Hi,

there is an update of Cast XMR available.

Cast XMR now supports CryptoNight-Heavy.

CryptoNight Heavy is different, the read/writes are smaller but use a larger "scratch pad" so the hash rates are not comparable with classic CryptoNight.

Here are some first performance values (tested with 18.3.4 drivers):

Vega 56  945 MemClk: 1450 Heavy Hashes/s
Vega 64 1025 MemClk: 1550 Heavy Hashes/s
Vega FE 1025 MemClk: 1640 Heavy Hashes/s
RX 570 Compute Mode: 850 Heavy Hashes/s (Polaris picks up!)

Regards,
glph3k

Cast XMR Version 0.9.5 (2018/04/14)
  - 0.5% performance improvement for the CryptoNightV7 kernel
  - support for CryptoNight Heavy set --algo=2 to mine:
    - Sumokoin (SUMO)
    - Haven (XHV)

Download Cast XMR 0.9.5 for Windows (64bit)

SHA256 hash of cast_xmr-vega-win64_095.zip:
46edbb6e52ada7eafaa05000e055fbbd5ce1f1cab134d763abdd8598c8610a09

Hi,

an Cast XMR update is available.

The main reason for the update is the integration of the PoW changes to CryptoNight for the Monero V7 network upgrade which are documented here:

https://github.com/monero-project/monero/pull/3253

The Monero V7 network upgrade is planned to be end of march, an exact and final block height is not set yet. So stay up to date!

Regards,
glph3k


Cast XMR Version 0.9.0 (2018/03/08)

  - support for the Monero V7 network upgrade. Read more about it here
  - the Monero PoW switch will be automatically detected, no restart of cast XMR is required. Only make sure to update cast XMR before the network upgrade is happening and that the used Monero pool supports the network upgrade
  - no decrease in hash rate for other currencies was introduced by the change
  - CryptoNightV7 will run at the same hash rate as CryptoNight
  - for any case there is the new --algo option to force which CryptoNight variant to use:
    - -1 = autodetect (default)
    - 0 = CryptoNight
    - 1 = CryptoNightV7 (Monero V7 network upgrade)


Download Cast XMR 0.9.0 for Windows (64bit)

cast_xmr-vega-win64_090.zip SHA256 fingerprint:
49fe7e231cf5d4434c5d0fcb69af261a6e4e7340441d4d6ad9f4022854e38027

Currently only one instance per card possible, but you can start multiple instance, each for a different card.


Working on a multi GPU version

Are the rejected shortly (1-2 seconds) after a new job was received? Something like this:


[08:11:00] New job received. Avg Job Time: 15.7 sec
[08:11:00] Difficulty changed. Now: 40000.
[08:11:01] Found Nonce, submitting...
[08:11:01] !Rejected, Nonce Invalid!


Then the reason is that there is already a new job and the calculated nonce are from the previous job.
A roundtrip takes 1-2 seconds in the GPU, an improvement would be to faster cancel the current job and throw the old results away. Currently it tries to submit them. Will try to fix in the next update.

The screenshot shows a Sapphire Vega 56 with following settings:

Core clock: -5%
MemClock: 990 Mhz
PowerLimit: +24
Fan Min Speed: 3000 RPM

The voltage control is still on auto.

The GPU consumes 260 watts.

Ok, maybe the Vega gets to hot, it has the habit to throttle down if it gets close to 70°. So to stay above 1900 you can increase the fan speed.
Flick the "Speed (RPM)" toggle an enter a value for min fan speed like 1200 and apply, the temperature should drop.

If you want to go for the 2000 and more you have to go all in. Turn up the min fan speed to "3000" RPM. Crank up memory clock to 1100 and increase power limit in steps, core frequency can stay. It gets really loud but you should reach 2000 hash/s and it will eat up another 50-100 watts. My suggestion of the above values are in my opinion the most efficient ones, but if you don't care about the power usage it is worth a try.

Also  one can edit the Voltage Control down, there are some threads where the experts are debating what the best settings are to bring the wattage down again... but please do on your own risk

1 GPU per instance, yes.

But you can run as many instances as you like.

So start a new terminal and run "cast_xmr -S [server] -u [wallet address] -G 1" to start a second instance mining on your second GPU and so on for 3rd GPU....

-G select the GPU

with

"--opencl x" one selects the openCL plattform, default x=0 which in your case seems to be the Intel OpenCL driver.

Please try:

"--opencl 1 -G 0"

Ok, planning to support multiple GPU per instance. Personally found that running on instance per GPU very stable and more flexible, but there seem some reports that it does influence the other GPUs. Should be no big deal to implement.

