How will this change the world of mining?? GTX 1080 / 1070 - page 8.

Amph

legendary

Activity: 3248

Merit: 1072

Quote from: djole on October 08, 2016, 01:05:15 PM

I also have a corsair PSU with Corsair Link enabled, and it reports about 70W more consumption than my wattmeter. So, Corsair link ~ 900, wattmeter 820.

Which one should i trust? Huh

900 w is the IN or the OUT? maybe that is the efficiency

djole

newbie

Activity: 26

Merit: 0

I also have a corsair PSU with Corsair Link enabled, and it reports about 70W more consumption than my wattmeter. So, Corsair link ~ 900, wattmeter 820.

Which one should i trust? Huh

Amph

legendary

Activity: 3248

Merit: 1072

Quote from: fredeq on October 08, 2016, 09:31:27 AM

Quote from: Amph on October 07, 2016, 01:33:37 AM

i was thinking the same at you until i actually measured it with my wattmeter, and the wattage were almost the same, so the tool is definitely working

wattmeter it's not needed anymore, there are very precise tool nowadays, like the nvidia one and the corsair link, tollerance is only +-5% max

if you don't believe me just try it yourself, use the nvidia tool, and then do the ultimate proof with the wattmeter

In nvidia-smi the values are the same as you reported, but its just a computed "current_tdp * 180W".
I dont state that those values are incorrect - they probably are on the card level.
Nothing internal however can know your PSU efficiency, which directly translates to power at the wall.

Looks like my efficiency is kinda low for lbry "135 / 160 = 0.84". Strange cause this is a gold corsair PSU at 50% load which should give around 90%.
For ETH the efficiency is pretty similar "99 / 115 = 0.86".

ah yeah when you measure with a wattmeter you need to account the efficiency, for that i use corsair link(you need a corsair psu with the "I"), which display the IN and the OUT, the OUT is the real usage of your system minus the efficiency for the psu

like this

so here i'm losing 70w at the wall, due to the efficiency

fredeq

legendary

Activity: 1537

Merit: 1005

Quote from: Amph on October 07, 2016, 01:33:37 AM

i was thinking the same at you until i actually measured it with my wattmeter, and the wattage were almost the same, so the tool is definitely working

wattmeter it's not needed anymore, there are very precise tool nowadays, like the nvidia one and the corsair link, tollerance is only +-5% max

if you don't believe me just try it yourself, use the nvidia tool, and then do the ultimate proof with the wattmeter

In nvidia-smi the values are the same as you reported, but its just a computed "current_tdp * 180W".
I dont state that those values are incorrect - they probably are on the card level.
Nothing internal however can know your PSU efficiency, which directly translates to power at the wall.

Looks like my efficiency is kinda low for lbry "135 / 160 = 0.84". Strange cause this is a gold corsair PSU at 50% load which should give around 90%.
For ETH the efficiency is pretty similar "99 / 115 = 0.86".

pokeytex

legendary

Activity: 1504

Merit: 1002

Quote from: pallas on October 07, 2016, 04:34:42 AM

Quote from: djole on October 07, 2016, 04:31:50 AM

Does anyone have rigs built with nVidia cards ? Which model/manufacturer is the best one?

Thanks

I tend to prefer EVGA 1070s. The SC one with backplate and single 8 pin power plug. YMMV

I second that! EVGA is all I use.

pallas

legendary

Activity: 2716

Merit: 1094

Black Belt Developer

Quote from: djole on October 07, 2016, 04:31:50 AM

Does anyone have rigs built with nVidia cards ? Which model/manufacturer is the best one?

Thanks

I tend to prefer EVGA 1070s. The SC one with backplate and single 8 pin power plug. YMMV

djole

newbie

Activity: 26

Merit: 0

Does anyone have rigs built with nVidia cards ? Which model/manufacturer is the best one?

Thanks

Amph

legendary

Activity: 3248

Merit: 1072

Quote from: fredeq on October 07, 2016, 01:57:41 AM

Quote from: Amph on October 07, 2016, 01:33:37 AM

i was thinking the same at you until i actually measured it with my wattmeter, and the wattage were almost the same, so the tool is definitely working

wattmeter it's not needed anymore, there are very precise tool nowadays, like the nvidia one and the corsair link, tollerance is only +-5% max

if you don't believe me just try it your self, use the nvidia tool, and then do the ultimate proof with the wattmeter

Ok will check it out

Can you still post your settings please? Wink

Oh and drivers. I am using the latest 372.9
Btw are your cards also starting with 1900 mem instead of 2000Mhz? This really annoys me.

yeah 3800, i'm not mining etheruem, i don't like that i need to OC the mem so much, it may be a bad thing in the long term

last time i was able to do 100w 29MH, i posted it here few weeks ago

ok i tested it now

this is with the gpu attached to the monitor(not with the mining rig but the gaming one), 1440p so there is some instability, because i'm using the pc also for browsing and video, that's why you see 24MH as a lowest, but the average is definitely 29MH+

go6ooo1212

legendary

Activity: 1512

Merit: 1000

quarkchain.io

I'll do some tests with my Palits and wattmeter during the weekend..

