CCminer(SP-MOD) Modded GPU kernels.

[Genoil]

Whirlpoolx is almost unique in how it was conceived: it really looks like it was made to favourite smart developers.
The large part of the optimisations come from "shortcuts", more than general algorithm speedup. In the end it's just reiterated whirlpool (which in turn is similar to groestl and other aes derived algos).
So I guess that a 970 with all the "shortcuts" in place, should be about as good as a 280x (500 Mh/s).

Thats why x11 is safer. Supported by Daesh..
I wonder why the poor NVIDIA miners are skipping the ETHER algo... Such Profit loss

my GTX 980 is hashing at 22mh/s which is not very interesting (less than a 280x). I am mining spreadcoin with your private miner, more profitable. Or sia (but without a pool and with their difficulty adjustement bitcoin like, this is only interesting from time to time : there are some big farms which jumps in and out giving incredible volatile hashrate).

I think an optimized ether ccminer/ethminer would be highly welcome : would need about 30mh/s for a GTX 980.

The maximum theoretical eth hash rate for GTX980 on stock clocks is 28MH/s (== 224GB/s / 8KB/hash). But even if the Keccak stages would be fully integrated, you would only win 15%, so it would max out at about 25MH.

[serg_25]

SP_ what 750 ti card model you have? (Link on ebay/amazon please)

I have checked power consumptuon on my gigabyte card and at 1420core/3300 mem it get 44-45watt. But at that clock it can make only 4.6mh on lyra.
Also i checked with -X1 ... -X17 and found that power consumption differs only 1-2 watt.

[ol92]

Whirlpoolx is almost unique in how it was conceived: it really looks like it was made to favourite smart developers.
The large part of the optimisations come from "shortcuts", more than general algorithm speedup. In the end it's just reiterated whirlpool (which in turn is similar to groestl and other aes derived algos).
So I guess that a 970 with all the "shortcuts" in place, should be about as good as a 280x (500 Mh/s).

Thats why x11 is safer. Supported by Daesh..
I wonder why the poor NVIDIA miners are skipping the ETHER algo... Such Profit loss

my GTX 980 is hashing at 22mh/s which is not very interesting (less than a 280x). I am mining spreadcoin with your private miner, more profitable. Or sia (but without a pool and with their difficulty adjustement bitcoin like, this is only interesting from time to time : there are some big farms which jumps in and out giving incredible volatile hashrate).

I think an optimized ether ccminer/ethminer would be highly welcome : would need about 30mh/s for a GTX 980.

The maximum theoretical eth hash rate for GTX980 on stock clocks is 28MH/s (== 224GB/s / 8KB/hash). But even if the Keccak stages would be fully integrated, you would only win 15%, so it would max out at about 25MH.

So mem bandwith is the limiting factor here ? Mining use P2 state for memory : 3000mhz : therefore overclocking memory/forcing P0 state would gain a lot?

On the other hand, I have tried with a GTX 980 ti (>+50% bandwith) and I have a very small gain in ethminer cuda...

[djm34]

Whirlpoolx is almost unique in how it was conceived: it really looks like it was made to favourite smart developers.
The large part of the optimisations come from "shortcuts", more than general algorithm speedup. In the end it's just reiterated whirlpool (which in turn is similar to groestl and other aes derived algos).
So I guess that a 970 with all the "shortcuts" in place, should be about as good as a 280x (500 Mh/s).

Thats why x11 is safer. Supported by Daesh..
I wonder why the poor NVIDIA miners are skipping the ETHER algo... Such Profit loss

my GTX 980 is hashing at 22mh/s which is not very interesting (less than a 280x). I am mining spreadcoin with your private miner, more profitable. Or sia (but without a pool and with their difficulty adjustement bitcoin like, this is only interesting from time to time : there are some big farms which jumps in and out giving incredible volatile hashrate).

I think an optimized ether ccminer/ethminer would be highly welcome : would need about 30mh/s for a GTX 980.