Which kind of API would you suggest? Is there an standard?

hm, is "AMD CrossFire" turned off?

I am testing with a Vega 56 and 64 in parallel, sorry, can't reproduce. Both keep there hash rates and don't mind if its neighbour is started or stopped.

Update to version 0.4

Cast XMR Version 0.4 (2017/10/13)

  - performance optimizations result in 1 - 2% higher hash rates
  - reduced dev fee to 1.5%

Thanks! The app should have no limit how many GPU used in parallel, probably more a driver issue.

Currently it is only one GPU per instance. One can select with "--opencl n" the OpenCL plattform to use (Nvidia, AMD, Intel) which number is the correct one is probably experimenting. And with "-G  x" which device to use. There is no programmed limit how many instances one can start,  One  process needs only 100 MB memory when running.



Oh, which version of the blockchain driver do you have? (I only know of the 17..30.1029 Beta from Aug 23).

Yes, only one GPU per instance (maybe gone change that) and only Vega GPUs, because of the simple fact I only have those and my goal is to get the most out of it. There are no plans to support any other cards, beside maybe the RX 480/580 but it never will be a generic miner replacement...

I first had a look at Dagger-Hashimoto for my OpenCL programming experiments... no luck yet.

Thanks, Sir.

Yes, the scheduling is very unorthodox, it uses 2 command queues which are started with an defined time offset. Still searching for an explanation why this faster than one larger queue. The scheduler does get the same work, but for some reason it likes it better that way. Probably very Vega specific but did not yet try it on different cards.

In the works. Ubuntu?

Afraid, yes. It is bug in the AMD drivers. It does not matter if HBCC is turned on or off, just a change seems to be needed so that the driver gets correctly initialized. HBCC is a feature to stream data from disk to GPU memory it is a feature for games with huge textures not needed for mining.

Could not find a way to fix the problem programmatically yet.

Does anybody know a solution?

On my Vega 56:

1550 at ~100 watts (power limit -50)
1800 at ~150 watts (power limit -25)
1950 at ~200 watts (power limit 0)
2050 at ~240 watts (power limit +20)

More power does not give better results because  further increasing the memory clock the card gets unstable.

Have a look at:
https://bitcointalk.org/index.php?topic=2256917

Ah, forget to list a feature:

- No viruses or malware included

Cast XMR - Highspeed CryptoNight (Monero/Bytecoin/DigitalNote/Loki/Haven/BitTube/Aeon) Miner for Radeon RX Vega and RX 470/480/560/570/580/590 GPUs

Update to version 1.6.6 on 2018/11/30



Features

- Full support for:

• CryptoNightV8 (CNv2)
  • Monero (XMR)
  • Graft (GRFT)
  • Lethean (LTHN)
  • Safex (SAFEX)
• CryptoNightV7 (CNv1)
  • Quantum Resistant Ledger (QRL)
  • DigitalNote (XDN)
• CryptoNight (Classic)
  • Electroneum (ETN)
  • Bytecoin (BCN)
• CryptoNightXTL
  • Stellite (XTL)
• CryptoNight-Heavy
  • Loki (LOKI)
  • Saronite (XRN)
• CryptoNightXHV-Heavy
  • Haven (XHV)
• CryptoNightV7-Lite
  • Aeon (AEON)
  • Turtlecoin (TRTL)
• CryptoNightTUBE-Heavy
  • BitTube (TUBE)
• CryptoNight-Fast
  • Masari (MSR)
• CryptoNight-SuperFast
  • Free Haven  (XFH)

- Fastest miner for AMD Radeon RX Vega 64 GPUs

• about 2000 CryptoNight/V7 Hashes/s
• about 1850 CryptoNightV8 Hashes/s (10 min average)
• about 1635 CryptoNight-Heavy Hashes/s
• about 4100 CryptoNight-Lite Hashes/s

- Fastest miner for AMD Radeon RX Vega56 GPUsl

• more then 2000 CryptoNight/V7 Hashes/s
• about 1825 CryptoNightV8 Hashes/s (10 min average)
• about 1550 CryptoNight-Heavy Hashes/s
• about 4260 CryptoNight-Lite Hashes/s

- Fastest miner for AMD Radeon Vega Frontier Edition GPUs

• about 2100 CryptoNight/V7 Hashes/s
• about 1930 CryptoNightV8 Hashes/s (10 min average)
• about 2120 CryptoNight-Heavy Hashes/s
• about 4260 CryptoNight-Lite Hashes/s

