AMD Mem Tweak XL - Read/modify timings/pp/straps on the fly

[spamster]

I can't run the program (gui). Get an excetpion:
StartService failed: The service cannot be started,
either because it is disabled or because it has no enabled devices associated with it
at WinFormsGUI.IOMap.Loaddriver()
at WinFormsGUI.Program.Main()



What could be the issue?

Zeo, I have the same problem on one machine I am trying to diagnose.  I have 2 identical machines, it works on one and I get that error on the other.  I have DDU'ed and reinstalled drivers on both machines.  Both machines have enabled onboard Intel graphics, disabling has no effect.  I will post more as I discover.

[1624552789]

1624552789

[1624552789]

1624552789

[1624552789]

1624552789

[1624552789]

1624552789

[1624552789]

1624552789

[1624552789]

1624552789

[pbfarmer]

@Eliovp - a couple bugs/issues to report (latest version, linux, gddr5):

1. setting --RAS reports as having set --PA2RDATA.  Looks like maybe just a display issue tho, as --current then reports RAS as having been set properly

2. setting --RC is also applying the value to CL - and since setting CL doesn't work (prevents card from being initialized by miner, reboot required) effectively RC cannot be set

3. you might be missing some bits for RC (or it's just incorrectly mapped to CL)?  strap says 65, but it reads same values as CL, and if i set > 31, i get an overflow (e.g. setting 32 results in 0).  

EDIT: Btw - any hints why DRAM1/2 values change under load, even when the core/mem state haven't changed?

[lstrike]

Gain looks to be ~0.7-0.8Mh/s (ethash) per card, or about 2-2.5%

(Polaris)
What is the power increase for that?
The Claymore boost increases exponentially the power consumption for similar extra hash, which is not effective...

[boxraider]

Hey, guys

Any tips on how to get the tool to make the timing settings stick? What programs are you guys running incl. this tool? I've got afterburner, and amd tweak. It's weird, I have to set my core clock in after burner and then tweak my voltages, memory and one-click timings in amd tweaker. But, 80% of the time itll crash or not apply one-click timings/manual timing changes when im running the same order of operations (weird 20%ish times my one-click will stick and reflect on the timing chart). It seems that I might have a bad card? Im not sure how you guys are able to adjust your timings and having them show up on your charts.

Right now, Im just getting 31mhs rather than crashing my pc 40x and settling for 28mhs (no 1 click timing/manual timing change) because Ive given up for the day lol.

[Eliovp]

@Eliovp - a couple bugs/issues to report (latest version, linux, gddr5):

1. setting --RAS reports as having set --PA2RDATA.  Looks like maybe just a display issue tho, as --current then reports RAS as having been set properly

2. setting --RC is also applying the value to CL - and since setting CL doesn't work (prevents card from being initialized by miner, reboot required) effectively RC cannot be set

3. you might be missing some bits for RC (or it's just incorrectly mapped to CL)?  strap says 65, but it reads same values as CL, and if i set > 31, i get an overflow (e.g. setting 32 results in 0).  

EDIT: Btw - any hints why DRAM1/2 values change under load, even when the core/mem state haven't changed?

I'm going to check this tomorrow, did everything in a rush without even testing  

Should clean a bit up because it's becoming a mess.
Already saw some issues going over the code.

And Laik2 pointed out that some bits are wrong, he's right (PMG for example.. 3-1-3-1-4-6-3..)
(See that Laik2, i'm not too "Godlike" to acknowledge my mistakes)

I'll see to add some extra timings that i didn't yet expose, will expose HBM2 Memory Manufacturer as well.

Anyhow, when i wake up i'll take some more time to test and clean up code because after adding hbm1 it's a total mess.

Cheers!

[pbfarmer]

Gain looks to be ~0.7-0.8Mh/s (ethash) per card, or about 2-2.5%

(Polaris)
What is the power increase for that?
The Claymore boost increases exponentially the power consumption for similar extra hash, which is not effective...

