, etc.. blablabla, then load a 'cmos reloaded' bank to go back for some comparison 14x275 @ same approx clock speed, etc... but now nothing works; 3dmark fails instantly, memtest is crapping all over my screen in red, etc... when it worked for days prior to the changes.

FWIW, my CPU can do 235FSB on 1.375v using 17x multi, which is over the ~3.9ish I am running at high FSB. Increasing CPU to 1.5v didn't make any difference, and with that @17x multi I run 4.2GHz, at least I did at one point :eek:

It is too bad this message board doesn't have enough emoticons, because if I wasn't laughing about all of this I guess possibly I could be upset, but why bother with all of that. After all it is only PC hardware, and I test enough of it on a daily basis that it almost bores me now. DFI has brought a unique twist to my mundane overclocking world of ASUS stability and features that actually work, such as on a P5P800. Incidentally that board gives the LP875T a good run performance wise, but is missing firewire and other stuff I disable anyways like printer port, firewire, on-board audio. Speaking of audio, has anyone had issues using SB Audigy2 ZS? I have mine in slot 4.