Bitcoin Forum

Alternate cryptocurrencies => Mining (Altcoins) => Topic started by: phobosq on November 02, 2017, 01:21:28 PM



Title: 1070 Ti benchmarked in Eth & Zec
Post by: phobosq on November 02, 2017, 01:21:28 PM
First test results can be found here - https://ithardware.pl/testyirecenzje/test_msi_geforce_gtx_1070_ti_gaming-4244-13.html (only in Polish, although charts speak for themselves). Results are inconclusive (a bit too low <30 MH/s ETH, <500 sol/s ZEC), but see for yourself.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: L4mie on November 02, 2017, 02:02:53 PM
Thank You for the fast test.  ;)
Sadly the results are a little bit disappointing.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: shibob on November 02, 2017, 02:06:21 PM
First test results can be found here - https://ithardware.pl/testyirecenzje/test_msi_geforce_gtx_1070_ti_gaming-4244-13.html (only in Polish, although charts speak for themselves). Results are inconclusive (a bit too low <30 MH/s ETH, <500 sol/s ZEC), but see for yourself.

It doesnt make sense for me why 1070Ti only hash <30Mh/s while 1070 could achieve 32Mh/s?


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: phobosq on November 02, 2017, 02:28:59 PM
Probably mems are limiting in ETH and extra CUDA cores don't help, especially if their 1070 were able to reach +1000 on mems, as their 1070 ti sample could do only +500.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Oakey22 on November 02, 2017, 02:36:44 PM
It's not looking good for the card, I will be getting some tomorrow so will post up some more info when I get them


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: gt_addict on November 02, 2017, 03:28:03 PM
So whats limiting it exactly? Can someone give a detailed explanation?  Ive got one on order but if its no better than the standard 1070 ill save some money.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: mojoxc on November 02, 2017, 03:38:41 PM
Not sure the results everyone wanted when the performance is limited by the memory type and bandwidth.  The reason you see a huge jump in the 1080ti ddr5x vs 1080 ddr5 on zec the change in memory type.  The 1070 vs 1070 ti use the same type of memory and we were already at the limit of the memory throughput for the stock timings on the ram therefore there is no significant change in speed.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: RealSwissMiner on November 02, 2017, 03:46:38 PM
I will also receive one tomorrow. I thought it would do better at equihash. But anyway the price was similar to a regular 1070 - they incresed the price today so Win-win anyway.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Undefined31415 on November 02, 2017, 04:00:31 PM
Hmm, so far the 1070ti doesn't seem worth it, unless Nvidia can pull something off with driver updates, which seems doubtful, but is certainly not impossible.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: lunobird on November 02, 2017, 04:40:23 PM
Yep I'm so glad I cancelled my 1070ti pre orders. I'm Going back to getting 1080 ti instead


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: zorvalth on November 02, 2017, 04:45:21 PM
Not sure the results everyone wanted when the performance is limited by the memory type and bandwidth.  The reason you see a huge jump in the 1080ti ddr5x vs 1080 ddr5 on zec the change in memory type.  The 1070 vs 1070 ti use the same type of memory and we were already at the limit of the memory throughput for the stock timings on the ram therefore there is no significant change in speed.

Funny, last time i checked 1080 always is been ddr5x :)


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: soothaa on November 02, 2017, 04:47:20 PM
41MHs for 1080 Ti on ETH? How? I can't even get 35..


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: flip4flop on November 02, 2017, 05:02:37 PM
Wow no 1070ti for me.  I am using the 1070 MSI Gaming and am pulling 480 sols/s with +150 core and +600 memory with Micron memory. The 1070 ti doesnt seem like it can even meet that threshold.  Unless there is some drastic difference with OCing which I dont think is possible it seems the 1070 is still king out of the 1070, 1070 ti, and the 1080.  Looks like I will seriously have to consider moving up to the 1080 ti this holiday season to swap out with one or two of my 1070s depending on how much my PSU can handle.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: butmuncher on November 02, 2017, 05:04:32 PM
440 sol 1070 , 473 sol 1070 ti, both in same system at stock
Eth speeds are the same for me.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Kompik on November 02, 2017, 07:11:59 PM
Bad test, using outdated miners. 1070 can do 485 sols @135 watt and almost all cards can go over 500 on max settings. I expect the 1070ti to go over 540 at worse.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: phobosq on November 02, 2017, 07:29:08 PM
Claymore 10.1 and EWBF 0.3.4 outdated? At 100% 1070 Ti did around 520 sol/s.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Cereberus on November 02, 2017, 08:04:13 PM
Very early conclusion. Nvidia is screwing up lately with all these new drivers, wait until they produce a good driver like in the old days and I am pretty sure the GTX 1070 ti can go over 520 sol/s easily and I expect it to reach about 35 Mhash in Ethereum. All those that canceled their pre order just because of "some test" made an error according to my opinion.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: lunobird on November 02, 2017, 08:50:42 PM
Very early conclusion. Nvidia is screwing up lately with all these new drivers, wait until they produce a good driver like in the old days and I am pretty sure the GTX 1070 ti can go over 520 sol/s easily and I expect it to reach about 35 Mhash in Ethereum. All those that canceled their pre order just because of "some test" made an error according to my opinion.

