Would it be naïve to sell my Vega 56 for $$$ to buy a better card "when" prices fall again?

[GoldenTiger]

I hated dropping $1799 on a Strix 3090 a year ago... but seeing where everything is at now, I'm really glad I did. GPU prices are bonkers now - if you can find one that is!

I feel like my evga 3080 FTW3 ultra hybrid at $779 in December was a steal . I'm jealous of the 3090's immense vram though.

 

[N4CR]

I'm also considering jumping to a 6600xt or 6700xt.
So, WTT?
V64 ref, late production run, excellent undervolter, never OCD and under 2k hours game time at~120-160w typically. Spent a year off and unused while I was overseas. Yet to clean dust out it's due I'd assume so will do for buyer.

 

[buttons]

We all have different needs, but i went from a GeForce 1080ti to a 9 year old firepro V7800 and didnt cause me any grief. Since last november I have won newegg shuffled 3 times (2 successful purchases) and currently using a 3060ti LHR to heat my office because i refuse to buy a space heater. Few weeks ago helped my friend build an AMD 5700G rig and really impressed. Its at least as good as my Firepro V7800 with ddr4 3600. I cant wait to see a 6700g with ddr5.

I use a 2560x1440 144hz monitor - -but mostly play old stuff. CS:GO, wow, overwatch, turn based 4x games.

 

[trick0502]

If you check best buy regularly you can find gpus at msrp. I was able to get a 3070 at msrp and sold my 5700xt to cover the cost. I was able to buy my 3070 and have $100 cash in hand.

 

[crazycrave]

Just sold the RX 5700 and RX 570 and the only real way to play this game of Cat and Mouse ..

 

[penn919]

It's super easy to setup Nicehash... and the clocks for the card you pretty much do an overclock/underclock for your specific card which you can also find everywhere on the internet.





There is also a feature where it will stop mining when it detects you're playing a game on STEAM:
https://www.nicehash.com/blog/post/new-nicehash-miner-feature-pause-mining-when-playing-steam-games

Alright, I've got a basic setup running. I'm surprised at the level of automation that's been devised. Some of the steps he (video poster) mentioned has already been simplified further. I installed all the miners (plugins) then ran the benchmark and I guess it decided itself which to run because it's already spitting out data via command text and my Vega is roaring full throttle:

I haven't bothered adjusting Core/Mem/Volt but I believe my Vega 56 was pre-flashed with the Vega 64 bios. I'm guessing this data might give someone a good idea of whether or not any of this is worth bothering with.

Let me know...

 

[legcramp]

Alright, I've got a basic setup running. I'm surprised at the level of automation that's been devised. Some of the steps he (video poster) mentioned has already been simplified further. I installed all the miners (plugins) then ran the benchmark and I guess it decided itself which to run because it's already spitting out data via command text and my Vega is roaring full throttle:

View attachment 416420

I haven't bothered adjusting Core/Mem/Volt but I believe my Vega 56 was pre-flashed with the Vega 64 bios. I'm guessing this data might give someone a good idea of whether or not any of this is worth bothering with.

Let me know...

Nice! Do a little digging online also to make sure you're getting the best hash-rate/energy usage for your specific card.

 

[penn919]

I'm also considering jumping to a 6600xt or 6700xt.
So, WTT?
V64 ref, late production run, excellent undervolter, never OCD and under 2k hours game time at~120-160w typically. Spent a year off and unused while I was overseas. Yet to clean dust out it's due I'd assume so will do for buyer.

Wait...were you asking me? I don't have either of those cards. I've got a Vega 56, RX 580, and GT 710....but I'm also looking for new card.

 

[kirbyrj]

Alright, I've got a basic setup running. I'm surprised at the level of automation that's been devised. Some of the steps he (video poster) mentioned has already been simplified further. I installed all the miners (plugins) then ran the benchmark and I guess it decided itself which to run because it's already spitting out data via command text and my Vega is roaring full throttle:

View attachment 416420

I haven't bothered adjusting Core/Mem/Volt but I believe my Vega 56 was pre-flashed with the Vega 64 bios. I'm guessing this data might give someone a good idea of whether or not any of this is worth bothering with.

