Vram addressing in double sided rtx 3090 pcb

[hallmark1730]

Hello, I know that normally the vram names are listed as in the photo. But in the RTX 3090, there are the same number of vrams on both sides. And the mats tells me that there is an error in the c1 and e1 vrams. Which side should I remove?

Mats log:

GPU 0 [01:00.0] dev.sub 0.0
----------------------------------------
DevInst : 0
PCI Location : 0x00, 0x01, 0x00, 0x00
GPU DID : 0x2204
PDI : 0x1cae29a3311eebb1
Raw ECID : 0x012030c00000005db04d09c4
Raw ECID (GHS) : 0x4705db04d0800000003018240
ECID : SAMSUNG-SNR1D2-03_x3_y9
Device Id : GA102
Revision : a1
Sub Revision : 1
NV Base : 0xfb000000
FB Base : 0xb0000000
IRQ : 11
WARNING: GFW boot did not complete. May be due to an invalid FS config
Boot status = 0x00000303
NV_PFB_FBPA_FALCON_MONITOR = 0xdeaddead
NV_PFB_FBPA_TRAINING_CMD = 0x24010000
NV_PFB_FBPA_0_TRAINING_STATUS = 0x00000000
NV_PFB_FBPA_1_TRAINING_STATUS = 0x00000000
NV_PFB_FBPA_2_TRAINING_STATUS = 0x00000008
NV_PFB_FBPA_3_TRAINING_STATUS = 0x00000000
NV_PFB_FBPA_4_TRAINING_STATUS = 0x00000008
NV_PFB_FBPA_5_TRAINING_STATUS = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(0) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(1) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(2) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(3) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(4) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(5) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(6) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(7) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(8) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(9) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(10) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(11) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(12) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(13) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(14) = 0x00000000
NV_PFBFALCON_FIRMWARE_MAILBOX(15) = 0x00000000
Error 000000000505 : Gpu.Initialize Memory address training failed [2.187 seconds]
Error 000000000505 : Global.PrintGpuInitError Memory address training failed [0.000 seconds]
Error 000000000505 : Global.InitializeGpuTests Memory address training failed [2.206 seconds]