The EPoX 4G4A+ seems very promising.

[Acer]

Originally posted by: senway
Just noticed something....the led that lights up with the POST codes, what is that supposed to read? I see C1 and C3 very quickly as soon as i boot, then it winds up to FF and stays there while in windows. Is that a prob?

That is what I get when in Windows. Those codes are significant if the puter hangs at a certain point during the booting process, after in windows it means nothing.

 

[anazoal]

Not to bash anyone, but I recommend that people read the mobo manual.



The 4G4AI looks like the perfect board (price + features) for those who don't want onboard RAID. At $100 from Newegg, I would have bought the 4G4AI and a PCI RAID card for greater flexibility.

 

[Acer]

I go along with reading the manual. Nothing wrong about mentioning it. I just found out that in order to populate dimm slots 2 & 3 at the same time, the memory stick has to be no larger than 512 mb and it has to be SINGLE sided. Otherwise, one stick won't register with the bios and Windows. Those two slots are shared slots.

 

[JingHola]

Hi, I am experiencing instability and freeze ups when using two Samsung 512MB DDR PC2700 CL2.5 mem sticks
populated in DIMM slots 1 and 2. With only one stick of 512MB mem in either DIMM slot 1 or slot 2 the system
runs stable. With two sticks, even though the system recognizes 1GB installed and runs... as soon as I
try to run 3DMark2001SE tests, the system locks up and I have to perform a reset.

My setup consists of:
Epox 4G4A+ (bought from Newegg.com)
Pentium 4 2.26Ghz, 533Mhz CPU (lbought at retail local computer store)
eVGA GeForce4 TI4200, 128MB RAM with VIVO feature (from Newegg.com)
1 stick of Samsung PC2700, CL2.5 rated DDR mem. (from OCSystens.com)
1 stick of Mushkin PC2700, CL2.5 rated DDR mem, which happens to be exactly the same as the Samsung chip. (NewEgg.com too)
Cheap Generic full tower case with generic 400Watt peak rated PSU (won off of eBay).

I have read about other peculiarities with using more than one DIMM slot, but only when it comes to
slot 2 and 3 usage and ro overclocking. However, I am using slots 1 and 2 populated properly with
these two DDR modules I have as far as I can tell from the specs.

So why can't these 845G boards use more than one slot at a time and remain stable?

I was running everything stock default too, no overclocking at all and the system would freeze up when
I had 1GB memory installed as two 512MB mem miodules in DIMM slot 1 and slot 2.

Sorry if this (my particular symtom) has been covered elsewhere before... I did not find anything mentioning
my case legally populating mem slots 1 and 2 which is supposed to work okay, no?

Has anyone else with this board experienced same? I would suspect that it could be video card or drivers
related except for the fact that all that stuff runs fine with one mem stick in slot 1 or in slot 2 alone.
Two mem sticks in there and it becomes totally unstable even at non- overclocked default settings.

Currently I am using one 512MB DDR stick in slot 1, overclocked MB at FSB=140 with Auto DDR settings
and Coolbits used to OC my GF4 TI4200 @ 300/550. All is stable and rock solid like this with one mem chip.
If I add the second one back in... freeze ups. Both mem chips are good chips for each used alone work
just as stable. So go figure... why the heck they put more than one DIMM slot when they cannot
accomodate their usage is beyond me. Seems like those DIMM sockets are there for looks and marketing
impression only~!

I also upgraded to latest EPOX 6/17/2002 BIOS and that improved the use of two sticks alittle bit, but still
get the freeze ups (not as often, but still get them). And the freeze ups seem to happen most easily when
you go to run graphics stuff like the 3DMark2001SE benchies.

Hope someone out there in the world figures this one out and Epox can put a new BIOS fix or hardware
modification out for this board. I lost hope in it for now.

 

[senway]

*sigh*

i did read the manual, and I only asked because I still didn't understand when those codes are significant (i.e. you should be concerned w/ them only when the comp won't post, or you should monitor them all the time). Acer's quick confirmation set me straight, thanks! The reason for my concern was that I was hoping that I might have had some unchecked problem that was the source of my "low" overclock...unfortunately I guess not.

 

[Acer]

JingHola: have you checked the output of your PS for 5/12? I mean measure it at the HD plug-in and at the ATX molar plug-in?

Since you really do have different vendor's memory even though Samsung, I would set the memory timing to manual and run it at 2.5, 7-3-3. See what happens then. let us know the outcome. BTW, you didn't say what your Vcore or Vmem voltages are set to.