Nothing - basically increasing REF should just be replacing refresh reads/writes with workload reads/writes, so you really shouldn't see any power use change, unless your RFC was so high that you have a bunch of NOOPs during the refresh period.

FYI, my power use is ~130w for 32.25Mh/s per card avg (which seems a bit high to me, given that i'm only at 1175 cclock, 2100-2200 mclock, and 837mv avg, tho it's a little warmer today, so I think my fans are pulling a bit more)

[pbfarmer]

Hey, guys

Any tips on how to get the tool to make the timing settings stick? What programs are you guys running incl. this tool? I've got afterburner, and amd tweak. It's weird, I have to set my core clock in after burner and then tweak my voltages, memory and one-click timings in amd tweaker. But, 80% of the time itll crash or not apply one-click timings/manual timing changes when im running the same order of operations (weird 20%ish times my one-click will stick and reflect on the timing chart). It seems that I might have a bad card? Im not sure how you guys are able to adjust your timings and having them show up on your charts.

Right now, Im just getting 31mhs rather than crashing my pc 40x and settling for 28mhs (no 1 click timing/manual timing change) because Ive given up for the day lol.

I load PPTs to set clocks, lock to the power states I use while mining (i use cclock 6 and mclock 2 for ETH on polaris), then apply timings w/ amdmemtweak (cli).  Changing to a different state (core or mem) at any point seems to lose the applied timings.

I'm on linux, but I would expect things to be basically the same in Windows...

[seefatlow]

vega samsung memory : ETH-CNR-BEAM

WinAMDTweak.exe --RAS 28 --RCDRD 12 --RCDWR 5 --RC 44 --RP 12 --RRDS 3 --RRDL 3 --REF 15600 --RFC 246 --FAW 14

Hi livada, thanks for all your posts and suggestions, I’m now mining 20-25% more hash at the pool because of you (and of course Eliovp) :-)

But when you run settings tests, you need to run them longer. 1 accepted share does not make a for a very good basis for overall long-term results. Valid results would be based on hours of mining, sometimes GPU crashes with new settings after a while, sometimes many hw errors occur over time, these may not present themselves until many hours of mining.

Cheers!

I test this  1-2 day. This is only SS for send..
With this amd mem tweak setup my 3*vega56 powercolor and 1*vega64LE (all samsung mem) work fine with many algo- I test  CRN-GRAFT- TRTL- MOX-WOW-MSR-LITEv7. any algo have better hr if i change mem tweak setup but with this  setings all algo work fine.

I have a Win 10 - Vega 56 rig

One Hynix card causes a dead GPU when trying 19.4.2, latest TR and have tried both of the suggested Hynix setttings.  Anyone have one that is working for them?   I've tried:

WinAMDTweak.exe --gpu 03 --RAS 24 --RCDRD 19 --RCDWR 4 --RC 35 --RP 13 --RRDS 4 --RRDL 5 --RFC 148 --REF 15600

and

WinAMDTweak.exe --gpu 03 --RAS 22 --RCDRD 17 --RCDWR 4 --RC 35 --RP 13 --RRDS 4 --RRDL 4 --RFC 148 --REF 15600

Both of these cause a dead gpu/restart

try this-

--rcdrd 19 --rcdwr 4 --rc 35 --rp 14 --rrds 4 --rrdl 5 --rfc 148 --REF 15600
or
--rp 14 --rc 42 --rfc 156 --rrd 3 --rcdrd 18 --rcdwr 3

not best but work

iff any have good setup for vega56/stockbios/hynix pls send ..

I got vega 56 strix hynix and i cant go more than 1800hs.. What is ur best results guys? Any good setup for this bad cards?

my 6*veg56 nitro+(hynix) work fine with this setings (+2000hr)- 1 card give 1 error every 1 hour but work fine. itest this with more algo.

