Hi, do you think I could also try it out on my R9 380 with Elpida memory after I convert it around?

Do you think I could use the same timings for Hynix AJR or could you suggest one for AJR?

My current R9 380 hynix 1500 strap looks like this: 7771332000000000CE516A3D9055111230964909004AE60022339D08740114206A8900A00200312 0150F292F94273116.

Many thanks!

Hello ! I have some problems with one rig who have 6 x AMD R9 380 (Sapphire, Asus Strix and MSI) GPUs on ETH . I saw people who get 23-25 MHs with this R9 380, but i can get only 20 MHs.

I try to mod bioses for all GPUs with hex editor, few versions of moded bios ... ( strings from 1500 Mhz to put up and strings from 1425 Mhz to put up ), but no more hashrate !

Hope to somene can help me ! Thanks in advance !

OS : ethOS
Miner : sgminer
Oc settings : Core - 1130 Mem - 1660 , Core - 1050 Mem - 1580 , and others more OC settings !

Edit :

Sapphire Nitro : Strings from 1500 Mhz to 1625 and 1750

(1500 Mhz) F0 49 02 02 77 71 33 20 00 00 00 00 CE 51 6A 3D 90 55 11 12 30 96 49 09 00 4A E6 00 22 33 9D 08 74 01 14 20 6A 89 00 A0 02 00 31 20 15 0F 29 2F 94 27 31 16

I had found a new better memory strap for Elpida RAM. More than 1kh/s cryptonight xmr !

RX 470 : 1020 h/s  , sgminer @1250 core, @2100 mem, 66W (gpu-z)

                   942 h/s , @1150 core, @2050 mem, 52W (gpu-z)

777000000000000022AA1C00AC615B3CA0550F142C8C1506006004007C041420CA8980A9020004C 01712262B612B3715

I don´t know the eth hash speed.   Maybe somebody can check it ?

I guess if the tool can suggest optimal values not only in timing straps but voltages as well, card dependent.
Also, is it possible to suggest a memory clock value based on ASIC quality of a card e.g. a card having higher asic quality can withstand higher clocks as well without tons of memory errors?

I hope I’m making sense here and apologies if not.

Perhaps it's a bit pedantic, but the training is to adjust the electrical interface for the gddr5 8-bit data bus.  It finds the optimal drive & termination characteristics in order to avoid errors, but it doesn't adjust timing.  So for example when your card heats up from 20C to 80C, the resistance of the traces increase, which means more attenuation of the electrical signals.  That may lead to more CRC errors, and when they reach a threshold the memory controller will initiate a retrain.  The training sequence itself will then read and write fixed data patterns on the bus, trying things like increasing the output driver strength to overcome attenuation issues.  Increased driver strength may lead to data reflections causing more errors, which could be corrected with lowering the termination resistance on the receiving end.

Another tweak I've wondered about is monitoring memory die temps.  GDDR5 chips have internal thermal diodes similar to the GPU, and they can be enabled by setting a bit in mode register 7.  Knowing the temperature of each memory die could provide useful diagnostic info for problems such as a heatsink not having good contact with the chip.

Anyone deciphered yet whats in those three seq_misc parts of strap?

I was just curious, as if thats the case, those would hold for example WRITE latency
and CAS latency, but it seems people never modify them, just use default value
(at least that seems to be the case for all the straps I saw circulating around)

This strap is bad. Both of my cards bsod hard with it. Also its slower and runs with more power. Trfx and ras2ras are too tight.

My bad, i didnt say that properly. The trfc and ras2ras for xryptonight is nearly useless, you lose about 10h/s when going to 150 for example. But the power cons drops by 5watts. Thats for cryptonight.

Also, all trcd values are not properly done, the A value should be the same as the non A one.

Good idea

And shame you have no merit, despite helping people a lot, so let me give you first one