I canceled my 1070 ti pre order.  No regrets at all the 1070 ti is a big let down.  Getting the big mac daddy 1080 ti.  "Proven" to get the most hash per dollar spent with least amount of electricity used and less overall components(risers) to get the desired hashrate.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Oakey22 on November 02, 2017, 09:53:21 PM
I just hope nvidia come up with something special, if not i will be after getting some of the 1080ti to complement these or my other cards.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: zorvalth on November 02, 2017, 10:07:27 PM
whats not to like 1070ti? perfect card it can mine dagger at 90watts with 32mh/s and mine all shitcoins algos (nist5, skunk, phi, neoscrypt, skein...) almost as fast as 1080. Its the perfect mining card. You can mine everything with it and its dead easy to cool. Cant imagine better card for mining. And I got almost evry card out there since 280x...


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jb7032 on November 02, 2017, 10:23:33 PM
Just found a fairly detailed review. Hopefully more to come.
http://www.legitreviews.com/nvidia-geforce-gtx-1070-ti-ethereum-mining-performance_199622


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: lunobird on November 02, 2017, 11:17:58 PM
Just found a fairly detailed review. Hopefully more to come.
http://www.legitreviews.com/nvidia-geforce-gtx-1070-ti-ethereum-mining-performance_199622


Yep what a let down,  Way more cost effective to just buy more regular 1070 gtx instead of the TI if doing ethereum.



Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 03, 2017, 02:07:42 AM
I currently have 1060s, 1070s, 1080s, and one 1080Ti.

I pre-ordered a few 1070-Ti.

I mine on ZPOOL on profit switching so... I think I will make a sheet showing the ZPOOL algos on each card vs. the cost... for my own reasons of course but I will share it when complete.

It may have a few specialties... for example the 1080s are great on a ton of algos where most folks don't consider them since they are bad at ETH.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: CjMapope on November 03, 2017, 02:15:38 AM
I currently have 1060s, 1070s, 1080s, and one 1080Ti.

I pre-ordered a few 1070-Ti.

I mine on ZPOOL on profit switching so... I think I will make a sheet showing the ZPOOL algos on each card vs. the cost... for my own reasons of course but I will share it when complete.

It may have a few specialties... for example the 1080s are great on a ton of algos where most folks don't consider them since they are bad at ETH.

agreed, for example, it would be interesting to see how the 1070ti does on LBRY algo.

Shit man, these benchs dont look good, i was really hoping the extra CUDA cores here would give a a solid mining boost, but i guess the mem bandwidth and such maxes out before those cores are even utilized fully i bet :(
Bah, back to 1080ti till Volta i guess for me too... maybe the 1070ti A8G's i ordered will be better... maybe.. ;p


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Cereberus on November 03, 2017, 09:17:17 AM
Very early conclusion. Nvidia is screwing up lately with all these new drivers, wait until they produce a good driver like in the old days and I am pretty sure the GTX 1070 ti can go over 520 sol/s easily and I expect it to reach about 35 Mhash in Ethereum. All those that canceled their pre order just because of "some test" made an error according to my opinion.

I canceled my 1070 ti pre order.  No regrets at all the 1070 ti is a big let down.  Getting the big mac daddy 1080 ti.  "Proven" to get the most hash per dollar spent with least amount of electricity used and less overall components(risers) to get the desired hashrate.

