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

[pbfarmer]

That is some crazy efficiency. I guess that is not even possible on windows, right? Maybe it is time for me to stop being scared and tried alternative operating systems.

I tested the base settings w/o modded timings on win10pro w/ 18.x drivers in the past, and saw the same results.  There may be differences w/ 19.x - I haven't tried yet.

That being said, a linux setup is pretty simple and equally performant for many algos these days.

[1675738669]

1675738669

[1675738669]

1675738669

[1675738669]

1675738669

[1675738669]

1675738669

[nars28]

Every combo I've tried on Polaris (even translating modded bios strap values) results in a crash to blue screen for me. At this stage, I'd say this tool is unusable for Polaris cards.

I have noticed that if you throw out --cl value it doesn't crash when applying string. So it must be some bug in applying --CL. Also the timings are reset to default when i run miner. If i apply them while mining - Bluescreen.

[pbfarmer]

Ill test ur CL, RAS and REF timing. Thx pbfarmer.

Results:
Cards : Vega 64 (aircooled, reference, samsung)
Miner : SRBMiner 1.8.3
Algo : Cryptonight litev7
Clocks : 1408/1107/875 (ODT)
Actual : 1354/1107/850 (HWinfo)

Stock                                                                                                                                                           : 4530
Todxx               --rp 10 --rc 44 --rfc 300 --rrd 3 --rcdrd 12 --rcdwr 5                                                                : 4800
My current        --rp 12 --rc 44 --rfc 250 --rrds 3 --rrdl 3 --rcdrd 12 --rcdwr 5                                                   : 4865 (7.4% inc over stock)
pbfarmer          --rp 12 --rc 44 --rfc 250 --rrds 3 --rrdl 3 --rcdrd 12 --rcdwr 5 --CL 19 --RAS 28 --REF 15600       : 4980 (9.9% inc over stock)

In all seriousness, was w8ing for ur numbers the most and u didnt disappoint  

Most of my settings are just using values which seem to be gathering consensus here - from people such as yourself  I honestly didn't put much time into tinkering, other than adjusting a couple 'derived' values based on what seemed more appropriate mathematically.  I'd like to look into more, esp CL/FAW/WR...

I mainly wanted to point out the outsized effect of increasing REF, and put some thoughts to what it does, and why it's ok to go from 3.9K to say 15-17K.

[clousian]

Every combo I've tried on Polaris (even translating modded bios strap values) results in a crash to blue screen for me. At this stage, I'd say this tool is unusable for Polaris cards.

I have noticed that if you throw out --cl value it doesn't crash when applying string. But without it the hashrate is even lower than stock. So it must be some bug in applying --CL.

Interesting. I guess my point still stands then. Unusable currently on Polaris.

[pbfarmer]

Results:
Cards : Vega 64 (aircooled, reference, samsung)
Platform : Windows 10, 18.6.1
Miner : PheonixMiner 4.2c
Algo : Ethash
Clocks : 1216/1107/875 (ODT)
Actual : 1187/1107/850 (HWinfo)

Stock                                                                                                                                                           : 44.75 mh/s
Current Timing  --rp 12 --rc 44 --rfc 250 --rrds 3 --rrdl 3 --rcdrd 12 --rcdwr 5 --CL 19 --RAS 28 --REF 15600       : 50.12 (12% inc over stock)

Finally broke the Ethash 50mh/s barrier in windows 

Ethash is even better than cn-trtl wrt being able to drop core clock on vega.  Where you may lose 5+% h/r on trtl by going to cclock p0 (852mhz), I saw zero h/r loss on ethash when doing the same, at least w/ stock timings.  Should allow you to get your voltage down significantly - closer to 800mv.