The maximum theoretical eth hash rate for GTX980 on stock clocks is 28MH/s (== 224GB/s / 8KB/hash). But even if the Keccak stages would be fully integrated, you would only win 15%, so it would max out at about 25MH.

So mem bandwith is the limiting factor here ? Mining use P2 state for memory : 3000mhz : therefore overclocking memory/forcing P0 state would gain a lot?

On the other hand, I have tried with a GTX 980 ti (>+50% bandwith) and I have a very small gain in ethminer cuda...

you'll gain a bit but not a lot.

[hashbrown9000]

submitted a pull request for fixing syslog output.  Helps with kopiemtu's stats webpage and external LED displays created by @induktor.

[bathrobehero]

So mem bandwith is the limiting factor here ? Mining use P2 state for memory : 3000mhz : therefore overclocking memory/forcing P0 state would gain a lot?

On the other hand, I have tried with a GTX 980 ti (>+50% bandwith) and I have a very small gain in ethminer cuda...

As far as I know memory access is the limiting factor and not bandwidth. Unfortunately tweaking memory timings is difficult though.

"Time to submit a share" doesn't depend on diff.
Nor does time to compute. Higher diff just means "less likely to find a share but with more value", so statistically speaking there is no disadvantage in higher diff.

I know what I missed now and yeah, it shouldn't be a problem on modern pools.

[myagui]

About memory intensive algos: any brainiac tried a LUT approach yet? In short, reducing memory access and doing additional compute work instead... (IIRC, the prime example of this is with scrypt-jane, and possibly some other flavours in cudaminer).

ETH can probably run much faster on the high end cards if some of the memory access can be replaced with computation.

[zTheWolfz]

submitted a pull request for fixing syslog output.  Helps with kopiemtu's stats webpage and external LED displays created by @induktor.

Nice, computer porn! Thanks for sharing!  

[sp_]

SP_ what 750 ti card model you have? (Link on ebay/amazon please)
I have checked power consumptuon on my gigabyte card and at 1420core/3300 mem it get 44-45watt. But at that clock it can make only 4.6mh on lyra.
Also i checked with -X1 ... -X17 and found that power consumption differs only 1-2 watt.

I have all the cards.. It's like the house of cards...

The lyra algoritm is already using bandwitdh. The -X doesn't make any difference

[sp_]

ETH can probably run much faster on the high end cards if some of the memory access can b replaced with computation.

The ET performance depends on the DAG  file.. Without a random file, every random acces isside the cached and the hashrate explodes

[djm34]

ETH can probably run much faster on the high end cards if some of the memory access can b replaced with computation.

The ET performance depends on the DAG  file.. Without a random file, every random acces isside the cached and the hashrate explodes

duh ?! there is no way to get rid of those... otherwise,as Myaguy was explaining, it may possible to calculate some of the dag element hence reducing memory usage

[serg_25]

SP_ what 750 ti card model you have? (Link on ebay/amazon please)
I have checked power consumptuon on my gigabyte card and at 1420core/3300 mem it get 44-45watt. But at that clock it can make only 4.6mh on lyra.
Also i checked with -X1 ... -X17 and found that power consumption differs only 1-2 watt.

I have all the cards.. It's like the house of cards...

The lyra algoritm is already using bandwitdh. The -X doesn't make any difference

Sorry i mean lyra2v2 algo. you have asked previously for power consumption with different -X parameters.

About all cards - nice, but i was interesting what card model can perform 5mhs at so low core 1388/2700?

[sp_]

SP_ what 750 ti card model you have? (Link on ebay/amazon please)
I have checked power consumptuon on my gigabyte card and at 1420core/3300 mem it get 44-45watt. But at that clock it can make only 4.6mh on lyra.
Also i checked with -X1 ... -X17 and found that power consumption differs only 1-2 watt.

I have all the cards.. It's like the house of cards...
The lyra algoritm is already using bandwitdh. The -X doesn't make any difference

Sorry i mean lyra2v2 algo. you have asked previously for power consumption with different -X parameters.
About all cards - nice, but i was interesting what card model can perform 5mhs at so low core 1388/2700?

