Gigabyte R180-F34 1U Server (2011-3) $94-109 + Shipping

ViciousXUSMC

Active Member
Nov 27, 2016
200
96
28
37
I don't have a SM SQ series to compare it to. I just received cpu/memory to power mine on yesterday and after the BIOS and BMC were updated, I was able to use the PWM offset setting under utilities in the BMC to slow the fans. Currently most fans are running either 2k or 3k rpms.

The BMC spins up the fans on CPU demand fairly quickly, but at idle it is a bit louder than a desktop machine. Much easier to deal with than the 2U quanta machine that I had to do a fan swap on.

On another note, one of my DDR4 memory sticks were faulty and caused the server not to post and stopped at B7 on the chipset initialization screen with no other error message. Once I had that under control and updated the BIOS the server actually specifies the stick and location that failed.
IMPI control of the BMC is exactly the reason I love and have held onto my Dell R210ii because with the fans down its just as quiet or more quiet than my desktop. I do voice over work and my server rack is only feet away and you cant even hear it. I am curious the process and commands you used to accomplish this and where you found them at?

Knowing that this can be done across more than one brand may open up a lot of options for me, as the Dell hardware is getting dated, and updating to newer hardware might have other benefits.
 

curley

New Member
May 3, 2020
4
3
3
Has anyone updated the BIOS/BMC etc on these v1.0 boards?
I have 2 of these systems and did update the BIOS and BMC. Also have the v1.0 boards as well.

Just checked one of them and I appear to have used the 4.88 BMC firmware version and R16 BIOS when I updated mine.
 

Zalouma

New Member
Aug 5, 2020
3
1
3
Hey Guys

I am new to this forum but though to share some of my experience using this server to help other folks here as I bought from this seller alone 23 of these, Also it was me who reported the Supermicro Rails CSE-PT8L to the seller thinking of helping other buyers like you as the OEM is impossible to find, the supermicro is a tight fit and you will have to remove the small purple plastic stopper on the server inner rails for it to fit well, of course you will loose the stopper option at this point, but it gets the job done and of course way cheaper than $90 rare to find OEM, another alternative is to remove the inner rail and get any of the universal rails for HP G6 universal one, super cheap around $20-25 on ebay and works well too!

Now If you update from the penguin BIOS to the one listed for v1.1 which is R16, you will get E5-v4 to work even on revision 1.0 (tested) however server wont be rock solid stable on all OS, will crash from time to time, you will have to tweak alot in BIOS to get to stable it (This applies to v3/v4 regardless for BIOS R16 you will still need to tweak BIOS to avoid random crashes)

On the other hand the penguin BIOS (some machines would have R09 or R11) is way far more stable, my advise if you dont need to upgrade it dont specially if you will stay with E5-v3, the MOBO itself support v3/v4 only need newest BIOS to support v4 if you check here even for MOBO revision v1.0 E5-v4 is listed
For the SATADOM concern someone listed here, another work around, I am using a PCI that takes NVMe and M.2, and just using this sata port (there is 2 sata port "red one" next to PCI slot, Port 4/Port 5) and it normally get the power from PCI, something like this one will work Amazon.com: Dual M.2 PCIE Adapter for SATA or PCIE NVMe SSD with Advanced Heat Sink Solution,M.2 SSD NVME (m Key) and SATA (b Key) 22110 2280 2260 2242 2230to PCI-e 3.0 x 4 Host Controller Expansion Card: Computers & Accessories or even skip this all together and add something like OWC ssd PCI and the board itself provide power through PCI to the extra SATA Drive, I have seen some of these by Starcom that can carry 2 M.2 and 1 NVMe in 1 single card PCIe M.2 SSD (NGFF) Adapter Card | SATA Cards | StarTech.com, so if you combined the 4 caddies + the 3 starcom and 1 OWC you are ending up with 6 SSD and 2 NVMe working drives, there is also ASUS and Supermicro version of these that support RAID you will just need to do some search for it, I can help with model number as i own the supermicro ones if anyone need it

Lastly the BMC latest version on Gigabyte website version 488 if you upgrade it through dos or shell command will brick your server, they have an issue with this release SOCFLASH command that will flash the 128MB version instead of the 256MB to the ASPEED and ending bricking it, if you want to upgrade to it just use the mergepoint through browser and use the 488.bin file and you will be all set, I am sure you can take a backup of the penguin rom and bios prior upgrading but i didnt try honestly and also didnt try to downgrade so i cant confirm it.. I have 13 of these upgraded and 10 i never did, the 10 runs flawless while the 13 i had to tweak alot with several random crashes to get it to work well

I hope this help some folks out there, cheers
 
Last edited:
  • Like
Reactions: ecosse

BeTeP

Well-Known Member
Mar 23, 2019
509
315
63
Oh, someone noticed
Kris, you can keep blaming Gigabyte for not making the difference between 2 images not clear enough but ultimately it was you who decided to flash the wrong image.