[exsexs1]

Hi Guys,

Zorg33 said to me share this set, because it's impossible, but working good for me:
--CL 12 --RAS 28 --RCDRD 12 --RCDWR 5 --RC 44 --RP 10 --RRDS 3 --RRDL 3 --RTP 5 --FAW 20 --CWL 7 --WTRS 4 --WTRL 9 --WR 16 --WRRD 1 --RDWR 19 --REF 17550 --MRD 8 --MOD 15 --PD 8 --CKSRE 10 --CKSRX 10 --RFC 250

Vega56 @ 64 sapphire samsung ram, Win10, 18.6.1 and 18.5.1 classic 1408/900core 1100/900mem

Turtle 22+
ETH 51+
Cr-Lite 10K khs (1600/990 1100/900, but crashed, 9600kh 1408/900 1100/900 super stable after 24h)

[rednoW]

Ethash is even better than cn-trtl wrt being able to drop core clock on vega.  Where you may lose 5+% h/r on trtl by going to cclock p0 (852mhz), I saw zero h/r loss on ethash when doing the same, at least w/ stock timings.  Should allow you to get your voltage down significantly - closer to 800mv.

why do you need such low gpu clocks if vega is perfectly capable of higher clocks with <800mv voltage?

it is vega56 pulse with hynix 20.5kh/s turtle btw
vega56 ref samsung can do the same ...

[Anwil]

Rx Vega 56 hynix memory  
1250/875mv   916/925mv
tartl cn 20.12 - 20.20  kh/s  L24+24
CN/r   2160   1450/900
60 min no errors.
WinAMDTweak.exe --rcdrd 19 --rcdwr 4 --rc 35 --rp 13 --rrds 4 --rrdl 5 --rfc 148 --REF 15600
Cool!!!

Looks good so far. About 1 kh/s boost to turtle. Didn't work with all my hynix tho. Need to figure out what to change to get to work with weaker card.

Edit: after longer time most gpu crash or have hashrate drop

two RX Vega56 had to increase the voltage and lower the memory to 900 , the worst 56 19.65 ,but stable. the rest 20.1

[Eliovp]

Pleases me to see you guys experimenting and having fun


About GDDR5, it makes no sense to set timings when the GPU isn't fully under load, as it won't work because you'll be setting the timings for the current active memory state, which is like 300Mh..

Ok, i have to admit, there are a decent amount of GDDR5 values in there that you'd better not touch, not exposing them wouldn't help either as you would have no idea what was going on. This also applies to HBM2, although there you can almost change em all without crashing the GPU.

I also noticed some of you already figured out that there are indeed some structures you can follow "RC=RAS+RP , ...."
(doesn't mean that thinking outside of the box won't help (such as bumping REF skyhigh ;-))) If RAS + RP is higher or lower than RC, it's not going to break your gpu..

Dram timings are calculated based on your clocks. That's also why they're not the same as a decoded strap.

Unfortunately, due to the fact that it's not possible to set GDDR5 timings before putting the GPU under load (such as on HBM2), except then in the vbios itself, it is indeed not possible to modify some values.

That said, on your (even modded Polaris) GPU, drop RFC for example.. just try it ;-)

Cheers!

[pbfarmer]

Ethash is even better than cn-trtl wrt being able to drop core clock on vega.  Where you may lose 5+% h/r on trtl by going to cclock p0 (852mhz), I saw zero h/r loss on ethash when doing the same, at least w/ stock timings.  Should allow you to get your voltage down significantly - closer to 800mv.

why do you need such low gpu clocks if vega is perfectly capable of higher clocks with <800mv voltage?

it is vega56 pulse with hynix 20.5kh/s turtle btw
vega56 ref samsung can do the same ...

Power use and heat...  Voltage isn't the only input into how much of both you consume/generate.  For the same voltage, I was able to log a 15-20w difference in power use by changing other settings under TRM.

[pbfarmer]

About GDDR5, it makes no sense to set timings when the GPU isn't fully under load, as it won't work because you'll be setting the timings for the current active memory state, which is like 300Mh..

Alternatively, couldn't you just lock out all other states (e.g. in ODNT or /sys under linux)?  Should force your mem to whatever state you want - w/o an applied load.

[Eliovp]

About GDDR5, it makes no sense to set timings when the GPU isn't fully under load, as it won't work because you'll be setting the timings for the current active memory state, which is like 300Mh..