We are talking about 1070 ti here and if we compare it with 1080 ti is an error, the 1080 ti costs the double of 1070 ti and 2x 1070 ti will provide us better hashrate and the same consumption of electricity at 180 watt both cards when mining Ethereum. Surely the hashrate of 2x1070 ti is better than 1x1080 ti both in Ethereum, Zcash and every other coin. I remain of the opinion that 1070 ti is one perfect card that just needs a new driver which I am sure will become available really soon.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: zorvalth on November 03, 2017, 09:29:53 AM
Very early conclusion. Nvidia is screwing up lately with all these new drivers, wait until they produce a good driver like in the old days and I am pretty sure the GTX 1070 ti can go over 520 sol/s easily and I expect it to reach about 35 Mhash in Ethereum. All those that canceled their pre order just because of "some test" made an error according to my opinion.

I canceled my 1070 ti pre order.  No regrets at all the 1070 ti is a big let down.  Getting the big mac daddy 1080 ti.  "Proven" to get the most hash per dollar spent with least amount of electricity used and less overall components(risers) to get the desired hashrate.

We are talking about 1070 ti here and if we compare it with 1080 ti is an error, the 1080 ti costs the double of 1070 ti and 2x 1070 ti will provide us better hashrate and the same consumption of electricity at 180 watt both cards when mining Ethereum. Surely the hashrate of 2x1070 ti is better than 1x1080 ti both in Ethereum, Zcash and every other coin. I remain of the opinion that 1070 ti is one perfect card that just needs a new driver which I am sure will become available really soon.

1080ti is 50-60% more expensive then 1070ti, not 100%.  Hashrate is pretty much proportional on the price because the cuda cores are also proportional on the price. 1080 ti takes less space, but 1070 ti is way easier to keep cool.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Oakey22 on November 03, 2017, 11:24:30 AM
Very early conclusion. Nvidia is screwing up lately with all these new drivers, wait until they produce a good driver like in the old days and I am pretty sure the GTX 1070 ti can go over 520 sol/s easily and I expect it to reach about 35 Mhash in Ethereum. All those that canceled their pre order just because of "some test" made an error according to my opinion.

I canceled my 1070 ti pre order.  No regrets at all the 1070 ti is a big let down.  Getting the big mac daddy 1080 ti.  "Proven" to get the most hash per dollar spent with least amount of electricity used and less overall components(risers) to get the desired hashrate.

We are talking about 1070 ti here and if we compare it with 1080 ti is an error, the 1080 ti costs the double of 1070 ti and 2x 1070 ti will provide us better hashrate and the same consumption of electricity at 180 watt both cards when mining Ethereum. Surely the hashrate of 2x1070 ti is better than 1x1080 ti both in Ethereum, Zcash and every other coin. I remain of the opinion that 1070 ti is one perfect card that just needs a new driver which I am sure will become available really soon.

1080ti is 50-60% more expensive then 1070ti, not 100%.  Hashrate is pretty much proportional on the price because the cuda cores are also proportional on the price. 1080 ti takes less space, but 1070 ti is way easier to keep cool.

I think the pick of the GPUs at the moment is still the 1070, no reason to go for the ti model at the moment unless Nvidia pull something out the bag with the drivers.



Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: azertyuiopvnx on November 03, 2017, 11:28:21 AM
440 sol 1070 , 473 sol 1070 ti, both in same system at stock
Eth speeds are the same for me.

Thank you for your bench.
I'll cancel my pre-orders of 1070 ti and buy some more 1070 (450 euros/gtx 1070 ti vs 380 euros/gtx 1070)


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Pils on November 03, 2017, 04:09:17 PM
EVGA 1070 Ti SC  75% PL (135w) +175 core + 550 mem (micron) 520 sols using DSTM's miner.

In comparison to:

MSI 1070 Armor 50% PL (98w) +160 core +350 mem (samsung) 430 sols using DSTM's miner (my best 1070, silicon lottery winner)
Palit 1080 Dual OC 70% PL (145w) +190 core +150 mem 535 sols using DSTM's miner.
EVGA 1080Ti FE 70% PL (180W) +200core +350mem 720 sols using DSTM's miner.

Neoscrypt appears to be semi-broken though, at least on windows 7. The new 1070ti driver has halved speed here on all my cards, the 1070ti is only doing 750kh. My 1070s now do 660kh instead of 1250kh, so by that maths the 1070ti producing 750kh would be doing about 1550kh on a working driver. Had to put the 1070ti's in with the 1080ti doing equihash and roll back drivers on my 1070 rig.

I also have one dodgy card (out of 6). Running +13c compared to the others, will have to disassemble later.

Not great, but not exactly a disaster either. Fills the void between the 1070 and the 1080ti better than the 1080 did anyway, and in the UK offers better hash/ than the 1080ti.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: shibob on November 04, 2017, 06:06:27 AM
EVGA 1070 Ti SC  75% PL (135w) +175 core + 550 mem (micron) 520 sols using DSTM's miner.