Let me know...

You can definitely tweak your memory/core/voltage for better results. Whattomine says they get 49MH/s at 170W. You'd be using 50 less watts and gaining extra hash rate (more money). Definitely seems worth it to keep playing around for a higher rate.

 

[penn919]

You can definitely tweak your memory/core/voltage for better results. Whattomine says they get 49MH/s at 170W. You'd be using 50 less watts and gaining extra hash rate (more money). Definitely seems worth it to keep playing around for a higher rate.

Thanks for the response. I'll try it out. I found a website that says I should set:

Core Clock: +100
Memory Clock: +600
Power Limit: 80

I'll start with that.

 

[penn919]

Idk where you got that, those are nvidia clocks...

Use the V56 bios (if it's currently flashed to V64 and has a switch), use teamredminer (not NBminer), then set: core clock to 1000, core voltage 850mv, mem clock 900mhz. Should be in the mid 40's for hashrate, and pull like 100w less than you're currently running. These things need a lot of fine tuning to run them at their full hashrate potential. These are very conservative clocks just to start with, then work your way up from there.

Im also in the camp of just selling it and getting something like a 3060ti. Roughly a 1:1 trade for a newer, better, and more efficient card that will still hash at 45mh/s without all the fine tuning BS of vega mining on windows.

Also if you're going to continue mining with the vega, use TRM to control the fans and target 75C mem temp. Around 80C is where you'l start to damage/degrade the HBM.

Source: miner since 2017 with dozens of vega's.

Thanks. I got it from here...it turns out they used an Nvidia 1070 illustration for what was supposed to be instructions for a Vega 56. Anyways, I'm heading your advice and I've paused the miner for now until I've got all the parameters in correctly. I also needed to update the driver anyways.

 

[funkydmunky]

Thats probably a good idea, at 77C core your mem is probably cooking at 85C+ (should throttle at 90-95C, iirc). The latest TRM version allows clocking with the miner itself, so you don't even need to run the nicehash program. You can just download TRM, then set it to mine with your nicehash address and setup the pool to the US ethash pool they (nicehash) have. I havent run the nicehash program in years, so idk if you can edit in custom commands or select specific miners, but anyways...

For the TRM commands, open up the Eth batch fine, setup your nicehash pool/address, then add this to the end:

--clk_core_mhz=1000 --clk_core_mv=850 --clk_mem_mhz=900 --fan_control=::75:50:25

If thats stable, increase the mem incrementally up to 960mhz. At that point if you want to, you can start playing with HBM timings for another 3-5mh/s or so.

I seem to have a high share # (MS) of around 90. Any advice for getting it down or is that speed fine?

 

[penn919]

Thats probably a good idea, at 77C core your mem is probably cooking at 85C+ (should throttle at 90-95C, iirc). The latest TRM version allows clocking with the miner itself, so you don't even need to run the nicehash program. You can just download TRM, then set it to mine with your nicehash address and setup the pool to the US ethash pool they (nicehash) have. I havent run the nicehash program in years, so idk if you can edit in custom commands or select specific miners, but anyways...

For the TRM commands, open up the Eth batch fine, setup your nicehash pool/address, then add this to the end:

--clk_core_mhz=1000 --clk_core_mv=850 --clk_mem_mhz=900 --fan_control=::75:50:25

If thats stable, increase the mem incrementally up to 960mhz. At that point if you want to, you can start playing with HBM timings for another 3-5mh/s or so.

That'll be my next stop because I've been having one HELL of a time trying to get your values to stick in either MSI After Burner or the built-in Radeon Software tuning feature. I settled with using the "under-volt" preset until I figure a better solution out.

I'm guessing both have to be disabled before trying TRM, right?