Alternatively, couldn't you just lock out all other states (e.g. in ODNT or /sys under linux)?  Should force your mem to whatever state you want - w/o an applied load.

Mmm, i might know a way to make this much easier and less confusing.

First something else

[heavyarms1912]

Vega 56 - Hynix.

2 x sapphire pulse vega 56 modded vs unmodded.
3+ KHs bump.  I am on 19.3.x drivers.  This might yield additional 1-1.5 Khs on 18.6.1 iirc.



At 1400 core
https://imgur.com/1mLlFNi

[Miner Addict]

Ethash is even better than cn-trtl wrt being able to drop core clock on vega.  Where you may lose 5+% h/r on trtl by going to cclock p0 (852mhz), I saw zero h/r loss on ethash when doing the same, at least w/ stock timings.  Should allow you to get your voltage down significantly - closer to 800mv.

why do you need such low gpu clocks if vega is perfectly capable of higher clocks with <800mv voltage?
https://i.imgur.com/8qnLNSB.png
it is vega56 pulse with hynix 20.5kh/s turtle btw
vega56 ref samsung can do the same ...

Wow, with Soft PP or AMD tweaking ?

Great info

[rednoW]

Ethash is even better than cn-trtl wrt being able to drop core clock on vega.  Where you may lose 5+% h/r on trtl by going to cclock p0 (852mhz), I saw zero h/r loss on ethash when doing the same, at least w/ stock timings.  Should allow you to get your voltage down significantly - closer to 800mv.

why do you need such low gpu clocks if vega is perfectly capable of higher clocks with <800mv voltage?

it is vega56 pulse with hynix 20.5kh/s turtle btw
vega56 ref samsung can do the same ...

Wow, with Soft PP or AMD tweaking ?

Great info

19.4.1 drivers allow you to control and disable (grey) all p-states via ODNT! (except mem p2 voltage)
Just set mem P2 voltage to 800 with softpowerplay ))

And make working freq ~1200
Than test stability, than apply timings  -rc 37 --rcdrd 12 --rcdwr 5 --rp 13 --faw 16 --rtp 4 --rrds 3 --rrdl 3 --CWL 6 --WTRS 4 --WTRL 9 --REF 12800

Use --cn_config L22+22 or --cn_config L20+20 for teamred ))

And last but not least - screen of overnight testing no hw errors and h/r on the pool is ok

[Miner Addict]

Ethash is even better than cn-trtl wrt being able to drop core clock on vega.  Where you may lose 5+% h/r on trtl by going to cclock p0 (852mhz), I saw zero h/r loss on ethash when doing the same, at least w/ stock timings.  Should allow you to get your voltage down significantly - closer to 800mv.

why do you need such low gpu clocks if vega is perfectly capable of higher clocks with <800mv voltage?
https://i.imgur.com/8qnLNSB.png
it is vega56 pulse with hynix 20.5kh/s turtle btw
vega56 ref samsung can do the same ...

Wow, with Soft PP or AMD tweaking ?

Great info

19.4.1 drivers allow you to control and disable (grey) all p-states via ODNT! (except mem p2 voltage)
Just set mem P2 voltage to 800 with softpowerplay ))
https://i.imgur.com/mN2YX7x.png
And make working freq ~1200
Than test stability, than apply timings  -rc 37 --rcdrd 12 --rcdwr 5 --rp 13 --faw 16 --rtp 4 --rrds 3 --rrdl 3 --CWL 6 --WTRS 4 --WTRL 9 --REF 12800

Use --cn_config L22+22 or --cn_config L20+20 for teamred ))

And last but not least - screen of overnight testing no hw errors and h/r on the pool is ok
https://i.imgur.com/p0aowce.png

Greatttt, thanks for the Enlightment

[clousian]

Pleases me to see you guys experimenting and having fun


About GDDR5, it makes no sense to set timings when the GPU isn't fully under load, as it won't work because you'll be setting the timings for the current active memory state, which is like 300Mh..

Ok, i have to admit, there are a decent amount of GDDR5 values in there that you'd better not touch, not exposing them wouldn't help either as you would have no idea what was going on. This also applies to HBM2, although there you can almost change em all without crashing the GPU.

