I had to set initial duty to 50 to get rid of it.
Thank you, that did the trick. I think any value below 50 triggers the ramping up and down.
Now I have yet more serious issues.
Around 30% of the times I am powering on, the systems starts and I can access it by kvm, but the intel arc gpu is not recognized so blank screen. This only gets fixed by a bmc reset (cut power).
Around 20% of the times I am powering on, I get 3-4 low-pitched loud beeps, and no system is started. The bmc reports in the logs:
ID: 138 Unknown sensor of type bios_post_progress logged a bios : System Firmware Error (POST error)(No system memory is physically installed in the system) was asserted
But on the DIMM inventory everything looks fine:
Memory Attributes
128 GB 128 GB 4 4 Single-bit ECC
Individual Memory Details
DIMM_P0_A0 : Avant Technology W724GU44J2320NAPresence Detected
DIMM_P0_A1 : Avant Technology W724GU44J2320NAPresence Detected
DIMM_P1_A0 : Avant Technology W724GU44J2320NAPresence Detected
DIMM_P1_A1 : Avant Technology W724GU44J2320NAPresence Detected
Anyway, overall 50% of my time I can do some work with this computer! Before it was only 0% , a great improvement for this sh***y motherboard /sarcasm.
I hope they actually fix things with Rev1.1 and not just solder 2 extra mosfets / vrms and call it a day.
I got a new problem now. I upgraded the firmware of optane p5800x 400gb u.2 (dell-branded) and now the host is unbootable when the optane is plugged in to direct pcie lanes. If I plug in instead this updated optane to chipset pcie lanes, the system boots and optane works fine.
Could it be the nfamous PCIE 4.0 AER errors not letting me boot? However, before the firmware update I didn’t have any AER errors and the host was operational on the direct lane.
This is a bizarre problem that I did not figure out yet.
Worse is that Dell does not allow me to downgrade the optane firmware, because I get:
“nvme status firmware activation prohibited: the image specified is being prohibited from activation by the controller for vendor specific reasons (0x4113)”
For downgrading, I guess the BMC should issue some Vendor Defined Message (VDM) to the pcie bus but information on how to do this is scarce. I doubt I will ever manage to downgrade the firmware
I have the same experience with a Intel E810 NIC - only boots in the bottom PCIe slot that’s attached to the PCH. Stupid. My support ticket about this has been open since early Jan.
It looks very related indeed. Does this NIC come with some kind of alternative firmware that you could flash? It is a noob suggestion, i dnt know anything abt NICs
I forgot to mention that when the optane is direct-lane attached, the BMC says that the “Host is Online” and reports an event log entry that CPU_TEMP has a critical temperature of 0 degrees Celcius ( ofcourse it doesn’t).
Good suggestion. I’ll have a look. I had read that other Intel NICs had this problem and you can change the boot behaviour on the NIC firmware/NVM - I did this but it made no difference. My old Intel CT/PT dual Gbe NIC works in all slots with no problem, so I’m not sure what the issue is. I wonder if this is a Intel PCIe device problem?
I think that is simply that the host crashed and it could not pull any data from it. I have also seen this before.
@TOMO1982 Did you try any of this: Disabling VT-d , Enabling CSM + Legacy Boot ?
Nevermind, I found your earlier comment.
I have a similar problem with a w580 motherboard and e810, xxv710 cards. I was trying to use a Xeon W 1300 and the system freezes promptly after the splash screen. I just installed an older 10900k and the problem went away… For the last 2-3generations it seems like intel consumer oriented cpus are not playing nice with w series chipset boards.
In reply to my PCIe card fails to boot ticket with Gigabyte eSupport they gave me a link to a new F11 BIOS download (which is not yet published on the website):
https://esupport.gigabyte.com/FileUpload/Answer/2023/3/1442359/MW34-SP0_F11.zip
Not sure if that link works for everyone who has an account on eSupport.