WinAMDTweak.exe --rp 14 --rc 35 --rfc 178 --rrds 4 --rrdl 5 --rcdrd 19 --rcdwr 5 --REF 15600

if not work try add --RAS 44(gives stability and lower HR) or chg --rfc 178 to --rfc 188(lower HR)

best card  have gpu= 1440mhz/866mv mem=935mhz/868mv
bad card have   gpu= 1400mhz/870mv mem=895mhz/875mv and give error every 1hour
http://imgbox.com/4Ast1m6p

Will be interesting to know if --REF 15600 still works for you in the summer months.

[Eliovp]

@Eliovp - a couple bugs/issues to report (latest version, linux, gddr5):

1. setting --RAS reports as having set --PA2RDATA.  Looks like maybe just a display issue tho, as --current then reports RAS as having been set properly

2. setting --RC is also applying the value to CL - and since setting CL doesn't work (prevents card from being initialized by miner, reboot required) effectively RC cannot be set

3. you might be missing some bits for RC (or it's just incorrectly mapped to CL)?  strap says 65, but it reads same values as CL, and if i set > 31, i get an overflow (e.g. setting 32 results in 0).  

EDIT: Btw - any hints why DRAM1/2 values change under load, even when the core/mem state haven't changed?

I'm going to check this tomorrow, did everything in a rush without even testing  

Should clean a bit up because it's becoming a mess.
Already saw some issues going over the code.

And Laik2 pointed out that some bits are wrong, he's right (PMG for example.. 3-1-3-1-4-6-3..)
(See that Laik2, i'm not too "Godlike" to acknowledge my mistakes)

I'll see to add some extra timings that i didn't yet expose, will expose HBM2 Memory Manufacturer as well.

Anyhow, when i wake up i'll take some more time to test and clean up code because after adding hbm1 it's a total mess.

Cheers!

Fixed the bitfield bugs, cleaned a bit up, added a few things.

More details

Only done so on linux version, will update the rest asap as well.

Direct link to linux cli binary: here

Greetings

[rednoW]

Didn't you miss my post? Loading fan speeds for Polaris cards from saved profile seems to be broken in beta 5.

[boxraider]

Hey, guys

Any tips on how to get the tool to make the timing settings stick? What programs are you guys running incl. this tool? I've got afterburner, and amd tweak. It's weird, I have to set my core clock in after burner and then tweak my voltages, memory and one-click timings in amd tweaker. But, 80% of the time itll crash or not apply one-click timings/manual timing changes when im running the same order of operations (weird 20%ish times my one-click will stick and reflect on the timing chart). It seems that I might have a bad card? Im not sure how you guys are able to adjust your timings and having them show up on your charts.

Right now, Im just getting 31mhs rather than crashing my pc 40x and settling for 28mhs (no 1 click timing/manual timing change) because Ive given up for the day lol.

I load PPTs to set clocks, lock to the power states I use while mining (i use cclock 6 and mclock 2 for ETH on polaris), then apply timings w/ amdmemtweak (cli).  Changing to a different state (core or mem) at any point seems to lose the applied timings.

I'm on linux, but I would expect things to be basically the same in Windows...

yeah its weird man, ill restart my pc and fire up amd tweak and phoenixminer, i left amd tweak run and then start mining. I have to set my timings first and then muck with my core/clock or else it wont apply haha. i'll set one click timing and 95% of the time itll revert back to default. when it does apply, ill just back away and try not to crash the pc.  it takes a lot of time for me to get one-click timings to apply let alone custom timings!haha im just curious as why the reason why just me or are others getting this same issue? hopefully my vega56 coming will be diferent! btw thanks Eliovp. youre a saint.

[Miner Addict]

Could GUI version running and apply at startup ?

Because it did not change anything yet if run at start up, so must apply it manually

CLi version only change REF, rfc and others not chamge

[Miner Addict]

Could GUI version running and apply at startup ?