Lastly the BMC latest version on Gigabyte website version 488 if you upgrade it through dos or shell command will brick your server, they have an issue with this release SOCFLASH command that will flash the 128MB version instead of the 256MB to the ASPEED and ending bricking it, if you want to upgrade to it just use the mergepoint through browser and use the 488.bin file and you will be all set
The latest version of the flashing script does not do any guessing - it requires user to select the memory size. The socflash binary just flashes the specified image. If user decides to flash the wrong image it would not matter if they used CLI or web UI - the end result would be the same.

Also it's the other way around - it's flashing 256Mb image instead of (proper for this system) 128Mb one what results in bricking.
 
Last edited:

Zalouma

New Member
Aug 5, 2020
3
1
3
Kris, you can keep blaming Gigabyte for not making the difference between 2 images not clear enough but ultimately it was you who decided to flash the wrong image.



The latest version of the flashing script does not do any guessing - it requires user to select the memory size. The socflash binary just flashes the specified image. If user decides to flash the wrong image it would not matter if they used CLI or web UI - the end result would be the same.

Also it's the other way around - it's flashing 256Mb image instead of (proper for this system) 128Mb one what results in bricking.
What you are saying is correct my friend but if you open the script file, both point to 128MB file regardless which version you choose, I reported it to Gigabyte even and i did brick one of my servers with it, had to jtag to fix it and its not 100% functional like before
 

Zalouma

New Member
Aug 5, 2020
3
1
3
I am looking at scripts in https://download.gigabyte.com/FileList/Firmware/server_firmware_ast2400.zip (which I downloaded when this thread just started) and I do not see that. Can you point out the script file you used?
Normally it should be under \server_firmware_ast2400\888_488\utility\fwud\dos if you will run the dos one, on this version it ask you which file you want to flash instead of it running the 1% test to determine which one to auto flash, this is where it brick the whole thing because 50% chance you select the wrong memory size

On the other hand on 487 this doesnt exist where you just run flashall and it run the 1% test in the script determine which correct flash size is your chip and flash the right .bin file to it, i believe this file is being use on several MOBO they offer so that has the AST2400 chip this is why the flashall script has this 1% test that normally should work, user shouldnt have to guess it

I reported it to them when i reached out to unbrick this one server and they are aware of this issue, however i find it Gigabyte careless to update this file on their site knowing this will brick it

Also to add, if you try to flash the 888.bin from mergepoint for example it will prompt you that the file mismatch and will stop the whole thing and only 488.bin will work which is the one that is correct, so it works on mergepoint if you try to flash the wrong one it wont take it, I invite you to try to flash 888.bin for example from the BMC UI wont take it with a mismatch error without you bricking it
 
Last edited:

BeTeP

Well-Known Member
Mar 23, 2019
509
315
63
The flashing scripts before vX.88 were trying to determine the appropriate image by using IPMI to query the current config. But the procedure was not reliable enough so they reverted to letting the users specify the memory size. In your previous posts you made it sound like the scripts would flash the wrong image even if the user selected the correct size. This is not the case.

My main point is that "having to choose out 2 options" does not mean that "there is a 50% to brick the board during flashing".
 

BeTeP

Well-Known Member
Mar 23, 2019
509
315
63
Luluzluzluzulzlz.
I successfully flashed my system on the first try without having to guess anything and you keep laughing.

During my career I had to fire a few people like you - very smart but pathologically unable to own their mistakes.
 

kjboughton

New Member
Dec 7, 2017
27
7
3
41
I successfully flashed my system on the first try without having to guess anything and you keep laughing.

During my career I had to fire a few people like you - very smart but pathologically unable to own their mistakes.
And you see that everybody? He did it all by himself.
 

ecosse

Active Member
Jul 2, 2013
374
64
28
Dumb q - is this a proprietary board or is it locked to the chassis? Just wondering whether its worth the hassle shipping to the UK via a redirector service.