Beep code 4-2-3-3 when ATI HD5770 inserted into Supermicro C2SBX board

[dsperber]

Currently running a Supermicro C2SBX with fanless Gigabyte ATI HD4850 inserted. It's been running fine for two years. Dual monitors.

I recently replaced my old CRT monitor with a new LCD monitor, so now I have two 24" Eizo monitors (each running at 1920x1200). Unfortunately, the HD4850 only has one DVI connector so I've been running the new second LCD in analog mode using the VGA cable. It still looks beautiful at 1920x1200 but I'd really like to run it digitally, meaning I need a second DVI connector.

So I purchased a Sapphire Vapor-X ATI HD5770 card, which has two DVI connectors. Seemed just about perfect, even though I was really wanting another fanless card. But this Vapor-X card has very positive reviews for silence.

Anyway, I cannot boot the machine with the HD5770 inserted into the same PCIe x16 slot that the HD4850 currently lives in. No other changes in the machine were made, so it's exactly the same except for the swapped video cards. With the HD5770 inserted there is no video output from either DVI connector, and instead the Supermicro C2SBX Phoenix BIOS (v2.0) beeps 4-2-3-3.

Beep code 4-2-3-3 means "extended block move", which suggest a problem with video memory (i.e. mapping it, conflict with onboard graphics, etc.) although none of the possible causes mentioned by other users with similar problems on the Internet seem relevant to my situation.

Furthermore, trying out the HD5770 in my second machine (which has an ASUS P5Q3 motherboard in it) sees it work perfectly! So the new video card is not defective, and in fact does work perfectly in that other ASUS P5Q3 motherboard. Obviously there is some issue with the Supermicro C2SBX board... although both boards support PCIe 2.0 which is what this HD5770 video card uses (I believe).

It's not a power issue I don't think, as the C2SBX machine has a 600W PSU, whereas the P5Q3 machine has a 500W PSU. Yes, there are additional drives and cards in the C2SBX machine but honestly I don't think this is power related. It's most likely some other BIOS incompatibility between the C2SBX and HD5770 that just is not a problem when using the HD4850.


Anybody have any thoughts or suggestions? Any BIOS engineer experts out there who have experience with 4-2-3-3 (extended block move) in Phoenix BIOS?

This is already the second HD5770 I've tried, since I RMA'd the first one believing it to be defective. However since this second card works perfectly in the P5Q3 I'm certain this card is not defective.

 

[bobkn]

At the risk of being insulting:

Did you remove the Catalyst suite before swapping cards?

I'm not sure whether the machine would properly recognize the swap if the old drivers stayed.

Does your 4850 have an HDMI connection? If so, there are HDMI to DVI adapters available. (These are cheap, passive adapters.) The nVidia Quadro card on my workstation at my job uses one to connect the second monitor (with a DVI port). That would have been a cheap alternative to swapping graphics cards.

 

[dsperber]

At the risk of being insulting:

Not offended. Any comments or help or ideas is appreciated.

Did you remove the Catalyst suite before swapping cards?

The machine will not boot!

I'm not talking about a Windows driver issue. I'm talking about a machine malfunction. When the ON button is pushed and the machine powers on, instead of the single beep, lights on the monitors changing from amber to green, and then VGA-mode output delivered to the monitors from the BIOS verbose output, instead the machine emits 4-2-3-3 beep codes and nothing. I can hear the rest of what I believe to be normal drive activity, and probably the standard BIOS boot process proceeding. but there is ZERO video output on either monitor (both connected with DVI to the HD5770's dual DVI output connectors) and the 4-2-3-3 beep code repeats.

It will not complete the machine power-on normal self-test and BIOS boot sequence showing output on the monitors.


Now as it turns out, I've got Catalyst 11.12 installed, should it ever get that far. And of course on my other ASUS P5Q3 machine when the HD5770 was installed there the machine booted fine, and Windows booted fine. On the ASUS machine I'm running with Catalyst 11.3, and no problems. The HD5770 is 100% supported (apparently) even by Catalyst 11.3.