Because it did not change anything yet if run at start up, so must apply it manually

CLi version only change REF, rfc and others not chamge

Just test it with Polaris, that did not work

[coinscrow]

VEGA64, PHOENIX-ETH, ETH MINING - SOLID 45MH

Core clock: 1030Mhz
Memory clock: 1030Mhz
VDDC: 800mV

Straps on the picture, but most key values for speed:

Code:

amdmemtool --REF 5400 --CL 14 --FAW 12 --RP 10 --RFC 300 --RCDRD 16 --RCDWR 5

(Who are trying REF values for RX560, 570, 580 seems --REF 30 were the best for me)
** EDIT: Increase --REF 6600 results in additional 0.6MH

[Eliovp]

Didn't you miss my post? Loading fan speeds for Polaris cards from saved profile seems to be broken in beta 5.

No sir, i didn't, check releases ;-)

Could GUI version running and apply at startup ?

Because it did not change anything yet if run at start up, so must apply it manually

CLi version only change REF, rfc and others not chamge

Just test it with Polaris, that did not work

Just tested it on Polaris as well, works fine

Cheers!

[rednoW]

Didn't you miss my post? Loading fan speeds for Polaris cards from saved profile seems to be broken in beta 5.

No sir, i didn't, check releases ;-)

Thank you very much!

[dragonmike]

@Eliovp
Would you possibly have looked at offering a similar tool for nVidia RTX GPUs with GDDR6?
Afaik the Ethlargement tool only works on GDDR5X cards...

[Eliovp]

Didn't you miss my post? Loading fan speeds for Polaris cards from saved profile seems to be broken in beta 5.

No sir, i didn't, check releases ;-)

Thank you very much!

Sure thing!

@Eliovp
Would you possibly have looked at offering a similar tool for nVidia RTX GPUs with GDDR6?
Afaik the Ethlargement tool only works on GDDR5X cards...

Nvidia made RTX regs read-only. There might be a way around it, who knows

[joaogti36]

release also to nvidia!!!
not all use claymore miner!!!

[SamAlackass]

Anyone has any idea about what could be causing this?

I have already verified it isn't driver related and most of the times I get zeros from all cards. Some times it will work normally, some times I will only get readings from some cards, as in the log below.

Needless to say I can't set any timings for the cards not giving proper readings.

Code:

GPU 1:  Vega 10 XL/XT [Radeon RX Vega 56/64]    Memory state: 0x0 (unknown)
Timing 1          CL: 0   RAS: 0          RCDRD: 0        RCDWR: 0
Timing 2          RCAb (RC): 0    RCPb (RC): 0    RPAb (RP): 0    RPPb (RP): 0
Timing 3          RRDS: 0         RRDL: 0         RTP: 0
Timing 4          FAW: 0
Timing 5          CWL: 0          WTRS: 0         WTRL: 0
Timing 6          WR: 0
Timing 7          RREFD: 0
Timing 8          RDRDDD: 0       RDRDSD: 0       RDRDSC: 0       RDRDSCL: 0
Timing 9          WRWRDD: 0       WRWRSD: 0       WRWRSC: 0       WRWRSCL: 0
Timing 10         WRRD: 0         RDWR: 0
Timing 12         REF: 0
Timing 13         MRD: 0          MOD: 0
Timing 14         XS: 0
Timing 16         XSMRS: 0
Timing 17         PD: 0   CKSRE: 0        CKSRX: 0
Timing 20         RFCPB: 0        STAG: 0
Timing 21         XP: 0   CPDED: 0        CKE: 0
Timing 22         RDDATA: 0       WRLAT: 0        RDLAT: 0        WRDATA: 0
Timing 23         CKESTAG: 0
RFC Timing        RFC: 0