- Optimized cores for Vega and Polaris based GPUs
- Automatic driver version detection and setup recommendations
- Multiple GPU support for upto 24 Vega GPUs
- Monitor temperature and fan speed of each GPU
- Full pool support
- Nicehash support
- Fast Job Switch option to minimize outdated shares
- Rate Watchdog to monitor each GPU performance
- Includes switch-radeon-gpu command line tool  to restart GPUs, switch on/off HBCC, Compute Mode and Large Pages
- Remote http access for statistics in JSON format
- only 1% dev fee

Requirements

  - Windows 8/8.1/10 64 bit or
  - Ubuntu Linux 16.04 or later
  - AMD Radeon RX Vega 56/64 GPU with at least 8GB RAM
  - or AMD Radeon RX Vega Frontier Edition with 16GB RAM
  - or AMD Radeon RX 470/RX 480/RX 560/RX 570/RX 580/RX 590 GPU with a least 4GB RAM
  - For up to 50% higher hash rates the Radeon Driver 18.5.1 or later has to be installed

Version History

Cast XMR Version 1.6.6 (2018/11/30)
- support for RX 590 GPUs added
- fixes for detecting Compute Mode on Polaris based GPUs
- support for CryptoNight-SuperFast (--algo=11) for mining Free Haven Protocol (XFH)
- updated switch-radeon-gpu to version 1.2.0, which includes bugfixes for newer versions of the AMD Radeon drivers

Cast XMR Version 1.6.0 (2018/11/07)
- Vega 56/64: 0.5 to 1% performance improvement on CryptoNightV8
- Vega 56/64: 2 to 3% speed improvement on CryptoNight-Heavy variants (Loki, BitTube, Haven)
- updated included switch-radeon-gpu to version 1.0.5
- support for RX 560 GPUs added

Cast XMR Version 1.5.0 (2018/9/27)
- Support for the Monero V8 Network Upgrade which introduces a new CryptoNight variant CryptoNightV8 aka CNv2. The Monero V8 update is scheduled for 18th October. In case you are mining Monero make sure to update Cast XMR before that date. Also make sure that the algo for Monero is not explicitly set to CryptoNightV7 but set to auto detect with --algo=-1 to automatically switch from CryptoNightV7 to CryptoNightV8 when the upgrade is happening. Please also note that the CryptoNight V8 algo has added complexity which decreases the hashrate up to 7% for Vega GPUs.

Cast XMR Version 1.4.0 (2018/8/28)
- 3% performance improvement for CryptoNightTUBE-Heavy when mining BitTube (TUBE)
- support for Festival Coin (FEST) upcoming update which introduces  a new CryptoNight variant: CryptoNightFEST. Use --algo=9 to continue mining Festival Coin after the network update happened
- subtle performance improvements
- Tested with Radeon Adrenalin Driver 18.5.1

Cast XMR Version 1.3.0 (2018/6/29)
- Performance improvement for job switches with multiple GPUs
- Support for BitTube (TUBE) V4 Pow Update which introduces the new algo CryptoNightTUBE-Heavy aka CN Saber. Stay on --algo=5 to switch automatically from CryptoNightIPBC-Lite to CryptoNightTUBE-Heavy when the update is happening (approx. 2nd July). Please note that the hash rate will drop as it is a switch from a Lite to a Heavy variant.

Cast XMR Version 1.2.5 (2018/6/19)
- Support for CryptoNight-Fast introduced with the Masari (MSR) update. Set --algo=8 to mine Masari (MSR). CryptoNight-Fast will produce higher hash rates than classic CryptoNight, typical hash rates are:
  - Vega64, 1025 Mhz MemClock: 3630 Hashes/s
  - Vega56, 945 Mhz MemClock: 3700 Hashes/s
  - VegaFE, 1025 Mhz MemClock: 3900 Hashes/s
  - RX 470/570, 2250 Mhz MemClock: 1550 Hashes/s

Cast XMR Version 1.2.0 (2018/6/12)
- Fix for "unauthenticated shares" pool error
- Support for the upcoming Haven (XHV) V3 update. In case you are mining Haven set --algo=7 now. Cast will then automatically switch to CryptoNightXHV-Heavy when the network update is happening.