Most important the BIOS in the ASUS P5Q3 with its PCIe 2.0 BIOS support apparently supports this PCIe 2.1 HD5770, whereas the BIOS in the Supermicro C2SBX with its PCIe 2.0 support apparently does NOT support this PCIe 2.1 HD5770.

===>> I'm certain this is a hardware malfunction in the C2SBX, or a deficiency in its Phoenix 2.0 BIOS, relating to the attempted use of the PCIe 2.1 HD5770 card.

Does your 4850 have an HDMI connection?

No, it only has DVI and VGA. One of each type. And the DVI (digital) connector is currently going to one of my two Eizo LCDs, with the VGA (analog) going to the other (through its VGA input).

If so, there are HDMI to DVI adapters available. (These are cheap, passive adapters.) The nVidia Quadro card on my workstation at my job uses one to connect the second monitor (with a DVI port). That would have been a cheap alternative to swapping graphics cards.

This is not the problem. I don't have the luxury of a second DVI or HDMI port on the HD4850, although the second Eizo monitor of course has a DVI port in addition to the VGA port I'm currently using to connect to the VGA output of the HD4850.


After talking with Supermicro several times by phone over the past week on this subject, yesterday they asked me to open a support ticket so that they could address it internally.

Today I received an email stating that they'd assigned it to a second-level engineer to see if he could reproduce the failure, and then take it from there. Still haven't heard back further on that yet.

 

[dsperber]

Just phoned Supermicro to follow up before the weekend.

They told me the name of the BIOS team engineer who the case was assigned to, and he's going to send me an email telling me where he currently is on this. I supposed they may not have an HD5770 in-house to play with, but I'm at least pleased that they truly are taking this seriously.

They did say that while the original BIOS came from Phoenix, that their own BIOS team made their own further modifications, and/or certainly has the capability to fix problems and/or make enhancements if they need to.


So... I'm hoping that this story may have a happy ending. But of course I'm only "cautiously optimistic", and it may turn out that they cannot even duplicate my symptom but that the HD5770 works perfectly in their own C2SBX.

That would be terrible for me, of course,... but I'm a realist.

Worst case is that I proceed with my ASUS/Supermicro complete "internal parts" machine swap, since I know that approach will work (although I'd end up with the 500W PSU in the bigger machine, and 600W PSU in the smaller machine, because swapping power supplies as well is definitely something I do NOT really want to do if I don't have to). But that's worst case.

 

[bobkn]

I guess that my reading comprehension isn't the best today.

If you can't get past POST, that's bad.

A semi-random suggestion: have you cleared the CMOS yet?

 

[dsperber]

A semi-random suggestion: have you cleared the CMOS yet?

No, I didn't.

But things probably got effectively reset the other day, when I upgraded the BIOS from 1.0b to 2.0 and was back to all firmware defaults. I had to go through everything and make the few changes that are appropriate for my particular setup.

I honestly feel it's something more fundamental. The other day I alternately tried HD4850 and HD5770 three times, and each time I reinserted the HD4850 the machine booted perfectly. Each time I reinserted the HD5770 it failed.

Still haven't heard from the Supermicro 2nd-level tech on where he stands.

I may spend some time this weekend trying to determine if it's an incompatibility with some other card in my machine. I will pull all of the non-critical devices out of their slots as well as all non-boot drives, and get down to the bare minimum required to boot.

If I still get 4-2-3-3 then it's clearly a C2SBX BIOS problem. If it now works, I'll research further to find out which of my expansion cards is causing the issue with the HD5770.

But I don't expect it to work.

 

[dsperber]

Just an update on the status of this ticket with Supermicro...

I had a phone conversation today with the 2nd-level engineer to whom the ticket was assigned last week. He advised me that he had requested information from Intel (who provides most of the firmware microcode functionality for the BIOS, since the X38 chipset on the C2SBX is from them) regarding the compatibility of a PCIe 2.1 video card with the X38 chipset. He was hoping they might have such an update which could be incorporated in their BIOS for the C2SBX to solve my problem.