 

[gsethi]

Originally posted by: JingHola
Hi, I am experiencing instability and freeze ups when using two Samsung 512MB DDR PC2700 CL2.5 mem sticks
populated in DIMM slots 1 and 2. With only one stick of 512MB mem in either DIMM slot 1 or slot 2 the system
runs stable. With two sticks, even though the system recognizes 1GB installed and runs... as soon as I
try to run 3DMark2001SE tests, the system locks up and I have to perform a reset.

My setup consists of:
Epox 4G4A+ (bought from Newegg.com)
Pentium 4 2.26Ghz, 533Mhz CPU (lbought at retail local computer store)
eVGA GeForce4 TI4200, 128MB RAM with VIVO feature (from Newegg.com)
1 stick of Samsung PC2700, CL2.5 rated DDR mem. (from OCSystens.com)
1 stick of Mushkin PC2700, CL2.5 rated DDR mem, which happens to be exactly the same as the Samsung chip. (NewEgg.com too)
Cheap Generic full tower case with generic 400Watt peak rated PSU (won off of eBay).

I have read about other peculiarities with using more than one DIMM slot, but only when it comes to
slot 2 and 3 usage and ro overclocking. However, I am using slots 1 and 2 populated properly with
these two DDR modules I have as far as I can tell from the specs.

So why can't these 845G boards use more than one slot at a time and remain stable?

I was running everything stock default too, no overclocking at all and the system would freeze up when
I had 1GB memory installed as two 512MB mem miodules in DIMM slot 1 and slot 2.

Sorry if this (my particular symtom) has been covered elsewhere before... I did not find anything mentioning
my case legally populating mem slots 1 and 2 which is supposed to work okay, no?

Has anyone else with this board experienced same? I would suspect that it could be video card or drivers
related except for the fact that all that stuff runs fine with one mem stick in slot 1 or in slot 2 alone.
Two mem sticks in there and it becomes totally unstable even at non- overclocked default settings.

Currently I am using one 512MB DDR stick in slot 1, overclocked MB at FSB=140 with Auto DDR settings
and Coolbits used to OC my GF4 TI4200 @ 300/550. All is stable and rock solid like this with one mem chip.
If I add the second one back in... freeze ups. Both mem chips are good chips for each used alone work
just as stable. So go figure... why the heck they put more than one DIMM slot when they cannot
accomodate their usage is beyond me. Seems like those DIMM sockets are there for looks and marketing
impression only~!

I also upgraded to latest EPOX 6/17/2002 BIOS and that improved the use of two sticks alittle bit, but still
get the freeze ups (not as often, but still get them). And the freeze ups seem to happen most easily when
you go to run graphics stuff like the 3DMark2001SE benchies.

Hope someone out there in the world figures this one out and Epox can put a new BIOS fix or hardware
modification out for this board. I lost hope in it for now.

What OS are you running ? Win 9x and ME wont work properly with anything over 512 MB. Also did you try slot 1 and slot 3 combo ?

 

[BlessidThug]

mcveigh: An earlier poster (Vsekh 06/23/2002 8:13PM) said that the 4G4a does have the NIC. It's only covered by the faceplate. Remove that and enable it in the Bios and you are good to go. Are there any 4g4a owners that can confirm this?

Originally posted by: mcveigh
anyone know anything about the epox 4G4AI ?
it seems to be the 4g4a plus a NIC,

 

[nightowl]

Whenever I run Prime95 I get a hardware error during test 6. I was getting them during test 3 but after I bumped up the Vcore .05V and Vdimm .1V I was able to go to test 6. Anyone have any suggestions to fix this. I am running Win2K with the following hardware:

Epox 4g4a
P4 1.6A w/ retail HSF and ASII temps are reading in 40's under load and idle
512MB Mushkinn PC3000 in Slot 2
8500LE
Intel NIC
Sb Live
Enermax 350w PSU

All of the voltages seem to be within spec and the Vcore voltage drops to 1.5V under load.

 

[nightowl]

DP, stupid slow dial up.

 

[Acer]

So, you are not o/cing. have you mesured your 5/12 at the HD plug and ATX plug? I would do that to see what the Enermax is really delivering to your board and HW.

Continue moving the Vcore and Vmem up.

 

[nightowl]

I am running a 2.13GHz.

Everything is going OK for now. I bumped the Vcore up to 1.575V and I am on test 12 right now.

/Crosses fingers

Edit: It failed at test 15. Max temp was 47C as read by MBM5. Windows booted fine at 2.13GHz with stock voltage.

 

[JingHola]

Acer, I am running all voltages at the default settings and I had also tried increasing the Vdimm memory
voltage up to the max +0.7 and still the system would freeze up with two memory sticks installed in
slot1 and slot2. I have not tried putting them in slot1 and slot3 combination and I also have not
tried manually setting the memory timings yet. I will give these two things a try and report back when
I find out the results.