520sol/s at 75% power limit is really great. Btw, have you been tested it with ETH mining?


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Bulletdodger on November 04, 2017, 12:21:22 PM
Disappointing.

I used to get 490+ stable on some models of 1070s (equihash), sometimes even hit 500, if cooled properly.
Ti version has almost no advancement over the original which is over year and a half old.



Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jyakulis on November 04, 2017, 12:31:27 PM
EVGA 1070 Ti SC  75% PL (135w) +175 core + 550 mem (micron) 520 sols using DSTM's miner.

520sol/s at 75% power limit is really great. Btw, have you been tested it with ETH mining?

Interesting. I'm running regular EVGA 1070 SC's at 80% power limit and it is a very stable 445 sol/sec. So, like a 15% improvement. Not bad. Disappointing about eth though. Always nice to have that option.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: BlockMedx on November 04, 2017, 01:03:03 PM
Slightly disappointing unfortunately.  I think people should be sticking to the 1080ti for ETH and ZEC.  The 1070ti just isn't cutting it, apparently


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: bstewart on November 04, 2017, 01:04:06 PM
Hmm. Didnt look all that great to me.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jimlite on November 04, 2017, 07:00:53 PM
Slightly disappointing unfortunately.  I think people should be sticking to the 1080ti for ETH and ZEC.  The 1070ti just isn't cutting it, apparently
How is it not cutting it? Are you insane? It cost $40 more than a normal 1070, it has the same memory as a 1070 so does the same ETH hashrate. It has more compute power so will be faster in EVERY other mining algo other than ETH and use almost exactly the same power watts. It is a dream mining card for anyone who pays over .10 c/kwh


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: QuintLeo on November 04, 2017, 07:04:42 PM
Slightly disappointing unfortunately.  I think people should be sticking to the 1080ti for ETH and ZEC.  The 1070ti just isn't cutting it, apparently

 The 1080ti is a BAD choice for ETH, but a decent one for ZEC.
 WAY too expen$ive for the 35ish Mhash it can manage on ETH, when AMD RX 470/480/570/580 cards can pull more than DOUBLE the hash/$ at recent pricing.

 1070ti is a GOOD ZEC card, in my testing of various cards it achieves THE highest efficiency (by a little) of any card in existance - though you can't build rigs as DENSE as you can with the 1080 ti.

 


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: dandreark1 on November 04, 2017, 07:22:57 PM
tune a 1080ti the right way and it makes these cards look dumb. Doesn't make sense to me, it's like pissing in the wind..



Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Marvell2 on November 04, 2017, 08:34:10 PM
whats not to like 1070ti? perfect card it can mine dagger at 90watts with 32mh/s and mine all shitcoins algos (nist5, skunk, phi, neoscrypt, skein...) almost as fast as 1080. Its the perfect mining card. You can mine everything with it and its dead easy to cool. Cant imagine better card for mining. And I got almost evry card out there since 280x...

best cardright now is the vega 56 costs under 400

mines eth at 42 mhs 140 watt, zec at 500 , cryptonight at 1500

it smokes the 1070ti


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jimlite on November 04, 2017, 09:10:38 PM
tune a 1080ti the right way and it makes these cards look dumb. Doesn't make sense to me, it's like pissing in the wind..



A 1080ti is $800, a 1070ti is $450. Two 1070tis get WAY MORE HASH for about the same watts as a 1080 ti. You need to mine for like a year to just ROI on an $800 gpu, ROIing on a $450 gpu is a little quicker.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: lunobird on November 04, 2017, 11:36:15 PM
tune a 1080ti the right way and it makes these cards look dumb. Doesn't make sense to me, it's like pissing in the wind..



A 1080ti is $800, a 1070ti is $450. Two 1070tis get WAY MORE HASH for about the same watts as a 1080 ti. You need to mine for like a year to just ROI on an $800 gpu, ROIing on a $450 gpu is a little quicker.

