Gateless Gate Sharp 1.3.8: 30Mh/s (Ethash) on RX 480! - page 4.

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: laik2 on September 03, 2018, 02:47:25 PM

Quote from: zawawa on August 30, 2018, 09:49:04 AM

I noticed all sorts of goodies are exposed through the I2C bus.
I got to be careful not to be carried away too much, though...

Did you notice that you forgot about linux users and mining distros? Imagine your miner running on ethOS, hiveOS, simplemining and etc...

CRACK THE VEGA/LINUX NUT--

If you solve the problems with VEGA card performance on Linux systems, it should be worth a small fortune, private or public. It would be nice to have a portable installation similar to that of NiceHash Excavator on Linux, or to co-ordinate with the Devs for inclusion on the various mining distros mentioned above. Each distro has thousands of miners, a small fee of any kind (from Devs or a constant percentage) should be beneificial to all.

I have 2 RX VEGA 64 cards and have refrained from buying more because I mine on Linux. In the USA, a 4-card rig will stay within the electrical limits of 110V houshold current. With 220V current, 8-card rigs will easily fit The VEGA cards are power-houses of hash rate, but for some reason they currently perform poorly on Linux drivers.

Find the fix, it may be woth your while. --scryptr

laik2

sr. member

Activity: 652

Merit: 266

Quote from: zawawa on August 30, 2018, 09:49:04 AM

I noticed all sorts of goodies are exposed through the I2C bus.
I got to be careful not to be carried away too much, though...

Did you notice that you forgot about linux users and mining distros? Imagine your miner running on ethOS, hiveOS, simplemining and etc...

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

I noticed all sorts of goodies are exposed through the I2C bus.
I got to be careful not to be carried away too much, though...

joaocha

full member

Activity: 254

Merit: 100

Quote from: zawawa on August 28, 2018, 11:50:10 PM

I was able to upload a PowerPlay table to Vega 56/64 on the fly.
No more messy registry hacks. This is getting interesting...

The new miner also has a (somewhat clumsy) way to disable some of
the compute units for reduced power consumption.
I need to run more experiments with that, though.

By the way, my new Ethash kernel made it into ethminer.
(I had no idea for weeks.

It seems that my kernel is possibly the fastest Ethash kernel out there.
I tweaked it a bit, but I don't think it would get any faster than as it is now.

Which ethminer have your kernel?

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: Iamtutut on August 29, 2018, 03:00:40 PM

Quote from: dragonmike on August 29, 2018, 04:22:50 AM

Quote from: zawawa on August 28, 2018, 11:50:10 PM

I was able to upload a PowerPlay table to Vega 56/64 on the fly.
No more messy registry hacks. This is getting interesting...

The new miner also has a (somewhat clumsy) way to disable some of
the compute units for reduced power consumption.
I need to run more experiments with that, though.

By the way, my new Ethash kernel made it into ethminer.
(I had no idea for weeks.

It seems that my kernel is possibly the fastest Ethash kernel out there.
I tweaked it a bit, but I don't think it would get any faster than as it is now.

Double high-five!
Cheesy

Count me in, high 20.

YOUR FEET SMELL--

But your heart is in the right place. Thanks, Zawawa! --scryptr

Iamtutut

full member

Activity: 1120

Merit: 131

Quote from: dragonmike on August 29, 2018, 04:22:50 AM

Quote from: zawawa on August 28, 2018, 11:50:10 PM

I was able to upload a PowerPlay table to Vega 56/64 on the fly.
No more messy registry hacks. This is getting interesting...

The new miner also has a (somewhat clumsy) way to disable some of
the compute units for reduced power consumption.
I need to run more experiments with that, though.

By the way, my new Ethash kernel made it into ethminer.
(I had no idea for weeks.

It seems that my kernel is possibly the fastest Ethash kernel out there.
I tweaked it a bit, but I don't think it would get any faster than as it is now.

Double high-five!
Cheesy

Count me in, high 20.

laik2

sr. member

Activity: 652

Merit: 266

Quote from: zawawa on August 28, 2018, 11:50:10 PM

I was able to upload a PowerPlay table to Vega 56/64 on the fly.
No more messy registry hacks. This is getting interesting...

The new miner also has a (somewhat clumsy) way to disable some of
the compute units for reduced power consumption.
I need to run more experiments with that, though.

By the way, my new Ethash kernel made it into ethminer.
(I had no idea for weeks.

It seems that my kernel is possibly the fastest Ethash kernel out there.
I tweaked it a bit, but I don't think it would get any faster than as it is now.

Yeah, thank you for opensourcing it!
Getting faster now its not that important, getting to consume less power is.

dragonmike

hero member

Activity: 1274

Merit: 556

Quote from: zawawa on August 28, 2018, 11:50:10 PM

I was able to upload a PowerPlay table to Vega 56/64 on the fly.
No more messy registry hacks. This is getting interesting...