Cast XMR Version 1.1.5 (2018/6/7)
- Fix for mixed up device order
- Support for Stellite (XTL) V4 Update. Add --algo=6 in case you are mining Stellite. It will automatically switch to CryptoNightXTL when the network update is happening.
- Stability fixes

Cast XMR Version 1.1.0 (2018/6/1)
- All new --intensity option to specify the memory allocation in the range of 0 to 10, upto 12 for Vega Frontier Edition. The default setting will correspond with the settings of version 1.0 and are displayed at startup.
- Vega Frontier 'Large Pages' support added, up to 30% higher hash rates for CryptoNight-Heavy, 5% increase for CryptoNight
- Includes the switch-radeon-gpu command line tool to restart GPUs, switch on/off HBCC, Compute Mode and Large Pages
- Nicehash supports now up to 24 GPUs
- --maxmem option is deprecated, will now be mapped to an corresponding intensity setting
- --forcecompute is deprecated, 'Compute Mode' of Polaris GPUs will be automatically detected. To overwrite use the --intensity option
- Default mining algo changed to CryptoNightV7, to still mine CryptoNight Classic add --algo=0
- Tested with Radeon Adrenalin Driver 18.4.1

Cast XMR Version 1.0.0 (2018/4/24)
- Support for Aeon (AEON) CryptoNight Lite variant set --algo=3 to mine
- Support for Turtlecoin (TRTL) CryptoNightV7 Lite variant set --algo=4 to mine
- Support for Interplanetary Broadcast Coin (IPBC) CryptoNight Lite IPBC variation set --algo=5 to mine
- Up to 3% performance improvement with new option --maxmem to allocate the maximum memory of the card. For Vega cards the "HBCC Memory Segment" has to be disabled in the Radeon Settings to use the option. Disadvantage is a higher risk of hash rate drops when the --maxmem option is used.

Cast XMR Version 0.9.6 (2018/04/17)
  - CryptoNight-Heavy fix for very low hash rates when HBCC Memory Segment is enabled
  - HBCC Memory Segment performance hint for Vega64 when mining CryptoNightV7
  - Improvement for 4GB versions of RX 470/480/570/580 when mining CryptoNightV7

Cast XMR Version 0.9.5 (2018/04/14)
  - 0.5% performance improvement for the CryptoNightV7 kernel
  - support for CryptoNight Heavy set --algo=2 to mine:
   - Sumokoin (SUMO)
   - Haven (XHV)

Cast XMR Version 0.9.2 (2018/03/29)
  - Improvements for RX Vegas that achieve 99% the performance of the Blockchain Compute Beta driver with the latest standard Radeon Driver 18.3.4

Cast XMR Version 0.9.1 (2018/03/22)
  - 0.5% overall performance improvement for Vega based GPUs
  - --ratewatchdog experimental option to monitor the hash rate, in case an occasional happening drop of the hash rate is detected the kernel will be resetted to restore the optimal hash rate
  - fully tested CryptoNight variant (CryptoNightV7) for the Monero V7 Network Upgrade PoW Change which is now scheduled for the 6th April.
  - In case you want to mine the Graft coin use the --algo=0 option, an automatic detection of the used CryptoNight variant is not possible for this coin. For all other known CryptoNight coins the hash variant seems to be detected correct
  - In case you are mining with nicehash.com or any other automatic coin switching pool that has the Graft coin in the mix the --algo=0 option is advised to prevent invalid shares on the Graft coin. Be aware to remove the option when the Monero V7 network upgrade is happpening otherwise it then will clash with mining Monero!

Cast XMR Version 0.9.0 (2018/03/08)
  - support for the Monero V7 network upgrade. Read more about it here
  - the Monero PoW switch will be automatically detected, no restart of cast XMR is required. Only make sure to update cast XMR before the network upgrade is happening and that the used Monero pool supports the network upgrade
  - no decrase in hash rate for other currencies was introduced by the change
  - CryptoNightV7 will run at the same hash rate as CryptoNight
  - for any case there is the new --algo option to force which CryptoNight variant to use:
    - -1 = autodetect (default)
    - 0 = CryptoNight
    - 1 = CryptoNightV7 (Monero V7 network upgrade)

Cast XMR Version 0.8.5 (2018/02/07)
  - 1% performance gain for Vega Frontier Edition with Blockchain driver (fastest GPU around)
  - 1% performance improvement with current stock driver (18.1.1) for RX Vega 56/64 (40% to go to reach Blockchain driver performance)
  - fix for random blocking console output
  - GPU now always logs with its device id not the order they are listed in the -G argument
  - --log option to log console output to a file
  - remote access includes CORS support in HTTP header


How To

cast_xmr has a command line interface. For a minimal configuration run:

cast_xmr -S [pool server] -u [username or wallet address]

For a complete list of configuration options run:

cast_xmr --help

For more info check www.gandalph3000.com

Download Cast XMR 1.6.6 for Windows (64bit)

SHA256 hash of cast_xmr-vega-win64_166.zip:
8ebd4aaf769a8089d0adbb7d2f3f01f9a682c82b2b34ae4c67534b2471b126b8

Cast XMR for Linux is available here:

Cast XMR for Ubuntu 16.04/18.04