fredeq

legendary

Activity: 1537

Merit: 1005

Quote from: Amph on October 07, 2016, 01:33:37 AM

i was thinking the same at you until i actually measured it with my wattmeter, and the wattage were almost the same, so the tool is definitely working

wattmeter it's not needed anymore, there are very precise tool nowadays, like the nvidia one and the corsair link, tollerance is only +-5% max

if you don't believe me just try it your self, use the nvidia tool, and then do the ultimate proof with the wattmeter

Ok will check it out

Can you still post your settings please? Wink

Oh and drivers. I am using the latest 372.9
Btw are your cards also starting with 1900 mem instead of 2000Mhz? This really annoys me.

Amph

legendary

Activity: 3248

Merit: 1072

i was thinking the same at you until i actually measured it with my wattmeter, and the wattage were almost the same, so the tool is definitely working

wattmeter it's not needed anymore, there are very precise tool nowadays, like the nvidia one and the corsair link, tollerance is only +-5% max

if you don't believe me just try it yourself, use the nvidia tool, and then do the ultimate proof with the wattmeter

fredeq

legendary

Activity: 1537

Merit: 1005

@Amph I wouldnt believe any software wattage tool for a second.
Its only good for noticing drops / raises, not actual values.

Your 100W are at the wall or from software?
Can you post exact readings(core, mem, v, settings) for that wattage please?

Amph

legendary

Activity: 3248

Merit: 1072

Quote from: go6ooo1212 on October 06, 2016, 04:34:29 PM

Quote from: Amph on October 06, 2016, 03:12:58 PM

Quote from: fredeq on October 06, 2016, 02:19:29 PM

Quote from: dexxa05 on October 06, 2016, 01:33:32 AM

Quote from: fredeq on October 05, 2016, 03:13:06 PM

Getting strange results(wattage) with Gigabyte G1 1070(samsung mem):

ETH: 29.7@115W - core using lowest curve possible@1165, Mem +700@2250, 55%TDP, 0.793V
LBC: 292@160W - core +200@2012, Mem -500@1650, 75%TDP, 0.931V
Measured at the wall.

Also memory is always 100Mhz lower than it should be according to settings. Starts at 1900Mhz. Is that normal?

Hi,

I have the same card but i'm at 24MHs can you help me to get around 29 ?

Thx in advance

Use latest msi afterburner beta with the details I have posted. To edit voltage-frequency curve press `ctrl+f` in the afterburner.

@Amph Could you please explain how did you get to 30Mhash@80W?
Did you measure that at the wall?
Cant decide between lbry and ETH... ETH has high memory OC, but lower wattage. Lbry lower OC, but higher wattage Wink

sry it was 90w, not 80w, 80w is not possible

it was not my setup, it was a setup of another user here, he reached something like 31.5MH at 90w

55% tdp, +100 on core and +1000 on the mem

Which miner , driver and OS you are reaching this score with , mate ?

it's not me it was another user, i'm at 100w last time i tried it, also i'm not able to put mem at 1000, without artifact, it depend o your 1070 model

Quote from: fredeq on October 06, 2016, 04:02:09 PM

Quote from: Amph on October 06, 2016, 03:12:58 PM

sry it was 90w, not 80w, 80w is not possible

it was not my setup, it was a setup of another user here, he reached something like 31.5MH at 90w

55% tdp, +100 on core and +1000 on the mem

Then my guess is he just multiplied 160 * 55% = 88W.
I am reading at least 115W with reduced core and *only* +700 on mem.
Thanks for info

no he used nvidia smi tool to monitor the wattage

go6ooo1212

legendary

Activity: 1512

Merit: 1000

quarkchain.io

Quote from: Amph on October 06, 2016, 03:12:58 PM

Quote from: fredeq on October 06, 2016, 02:19:29 PM

Quote from: dexxa05 on October 06, 2016, 01:33:32 AM

Quote from: fredeq on October 05, 2016, 03:13:06 PM

Getting strange results(wattage) with Gigabyte G1 1070(samsung mem):

ETH: 29.7@115W - core using lowest curve possible@1165, Mem +700@2250, 55%TDP, 0.793V
LBC: 292@160W - core +200@2012, Mem -500@1650, 75%TDP, 0.931V
Measured at the wall.

Also memory is always 100Mhz lower than it should be according to settings. Starts at 1900Mhz. Is that normal?

Hi,

I have the same card but i'm at 24MHs can you help me to get around 29 ?

Thx in advance

Use latest msi afterburner beta with the details I have posted. To edit voltage-frequency curve press `ctrl+f` in the afterburner.

