DFI SLI-DR Expert, no support for Athlon 64 3400+?

I was fighting for probably close to an hour with some Athlon 64 3400+ CPU's trying to get them to work in an SLI-DR Expert.

Most wouldn't POST at all, system would turn on, no beeps and would just sit there. Another odd thing was that the CPU fan wouldn't spin at all with these inserted, except for in one odd case when I actually managed to get it to the desktop, at which point the system locked up completely. (not surprisingly, I must say)

I've tried an ADA3400DAA4BZ (E3 stepping Venice) and two ADA3400DAA4BY's (E6 stepping Venice).

It's really odd, like DFI just chose to skip from the 3200+ to the 3500+ and left out the 3400+ for reasons unknown. Which is just plain weird, because as far as I know every other socket 939 CPU ever made is on the supported CPU list, save for these.

Then, I went looking for a CPU support list and found one here. I must say I was quite surprised to find the 3400+ wasn't on the list at all, and not even any note about limited support (or no support at all).

Anyone care to explain the reason for this? RGone, you around? (I'm assuming he'd be the best person to ask, but anyone else please feel free to chip in if you know the reason why this isn't working.)

I must say this is rather annoying, since I bought them with the expectation of being able to do some benching with them. Now it appears as though I will need to find another board if I hope to run (or test) them at all.

3400+ was to be OEM only for HP systems I believe. HP probably over-estimated the numbers needed and then they trickled into the retail market and thus they are not supported on the Expert.
RGone...

RGone said:

3400+ was to be OEM only for HP systems I believe. HP probably over-estimated the numbers needed and then they trickled into the retail market and thus they are not supported on the Expert.
RGone...

Click to expand...

Okay, thanks for the quick response and explanation, RGone.

After scratching my booty a little more, I think the 3400+ would work in Ultra-D, Sli-D and Sli-DR. I think the bios was modded at one time for that series of socket 939 boards. But man it has been a long time and remembering that one odd 3400+ strained my brain. Hehehe.
RGone...

that explains alot... had a couple of 3400's myself which i couldn't get to work, i got them for free untested so i just assumed they were dead

Had a couple of used s939 A64 3400+ here, E6 I recall.
They worked flawlessly in Foxconn NF4, MSI NF3 Ultra, RS480 and VIA K8T800Pro boards, from the getgo.
Did not have to look up any BIOS updates.
Basically same chip as the A64 3500+. BIOS and OS reported them as 3400+.

mammut said:

Had a couple of used s939 A64 3400+ here, E6 I recall.
They worked flawlessly in Foxconn NF4, MSI NF3 Ultra and VIA K8T800Pro boards, from the getgo.
Did not have to look up any BIOS updates.
Basically same chip as the A64 3500+. BIOS and OS reported them as 3400+.

Click to expand...

Correct but you did not name the DFI boards that it did not work on.
RGone...

^And how many s939 3400+ did you try with? I am also curious what would have made DFIs so special in this regard as not even being able to POST. Since the fan would not even spin seem to indicate the chip is broken? I guess the beep codes should be possible to lure out if the chip is not supported.

at OP, guess you did a thorough clear CMOS between and after swaps and tried with one RAM stick only? If you get into BIOS disable CnQ, in case it is the P-states that cause the hiccups. Do the stock Vcore look sane in BIOS/PC Health?

EDIT: If you manage to get into BIOS set HT to 4x (800MHz). Recall seeing a s939 hp OEM 3400+ that defaulted HT to 800MHz. The 3500+ goes with 1000MHz on s939.

Like I have already said in post #4, if I am remembering stuff that is way old, there was a bios that let those three DFI boards work with OEM 3400+. Beyond that I never really cared since I was not going to run a 3400+. I got one for testing only and then put my dual cores back in the socket until the old s939 gave way to later AMD chipset boards not nVidia chipset boards which became real pain to live with for me.
RGone...

mammut said:

^And how many s939 3400+ did you try with? I am also curious what would have made DFIs so special in this regard as not even being able to POST. Since the fan would not even spin seem to indicate the chip is broken? I guess the beep codes should be possible to lure out if the chip is not supported.

at OP, guess you did a thorough clear CMOS between and after swaps and tried with one RAM stick only? If you get into BIOS disable CnQ, in case it is the P-states that cause the hiccups. Do the stock Vcore look sane in BIOS/PC Health?

EDIT: If you manage to get into BIOS set HT to 4x (800MHz). Recall seeing a s939 hp OEM 3400+ that defaulted HT to 800MHz. The 3500+ goes with 1000MHz on s939.

Click to expand...