The new miner also has a (somewhat clumsy) way to disable some of
the compute units for reduced power consumption.
I need to run more experiments with that, though.

By the way, my new Ethash kernel made it into ethminer.
(I had no idea for weeks.

It seems that my kernel is possibly the fastest Ethash kernel out there.
I tweaked it a bit, but I don't think it would get any faster than as it is now.

Double high-five!
Cheesy

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

I was able to upload a PowerPlay table to Vega 56/64 on the fly.
No more messy registry hacks. This is getting interesting...

The new miner also has a (somewhat clumsy) way to disable some of
the compute units for reduced power consumption.
I need to run more experiments with that, though.

By the way, my new Ethash kernel made it into ethminer.
(I had no idea for weeks.

It seems that my kernel is possibly the fastest Ethash kernel out there.
I tweaked it a bit, but I don't think it would get any faster than as it is now.

MadKernel

newbie

Activity: 25

Merit: 1

Quote from: zawawa on August 24, 2018, 09:07:08 AM

Quote from: MadKernel on August 23, 2018, 03:14:50 PM

Quote from: zawawa on August 23, 2018, 01:51:43 PM

See, advertising your miner in my thread is the surest way to invite a really stiff competition, you know...

I Will opt, if possibile to start a cooperation 😁

This is how I perceive the Cryptoland. It's ll about Darwinian natural selection...

I don't see the Hokuto Man here Grin

, so you are really from Nippon.

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

Quote from: MadKernel on August 23, 2018, 03:14:50 PM

Quote from: zawawa on August 23, 2018, 01:51:43 PM

See, advertising your miner in my thread is the surest way to invite a really stiff competition, you know...

I Will opt, if possibile to start a cooperation 😁

This is how I perceive the Cryptoland. It's all about Darwinian natural selection...

MadKernel

newbie

Activity: 25

Merit: 1

Quote from: zawawa on August 23, 2018, 01:51:43 PM

See, advertising your miner in my thread is the surest way to invite a really stiff competition, you know...

I Will opt, if possibile to start a cooperation 😁

Iamtutut

full member

Activity: 1120

Merit: 131

Quote from: zawawa on August 23, 2018, 10:14:05 AM

The C++ version is coming along nicely, so I should be able to run some experiments with it pretty soon.
I might rewrite the kernel driver if I have time. Alternatively, I could poke atikmdag.sys.
It seems to have pretty interesting stuff in there...

I'll be so happy if you could release a separate fix for the atikmdag.sys, I can't use the GPU cores of my Ryzen 2400G because if I patch the driver to use my discrete GPUs for computing, the driver (18.6.1 with W10 X64 pro 1803) gives me a code 43 for the Ryzen Vega cores. Only solution is to use the Ryzen as a "microsoft basic video card" and it s**ks.

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

See, advertising your miner in my thread is the surest way to invite a really stiff competition, you know...

MadKernel

newbie

Activity: 25

Merit: 1

Quote from: dragonmike on August 23, 2018, 10:25:19 AM

Quote from: zawawa on August 23, 2018, 10:19:51 AM

Quote from: dragonmike on August 23, 2018, 06:40:29 AM

FYI, tried to use the "boost" function on my 6x Vega56 @ 64 rig (for x16r) but it crashed almost instantaneously. Tongue

Not going to try to "optimise" it for now either. It hashes fairly well (albeit at clocks that are lower than they should be, imo) so I'm not jeopardising the rig at the moment (until I'm certain the functionality works - afaik it's not officially supported on these Vega cards, especially on driver 18.6.1).

I know, I know... The plan with the C++ version is to outsource testing and optimizing default settings.
My sponsor set up a pretty interesting arrangement for the project. We will see.

Fingers crossed!
x16r is an algo I am sure you can still optimise greatly. The nVidia miners have gained about 50% speed over the last months, purely through software optimisation. [subtle hint]Currently my 6x Vega rig hashes slower than my 5x GTX1080 rig.[/subtle hint]

Wink

I suggest you to test the version 06x of my miner...20% improvement than other implementations.

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

Quote from: dragonmike on August 23, 2018, 10:25:19 AM

Quote from: zawawa on August 23, 2018, 10:19:51 AM

Quote from: dragonmike on August 23, 2018, 06:40:29 AM

FYI, tried to use the "boost" function on my 6x Vega56 @ 64 rig (for x16r) but it crashed almost instantaneously. Tongue

Not going to try to "optimise" it for now either. It hashes fairly well (albeit at clocks that are lower than they should be, imo) so I'm not jeopardising the rig at the moment (until I'm certain the functionality works - afaik it's not officially supported on these Vega cards, especially on driver 18.6.1).