But he's not optimistic, since X38 was a 2007-era chipset and might have simply been "orphaned" as "legacy". Support for the GDDR5 memory on these newer PCIe 2.1 video cards just might not be possible with X38.

Or, as I mentioned to him that my ASUS P5Q3 with its P45 chipset (also from Intel, but newer than the X38) DID support the HD5770, I suppose it might be possible to try and use the appropriate code which works for another chipset (e.g. P45) with X38 even if Intel hadn't officially done it for themselves.


Anyway, he's still awaiting a reply from Intel. Then he'll notify me and we can go from there.

Depending on what I learn, I will or will not proceed with my own "Plan B", which as described earlier is to do a complete "internals" machine swap of my two systems, so that the ASUS P5Q3 ends up at the location which requires the new HD5770 dual-DVI video card but reconfigured to internally contain all of the parts and peripherals previously at that location on the C2SBX machine.

 

[dsperber]

Well, just a final post-script to close out this thread.

I never did get an update from the engineer at Supermicro as to whether or not Intel did have a microcode update for the X38 chipset which would allow a PCIe 2.1 video card to be used with the C2SBX board. I called several times, left voicemails and sent emails, and went through the main switchboard to try and get to him.

Obviously it's not happening. And even if it were available it would take development time (assuming they wanted to even do it), etc., and all in all this just seemed like a poor solution on the whole.

So I fell back to my "plan B", which was to use the HD5770 in my second ASUS P5Q3 machine instead, and then swap the other incompatible internals of the two machines so that I could also swap their physical locations and re-purpose each machine to do what the other was currently doing.


And now I can report that the "dual plastic surgery" has been completed. Both machine/patients have "survived" and are doing perfectly.

In the process I also made some upgrades to the machines to make them even more identical than they were before (e.g. added a USB 3.0 PCIe adapter card to one machine which had already been in the other, replaced a CD/DVD burner drive in that machine with a CD/DVD/BD burner drive as had already been in the other).

It also gave me an opportunity to do some cleanup and re-cabling inside each box, replacing an old long 5-device SCSI ribbon cable in one box and a 7-device SCSI ribbon cable in the other box, neither of which were needed any longer as there are only two SCSI devices remaining in both. So one short 27" 2-device SCSI ribbon cable now lives in both boxes. Very much cleaner, and better airflow in both.

Anyway, the important thing is that I now do have the HD5770 dual-DVI video card fully operational in the rebuilt ASUS P5Q3 machine, living and operating at the location which formerly was the domain of my rebuilt Supermicro C2SBX. And vice versa. The P5Q3 now has the four internal hard drives and my two TV tuner cards (ATI TV Wonder 650 PCI for 1-tuner OTA ATSC, and Ceton InfiniTV 4-tuner cablecard-enabled for TWC/LA) and dual-DVI 24" Eizo 1920x1200 LCD monitors and is my home's HTPC, and the C2SBX (still with ATI HD4850 video) now has the three internal hard drives and is my office PC with a single 19" IBM P92 CRT monitor.


Notes from "the surgery".

(1) On Saturday, my first attempt at installing the HD5770 into the ASUS P5Q3 resulted in a long-short-short-short beep code from the AMI BIOS!! This indicates a "graphics card problem"! I WAS HORRIFIED, as I'd run this very experiment a week before and it worked fine.

Well, nothing I could do seemed to solve the problem... just as swapping the HD4850 with the HD5770 in my Supermicro C2SBX had accomplished nothing a week earlier. HD4850 -> GOOD. HD5770 -> BAD. Well now with my rebuilt ASUS P5Q3, it seemed the same results occurred: HD4670 -> GOOD. HD5770 -> BAD!!

How could that be? In frustration I spent the rest of Saturday putting the machines back the way they were before the surgery. A total of 14 hours spent first performing and then un-performing the machine-swap. I couldn't believe I'd had good luck a week earlier in my tryout, but that the actual surgery was a failure.