 

[funkydmunky]

90ms is "acceptable", but is that via a ping test or from the miner reported shares? There will be some added latency from the pool itself, so you might actually be seeing 50ms from a ping test to the pool, but then the miner is reporting a share rate of 90ms. I noticed this first when I was on nanopool, where my tested ping was great, but the share latency reported by the miner was crazy high (150-200ms). Ethermine/hiveon report pretty close to actual ping, fwiw.

Things to try and test are different pools and pool locations (physical location obviously matters, and most pools have east/west coast servers). If your rigs are on a shared connection like a home network, a large download and poor (or no) QoS will send your share latency through the roof, spike your stales, and really skew your 24hr stale rates.

TLDR: 90ms isn't bad, but look at your stale rate, if your pool reports it. 1% or less is ideal, up to 2% is "acceptable" but has room for imporovement. Ive run large farms with sub 20ms ping that get .5% (or less) stales, then my home connection is around 50-60ms and I see pretty much exactly 1% stales. 1-2% is likely only pennies a day in lost revenue (unless youre mining at scale), and nothing to lose sleep over.


Yeah I never had much luck using either of those programs either, I used overdriveNtool when I had to use windows and AMD stuff (it works well with batch files for easy startup scripting too). It worked great as long as it was compatible with certain drivers, but you still have to do some tinkering with it. Ive always had better luck using miners (like Phoenix or now TRM) to set values.

Also a heads up, youl likely need to run TRM as admin (right click the exe > properties, check set to run as admin) for it to be able to actually adjust the clocks.

Such thanks bro. Quite the response thank you.
Ya all my PC's share the one.
I am way under the 1% even with my VPN which I tested with/without.
Not sure how to try another pool via NiceHash. I thought the point was I have no choice
I am not complaining as this has been the nicest mining experience yet. Well when I mined the "F" outta Dodge back in the day only to throw it in the garbage when it went down. Lesson learned!

 

[penn919]

Yeah I never had much luck using either of those programs either, I used overdriveNtool when I had to use windows and AMD stuff (it works well with batch files for easy startup scripting too). It worked great as long as it was compatible with certain drivers, but you still have to do some tinkering with it. Ive always had better luck using miners (like Phoenix or now TRM) to set values.

Also a heads up, youl likely need to run TRM as admin (right click the exe > properties, check set to run as admin) for it to be able to actually adjust the clocks.

I modified the bat, but nothing is launching, so I'm guessing I did something wrong. Care to look it over? I've already set "teamredminer.exe" to run as administrator via properties. The US ethash pool address is one I got from here

:: These environment variables should be set to for the driver to allow max mem allocation from the gpu(s).
set GPU_MAX_ALLOC_PERCENT=100
set GPU_SINGLE_ALLOC_PERCENT=100
set GPU_MAX_HEAP_SIZE=100
set GPU_USE_SYNC_OBJECTS=1


:: Example batch file for starting teamredminer.  Please fill in all <fields> with the correct values for you.
:: Format for running miner:
::      teamredminer.exe -a <algo> -o stratum+tcp://<pool address>:<pool port> -u <pool username/wallet> -p <pool password>
::
:: Fields:
::      algo - the name of the algorithm to run. E.g. lyra2z, phi2, or cnv8
::      pool address - the host name of the pool stratum or it's IP address. E.g. lux.pickaxe.pro
::      pool port - the port of the pool's stratum to connect to.  E.g. 8332
::      pool username/wallet - For most pools, this is the wallet address you want to mine to.  Some pools require a username
::      pool password - For most pools this can be empty.  For pools using usernames, you may need to provide a password as configured on the pool.


