If you installed lm-sensors, you might be able to use sensors to report your cpu voltage level. Don't know anything about Biostar computers or which SIO monitoring chip they use. If it is a common one sensors should be able to poll it for current motherboards sensor readings.

Found my notes on the lm-sensors. Tried the install and the sensor probe didn't find anything :(

---

A proud member of the OFA (Old Farts Association).

And apparently there is no reliable bios settings to get 6 gpus running. You just have to start with the default 6X miners mode and keep trying till you get all of them working.

Tom

So far I have been able to run 6 gpus with 6X Mining mode on. Didn't matter if I had the "Above 4G" toggled on or off. So it is now on. Parked under something like the IGF is a gen1, gen2, gen3 speed setting so its now on Gen3. I think Mining mode is disabling the CPB (cpu boost) on the bios menu.

I am running without -nobs and with less than a core for each gpu.
The task manager shows many cpu cores running at 6% of the 100% load. It also shows 5 cpus running at 5% while the 5 gpus are showing 1% and cycling on and off. Can't really see much processing time difference and this allows me to run more CPU threads.

Tom

---

A proud member of the OFA (Old Farts Association).

If you installed lm-sensors, you might be able to use sensors to report your cpu voltage level. Don't know anything about Biostar computers or which SIO monitoring chip they use. If it is a common one sensors should be able to poll it for current motherboards sensor readings.

Found my notes on the lm-sensors. Tried the install and the sensor probe didn't find anything :(

It found something, it just couldn't automatically do anything about it. Sensors always finds the motherboard sensor chip or you wouldn't be able to view sensors in the BIOS. It has to be physically present. If you look closer at the sensors output it will have likely printed out a value labelled biostar-isa-000 or some other value such as:
Probing for Super-I/O at 0x2e/0x2f
...
Trying family `ITE'... Yes
Found unknown chip with ID 0x8665
(logical device 4 has address 0x290, could be sensors)
The isa-xxx value is the address the SIO sensor monitoring chip is located. In the printout of the sensors-detect it will have found the SIO chip and labelled it as "unknown" or even the manufacturer but was unable to automatically install the matching driver for the chip. Only the oldest most common SIO chips have drivers preloaded in the modules directory. If the driver is not there, then sensors fails with no detected sensor chip message.

For example on this ASUS Crosshair VII Hero motherboard the SIO chip is an ITE8665E chip which is a new chip and doesn't have any documents available to the public for developers to write a driver for. It is identified in the sensors output as asus-isa-0290 and exists at ISA address 0x290. You need to see what address your SIO chip is using and then Google which SIO chips use that address. When I did that I discovered a developer that had written a driver for the ASUS ITE 8665 chips used on most of ASUS' boards named it87. All I had to do was download the driver source code make and install it with a modprobe command. Then I had access to some of the motherboard sensors. Later another developer created an even better driver that interfaces with the WMI protocol of ASUS motherboards that have a WMI enabled BIOS. That driver asus-wmi-sensors has all the same sensor outputs that are available in the BIOS and matches all the outputs seen in Windows.

So first identify the SIO chip that BIOSTAR uses on your motherboard and see if there is a driver available for it. For example on my system in the /lib/modules/4.15.0-47-generic/kernel/drivers/hwmon directory I have 151 preloaded drivers that handle SIO chips from
ADC, ADT, ABIT, ASPEED, DELL, LTC, LM, Maxxon,NCT, SuperMicro, VIA and Winbond for example. Look in your motherboard manual or online docs and see if it lists the SIO chip brand for your board. Then see if a driver exists or it and identify what address it is located at. If you have the driver already you can modprobe it in. You might have to --force it in at the detected address. You can use the isadump utility to probe for chips

sudo isadump -k 0x87,0x01,0x55,0x55 0x2e 0x2f 7
sudo isadump 0x295 0x296

which will help pin down the address where the chip is located.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

So if you purchased the BIOSTAR TB350-BTC motherboard in your previous post, I looked at the motherboard image for the product on Biostar's product page. You can clearly see the ITE Super I/O monitoring chip in the middle bottom of the board. So I would bet you can use the it87 driver for the chip.
This is the location of the driver.

https://github.com/a1wong/it87

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

So if you purchased the BIOSTAR TB350-BTC motherboard in your previous post, I looked at the motherboard image for the product on Biostar's product page. You can clearly see the ITE Super I/O monitoring chip in the middle bottom of the board. So I would bet you can use the it87 driver for the chip.
This is the location of the driver.

https://github.com/a1wong/it87

Thank you.

---

A proud member of the OFA (Old Farts Association).

It's now north of 100,000 RAC. The CPU is running at about 3.7GHz. I think I probably have a stable setup unless I wan to start a GPU upgrade cycle.

Tom

---

A proud member of the OFA (Old Farts Association).

It's now north of 100,000 RAC. The CPU is running at about 3.7GHz. I think I probably have a stable setup unless I wan to start a GPU upgrade cycle.

Tom

It is running reliably. It is acting like a 24/7 motherboard. So I am going to formally recommend that anyone who wants to run more than 4 gpus on an AM4 socket motherboard should consider the Biostar TB350-BTC. Its cheap. It runs any AM4 socket CPU. It runs up to 6 gpus (maybe 7). And it is reliable and runs at a pretty high overclock.

Tom

---

A proud member of the OFA (Old Farts Association).

Spent a couple of days transplanting the guts of the AMD box which uses a Biostar TB350-BTC motherboard into a 6 gpu mining frame that was recommended to me as a "easy assemble" and it sure was. Probably the main reason is a had a "step by step" guide which the last mining frame I bought didn't.

