Please post your results BEFORE and AFTER using rxboost tool  Show some pictures  You really should credit Eliovp for allowing you to be able to do this. Got to give credit where credit is due. Yeah well . ... doesn't surprise me :p Here, for the people on Windows --> LinkWill upload CLI version for Linux users asap as well  |
|
|
|
New Release - Added FIJI support
- Added Refresh Rate timing (RX boost )
- Overdrive fixes
- Added memory manufacturer detection for G5, HBM & HBM2
- .....
Link: HereWill upload CLI versions asap as well! I noticed some people have figured out that there is still room to improve on RX 4**/5** cards.
Just not many people saying how..
Well, open up the tool, and raise Refresh Rate (REF) i created a separate textbox for it to make it easy.
Raise it from 5 to 7/8 (just an example) and you'll see improvement already.
Slightly dropping RFC will increase your hashrate as well.Greetings |
|
|
|
Eliovp, so only with 19.4.x driver i could get the tool to to run, with 560/570/580 Tool is prompt when i tested, got hash increase. However, after a disconnect and miner restating or anything similar, i can't apply any timings  not sure why, also, this 19.x drivers doesn't play well with most miners. I also lost temperature control with overdriven or your tool, on those drivers. By newish drivers as a requirement, which do you can do? 18.6.1/18.7.1 ? also, thats a technical question, why do memory in action, change Dram and Dram timing 2 also EDC  I'm on windows 10 pro 1803 There's like 50 different cases to check, driver version, overdrive compatibility, gpu's, ... if i had to add support for every driver, gpu, overdrive api, i'd go insane. Regarding dram, dram values are calculated upon boot based on clocks and vbios values. So for everyone else with "similar" issues. In case you experience issues with the GUI version, update drivers, use DDU to remove old ones first.
If that still gives you issues, simply use the CLI version which does not have overdrive support (Clock/Fan/.. control)That said. I will release a new version (probably today) with Fiji support, some new timings, some bug fixes, memory manufacturer for HBM2, etc... Cheers! |
|
|
|
Is there any solution for the non ref 56 cards that have a default of 800 mem and can’t go past 900-950?
I got four of these crap cards any thoughts ? Three are red devil 56 power colors and one is a sapphire pulse
If you're lucky, you might find a Vega 64 vbios for that same card (if they made em both 56 & 64), which is unique.. not much out there, Pulse, only 56 afaik. That however doesn't mean they're crappy, i've got a Sapphire Pulse with Hynix memory and it's running smooth at 47~Mh (memory at 975). Got a Samsung Vega56 too and hits 48+Mh.. The thing is, they only have 6 SOC states, so it gets auto forced into a lower memory state (800MH), you can see this in the tool.. Vega 64 has 8 SOC states, therefor it goes to the highest memory state (1000Mh). is there way to apply straps like the claymore miner ?
for polaris
GUI version has an export/import feature, you can use that.. "kind" of the same. Hello,
I have strange problem with one rig with 3 vega 64 liqued cooling.
I try many memory tweaks with latest memtweak cli and got 51,5 Mh/s on each vega, but pools (ethermine, 2miners, miningpoolhub) show disconected status. But if I lucky and pool accept connections -> low end pool hashrate
HBM2 1100 and CORE 1050 on all three cards (Win10x64 1809). Tried claymore and phoenix. Coins - etc, eth, exp
When i reset timings - all card hashing without problems on pool.
Thank you in advance for any advice.
I'm not able to reproduce this, just to be sure i let my Vega 64 run at 53Mh for half an hour, pool hashrate was spot on. What are your timings? Cheers! |
|
|
|
Thank you both for the suggestions.
Zorg33, I checked the files and they seemed alright (date, size, everything). I still went ahead and copied the file from C:\AMD just to make sure, still no change I'm afraid. I also uninstalled everything w/DDU, double-checked the files were deleted and installed the driver again. Again, no change.
naska21, could you elaborate a bit? The system is running fine, mining for days non-stop, I just can't make AMD Memory Tweak work for whatever reason. I could be missing something of course.
edit: forgot to mention, no conflicts in device manager.
Have you tried the CLI version? --> hereStrong feeling it's an ADL issue.. |
|
|
|
Thank you for some directions. Now it's time to play with that. One question though: I use the GUI. What happens after system restart or gpu crash? Everything stays the same or do i have to apply the values again?