Gsethi, I am using Windows XP Pro with all the updates, so I don't have the WinME and Win2000 512MB
bug limitation here.

Thanks for the replys, I will let you all know what if any progress I ever make on this dilema.

 

[Acer]

Don't forget to raise the Vcore too.

 

[JingHola]

Okay I just got done testing out my system with two 512MB PC2700 (DDR333) CL2.5 rated Samsung memory
modules in slots 1 and 3. It did not take long for my WinXP system to freeze up. It booted a couple of times
into WinXP successfully but as soon as I started IE6 browser and start doing things, it freezes up dead. I have
to hit reset to reboot.

Then I tried manually setting the RAM timings. I set it to 2.5,7,3,3. In auto SPD it was 2.5,6,3,3.
That did not work either. Same behaviour. Boot into WinXP, start doing normal stuff and eventually the
system freezes and locks up. Have to hit the reset button.

So none of this stuff works. BTW this time I also had the MB Overclocked with FSB=140 and Auto DDR which
gave me a DDR of around 350 I think (cannot exactly remember it now). But this overclock setting always
worked rock stable with only one memory stick installed in either slot1 or slot2. Since I know that changing the
Vcore and Vdimm voltages upwards did not make any difference when running the two memory sticks together,
I did not bother repeating increasing these voltages this time and I left them all at the defaults (no increases).

I finally gave up and went back to one stick of 512MB DDR333 RAM installed and all was rock stable again.

BUT then I had a brainstorming session (whilst I was outside having a relaxing miniature cigar break)... I recalled
that in all the test reviews that I have read, they all mentioned that Intel says that these i845G chipsets DO NOT
officially support DDR333 memory bus speeds, even though they let other manufacturers design their boards with
BIOS to go up to these DDR333 speeds and beyond. This board as well as other i845G boards are only specified
to officially support up to PC2100 DDR RAM. The ability to OC them to use DDR333 RAM is just gravy and a selling
point (as in a gimmick). So, guess what I decided to do to see if my Epox 4G4A+ board was working properly or not?

Even though I am using two 512MB DDR333 modules and the EPOX BIOS correctly reads and detects them as such,
I figure Intel did not condone(-sp) official support of DDR333 RAM for their 845G chipsets for no reason. This I am
guessing is one of the reasons why Intel did NOT Officially say the 845G chipset supports DDR333 speed RAM.
It must be something in their 845G memory bank switching that cannot reliably switch between the DIMM socket
memory banks at DDR333 speeds! Using one DDR333 works fine, stick another bank into the second slot and the
i845G circuit cannot keep pace switching the address lines.

Anyway, that's my take on it other than maybe I got some bum slow memory modules that don't work exactly well
together (like Acer stated). Proof it works at PC2100 speed? I changed my memory ratio to 1:1 so that at FSB=140 overclock I get a DDR=280 speed setting. (before I had always had it set to 4:5 ratio or AUTO DDR which gave me a DDR350 or higher
setting). I am using my 1GB of RAM (two separate 512MB DDR333 CL2.5 RAM modules in slot1 and slot2) right now
and it has not choked or frozen up on me yet. I ran through one test run of 3DMark2001SE and it passed with a
3DMark 0f 10360. I am using a eVGA brand GeForce4 TI4200, 128MB video card overclocked with Coolbits to 290/525
right now. So did I solve my dilema of how to get this Epox 4G4A+ MB to accept my two sticks of DDR333 RAM?
I hope so. I basically have my PC overcloked at FSB=140 and 1:1 ratio to give me PC2100 DDR280 speeds. This
falls within the officially supported specifications of the Intel i845G as per Intel and the MB makers.

I will load test this setup with these settings and even try to OC the CPU higher whilst trying to maintain the
DDR memory at PC2100 levels and see if it all holds up.

Hopefully I might have solved this case... stay tuned, for I will post any corrections to my theory if anything fails
with this setup!

Thanks for listening.

I edited this in case my main point was not clear from all my rambling... if you are having the same probs using two
512MB PC2700 rated mem modules, change your memory speed ratio to 1:1 or whatever it takes to get the DDR speed
down to or near PC2100 spec speeds. Then the two memory modules will work reliably... if I find out that this is not
the case with my system after further testing, I will be sure to let you all know here!

 

[gsethi]

Interesting find there......

yep...the memory addresses issue might be the reason for 2 chips not working together. maybe thats why Intel still didnt officially supported the DDR 333 since they would have to rewrite the entire code for their chipset. thats just my thought.