This card:

http://www.amazon.co.uk/Graphics-Gainward-nVidia-GeForce-192-Bit/dp/B00IIMVBPA

windows 7 64bit (sp-mod 68 x86) latest drivers


Yes. thanks for testing the watt. From 44 watt to 42 watt is a 4.5% reduction of the powerbill.

[alz]

oops wrong thread 

[flipclip]

Nice display HB9K!

[hashbrown9000]

Nice display HB9K!

I cant' take credit for the added LED displays. Those are @Induktor's rigs.  However, outputting info to syslog (then outputting data thru a serial port) enables those displays to come alive.

If anyone else is interested, his schematics and build info is here:

https://litecointalk.org/index.php?topic=16800.msg277720#msg277720

[flipclip]

Nice display HB9K!

I cant' take credit for the added LED displays. Those are @Induktor's rigs.  However, outputting info to syslog (then outputting data thru a serial port) enables those displays to come alive.
If anyone else is interested, his schematics and build info is here:
https://litecointalk.org/index.php?topic=16800.msg277720#msg277720

Props to @Induktor then.  Great rig.  With a fourteen-segment display can we get a Vacuum Tube display next?

[hashbrown9000]

i think a Nixie display is better

[bensam1231]

Whirlpoolx is almost unique in how it was conceived: it really looks like it was made to favourite smart developers.
The large part of the optimisations come from "shortcuts", more than general algorithm speedup. In the end it's just reiterated whirlpool (which in turn is similar to groestl and other aes derived algos).
So I guess that a 970 with all the "shortcuts" in place, should be about as good as a 280x (500 Mh/s).

Thats why x11 is safer. Supported by Daesh..
I wonder why the poor NVIDIA miners are skipping the ETHER algo... Such Profit loss

I was mining Eth instead of Vert for awhile, but right now a 280x mines almost twice as fast as 970, so after all the AMD miners hoped on board it was no longer all that profitable. It's about break even with Vert right now though, still sometimes more profitable, regardless of it being slower.

ETH can probably run much faster on the high end cards if some of the memory access can b replaced with computation.

The ET performance depends on the DAG  file.. Without a random file, every random acces isside the cached and the hashrate explodes

duh ?! there is no way to get rid of those... otherwise,as Myaguy was explaining, it may possible to calculate some of the dag element hence reducing memory usage

Have you guys considered more compression?

i think a Nixie display is better

Nixie tubes are super expensive, but look cool.

[theotherme]

Whirlpoolx is almost unique in how it was conceived: it really looks like it was made to favourite smart developers.
The large part of the optimisations come from "shortcuts", more than general algorithm speedup. In the end it's just reiterated whirlpool (which in turn is similar to groestl and other aes derived algos).
So I guess that a 970 with all the "shortcuts" in place, should be about as good as a 280x (500 Mh/s).

Thats why x11 is safer. Supported by Daesh..
I wonder why the poor NVIDIA miners are skipping the ETHER algo... Such Profit loss

I was mining Eth instead of Vert for awhile, but right now a 280x mines almost twice as fast as 970, so after all the AMD miners hoped on board it was no longer all that profitable. It's about break even with Vert right now though, still sometimes more profitable, regardless of it being slower.

ETH can probably run much faster on the high end cards if some of the memory access can b replaced with computation.

The ET performance depends on the DAG  file.. Without a random file, every random acces isside the cached and the hashrate explodes

duh ?! there is no way to get rid of those... otherwise,as Myaguy was explaining, it may possible to calculate some of the dag element hence reducing memory usage

Have you guys considered more compression?

thinking about it, but it isn't really obvious it would help... need to find a compression algorithm which works on it (I tried zipping the file, just to have an idea of what could be done and ended up with a larger file...   ) so it seems there isn't much to compress (well 1.2Gb of integers...). For info, for a 780ti to work correctly on windows 8.1, only 1.3% less dag is required (but that 1.3% is still too large to store anywhere that amount of data (too large for either register or texture/shared )