I know, I know... The plan with the C++ version is to outsource testing and optimizing default settings.
My sponsor set up a pretty interesting arrangement for the project. We will see.

Fingers crossed!
x16r is an algo I am sure you can still optimise greatly. The nVidia miners have gained about 50% speed over the last months, purely through software optimisation. [subtle hint]Currently my 6x Vega rig hashes slower than my 5x GTX1080 rig.[/subtle hint]

Wink

I haven't done any optimizations for X16R, so I don't think it is difficult to catch up with CUDA miners.
I have to take care of other algorithms first, though: CryptoNightHeavy, Lyra2REv2, NeoScrypt, and CryptoNightV7.

dragonmike

hero member

Activity: 1274

Merit: 556

Quote from: zawawa on August 23, 2018, 10:19:51 AM

Quote from: dragonmike on August 23, 2018, 06:40:29 AM

FYI, tried to use the "boost" function on my 6x Vega56 @ 64 rig (for x16r) but it crashed almost instantaneously. Tongue

Not going to try to "optimise" it for now either. It hashes fairly well (albeit at clocks that are lower than they should be, imo) so I'm not jeopardising the rig at the moment (until I'm certain the functionality works - afaik it's not officially supported on these Vega cards, especially on driver 18.6.1).

I know, I know... The plan with the C++ version is to outsource testing and optimizing default settings.
My sponsor set up a pretty interesting arrangement for the project. We will see.

Fingers crossed!
x16r is an algo I am sure you can still optimise greatly. The nVidia miners have gained about 50% speed over the last months, purely through software optimisation. [subtle hint]Currently my 6x Vega rig hashes slower than my 5x GTX1080 rig.[/subtle hint]

Wink

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

Quote from: dragonmike on August 23, 2018, 06:40:29 AM

FYI, tried to use the "boost" function on my 6x Vega56 @ 64 rig (for x16r) but it crashed almost instantaneously. Tongue

Not going to try to "optimise" it for now either. It hashes fairly well (albeit at clocks that are lower than they should be, imo) so I'm not jeopardising the rig at the moment (until I'm certain the functionality works - afaik it's not officially supported on these Vega cards, especially on driver 18.6.1).

I know, I know... The plan with the C++ version is to outsource testing and optimizing default settings.
My sponsor set up a pretty interesting arrangement for the project. We will see.

zawawa

sr. member

Activity: 728

Merit: 304

Miner Developer

Quote from: ?? on ??

Quote from: zawawa on August 03, 2018, 01:34:27 PM

Quote from: ?? on ??

Quote from: zawawa on August 02, 2018, 01:31:42 PM

Quote from: ?? on ??

Quote from: zawawa on July 30, 2018, 05:55:29 PM

Quote from: KL0nLutiy1 on July 30, 2018, 09:46:27 AM

Quote from: KL0nLutiy1 on July 11, 2018, 04:47:56 PM

Hi, @zawawa, any update on vega timings modding?

@zawawa, could you please answer?

I cannot give you details right now, but it seems like timing mods on Vega seem a whole lot harder than those on pre-Vega ASIC's.

Somewhat. There's perks, too.

That's at least an insider told me.
On a related note, I meant to ask you if you figured out how to modify microcodes for memory controllers.
That would be an ultimate hack...

Heh, hilarious tidbit. I *think* I can. But since I've no fucking clue what it wants to read...

Those microcodes were what I was trying to figure out before I started working on the C++ version of Gateless Gate Sharp.
If you haven't figured them out yet, I think it is safe to assume nobody else did...

Actually, the microcode to this, I'm beginning to believe is irrelevant. Unless it's designed how I think, and the SOCCLK is referring to the same thing being used to control the memory, or at least affecting it. Unexplained hashrate drops with low SOCCLK are what I'm hearing from others, on algos which are mostly memory-based. (Yes, I know exactly what this component is; no, not saying. It does bug me, though, why they would do it like this. Just odd.)

The C++ version is coming along nicely, so I should be able to run some experiments with it pretty soon.
I might rewrite the kernel driver if I have time. Alternatively, I could poke atikmdag.sys.
It seems to have pretty interesting stuff in there...

dragonmike

hero member

Activity: 1274

Merit: 556

FYI, tried to use the "boost" function on my 6x Vega56 @ 64 rig (for x16r) but it crashed almost instantaneously. Tongue

Not going to try to "optimise" it for now either. It hashes fairly well (albeit at clocks that are lower than they should be, imo) so I'm not jeopardising the rig at the moment (until I'm certain the functionality works - afaik it's not officially supported on these Vega cards, especially on driver 18.6.1).

Topic: Gateless Gate Sharp 1.3.8: 30Mh/s (Ethash) on RX 480! - page 4. (Read 214425 times)