You're welcome. You have to apply the values again yes. See this as a positive thing ;-) Cheers! |
|
|
|
im running into the issue with my vega 64's . I'm using 18.6.1 driver on srbminer because you "cant" use the updated drivers but the Tweaker tool gui needs new drivers can someone explain. Do i just use pixel patcher and update the drivers and srbminer to newest? to then beable to use the lastest Tweaker tool? I was seeing my voltage changing up and down in the old tool going from 900-1100. Is that because of old drivers? And now the tweaker tool wont open.
The GUI version first checks if the installed driver has support for the overdrive functions in the tool, if not, in theory, you should just see empty space where all the clock control is. Or.. an error message..  As stated before, to solve that, update drivers to 19.4.*, a lot of earlier versions will work as well. As long as the overdrive API is supported. If that all still doesn't work, there's still the CLI version you can download and run. It's purely written for timing modifications and nothing else. Hei.
Anybody used this tool on RX 570 4GB?
I have a bunch of them and was playing around a little but not quite sure what to change. Would be nice to hear what i should change and what to leave alone so i don't mess anything up.
Thanks:)
Try to work with small steps. things such as tRFC, drop those slightly, benchmark for a bit, do this in steps until your GPU crashes, then just revert back to latest working modifications. This will already give you an increase. Have fun! Cheers |
|
|
|
@eliovp sorry I do not know how quotes works on this forum.
Anyhow HBM2 on vega64 with Samsung. It does not seem to matter what setting you put in. I can do some screen shots. I mean the tool reads the setting you imput...but it does not have any actual effect on anything...as if it is being ignore.
Yeah, i'm aware ;-) Check my post above this part here For HBM2 you can raise tCL to offer a little bit of extra stability.
It is however a fact that changing this, or lowering this to a crazy number won't have any effect.
I read somewhere that there are two places these timings are set, on the ASIC and on the RAM side.
The ASIC side is the one being changed here, which is not a problem, however, the RAM value is the one being used. (This only applies to Cas Latency and Write Latency i believe)
Thus, to be able to change this, we're going to have to change vbios strap or do this before the timings get loaded into the registers. |
|
|
|
Eliovp - can you possibly add a version somewhere on the gui?
I know its a pain, but I am looking at multiple version and comparing things and i get lost.
Thank you!
Hey, Yeah, i can do that. There has been a request to add a --version flag to the cli version as well. It's on my todo list! For now, you could always check assembly info (right click exe), you can find version info in there. @eliovp can you check something tCL does not seem to change anything. This was pointed out by someone else on reddit. Anyhow I tested it, and it does indeed seem to be true. You can set the tCL value to 0, 1 up to 50.
Are you referring to HBM2 or GDDR5? For GDDR5 i disabled this (for the GUI) on purpose. You can create a strap for the vbios with tCL being at 24 (decimal) for example and it will work just fine. Now the strange thing about this is the fact that it should in theory be a value between 0 and 20. I have to be honest here, i believe programming it on a live system with a value between 0 to 20 might just be possible, it should! , i have only tested this with a default Cas Latency of higher than 20. So that might have been the reason for the GPU to bail on me. I will look into this in more depth. For HBM2 you can raise tCL to offer a little bit of extra stability. It is however a fact that changing this, or lowering this to a crazy number won't have any effect. I read somewhere that there are two places these timings are set, on the ASIC and on the RAM side. The ASIC side is the one being changed here, which is not a problem, however, the RAM value is the one being used. (This only applies to Cas Latency and Write Latency i believe) Thus, to be able to change this, we're going to have to change vbios strap or do this before the timings get loaded into the registers. In a way, perhaps a bit comparable to the dram values on the GDDR5 side. These get loaded into the registers and based on clocks, they change. I wish i had all the answers, but unfortunately, not "yet" Cheers Edit: I made a good set up for my two vega 56@64,i get 2350 hash each, but my pc lag, stuck when I do anything even when I move mouse, I use red team miner, the amd mem tweak tool, win 10 pro 64,drivers 19.4.1,19.42.
what's your setup, can you share it, and what card you own? happy mining my friend!thanks to EviolpEliovp! Try latest GUI mate, you're using an old version. You won't need overdriventool anymore, can do it all in the gui. Cheerio! |
|
|
|
Eliovp:
RX series modifications suggestion?
quick question. The order of the listings vs the order of the timings list is there away you can flip the positions?
when you run a timings Query the order is PMG, RAS, CAS... but the tools GUI goes PMG, CAS, RAS... just curious.
and is there a possible typo on the CAS timings? there's tCCLD however in the query there's tCCDL is this a typo or a different setting?