(2) After a night's sleep, I decided to try this one more time on Sunday... this time once again replacing just the HD4670 video card in the P5Q3 with the intended HD5770, to see if I could reproduce my prior week's success. Well, surprisingly, I COULD!! The HD5770 once again worked perfectly!

Then, one card at a time, I removed the two TV tuner cards from the C2SBX and placed them into the P5Q3 where they were intended to go. And sure enough, the machine still booted fine. My original suspicion that it was an overall card/P5Q3/HD5770 incompatibility which caused Saturday's problems, well didn't seem to pan out. Could it be the hard drive swap that put it over the edge? Maybe overloading the 500W PSU? Didn't seem plausible, but who knows?

Armed with semi-confidence, and with all the cards now successfully transplanted and screwed down firmly into the P5Q3 I repeated on Sunday exactly the same remaining hard drive transfer I'd done on Saturday. With the P5Q3 now looking exactly the same as it had the day before, for some reason on Sunday the machine now worked perfectly!!! I'd simply done it "slower" on Sunday, confirming at each step that everything was working fine.

I can only believe I must have "knocked something out" on Saturday which caused the long-short-short-short beep code, unrelated to the compatibility of the HD5770 itself with the motherboard. You'd think the multiple swaps of HD5670/HD5770 would have eliminated this mistake, but apparently it must have been something else which remains a mystery, which explained Saturday's failure.

(3) Anyway, on Sunday I had 100% success, whereas on Saturday I had had unexpected 100% failure!!

So I completed the rebuild of the C2SBX and installed both machines in their new reversed location homes.

And now, after a bit of software reversal because of the reversed machines, setting up Windows Media Center and all my recordings in the new ASUS HTPC, nightly backup setup reversal in both machines, etc., I'm now fully operational on both machines!

And as was the ultimate goal of all of this, the location with the two 24" Eizo LCD monitors at 1920x1200 is now supported with the dual-DVI Sapphire Vapor-X HD5770 (which I might mention is as I'd read in advance, EFFECTIVELY SILENT despite the presence of a fan, much to my great glee).

Mission accomplished!

(4) The initial Catalyst 12.1 video drivers running in the rebuilt ASUS P5Q3 (which I'd installed days earlier, in anticipation of the machine upgrade pending) were extremely unstable with the HD5770 installed. I had no problem with the HD4670 (VGA out to my CRT) but BSOD every 5 minutes with the HD5770 (dual-DVI out to two LCD's).

Maybe it was just the "new hardware detected" auto-transition which was responsible, don't know. But after 20 BSOD situations I decided it was just unacceptable, and uninstalled the ATI 12.1 drivers in order to revert temporarily to the vanilla generic MS video support so that I could research the problem.

I learned that the newly released 12.1 has been extremely unstable for many users, and that upgrading to the 12.1a preview driver is a required solution to eliminate BSOD symptoms.

So that's what I did... upgraded to the ATI 12.1a driver.

And, sure enough... LIKE A MIRACLE all of the video instability disappeared, and rock-solid video stability returned!! My new HD5770 and dual-DVI twin 24" LCD monitors at 1920x1200 was working perfectly!!!


(5) So that's the story. Happy ending.

I still have a few more hardware/software-reversal items to go through (e.g. reversing the previous system backup datasets living in the external USB drives at each location, uninstalling/reinstalling some software which I'd had at only one machine and not both machines, etc.). But essentially THIS NIGHTMARE IS OVER!!!

It's apparently true that the X38/ICH9 chipset on the Supermicro C2SBX just wouldn't support the PCIe 2.1 HD5770 whereas the P45/ICH10 chipset on the ASUS P5Q3 would.

It's apparently true that either the original 600W PSU in the C2SBX machine or the current 500W PSU in the P5Q3 machine is perfectly able of handling the less than total 325W draw through the UPS battery-backup sockets (which includes the two LCD monitors along with the PC itself).