I don't normally reset the BIOS between swaps because it's a major pain in the *** having to redo all of my bios settings over and over if I'm testing multiple CPU's. DFI boards are usually good about detecting new CPU's and applying the proper settings anyway, so I rarely bother with it.

But, of course I tried that too after nothing else worked, and it didn't help either.

Yes, also tried with one stick of ram (from several different models), didn't work.

Cool 'n Quiet is disabled in the bios by default, so that wouldn't make a difference.

Also, an update from me. I tried all three chips in an Asus A8N-SLI Premium board I had here tonight. All of them fired up and ran just fine. Seems the DFI just has no support for them, as it's not bad CPU's.

Interestingly, the Asus A8N-SLI Premium and Deluxe also do not have the 3400+ listed on their compatibility lists either (they're essentially the same board), but seems to run just fine in the Premium (haven't tried in a Deluxe, but would expect the same results).

Why these have an 800MHz HT Link speed is beyond me (that's what they defaulted to on Auto settings), that's a step backwards to the S754 days, as all other S939 CPU's default to a 1000MHz HT Link speed.

If I try this in a DFI again I may have to manually set the CPU settings (HT Link mainly, and perhaps vCore) with another CPU in the board before I attempt to swap for a 3400+. The vCore is not a major concern to me though, as it's the fairly standard 1.4v that most of the Athlon 64 single core chips use (the non-Clawhammer ones anyway).

If I try this in a DFI again I may have to manually set the CPU settings (HT Link mainly, and perhaps vCore) with another CPU in the board before I attempt to swap for a 3400+. The vCore is not a major concern to me though, as it's the fairly standard 1.4v that most of the Athlon 64 single core chips use (the non-Clawhammer ones anyway).

Click to expand...

I second that. The reason I mentioned Vcore is that if the DFI does not know what do with that processor, then it is better to set some fixed value (~1.35V) in BIOS prior to swap.

RGone said:

After scratching my booty a little more, I think the 3400+ would work in Ultra-D, Sli-D and Sli-DR. I think the bios was modded at one time for that series of socket 939 boards. But man it has been a long time and remembering that one odd 3400+ strained my brain. Hehehe.
RGone...

Click to expand...

If you can recall what site hosted the modded bios, it could probably be found on the internet archive http://archive.org/index.php
I've had lots of luck locating relics of the past with that wonderful tool.

rescuetoaster said:

If you can recall what site hosted the modded bios, it could probably be found on the internet archive http://archive.org/index.php
I've had lots of luck locating relics of the past with that wonderful tool.

Click to expand...

I cannot remember everything about all particulars since that cpu was never really to be sold thru consumer channels and was made up as a proprietary cpu for HP, I think it was. I know DFI made up a bios for the Ultra-D type boards and the SLi versions I believe. Never the Expert or Venus that I remember though.

I had one of those 3400+ cpus for testing and it would not boot on my Sli-D board and they sent a bios to me that did work. Then they uploaded it. Don't remember if it was a Beta only bios. Seems the fix was incorporated in the later bioses of all the Ultra-D, the SLi-D and DR boards. But man I just do not remember, that all happened nearly 7 years ago.

Again as I said before, the 3400+ cpu never had a bios made up for it and Expert or Venus motherboards. At least nothing in these old brain cells remembers that cpu made to run on the Expert or Venus. Other cpus had come along by then.
RGone...

I just looked at what is still up of DFI's download page, since I was trying to 'jog' my memory about the old 3400+ thing.

I did go to the Beta bios download section and found something I never realized before >> Date Code 2007-11-02 << I don't guess I ever even looked to see if there were any later bioses beyond 2006 for the NF4 boards like Ultra-D, Sli-D, Sli-DR, Expert or the like. But there was a 2007 bios released in BETA only for those few boards. I still have a couple of those boards, I guess time permitting I might try that last beta if I get the dirver first since it changes out the raid rom for that later one everyone seemed to want.
RGone...

I didn't know there was a bios release in 2007 either.

I think I've got the 2006-04-06 bios on my boards.

I've never really tried looking up the beta bioses since I tend to stick with the official releases.

Well knowing at the time something of what was in the DFI bios, I probably ran a Beta bios more than any other kind. Hehehe. When they did the release of the 2006-04-06, I did finally move to it for most of my NF4 s930 boards.
RGone...

Well, just found out my SLI-DR (non-expert) will run the Athlon 64 3400+ (E6) just fine. I swapped out the 3000+ Venice for the 3400+ Venice and it detected it just fine.

Weird, no idea why this would work, but it wouldn't work in the Expert.

i remember using a 3400+ in my ultra D just fine, didnt stick with it and wound up using an opteron as it was more fun.

good place for bios on boards and companies that are long gone...

http://www.rebelshavenforum.com/sis-bin/ultimatebb.cgi