I also noticed some of you already figured out that there are indeed some structures you can follow "RC=RAS+RP , ...."
(doesn't mean that thinking outside of the box won't help (such as bumping REF skyhigh ;-))) If RAS + RP is higher or lower than RC, it's not going to break your gpu..

Dram timings are calculated based on your clocks. That's also why they're not the same as a decoded strap.

Unfortunately, due to the fact that it's not possible to set GDDR5 timings before putting the GPU under load (such as on HBM2), except then in the vbios itself, it is indeed not possible to modify some values.

That said, on your (even modded Polaris) GPU, drop RFC for example.. just try it ;-)

Cheers!

So you're saying to ignore all other values except the RFC which should produce results, either positive or not. Do you have a percentage change you would suggest? -25%?

[e97]

Update:

Decided to release a Winblows supported version.

I've tested it on 2 systems with Latest adrenaline version (one with a Vega gpu, the other with a 570) and confirmed working.

Check OP for link.


Please post results here. That's all i'm asking

Cheers guys!

@Eliovp, once again thanks for your efforts here getting it onto windows. Much easier for traditional Vega Users with PPTs and Overclocking tools to really test the limits of your tool!

For my part, I'm getting pretty decent results on ETH; Reference Powercolor Vega 64 on latest Adrenalin 19.3.3 drivers.
Thanks to Windward for getting me started on values for parameters.

https://i.imgur.com/6bedvOO.jpg

Given the generosity of Eliovp in even making this tool available, I'm not gonna be the guy that takes and doesn't give back.

For those curious, here's what I put in a batch script after I honed in on parameters.

(PATH to folder where MemTweak is stored)\WinAMDTweak.exe --CL 16 --RAS 30 --RCDRD 14 --RCDWR 14 --RC 44 --RP 14s --RRDS 3 --RRDL 5 --RTP 5 --FAW 20 --CWL 7 --WTRS 4 --WTRL 9 --WR 16 --WRRD 1 --RDWR 19 --REF 17550 --MRD 8 --MOD 15 --PD 8 --CKSRE 10 --CKSRX 10 --RFC 244

Run before starting the miner. For some reason, injecting the timings when mining would crash the miner if FAW>17, but FAW>17 is okay if injected prior. Also, RC<44 led to incorrect shares being found. No guarantees of stability... I was simply looking for accepted shares w/o rejects in short runs. I need to read up as well on compatible RAM timings... lower RC seems to drive HR the most.

Happy to answer any Q's; also, memclock was 1107 and was using PhoenixMiner4.2c
                  

Thanks @Eliovp for your research and releasing this tool! I'm liking your 12 GPU Vega system -- how does it fare for compute use? I worry not all the slots are x16.

Thanks for sharing @N2DCRYPT!

19.4.2 got me stuck at 37 Mh/s and temp was in the 70s and climbing! I'd advise anyone not to upgrade to those.

I'm getting 48.5 Mh/s with a Reference Gigabyte Vega 64 on water @ 59 C with 19.3.3 drivers.
Was getting around the same @ 54 C with the block chain drivers, I'll probably downgrade back to those.

[dragonmike]

That said, on your (even modded Polaris) GPU, drop RFC for example.. just try it ;-)

Cheers!

Dropping RFC from 260 to 250 results in much lower hashrate on pool for my Vega rig.
At least at 1107 MHz HBM clocks on CN_Turtle.

So far I'm quite happy with --CL 16 --RAS 30 --RCDRD 14 --RP 12 --RC 44. Reported pool hash is spot on - and ultimately, that's what matters.

[rednoW]

That said, on your (even modded Polaris) GPU, drop RFC for example.. just try it ;-)

Cheers!

Dropping RFC from 260 to 250 results in much lower hashrate on pool for my Vega rig.
At least at 1107 MHz HBM clocks on CN_Turtle.

So far I'm quite happy with --CL 16 --RAS 30 --RCDRD 14 --RP 12 --RC 44. Reported pool hash is spot on - and ultimately, that's what matters.

It was said for Polaris cards ))