Stop inflating those numbers. 1080 ti is $730 msi gaming (decent quality)
https://www.amazon.com/MSI-GTX-1080-TI-GAMING/dp/B06XVG7M23/ref=sr_1_6?s=pc&ie=UTF8&qid=1509838282&sr=1-6&keywords=1080+ti&dpID=516O0cdlnYL&preST=_SX300_QL70_&dpSrc=srch

1070 ti of same quality is $470
https://www.amazon.com/EVGA-GeForce-Scanner-Graphics-08G-P4-5671-KR/dp/B076S4RH6K/ref=sr_1_2?s=electronics&ie=UTF8&qid=1509838319&sr=1-2&keywords=1070+ti+gaming&dpID=51PLpT-02fL&preST=_SY300_QL70_&dpSrc=srch

1080 ti is only  36 percent more expensive and im sure it hashes at least 36 percent better than the 1070 ti if not better for equihash.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: otto_diesel on November 05, 2017, 05:40:06 AM
nice test thank you

but i dont get why 1070 ti worse than 1070 ? because gtc ti gpu's are better than normal gtx, it just    30-32MH/s   450-465 Sol/s. the difference is very little. it makes 1070 ti worse than 1070  if compared  the price/performance criterion.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Estomid on November 05, 2017, 08:59:53 AM
Slightly disappointing unfortunately.  I think people should be sticking to the 1080ti for ETH and ZEC.  The 1070ti just isn't cutting it, apparently

 The 1080ti is a BAD choice for ETH, but a decent one for ZEC.
 WAY too expen$ive for the 35ish Mhash it can manage on ETH, when AMD RX 470/480/570/580 cards can pull more than DOUBLE the hash/$ at recent pricing.

 1070ti is a GOOD ZEC card, in my testing of various cards it achieves THE highest efficiency (by a little) of any card in existance - though you can't build rigs as DENSE as you can with the 1080 ti.

 

I agree the 1080Ti is still the best for ZEC mining if you consider the total rig cost.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jimlite on November 05, 2017, 05:49:34 PM
lunobird,
If all you are mining is ZEC and your electric cost is very low, then maybe 1080 ti makes more sense then 1070 ti.
But for ETH the 1070ti is same speed due to lower latency memory which eth likes. And if you are going to be mining many different algos, the 1070ti is the most efficient and diverse card. I own a couple 1070, 1080 and 1080ti cards and know that 1080 sucks for eth due to its memory, the 1070 is faster at eth. The biggest issue is the 1080ti is about $250 more than a 1070ti for the almost the same performance in every algo but zec, or a little slower, but faster at eth. The HUGE issue is that any of these 3 gpus will make you about $1.50-$2.50/day, then subtract energy costs of $.50-$1/day and you get a profit of $1-$1.50/day per gpu. So it would take 200 to 250 days to make up the difference in cost between the 1070ti and 1080ti, that is why I hardly think a 1080ti is a good choice unless you are slot or motherboard limited. A 1080ti is going to take over a year just to ROI itself.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: umine on November 05, 2017, 06:27:30 PM
According the results GTX1080 seems better than GTX1070Ti for core based algos like Equihash. The price of both a practically the same, power consumption also the same. Now my choise is GTX1080 not 1070Ti


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jimlite on November 05, 2017, 07:00:37 PM
But 1070ti will be better at memory based algos like ETH / dagger than 1080. So the 1070ti is still the best all around every algo card for the price and watts. Yes if you are ONLY mining ETH go for 1070 or 1070ti, if you are only mining zec go for 1080 or 1080ti, but if you are mining many different algos over the next few years, 1070ti seems the best bet and value.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Bulletdodger on November 05, 2017, 07:30:47 PM
But 1070ti will be better at memory based algos like ETH / dagger than 1080. So the 1070ti is still the best all around every algo card for the price and watts. Yes if you are ONLY mining ETH go for 1070 or 1070ti, if you are only mining zec go for 1080 or 1080ti, but if you are mining many different algos over the next few years, 1070ti seems the best bet and value.

I think the improvement over the regular version is marginal.
Not an important car IMHO.

I'm sticking with 1080tis.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: QuintLeo on November 05, 2017, 11:13:30 PM
tune a 1080ti the right way and it makes these cards look dumb. Doesn't make sense to me, it's like pissing in the wind..


 I like my 1080 ti cards but show me ONE that can manage better than 4.2 efficiency - and the settings to get there.

 1080 ti is the hashrate king hands down (unless you want to spend crazy money for a Titan or a Tesla) but it's NOT the effciency king.

 Hash/$ at the SYSTEM level it's pretty close either way between the 1070, 1070 ti , 1080, and 1080 ti - unless NVidia decides it NEEDS to answer the recent Vega price drop and drops the 1070/1070ti/1080 prices to stay competative.



Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 08, 2017, 03:37:10 PM
I currently have 1060s, 1070s, 1080s, and one 1080Ti.