:: Example steps:
:: 1. If you prefer a different pool, change the pool server address.
::
:: 2. Replace the example wallet with your own wallet(!).
::
:: 3. Name your worker by changing "trmtest" to your name of choice after the wallet below.
::
:: 4. You're good to go!


teamredminer.exe -a ethash -o stratum+tcp://daggerhashimoto.usa-east.nicehash.com:3353 -u <REDACTED> -p x --clk_core_mhz=1000 --clk_core_mv=850 --clk_mem_mhz=900 --fan_control=::75:50:25

 

[agentdomo]

I did that with my 5700xt and then went on one of those stock drop discords and got a 3080 FE from bestbuy @ retail

 

[penn919]

For sure man, if you want to delve even further into it, this is even more detailed of a response and has a neat tester to check "real" pool latency. Nicehash has east and west servers now (assuming you're US based), so it might be worth checking if you're using the appropriate one.
https://2miners.com/blog/check-the-real-ping-to-the-mining-pool-server-with-stratum-ping-tool/


For starters, place a "pause" at the bottom of the batch file (without quotes) so you can catch if it spits out an error. But secondly, try removing the = sign from the fan control, then try again. If that doesn't help, try removing the = signs from all the other paramters. If that doesnt work, remove everything after the "-p x" line and add paramters one by one. Wish I could help ya out more but I don't run any AMD cards under windows anymore to test it out. The fan control "should" be correct though, according to the TRM usage file:

--fan_control(=CFG1,CFG2,...)  This argument enables gpu fan control by the miner.  TRM supports
                              auto-adjusting the gpu fan speed(s) based on core, junction and/or mem temp.
                              We also support setting a static fan speed in percent.

                              Each gpu type has a fan configuration that will be used by default.  We support
                              both overriding the default configuration per gpu type as well as setting a
                              specific config per gpu in the rig.

                              A fan config consists of (max) six values, separate by the ':' (colon) char.  The values
                              are: core target temp, junction target temp, mem target temp, initial fan speed in
                              percent, min fan speed in percent, max fan speed in percent.  Any value can be left empty.
                              These are a few examples:
                              ::70:50:25   Target mem temp to 70C, start fan at 50% speed, min fan speed 25%.
                              55::::20:80  Target core temp to 55C, start fan at default configuration's speed,
                                           always keep fan between 20% and 80%.
                              55::75:80    Adjust fan so that core temp is <= 55C and mem temp is <= 75C, start
                                           fan at 80%.
                              :::100       Set static fan speed to 100%, never adjust based on temps.

                              The fan configuration used for a gpu is derived in the following precedence order:
                              1. The --fan_control argument was passed a list of configurations and included an
                                 entry for gpu X.
                              2. The default fan configuration for this gpu type was overridden with a
                                 --fan_default_<type>=... parameter.
                              3. Use the built-in default fan configuration for this gpu type.

                              In the list above, if the chosen configuration is missing the start value for the fan
                              the setting from the next configuration is used.

Apologies, it's been like 2 years since I played around with TRM under windows..

Edit: You could also just simplify fan control and use this, it will run the fans a bit louder though as it will target 70C mem instead of 75C:

--fan_default_vega=CFG

Info from here: https://github.com/todxx/teamredminer/blob/master/USAGE.txt

Well the good news is that I figured out the problem. When I placed "pause" at the end of the script, I discovered that it wasn't recognizing teamredminer.exe as a valid file. That's when I remembered that I had been running the batch file as administrator as well which automatically runs from a different directory....(facepalm). Double clicking the batch file does launch the exe as expected, but....

The exe launches and maybe for 1 sec I see text but then it inexplicably closes before I get the chance to read anything. Is there a similar "pause" feature for the teamredminer.exe? I already checked to see if windows was blocking the program; It isn't. No blocked threats are being reported in the Windows Security center. I've got no third party security installed on this system.

I thank you for all the help and I hate to be a bother, but I think I can figure this thing out if I can actually see exactly what's happening in the TRM window.

 

[penn919]

So I tried out mining for a bout a month which netted about $70 bucks worth of bitcoin; however, I eventually opted to sell the GPU before it's too late to capitalize on the current prices. Even if prices don't come back down to normal, I doubt a second hand reference RX Vega 56 will be worth $700 (the amount I sold it for) for much longer. It looks like the RX 6600XT is nearly 50% faster. I guess that 'll be the card I'll shoot for as a replacement, but for the time being and I'll wait and see if current prices are as fixated as many people seem to believe. I suppose it couldn't hurt, right?