So I am now running 6 gpus again. However I had a series of USB 3.0 cables apparently die on me. It took me a couple of days to be willing to bite the bullet and try the USB 2.0 cables I already had.

So when you are using riser cards for your gpus and driving them with USB 3.0 cables, how LONG have your cables been lasting?

I am probably going to order a dozen more Monday.

Tom

---

A proud member of the OFA (Old Farts Association).

I have been slowly increasing the cpu multiplier of my Ryzen 7 2700. In theory it should be able to turbo to near 4 Ghz.

It crashed at 3.5GHz with the "auto" setting. It had cpu voltage "offsets" on Auto.

I now have cpu voltage offset it up to the highest "yellow zone".

What CPU voltage are you using on your 2700/2700x and what kind of CPU multiplier?

Thank you.

Tom

---

A proud member of the OFA (Old Farts Association).

try 1.35V cpu core fixed and 4.0 GHz on all cores manually. Should be fine, but needs to be tested. If its stable, lower the cpu voltage.
Hopefully you have some good cooling on the CPU though.

---

I have been slowly increasing the cpu multiplier of my Ryzen 7 2700. In theory it should be able to turbo to near 4 Ghz.

It crashed at 3.5GHz with the "auto" setting. It had cpu voltage "offsets" on Auto.

I now have cpu voltage offset it up to the highest "yellow zone".

What CPU voltage are you using on your 2700/2700x and what kind of CPU multiplier?

Thank you.

Tom

I just let the board and cpu do their thing on Auto. I set a manual multiplier of either 40 or 40.25. With 40 multiplier the steady state load Vcore is 1.29V. With 40.25 multiplier the steady state Vcore is 1.33-1.34V.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

https://www.pcgamer.com/if-this-leak-is-accurate-a-16-core32-thread-ryzen-cpu-is-headed-to-desktops/

I don't know if I could afford a new one, but this leak, if accurate. would allow me to approach or equal the cpu processing power of my Intel box. That is assuming it will run full tilt (except for Pcie Gen 4) on my AMD Motherboard (Biostar TB350-BTC).

I am seeing reports that not all AM4 Motherboards will be able to run the Ryzen 3000 series AM4 cpus. Either they won't get a bios update, or it simply isn't possible on some of the lowest end/entry level chipsets.

I am "worried" that the 300 series chipsets (which presumably the Biostar MB is) won't really get 3000 support.

Tom

---

A proud member of the OFA (Old Farts Association).

I saw reports that specifically the B320 chipset would not be supported.

Biostar recently announced that it plans to launch the fourth generation of its Biostar Racing boards (presumably X570) to support AMD's new chips

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Would you believe 9 gpus on at TB350-BTC motherboard?

The bios mining mode has "disabled, 6x, and above 6x".

So if you toggle it to the above 6x and put 1 expander card in slot 3, counting from the cpu,

It recognized and booted with all 9. This is a mix of 1 gtx 1070, 7 gtx 1060 3GB and a "spare" gtx 750ti.

I am now seeing if it will "run". So if it runs 24 hours without a stop I am going to be hopeful.

Tom

---

A proud member of the OFA (Old Farts Association).

It started "losing" gpus so I replaced 8 out of 9 USB cables with USB 3 1'6" cables that are "speced" for hard drive data transfer. The brand is UGREEN from Amazon.

It hasn't lost a GPU since. (So far).

Tom

---

A proud member of the OFA (Old Farts Association).

Why keep doing the same thing over & over expecting different results????
You already know that Cheap x1 splitter won't work for long!
If you want to try something with any chance of success you're going to have to pay for it.
This was around $200 back when they listed the price, people who've tried it say it works. I haven't tried it.
I suspect it all depends on which board you connect it to.

https://amfeltec.com/products/flexible-x4-pci-express-4way-splitter/

To be fair, he is trying something new. New cables. Which I recommended he try several times now. In my experience, 99% of the problems are due to poorly made and improperly shielded USB cables. The splitter could be problematic as well. But unless he gets good cables he’ll have problems, even with the expensive splitter.

New cables is a good first step.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

Right.... cables....
Here, go over here and ask anyone there about that ~$35 x1 splitter, https://render.otoy.com/forum/
See what they tell you. I have one of those ~$35 splitters.... it works for a couple of hours, it you're lucky.

Right.... cables....
Here, go over here and ask anyone there about that ~$35 x1 splitter, https://render.otoy.com/forum/
See what they tell you. I have one of those ~$35 splitters.... it works for a couple of hours, it you're lucky.

I have had some pretty good USB 3.0 cables last several months.

But they did eventually quit giving me the classic "lost gpu" symptoms.

However, this was a "new" trial. And apparently running 1' 6" USB 3.0 cables in 8/9 slot positions got rid of the "lost gpu"symptoms.

I agree, I may have to upgrade to the expander you proposed if the new(er) cables don't have reasonable life spans of more than a couple of months.

Tom

---

A proud member of the OFA (Old Farts Association).

It started "losing" gpus so I replaced 8 out of 9 USB cables with USB 3 1'6" cables that are "speced" for hard drive data transfer. The brand is UGREEN from Amazon.

It hasn't lost a GPU since. (So far).

Tom

Thank you Ian&SteveC for the lead on the UGREEN cables.

Since I was apparently feeling ambitious I tried the "other moonshot" and borrowed a couple of gtx 1060 3GB's from my other box.

Tried for 11. Either it only would recognize 9 gpus. Or the gui wouldn't start (displaying the command line).

Since I have already spent plenty of time and failures trying to get past the gui won't start if I have too many gpus connected with previous experiments on other systems, I threw up my hands.

I am back "only" 9 gpus :)

Tom

---

A proud member of the OFA (Old Farts Association).