I pre-ordered a few 1070-Ti.

I mine on ZPOOL on profit switching so... I think I will make a sheet showing the ZPOOL algos on each card vs. the cost... for my own reasons of course but I will share it when complete.

It may have a few specialties... for example the 1080s are great on a ton of algos where most folks don't consider them since they are bad at ETH.

Hey guys,

I did some cost vs. hash power analysis of all my system types...

4x 1060
3x 1070
3x 1070 Ti (currently just one installed... I multiplied it out for the total system cost / production)
3x 1080

I do have one system with a single 1080 Ti that I'll throw in the mix later (it's at home so I don't have the data in front of me).

This is JUST *cost* per "work unit" (whatever that is per algo) -- not factoring in power cost (I'll add that in later when I have time).

Basically... the 1060 systems clean house EVEN when I figure in the cost of the computer itself in nearly every category in terms of up-front costs per work unit.

And a surprise... the 1070 Ti is a *better* value than the regular 1070 at current pricing on almost all algos (except Equihash). I listed the price I used per card which is the best I can find at this time per card type.

Now... I didn't agonize over optimized every card. I simply gave them all the same 85% power rate and some basic overclocking... I did this for my reference mostly as I wanted to be sure building with 1060s is what I wanted to keep doing (and it is).

https://image.ibb.co/fdedLw/GPU_Comparison.jpg


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: jimlite on November 09, 2017, 12:37:12 AM
Nice work, but that's a lot of data without much explanation. Correct me if I am wrong but buying a 1060 3gb for $200 or a 1060 6Gb for $300 is pretty pointless. It will not even do ETH or get terrible speeds due to not enough memory for DAG. I am sure it will also struggle on other memory needed algos like Keccak, ect. And bottom line is Why pay $325 for a 1060 6gb when you can just pay $460 for a 1070 Ti 8Gb which will CLEARLY give you way more hash (>25% more hash and card costs 25% more)? Plus it is more future proof. The only reason the 1060 looks cheaper to run is because it is a lot slower and uses less electricity. But even if you are paying high electric rates, a 1070Ti at like 70% power is only gonna pull 100-125w and give you like double the hash of a 1060 that is pulling about 75w or so. So Please explain how a 1060 rig makes any sense?

P.S. The ONLY algo that is not 2 to 3 times slower than a 1070ti or 1080/ti is ETH on a 6gb 1060. But you could buy one 1080Ti or two 1060 6gb cards for the same price and the 1080ti will beat both of the 1060's combined in every algo except ETH, then when ETH dag need more than 6 gb the 1060 will be useless. Unless you have a ton of motherboards, psus, cpus, ect just lying around and only a little cash to buy cards, the 1060 is a horrible choice, unless all you care about is ETH for a few months.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: kittiewat on November 09, 2017, 02:00:12 AM
nice test thank you

i think get 1070 in low price is better then?


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: phuocduong on November 09, 2017, 02:40:29 AM
i think better price/ speed than 1080


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 09, 2017, 03:11:58 AM
Nice work, but that's a lot of data without much explanation. Correct me if I am wrong but buying a 1060 3gb for $200 or a 1060 6Gb for $300 is pretty pointless. It will not even do ETH or get terrible speeds due to not enough memory for DAG. I am sure it will also struggle on other memory needed algos like Keccak, ect. And bottom line is Why pay $325 for a 1060 6gb when you can just pay $460 for a 1070 Ti 8Gb which will CLEARLY give you way more hash (>25% more hash and card costs 25% more)? Plus it is more future proof. The only reason the 1060 looks cheaper to run is because it is a lot slower and uses less electricity. But even if you are paying high electric rates, a 1070Ti at like 70% power is only gonna pull 100-125w and give you like double the hash of a 1060 that is pulling about 75w or so. So Please explain how a 1060 rig makes any sense?

P.S. The ONLY algo that is not 2 to 3 times slower than a 1070ti or 1080/ti is ETH on a 6gb 1060. But you could buy one 1080Ti or two 1060 6gb cards for the same price and the 1080ti will beat both of the 1060's combined in every algo except ETH, then when ETH dag need more than 6 gb the 1060 will be useless. Unless you have a ton of motherboards, psus, cpus, ect just lying around and only a little cash to buy cards, the 1060 is a horrible choice, unless all you care about is ETH for a few months.