 

[Acer]

All sounds very logical. A lot has to do with manufacturer. Samsung is NOT on Epox's list of recommended memory. Food for thought!

 

[JingHola]

Acer, you are right again, I forgot about that fact... I remember looking at the Epox recommended compatibility
listings awhile back and did not see the P4 2.26Ghz, 533Mhz CPU processor listed for this 4G4A+ board either!
I said to myself, how the heck could a MB maker not test and list all these various popular CPU's to validate they work on
their MB's when it is in their best interest to make and increase sales. Same for the memory chips, I either did not
remember that Samsung is not listed there or I assumed that it was just not tested by Epox nor verified compatible
yet but would be... (this is a BIG assumption really by me, because Epox as smart as they appear to be, should
have done this homework already)... so maybe they did and maybe that is why they did not list this Samsung memory
that I am using as recommended for the 4G4A+ board?...! Anyway, if not, maybe I just helped them... they can leave
this Samsung DDR333 memory off the list for those who want to OC and use more than one mem module slot! Are you
listening Epiox?

Anyway, here is my latest update... my system (with the two Samsung 512MB DDR333 RAM modules) is still running
okay with no freeze up's at. It ran for hours last night running video game demos and 3DMark2001SE bench tests.
Woke up this morning and fired it up cold and it is stable and ran 3DMark2001SE all the way through no probs giveing
about a 10440 3DMark score. Still running stable with FSB=140 and 1:1 ratio yielding DDR=280 speed. Stable.

BTW, this trick ought to work with all brands of Motherboards if my basic theory is correct. So all the other people
who've had similar headaches with trying to run two fast 512MB DDR333 rated modules and had system stability problems
ought to try what I did and force the BIOS into clocking memory at or near the PC2100 speeds and NOT at the PC2700
speeds... of course this is a compromise for those wanting to OC and use two memory modules at the same time...
with these current generation i845G boards, it seems we are stuck with this quirk until they improve and release
the next faster cnipset that will officially support faster DDR333 and faster speed memory bank switching and access times.

Still running stable at FSB=140, DDR=280 with two 512MB Samsung PC2700. CL2.5 rated RAM modules stuffed into slot 1 and 2 of
this Epiox 3G4A+ board...!

 

[Acer]

At least you found your comfort zone and can get some realistic performance from it now.

I built this system for my son and I didn't read the manual before buying or asked the question of 2 sticks versus 1 stick. We now know that slot 2 & 3 are shared and only single sided memory can be in there if you wan to use both slots. Oh well, this is life of hard knocks and real live learning experiences! hehehehe

 

[Antoneo]

Is everyone running revision 1.0 of this board (4G4A+, 4G4AI, 4G4A)? That's what I have and am wondering if if got an "old" one.

Also, what is the general consensus on the version of the bios to use? The one that is already on board or the updated one on their website? I've been following this thread for a while and haven't really reached a clear cut answer. Or perhaps there isn't one... . If it ain't broke don't fix it? Thanks guys!

 

[terminalmind]

I've got the revision 1.0 motherboard and I'm using the original bios version 4/24

 

[legendma]

I've had this strange problem , so thought maybe someone can explain this or if anyone else has experienced it. I've had the system not bootup twice with a Error Post Code of 25.. I emailed Epox about this and their tech guy replied that the board was not recognizing the Bios chip?? I tried clearing the CMOS and that didn't even help.. Only way I got the system to post again was to remove the heatsink and CPU, which was a pain in the ass!! Wondering if anyone else ever saw this problem ??

 

[Acer]

Sounds like RMA time, but you could try flashing the bios to see if you can straighten that out.

 

[Muadib]

Originally posted by: MistaEng
Is everyone running revision 1.0 of this board (4G4A+, 4G4AI, 4G4A)? That's what I have and am wondering if if got an "old" one.

Also, what is the general consensus on the version of the bios to use? The one that is already on board or the updated one on their website? I've been following this thread for a while and haven't really reached a clear cut answer. Or perhaps there isn't one... . If it ain't broke don't fix it? Thanks guys!

I don't think a revision as been made on the board yet. I'm back to the 4/24 bios too. I was using a beta dated 6/20 that was sent to me by EPOX, because the 6/17 one on their site took away raid. Since most here were on 4/24, I went with it.

I think I may have to rma my ram, because I can't go beyond 133, and remain stable. This XMS stuff has been overated crap so far.:disgust:

 

[Antoneo]

This XMS stuff has been overated crap so far.

Those are handpicked Samsung chips on Corsair PCB correct? Should be good ... I'm just waiting around for my RMA'ed GF4 Ti4200 to arrive. Then I'll see what my Samsung Original 512 can do . Are you sure it's not the chip that's limiting you?