GPU 2:  Vega 10 XL/XT [Radeon RX Vega 56/64]    Memory state: 0x0 (unknown)
Timing 1          CL: 0   RAS: 0          RCDRD: 0        RCDWR: 0
Timing 2          RCAb (RC): 0    RCPb (RC): 0    RPAb (RP): 0    RPPb (RP): 0
Timing 3          RRDS: 0         RRDL: 0         RTP: 0
Timing 4          FAW: 0
Timing 5          CWL: 0          WTRS: 0         WTRL: 0
Timing 6          WR: 0
Timing 7          RREFD: 0
Timing 8          RDRDDD: 0       RDRDSD: 0       RDRDSC: 0       RDRDSCL: 0
Timing 9          WRWRDD: 0       WRWRSD: 0       WRWRSC: 0       WRWRSCL: 0
Timing 10         WRRD: 0         RDWR: 0
Timing 12         REF: 0
Timing 13         MRD: 0          MOD: 0
Timing 14         XS: 0
Timing 16         XSMRS: 0
Timing 17         PD: 0   CKSRE: 0        CKSRX: 0
Timing 20         RFCPB: 0        STAG: 0
Timing 21         XP: 0   CPDED: 0        CKE: 0
Timing 22         RDDATA: 0       WRLAT: 0        RDLAT: 0        WRDATA: 0
Timing 23         CKESTAG: 0
RFC Timing        RFC: 0

GPU 3:  Vega 10 XL/XT [Radeon RX Vega 56/64]    Memory state: 0x118 (800MHz)
Timing 1          CL: 17          RAS: 27         RCDRD: 13       RCDWR: 8
Timing 2          RCAb (RC): 39   RCPb (RC): 39   RPAb (RP): 12   RPPb (RP): 12
Timing 3          RRDS: 4         RRDL: 5         RTP: 5
Timing 4          FAW: 13
Timing 5          CWL: 6          WTRS: 3         WTRL: 7
Timing 6          WR: 13
Timing 7          RREFD: 7
Timing 8          RDRDDD: 3       RDRDSD: 2       RDRDSC: 1       RDRDSCL: 3
Timing 9          WRWRDD: 3       WRWRSD: 2       WRWRSC: 1       WRWRSCL: 3
Timing 10         WRRD: 1         RDWR: 17
Timing 12         REF: 3120
Timing 13         MRD: 8          MOD: 12
Timing 14         XS: 216
Timing 16         XSMRS: 6
Timing 17         PD: 6   CKSRE: 8        CKSRX: 8
Timing 20         RFCPB: 128      STAG: 7
Timing 21         XP: 6   CPDED: 2        CKE: 6
Timing 22         RDDATA: 16      WRLAT: 4        RDLAT: 14       WRDATA: 1
Timing 23         CKESTAG: 2
RFC Timing        RFC: 208

GPU 4:  Vega 10 XL/XT [Radeon RX Vega 56/64]    Memory state: 0x118 (800MHz)
Timing 1          CL: 17          RAS: 27         RCDRD: 13       RCDWR: 8
Timing 2          RCAb (RC): 39   RCPb (RC): 39   RPAb (RP): 12   RPPb (RP): 12
Timing 3          RRDS: 4         RRDL: 5         RTP: 5
Timing 4          FAW: 13
Timing 5          CWL: 6          WTRS: 3         WTRL: 7
Timing 6          WR: 13
Timing 7          RREFD: 7
Timing 8          RDRDDD: 3       RDRDSD: 2       RDRDSC: 1       RDRDSCL: 3
Timing 9          WRWRDD: 3       WRWRSD: 2       WRWRSC: 1       WRWRSCL: 3
Timing 10         WRRD: 1         RDWR: 17
Timing 12         REF: 3120
Timing 13         MRD: 8          MOD: 12
Timing 14         XS: 216
Timing 16         XSMRS: 6
Timing 17         PD: 6   CKSRE: 8        CKSRX: 8
Timing 20         RFCPB: 128      STAG: 7
Timing 21         XP: 6   CPDED: 2        CKE: 6
Timing 22         RDDATA: 16      WRLAT: 4        RDLAT: 14       WRDATA: 1
Timing 23         CKESTAG: 2
RFC Timing        RFC: 208

No change on that front, not sure if it should make any difference but I did try beta 5 and 6 and the latest cli. Beta 5 did seem to make a difference at first but went back to same behavior after a restart.