You have a point, they are switched, i changed that in latest release. Also fixed that typo ;-) Also dunno if you or anybody else knows.. if you zero out all the P states on the core control it'll actually remove the P state out of Wattman. which is pretty sick I love it.
If you're running the memory tweak also noticing that when you set the P state setting it'll fight you using the Wattman so this tool will replace your management for ALL of your settings.
That's indeed the Idea, to not have to use multiple tools.. Previously i thought you had to setup batch files to run for the cards, which you can do, however you can also just export the settings and import them.
I understand this is probably stating the obvious, however just thought i'd point it out anyways in case people didn't see it.
However when you import the Fan control settings are blank and you have to put them in manually every time. Can you fix this?
Fixed in latest release Once again... Eliovp.. thank you for making such a kick ass tool!!
You're welcome! IF You have an on-board video device or a non AMD card in your system you will get those errors. The GUI is based off of Overdrive N Tool code. As a result it will ONLY work with an all AMD environment. Yes and no, the control part (Clocks, fans, etc) is based off of public Overdrive ADL functions. Anyone can do this. --> https://github.com/GPUOpen-LibrariesAndSDKs/display-libraryI am not sure if Eliovp has a solution to that particular issue, so for now as you get to test and run your settings be sure to disable your onboard adapter and run your system with 2 cards, 1 to view your screen and the 2nd to test and configure your settings.
Mmm, i'll look into this.
There are a few things that are incorrect in the GUI. Such as the following:
Under the CAS Timing. - tR2R is actually TCCDS / tCCLD is actually TCCDL
There's nothing incorrect in the GUI (anymore). The CCLD typo, that was indeed wrong. Fixed in latest release. Further, here's the structure for MC_SEQ_CAS | Field | Bit | | tNOPW | 1:0 | | tNOPR | 3:2 | | tR2W | 8:4 | | tCCDL | 11:9 | | tR2R | 15:12 | | tW2R | 20:16 | | tCL | 26:24 |
You will also note that on the bottom right there is a Timing level setting.
Default (0) - will get you 18 to 20 mhs
Level 1 - 21 to 25 mhs
Level 2 - 27 to 30 mhs
SO you will want to set your timing level to 2.
Don't forget, these are predefined by AMD.. Anywho, latest gui with some of these fixes on github! Cheers! |
|
|
|
Hi All!
Trying to start WinGUI version, but get:
Driver Version 18.10.21.01-180607a-329759E-RadeonSoftwareAdrenalin
ADL version 7.18.10.1327 build 11021
ADL dll path C:\WINDOWS\SYSTEM32\atiadlxx.dll
ADL2_OverdriveN_SettingsExt_Get not found
ADL2_OverdriveN_SettingsExt_Set not found
ADL2_OverdriveN_MemoryTimingLevel_Get not found
ADL2_OverdriveN_MemoryTimingLevel_Set not found
in logs and message "Start service failed..."
.Net FW 35 is on, drivers 18.6.1
Please, help!
Hey, Yeah, you're using an older driver that doesn't support the ADL functions being used in the GUI tool. I suggest to either upgrade your drive or use the CLI version which will work perfectly fine. That said. New GUI version released. Link- Many Bug Fixes
- Overdrive 8 added (R7)
- More checks for supported overdrive versions
- Changed values appear bold
- Stability upgrades
- .....
 Cheers! |
|
|
|
Vegas are shit lol, what’s the point of all this effort to add 5 to 10 percent extra hash and most likely have to reset all your low power use power play settings and stuff?