I don't mine ETH at all anymore -- so that's off the table. I also don't buy 6GB cards anymore for this same reason (I do have some that I got before price inflation). I also only buy 1060 3GB that are $189.99-$199.99 per card.

I will include power data too soon as well to really get to the bottom of it & dig deeper.

Just keep in mind I have no agenda here -- I make nothing based on sharing my data... I am building a mining farm and I am *actually* buying hardware based on my data. I assume others may want to see it. As I collect more data I'll put it all out there =)


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 09, 2017, 03:13:04 AM
nice test thank you

i think get 1070 in low price is better then?

The 1070 would look way better at a regular cost... the 1070s I own I paid no more than $329 for them (the majority I paid $299). But it's impossible now to get that.

I would buy them again at $329 or less.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: VoskCoin on November 09, 2017, 03:33:17 AM
Very early conclusion. Nvidia is screwing up lately with all these new drivers, wait until they produce a good driver like in the old days and I am pretty sure the GTX 1070 ti can go over 520 sol/s easily and I expect it to reach about 35 Mhash in Ethereum. All those that canceled their pre order just because of "some test" made an error according to my opinion.

I canceled my 1070 ti pre order.  No regrets at all the 1070 ti is a big let down.  Getting the big mac daddy 1080 ti.  "Proven" to get the most hash per dollar spent with least amount of electricity used and less overall components(risers) to get the desired hashrate.
This might partially change your mind? 4.7 sols per watt O_o
https://youtu.be/jtp4plChU9Y
https://i.imgur.com/paGt3yym.png


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: pfft on November 09, 2017, 01:06:16 PM
should i buy 1070ti or 1080 ? same price in my country ?


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 09, 2017, 01:51:55 PM
Nice work, but that's a lot of data without much explanation. Correct me if I am wrong but buying a 1060 3gb for $200 or a 1060 6Gb for $300 is pretty pointless. It will not even do ETH or get terrible speeds due to not enough memory for DAG. I am sure it will also struggle on other memory needed algos like Keccak, ect. And bottom line is Why pay $325 for a 1060 6gb when you can just pay $460 for a 1070 Ti 8Gb which will CLEARLY give you way more hash (>25% more hash and card costs 25% more)? Plus it is more future proof. The only reason the 1060 looks cheaper to run is because it is a lot slower and uses less electricity. But even if you are paying high electric rates, a 1070Ti at like 70% power is only gonna pull 100-125w and give you like double the hash of a 1060 that is pulling about 75w or so. So Please explain how a 1060 rig makes any sense?

P.S. The ONLY algo that is not 2 to 3 times slower than a 1070ti or 1080/ti is ETH on a 6gb 1060. But you could buy one 1080Ti or two 1060 6gb cards for the same price and the 1080ti will beat both of the 1060's combined in every algo except ETH, then when ETH dag need more than 6 gb the 1060 will be useless. Unless you have a ton of motherboards, psus, cpus, ect just lying around and only a little cash to buy cards, the 1060 is a horrible choice, unless all you care about is ETH for a few months.

I don't mine ETH at all anymore -- so that's off the table. I also don't buy 6GB cards anymore for this same reason (I do have some that I got before price inflation). I also only buy 1060 3GB that are $189.99-$199.99 per card.

I will include power data too soon as well to really get to the bottom of it & dig deeper.

Just keep in mind I have no agenda here -- I make nothing based on sharing my data... I am building a mining farm and I am *actually* buying hardware based on my data. I assume others may want to see it. As I collect more data I'll put it all out there =)

So I've started the power data.

4x 1060 setup uses ~425 watts on the highest draw algo (at the wall).
3x 1070 Ti setup uses ~500 watts on the highest draw algo (at the wall).

So using WhatToMine right now... let's use Xevan on BitSend for example (my systems do that Algo alot)...

GTX 1060 System = $1609.40 / Year - $259.15 (power) = $1350.25
GTX 1070Ti System = $2383.30 / Year - $306.60 (power) = $2076.70
System Cost Difference = $578.69
GTX 1070Ti Extra Profit = $726.45
Time for 1070Ti system to win in profit = 290 days

So the 1070Ti system is going to win out with power to production over time -- at least based on the figures I measured with my quick & dirty method.

So next I will need to fully tweak both setups for max production per watt and see what I come up with. For my table I simply set them all at 85% TDP with a basic 100 core / 200 memory overclock.

I plan to use / re-sell my cards on a 1-year cycle as well. So when I go to refresh I may, in fact, switch to the bigger cards in more systems.