It's apparently true that the Sapphire Vapor-X 1GB (GDDR5) HD5770 is a stunningly quiet excellent video card, and has superb dual-DVI support for my two 1920x1200 monitors. The newer Eizo S2433WFS looks so much better in digital mode with the HD5770 than in the analog mode I'd been running it as with my old HD4850 from the C2SBX machine. Highly recommend this video card.


THIS CASE IS CLOSED!!! (no thanks to Supermicro, but a success nevertheless)

 

[dsperber]

Just one more item worth mentioning...

THANK HEAVENS I'D HAD THE FORESIGHT TO INSTALL A CAT6/GIGABIT-ROUTER WIRED ETHERNET LAN SETUP IN MY HOME!!

Turns out Windows Media Center and DRM causes "copy protection" limits and restrictions on TV programs delivered by your cable system marked with "copy-once". Here in TWC/LA virtually all programs are so marked.

This means they (a) can only be played back by the Win7 machine and installed OS that did the recording in the first place, and (b) cannot be played back by another WMC running on another Win7 system on the LAN.

Well, I now had TWO "operational" WMC machines... one (ASUS P5Q3) which really now had the TV tuner cards to be used for new programs to be recorded into a "Recorded TV" folder already populated with 300GB of recordings from a previous WMC (on the Supermicro C2SBX machine), and the second (C2SBX) which had previously made the recordings and thus was the only one eligible to play them back.

Fortunately, WMC supports defining the "Recorded TV" media library to be either local or on a mapped network drive or even on a non-mapped other computer. So I simply modified the previously "local" definition on the now relocated C2SBX machine to instead be the "mapped network drive" location of that folder now hosted by the newly relocated and WMC-enabled P5Q3 machine.

And, using that one common 1TB "Recorded TV" folder hosted by WMC on the newly relocated P5Q3, all copy-once programs previously recorded by the C2SBX are now still playable only by WMC on the C2SBX but from its new network location. All copy-once programs newly recorded by WMC on the newly relocated P5Q3 are playable only by the P5Q3. And all copy-freely content recorded (a) previously by the C2SBX or (b) newly by the P5Q3, well all copy-freely content is now playable by EITHER WMC.

In addition to being able to play eligible TV content on the monitor(s) attached to either WMC machine, I also have two WMC extenders around my house... to support the HDTV's attached to each extender at those two locations. An extender is conceptually attached to one WMC machine, but it can be easily reassigned to the other WMC machine. But it can only be connected to one WMC machine at a time.

So for convenience, and because it works out fine, I assigned one extender (in the same office room where the C2SBX now lives) to be attached to that C2SBX WMC. This allows me to watch-down all old copy-once recordings made on the C2SBX machine using that extender and HDTV. And the second extender I assigned to the new P5Q3 WMC, so that I can watch all newly made copy-once recordings (from the P5Q3 machine) on that extender and HDTV. Any previous or new copy-freely content I can watch on EITHER extender and attached HDTV.

And this is all possible because of (a) shared "Recorded TV" folder, hosted by one PC and accessible to WMC on that local PC as well as WMC on the network PC, and (b) GIGABIT WIRED ETHERNET NETWORK CAPABILITY.

For example, I was watching a program last night through the C2SBX WMC and extender. That was 20Mb/s delivering the stream from Win7 on the P5Q3 machine through the router and to the C2SBX WMC, and then another 20Mb/s from the C2SBX WMC to the router and to the DMA2100 extender and HDTV.

Each length of wired ethernet cable running between PC and router and extender is CAT6, and the router is a Netgear WNR3500 gigabit router. I don't think I would be able to do this without wired gigabit capability.

Certainly if I wanted to run two or more extender/HDTV pods simultaneously, and especially if I had "Recorded TV" folder shared like this between two WMC machines, or even worse if I had network-supplied TV tuners (e.g. HDHomeRun devices) feeding tuner source to the recording WMC... well an ordinary CAT5 network supporting 10/100 router... just could not be done.

Word to the wise: upgrade your home to gigabit LAN if given the opportunity.