I just don’t see why it’s worth all the effort. Now if these on the fly timing settings work with Polaris cards which are actually good mining cards on a price / watt basis this would be somewhat interesting
Bro - did you even try? ;p Seriously, I guess you haven’t, because changing timings doesn’t require any changes to pp settings. It causes your mem package to draw a little more power, which is on a fixed voltage (set in bios) with plenty of headroom. It may, depending on how downclocked/undervolted you are, require a tiny bump to take full advantage of the available extra performance - like I had to for my 56 on CN-turtle, but I’m generally way more dc/uv’d than most people. And btw, Vegas are much more efficient than Polaris on a h/w basis... maybe not so much on hash/purchase cost, though haven’t checked prices in a while He's not worth the trouble.. I remember him calling me out, calling me names, just because i was merely defending the work that was stolen from me in the first place. Here's his crazy post, and a few others under that. Funny how it all turned out.. theft, scams, lawsuits, etc.. (and it's not the end of it..) (from that awesome "company") .. Link here and here. So again, don't pay attention to him. To answer his post. "What's the point of having 5-10% hash increase" Uhm... Lets just imagine you have 1000 Vega's running (which is not abnormal because i know farms with a ton more), add 10% (free) hash to that.. From 44Mh/s per GPU (44Gh) to (51Gh) that's 4000 USD more per month. Yeah.. just saying On another note: will be releasing a new version very shortly, tons of updates/upgrades. (UI version..) Will release a CLI version update soon as well! Keep tweaking guys! Oh right.. Polaris live timing mods does work, there are indeed some values that can and will crash your GPU. Working on making this more user friendly. Cheers! |
|
|
|
Check releases guys, in case it's not opening, there's another version. Working on a newer release with upgrades and fixes to potential issues. Here's the link to a working version: Link |
|
|
|
Hey all! Just to let you know that i've released a new GUI version with muchos improvements and options. A few of those.. - Bug Fixes
- Clock Control (Both Core and Memory)
- Fan Control (Curve)
- Export/Import values
- Reset option
- Removed EIO requirement
- ....
LinkSome shots   |
|
|
|
Installed all NetFrameWorks still cant start GUI - Win7 and 10
Make sure you copy those 3 included files to the same directory, (driver, pretty important), i just noticed i didn't include them all haha Release has been updated. Also, run as admin. Will release a new version soon without EIO.dll and manifest added so it auto asks to start as admin I'll update post asap! Cheers! |
|
|
|
Hey all, For those that want to try out my GUI version of this. LinkHas: - Memory Clock / State control (change clocks, voltages, disable/enable states) - Refresh values on screen - Save all values - Load saved values - .. Some shots   |
|
|
|
Hey Mashy, sorry sir, no XMR address.
You can always use a shift service if you want to tip.
coinswap, changelly.. etc..
Get yourself one, I'll be happy to mine some XMR for you Haha, ok ok, you convinced me ;-) Check OP And thank you! |
|
|
|
Thanks alot Eliovp for your efforts for the windows version. I have been learning this vague ubuntu thing for the last week until you released windows version. Now, for the love of god, this 18.50 driver with Ubuntu 18.04 is installable?  As a desktop ubuntu NOT headless (i use the machine mainly to experiment on mining, and hell yes this linux is WAY stable than windows.) Can anyone tell me how i can get 18.50 GPU pro running? i tried installing amdgpu install and opencl only from amdgpu pro, tried alot and mostly end up in a stuck boot. 2nd issue! Windows, mate, i tried changing timings simply tRRD of elpida while its running..... (mining Graft with TRD 0.4.3) it succeds in changing tRRd to 4 for example, but when i use --current its back to old value (5) also tCKE/Pulse of Hynix AJR is wierd! 13/13 underload.... oO same with other wierd readings like bus_turn. are these things getting changed dynamically? also my gpus give memory errors while mining so that might explain failure to alter timings? I recommend to go online and search for a simple tutorial on how to install amdgpu drivers. It's not that hard but i can imagine if it's your first time, that it might be a bit tricky. I'm working on something to make this process easier. ;-) Great tool. I have been playing a bit mainly with list timings in previous post and have found something that works for each type of card I have, V56 V56 flashed to V64 and V64.
Thank you soo much for releasing this tool.
Do you also have an XMR address, would be an idea to post up and maybe we can all throw some hashpower your way aswell.
you should be greatly rewarded for your work.
of topic:
I am also using a hynix memory strap you once posted on my rx550's and it is amazing
Hey Mashy, sorry sir, no XMR address. You can always use a shift service if you want to tip. coinswap, changelly.. etc.. Using TRM 0.4.4 and last driver ;
Working great with my vega 64/56 reference card ! Around 2200h/s avg..
But summer is coming , is there a way to decrease consumption with this this tool ?
For now my card are running at 1407/1100 or 1407/950 at 875mV for most of them and 900-930 mV for bad ones.
Who knows.. perhaps something is coming  |
|
|
|
Hey All! Thank you for being so active and getting such nice results! Lovin it! I tried my best to read through the comments, noticed some people with issues, most of them got resolved real quickly, so that's ok. Due to that i'm currently busy completing some changes. There's a bug in the code for CL (GDDR5), so that indeed won't work, even if the bug wasn't there, it still wouldn't really make a difference unfortunately. The way those timings get applied/loaded differs very much from HBM2. Next release will be a nice change Continue tweaking! Cheers! Elio |
|
|
|
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 |
|
|
|
|
Show Posts