I will convert the table into something a bit more "in depth" once I am done collecting all the data together =)

PS: Here is some data on how I build my systems:

https://bitcointalk.org/index.php?topic=1955358.0


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: QuintLeo on November 09, 2017, 08:34:23 PM
nice test thank you

i think get 1070 in low price is better then?

The 1070 would look way better at a regular cost... the 1070s I own I paid no more than $329 for them (the majority I paid $299). But it's impossible now to get that.

I would buy them again at $329 or less.


 Must be nice - I've NEVER seen 1070s in that kind of price range, I think the LOWEST I ever saw them at NEW was $369 or $379.

 Are you talking new cards or used?


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 09, 2017, 08:37:57 PM
nice test thank you

i think get 1070 in low price is better then?

The 1070 would look way better at a regular cost... the 1070s I own I paid no more than $329 for them (the majority I paid $299). But it's impossible now to get that.

I would buy them again at $329 or less.


 Must be nice - I've NEVER seen 1070s in that kind of price range, I think the LOWEST I ever saw them at NEW was $369 or $379.

 Are you talking new cards or used?

Made a mistake above (checked my records).

The lowest period was $319.99 for refurb EVGA 1070 full length cards -- early June this year direct from EVGA.

The lowest new was $329.99 for ZOTAC 1070 Minis -- mid May this year from Newegg.

Really wish I had gotten more, hah!


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: QuintLeo on November 09, 2017, 08:41:53 PM

3x 1070 Ti setup uses ~500 watts on the highest draw algo (at the wall).


 You are running them at an inefficient operating point.

 Try +200 core +700 memory at 60% TDP (106 watts) for an EFFICIENT operating point on the GTX 1070 ti, then tweek around that starting point.
 





Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: QuintLeo on November 09, 2017, 08:43:05 PM

The lowest new was $329.99 for ZOTAC 1070 Minis -- mid May this year from Newegg.

Really wish I had gotten more, hah!

 Makes me wish I'd been LOOKING at NVidia cards harder that far back - though even that price had to have been a short-lived sale price.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: sundownz on November 09, 2017, 10:24:13 PM

3x 1070 Ti setup uses ~500 watts on the highest draw algo (at the wall).


 You are running them at an inefficient operating point.

 Try +200 core +700 memory at 60% TDP (106 watts) for an EFFICIENT operating point on the GTX 1070 ti, then tweek around that starting point.

Indeed. I will be doing alot of TDP testing here soon -- mainly have been working on getting as much online as possible first =)


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: Pils on November 10, 2017, 08:38:12 PM
Around 115w seems to be the best balance of hashrate, temperature and efficiency. Approx 4.3 sols/w on all cards (505 - 515 sols).

Remember you can really push the core clocks with these, they're all clocked low by default. My cards all do +200 core +550 mem.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: bitgameSolo on November 13, 2017, 03:14:01 PM
eth only on claymore... shouldn't need the +200 core , right? also anyone know if any of the previous drivers work on the 1070 ti or only the most recent?


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: crocozino on November 13, 2017, 04:26:27 PM
so it looks like that a good overclocked  GTX1070ti is a perfect choice if you are in the market for a new card!
and especially if you are choosing between GTX1070ti and GTX1080 so in that case, no doubt  my choice would be GTX1070ti


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: b20lopez on November 13, 2017, 04:52:00 PM
I wish i have decent video cards for mining.. and gaming xD xD


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: gt_addict on November 13, 2017, 08:22:57 PM
eth only on claymore... shouldn't need the +200 core , right? also anyone know if any of the previous drivers work on the 1070 ti or only the most recent?

The newest driver (388.13) only for the 1070ti as the older driver wouldnt recognise it. It also works with my older 1060 6gb aswell(more reliable than the last driver).

Currently getting 32.9mh ETH with 65power, +150core, +825mem with micron memory.


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: crocozino on November 14, 2017, 05:22:00 AM
there is no performance boost comparing to the old drives
so I think nvidia squeezed all of it and we will not see any more speed gains, sadly
another 5-10% performance boost would be highly appreciated !


Title: Re: 1070 Ti benchmarked in Eth & Zec
Post by: QuintLeo on November 14, 2017, 05:26:57 AM
Odd thing - the Nvidia LINUX driver most recent version that supports the GTX 1070 ti is 384.something.

 Gotta wonder if they were DELIBERATELY aiming this card at least in part at miners since the Windows version took so much longer to get support into it (388.13 is the FIRST for Windows with 1070 ti support).