The new RX6600 (tested)

[Dojo76]

I suppose it’s your setup while in my case it’s the mainboard to blame because I’m super stable with 6 cards connected to the same 6 ports bifurcation cards (11 GPUs total), same chipset as yours, my problem is that if I connect 12 GPUs it boots directly into UEFI BIOS like there’s no boot device connected (but there is, I tried M.2 card and USB drive as well).

I want to try your ideal configuration (4+4) because Intel mainbaord chipset (H510), according to Intel, supports up to 16 PCI-E lines and my CPU supports 16 as well (Pentium Gold). So why the hell I’m stuck to 11? No idea.

I also have an integrated USB riser port, I think I can bifurcate it as well, not sure if that would work:

what do you think? Basically I want to go from mainboard USB to 4 PCI-E card (no PCI-E adapter)

[Batfink]

@Dojo76. Motherboard USB ports (or any other USB port) use USB signalling/protocol. PCIe risers (over USB connectors and cables) use PCIe signalling/protocol. The two are not compatible. So unfortunately plugging a PCIe riser or Splitter into a motherboard USB port won’t work.

It does sound like you may be out of PCIe lanes with 12x GPUs (it won’t boot properly). You can try using a USB3 Network Adapter (the Amazon Basics one works with HiveOS straight out of the box) in the USB3 port that comes directly from the CPU (this will be in your motherboard manual), you can then disable the on-board LAN port and see if it frees up a PCIe lane (it should). You can also do the same thing with a USB3-to-SATA adapter and run your SATA SSD/HD again off a CPU USB3 port. This may save you another PCIe lane because the onboard SATA controller won’t be used. I do not guarantee either of these will work but if you have both items lying around it might be worth a try.

Bifurcation cards split up a PCIe slot into seperate groups of PCIe lanes (or logical PCIe slots). So a 16x PCIe slot can be split into two 8x slots or four 4x slots (depending on your motherboard BIOS support for bifurcation). Likewise a 8x PCIe slot can be split into two logical 4x slots, etc.

Splitters / Packet Switches operate differently and I guess are just multiplexers at the end of the day.

BTW, a M.2 slot SSD will likely consume four PCIe lanes so stick to SATA SSDs (HiveOS does not need super fast disk access). Your motherboard manual should state how many lanes are used by the M.2 slot. You can use the M.2 slot with a M.2-to-PCIe 4x adapter and then stick a 4-port Splitter in it. Hope this helps…

[Dojo76]

I don’t want to bifurcate a normal USB port but a USB mining port, ASRock mainboard has one mining port integrated. Tomorrow I’ll try the board to extrend from 1 to 4 and I’ll let you now if it works.

I tried the AsMedia splitter in 1x and 16x port, no difference, even GEN1, GEN2, GEN3 are not a factor, I’m always stuck at 11. Tried to disable ethernet card and SATA controller as you suggested, still no joy but the idea was very good, makes sense. So basically it seems it’s not a problem of not available PCI-E lines but a boot issue with more than 11 GPUs, very difficult to solve.

Are you sure you’re not stable because of an under performing power supply? It’s not easy to balance all risers power.

[niexasadas]

man, dont leave cards on auto, boost fast at least 50%

[sousajgg]

I just started dual mining. I wonder if my OC is pushing my gpu to much.
What do you think guys?
team red miner 0.9.2.1

[niexasadas]

Does anyone have Sapphire GPRO X060? hiveos recognizes as rx6600 and wouldnt let to overclock memory more than 950, while card should be able to reach at least 1000.

[DEJANC]

Did you find any solution? I have same Micron and same problems.

[niexasadas]

1150 MVDD works for me

[babycicak]

More speed with teamblackminer …
Hope will be stable!! uptime 4 hour now.

[Pinfy]

Hello,
I used to have the same problem as you for about 3 or 4 weeks with two GPU - Radeon Sapphire 6600 with Micron memory. i tried every possible oc setting with no result. Finally i discovered that the problem has gone after i changed the riser. The miner uptime is about 2 days already with no problem. I hope will continue in this manner.

At the same time, before changing the riser, I replaced the PSU and powered the riser through the 6-pin rear plug, not the one next to the molex and the problem occurred less often. I hope the information will help you and fix the problem.

Ill wait for your feedback.

[Denrau]

Since I got like 20 RX6600 in 3 different rigs I’d like to share my opinion:

1. All issues coming fromOC, CPU + motherboard combo. You should always start from OC and only after that you need to go to the hardware side. The same cards act differently when you replace mobo+CPU. Radeon get’s picky and newer cards only getting worst.

2. You will see one card that’s acting up like a fan won’t start when others already run, runs hotter but it’s not in the middle of the rig, etc. Look for something that’s odd. Take that cards and see if it helps. If yeas you know what’s the problem. Recently I got a new Gigabyte Eagle 6600 - 27Mh max no matter how hard you try. Had to send it back. Asus and MSI Mech can be hit or miss. ASRock and XFX are running strong!

3. I still have that rig with 12 USB/PCIe slots (Chinese bord). I’ve replaced everything (meaning every wire, SSD, raiser, power supply (x2), LAN cable, etc, etc, etc. Now it’s only 9 GPUs down from 11. Two weeks of different overclock individually for each card every single day… Still gets “frozen” every 24h. Now the grand finale: I’m replacing mobo You see where I’m going - don’t buy cheap boards haha. Should I bring my hammer and create a viral video?!

Anyways if you’re like me with a lot of RX6600 please share your experience. I’m regretting buying some cards and I wish I would read about that the same way you’re doing it right now. On the mission to find the block before the end of March
Wish me lower luck!

[SerialG]

Good evening to all, I bought 2 RX6600 a few days ago and to my great surprise, my memory is Micron GDDR6 and not Hynix GDDR6 like my 2 previous ones and I can’t find any overclocking that is stable, if someone manages to tame these cards, please come and help me

[SerialG]

My last overclocks on the two RX6600 micron GDDR6 were : core 901 ; VDD 630 ; VDDCI 650 ; MVDD 1150 ; MEM 950
and still before this one :
core 930; VDD 630; VDDCI 650; MVDD 1130; MEM 950
andddd … GPU 0: detected DEAD (03:00.0), will execute restart script watchdog.sh

[recklezzz06]

What miner are you using?

[SerialG]

Teamredminer , i make screen with teamredminer logs

[recklezzz06]

Lol oops. Try core 900, core voltage 620, mem controller 630, mem 950, mem voltage 1200, PL 0, SoC 534, SoC VDD 850. It worked in Teamredminer for me before I switched to gminer, but wattage will be around 45w-48w in hiveos

[SerialG]

Thank you so much I will test this all night I cross my fingers that it does not crash …

[SerialG]

Hello, finally with the oc that you provided me I did not have a crash, only negative point is the power consumption that I find really high for this type of graphics card …

[Nebulaflood]

Hi, I see your hard work
And I wonuder if you tried any other OS like raveos or minerstat?

[Denrau]

Thanks, I did try Windows when I had only a few GPUs. Now when I got 35 GPUs I don’t wish to go “somewhere” and try “something”. If you have any positive experiences let me know. I’m open to trying it out.