Release Notes Say:
Issues fixed in this version
F11
- Update RC code (RC0C.00.75.10(3385.00)), Microcode (321) and ME (16.1.25.1932)
for 2023.1 security update.
(CVE-2022-30704)
I will test some time today hopefully and will report back.
Any further requests for RMA services please use our esupport portal, and your local RMA hub will assist with the process. So here forward, any discussion about MW34 service support should follow our esupport process, and don’t reach out to me, as it has been made clear to me to not engage in RMA matters, as it isn’t one of my job duties in marketing, But why did I post? To show support for L1, but in doing so I have received a lot of internal flack.
I also want to add…Wendell is a valuable contributor and educator in the enterprise community. This you must know if you are here. But did you know…the products we ship him for review, they are shipped back to us once he is done, and no money is exchanged? He’s not a paid reviewer, he’s just someone passionate about technology. So, I hope you all will get an amicable resolution. Unfortunately, defective units is a part of manufacturing, and our % of defective units is very good compared to other companies.
Good News!
F11 gets the motherboard to boot nicely with my Intel E810 25G NIC!
Woohooo!
I did not test with my other Intel 40G (XL710 I think) NIC, but for others with NICs and Dell U.2 drives not booting, F11 could be worth a try.
@LiamGigaComputing I am sorry to hear about this internal flack. You and @wendell have given us way more support than what Gigabyte eSupport has given us all this time.
Whenever I contact Gigabyte eSupport I do not get any real technical support.
I was even fighting for months with Gigabyte eSupport denying that they promised us Raptorlake support.
I only get copy-pasted messages to RMA the board to my dealer; my dealer does not want to do an RMA anymore, saying the board is fine and to contact Gigabyte instead.
Obviously this Gigabyte Rev1.0 board is not fine and has flaws; I hope the Rev1.1 fixes those.
About @wendell being a valuable contributor; his great reviews are the whole reason I bought this board in the first place. I wrongly assumed it was a paid review and that’s my mistake, I admit. But as a reviewer and valuable contributor to the community it is worthwhile also to point the flaws to the reviewed product. This Rev 1.0 board has flaws and if the community did not voice its criticism in this thread, we would not be getting any Rev1.1 replacements from Gigabyte eSupport; the eSupport and dealers were simply denying that there is anything wrong with the board.
Thank you @TOMO1982 . I will give F11 a try. I wrote two lengthy reports about the dell drive issue to Gigabyte eSupport last week explaining the problem.
What I got back was a copy-pasted response to RMA my Rev1.0 board again to my dealer! The same board that I have RMA’ed to my dealer two times already and who does not want to RMA it anymore!
I’ve already contacted esupport twice, I’ve sent the board to the RMA service in Netherlands (OGS Europe BV), they sent me back the board reworked to rev 1.1, but the BMC still not start after a power loss.
So what should I do to have a board that works as expected ?
My board has also huge issues when restarting with or without prior power loss.
I try to never restart it as much as possible.
After reboot, at best I get a very delayed boot (couple of minutes). On other occasions I get a responsive bmc but with error messages in the event log saying the host has no ram installed. At worst I get an unresponsive bmc (and no host boot).
I had a strange experience with the “will not power up after power reconnected” problem. I updated the board to F11 and powered off and removed it from the rack to put in my E810 NIC and I had to bridge the clear CMOS pins for it to boot. Then after seeing that it was booting with the NIC I powered it down and removed the power and put it back in the rack and reconnected it. It powered straight up with no problems. I think it was only probably 2-3mins without power. So maybe F11 solved this aswell?
I see that F11 is still not published - if someone is desperate I will try to upload it here. Please let me know if you want me to do that.
Thanks for getting involved in this thread and trying to help people get this board running up to its potential. Can you please determine if the new production run boards will be clearly marked and or serialized so that I can buy one of the new boards with confidence? I’ve been dragging my feet on spending $500 for a board because of this thread and the poor availability…and despite the fact that I wanted this board a year ago, I still could use one!