@Amph Could you please explain how did you get to 30Mhash@80W?
Did you measure that at the wall?
Cant decide between lbry and ETH... ETH has high memory OC, but lower wattage. Lbry lower OC, but higher wattage Wink

sry it was 90w, not 80w, 80w is not possible

it was not my setup, it was a setup of another user here, he reached something like 31.5MH at 90w

55% tdp, +100 on core and +1000 on the mem

Which miner , driver and OS you are reaching this score with , mate ?

fredeq

legendary

Activity: 1537

Merit: 1005

Quote from: Amph on October 06, 2016, 03:12:58 PM

sry it was 90w, not 80w, 80w is not possible

it was not my setup, it was a setup of another user here, he reached something like 31.5MH at 90w

55% tdp, +100 on core and +1000 on the mem

Then my guess is he just multiplied 160 * 55% = 88W.
I am reading at least 115W with reduced core and *only* +700 on mem.
Thanks for info

Amph

legendary

Activity: 3248

Merit: 1072

Quote from: fredeq on October 06, 2016, 02:19:29 PM

Quote from: dexxa05 on October 06, 2016, 01:33:32 AM

Quote from: fredeq on October 05, 2016, 03:13:06 PM

Getting strange results(wattage) with Gigabyte G1 1070(samsung mem):

ETH: 29.7@115W - core using lowest curve possible@1165, Mem +700@2250, 55%TDP, 0.793V
LBC: 292@160W - core +200@2012, Mem -500@1650, 75%TDP, 0.931V
Measured at the wall.

Also memory is always 100Mhz lower than it should be according to settings. Starts at 1900Mhz. Is that normal?

Hi,

I have the same card but i'm at 24MHs can you help me to get around 29 ?

Thx in advance

Use latest msi afterburner beta with the details I have posted. To edit voltage-frequency curve press `ctrl+f` in the afterburner.

@Amph Could you please explain how did you get to 30Mhash@80W?
Did you measure that at the wall?
Cant decide between lbry and ETH... ETH has high memory OC, but lower wattage. Lbry lower OC, but higher wattage Wink

sry it was 90w, not 80w, 80w is not possible

it was not my setup, it was a setup of another user here, he reached something like 31.5MH at 90w

55% tdp, +100 on core and +1000 on the mem

fredeq

legendary

Activity: 1537

Merit: 1005

Quote from: dexxa05 on October 06, 2016, 01:33:32 AM

Quote from: fredeq on October 05, 2016, 03:13:06 PM

Getting strange results(wattage) with Gigabyte G1 1070(samsung mem):

ETH: 29.7@115W - core using lowest curve possible@1165, Mem +700@2250, 55%TDP, 0.793V
LBC: 292@160W - core +200@2012, Mem -500@1650, 75%TDP, 0.931V
Measured at the wall.

Also memory is always 100Mhz lower than it should be according to settings. Starts at 1900Mhz. Is that normal?

Hi,

I have the same card but i'm at 24MHs can you help me to get around 29 ?

Thx in advance

Use latest msi afterburner beta with the details I have posted. To edit voltage-frequency curve press `ctrl+f` in the afterburner.

@Amph Could you please explain how did you get to 30Mhash@80W?
Did you measure that at the wall?
Cant decide between lbry and ETH... ETH has high memory OC, but lower wattage. Lbry lower OC, but higher wattage Wink

dexxa05

member

Activity: 87

Merit: 10

Quote from: fredeq on October 05, 2016, 03:13:06 PM

Getting strange results(wattage) with Gigabyte G1 1070(samsung mem):

ETH: 29.7@115W - core using lowest curve possible@1165, Mem +700@2250, 55%TDP, 0.793V
LBC: 292@160W - core +200@2012, Mem -500@1650, 75%TDP, 0.931V
Measured at the wall.

Also memory is always 100Mhz lower than it should be according to settings. Starts at 1900Mhz. Is that normal?

Hi,

I have the same card but i'm at 24MHs can you help me to get around 29 ?

Thx in advance

PovertyByte

sr. member

Activity: 420

Merit: 250

Quote from: Kompik on September 23, 2016, 05:22:18 PM

I managed to OC my EVGA 1070 with samsung memory to 9216 Mhz. It seems stable, but is it safe? With undervolted GPU to 62% of TDP - about 93watt per card at the wall.

I have been doing 9420 mhz memory for days with my FTW edition dual mining EHT/LBRY with the dcri at 500 (84% power target), and I had the thing running at 9700mhz just fine when single mining ether only on 50% power target

What is this Samsung memory people are mentioning?

fredeq

legendary

Activity: 1537

Merit: 1005

Getting strange results(wattage) with Gigabyte G1 1070(samsung mem):

ETH: 29.7@115W - core using lowest curve possible@1165, Mem +700@2250, 55%TDP, 0.793V
LBC: 292@160W - core +200@2012, Mem -500@1650, 75%TDP, 0.931V
Measured at the wall.

Also memory is always 100Mhz lower than it should be according to settings. Starts at 1900Mhz. Is that normal?

Topic: How will this change the world of mining?? GTX 1080 / 1070 - page 8. (Read 134124 times)