Hi,
I have this board, where I have exchanged the LGA1151 socket. During testing I discovered, that the large PCIEX16 slot won't by no force work in x16 mode, but keeps to work in x8 mode only.
I tested this with GPU-Z, also with Render Test (but tried also Furmark and Heaven Benchmark - no effect).
Tested on Core i5-9600K as well as on an Pentium G5420 and MSI NVIDIA GTX1070 and 2 other cards, including RTX3060Ti.
Board architecture definitely allows x16 mode, as all 16 PCIe lanes are present between CPU and main GPU slot (nearest to CPU).
What I have done by now is:
1. Upgraded bios to latest F3d version - no change
2. Upgraded Management Engine to latest available version v11.8.96.4657 - no change
3. Tested if all of the PCIe lanes are physically conducting between socket and CPU, both by a PCIe tester, and measuring physically each connection by a meter in reverse diode mode and inserted CPU.
4. Tested DMI bus between CPU and PCH with a DMI bus tester - all OK
5. Examined the exchanged socket for any bent pins and cleaned it once again - no best pins, cleaning did not help
6. Tested GPU, CPU and the NVME SSD with test Windows installation on another board - All OK - all working in x16 mode
7. Tested whether any of the PCI lanes have a short between each other - no short
8. Tested all the 220nF caps on the PCIe lanes for shorts and nominal capacity - all OK
9. Tried another bios form a similar board to rule out bios bug (GIGABYTE Z370M D3H) - board boots w/o problems with this alien bios, but same effect, x8 only
10. Tried GPU and CPU on an MSI B365 board to rule out that the B365 chipset does not support x16 mode - it does work normally in x16 mode on the MSI board
11. Tried to disable SATA controller and all SATA ports to rule out that PCIe lanes are stolen from PCIEX16 slot for SATA ports - no effect (BTW, the architecture of the board does not allow this so or so, there is no CPU PCIe bus switch on the board which could do this).
12. Tried to boot Windows from USB stick, to rule out M.2 ports steel any PCIe lanes from CPU, tried both M.2 port for NVME SSD - no change
13. Tried to boot Windows from an M.2 PCIe extension card from all remaining PCIe slots, no effect
14. Looked on all the CPU control signals to exclude there is any signal that could restrict the CPU PCIe bus from working in x16 mode - did not find any signal which could have anything to do with this issue
15. Looked for any irregularities in the bios with AMIBCP related to PCIe and GPU port width control - didn't find anything relevant
16. Examined the board for any physilcal demages and parts missing - all seems OK
17. Tried to boot board with bios options "Other OS", "Windows 8/10", with CSM Support, w/o CSM Support - no change
18. Looked for any bifurcation settings in the bios, but seems there is nothing that can be altered here related to bifurcation - this bios seems to do here everything automatically (black box principle) - maybe this does not work correctly, as this board seems to have had very low attention by Gigabyte, there are only 3 bios versions available.
I'm out of ideas, any suggestions are appreciated.
Maybe this is just normal for this board, and I expect too much from such an quite old board ?
I have this board, where I have exchanged the LGA1151 socket. During testing I discovered, that the large PCIEX16 slot won't by no force work in x16 mode, but keeps to work in x8 mode only.
I tested this with GPU-Z, also with Render Test (but tried also Furmark and Heaven Benchmark - no effect).
Tested on Core i5-9600K as well as on an Pentium G5420 and MSI NVIDIA GTX1070 and 2 other cards, including RTX3060Ti.
Board architecture definitely allows x16 mode, as all 16 PCIe lanes are present between CPU and main GPU slot (nearest to CPU).
What I have done by now is:
1. Upgraded bios to latest F3d version - no change
2. Upgraded Management Engine to latest available version v11.8.96.4657 - no change
3. Tested if all of the PCIe lanes are physically conducting between socket and CPU, both by a PCIe tester, and measuring physically each connection by a meter in reverse diode mode and inserted CPU.
4. Tested DMI bus between CPU and PCH with a DMI bus tester - all OK
5. Examined the exchanged socket for any bent pins and cleaned it once again - no best pins, cleaning did not help
6. Tested GPU, CPU and the NVME SSD with test Windows installation on another board - All OK - all working in x16 mode
7. Tested whether any of the PCI lanes have a short between each other - no short
8. Tested all the 220nF caps on the PCIe lanes for shorts and nominal capacity - all OK
9. Tried another bios form a similar board to rule out bios bug (GIGABYTE Z370M D3H) - board boots w/o problems with this alien bios, but same effect, x8 only
10. Tried GPU and CPU on an MSI B365 board to rule out that the B365 chipset does not support x16 mode - it does work normally in x16 mode on the MSI board
11. Tried to disable SATA controller and all SATA ports to rule out that PCIe lanes are stolen from PCIEX16 slot for SATA ports - no effect (BTW, the architecture of the board does not allow this so or so, there is no CPU PCIe bus switch on the board which could do this).
12. Tried to boot Windows from USB stick, to rule out M.2 ports steel any PCIe lanes from CPU, tried both M.2 port for NVME SSD - no change
13. Tried to boot Windows from an M.2 PCIe extension card from all remaining PCIe slots, no effect
14. Looked on all the CPU control signals to exclude there is any signal that could restrict the CPU PCIe bus from working in x16 mode - did not find any signal which could have anything to do with this issue
15. Looked for any irregularities in the bios with AMIBCP related to PCIe and GPU port width control - didn't find anything relevant
16. Examined the board for any physilcal demages and parts missing - all seems OK
17. Tried to boot board with bios options "Other OS", "Windows 8/10", with CSM Support, w/o CSM Support - no change
18. Looked for any bifurcation settings in the bios, but seems there is nothing that can be altered here related to bifurcation - this bios seems to do here everything automatically (black box principle) - maybe this does not work correctly, as this board seems to have had very low attention by Gigabyte, there are only 3 bios versions available.
I'm out of ideas, any suggestions are appreciated.
Maybe this is just normal for this board, and I expect too much from such an quite old board ?