Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 86.

ajolly87

newbie

Activity: 3

Merit: 0

Quote from: ajolly87 on September 07, 2019, 10:42:22 PM

All,

Quick question, is anyone else experiencing instability with the RX5700XT?

It will sit at 51mh/s for a few rounds and then drop down to 21, 8, 9 before going back up to 51mh/s

Power Limit -45%
Memory Max Frequency 910
Core Clock Frequency Auto

Checked Board, Riser and PSU with no luck

On latest drivers and Claymore version

Anyone had any luck with this?

Not crashing but the hash rate isn’t consistent?

Reference card 5700xt, Tried using afterburner and wattman to adjust clocks and power with no luck.

aknataga

full member

Activity: 387

Merit: 102

Quote from: strelly83 on September 13, 2019, 08:02:10 AM

Hi everyone, at this moment what are the best cards in terms of price / performance based on RX 5700 for mining eth?
I plan to build a 4-card rig soon and I am undecided between the purchase of reference cards or some 2-fan variants (for example Sapphire Pulse RX 5700 or MSI RX 5700 MECH)...thank you!

In my opinion everything is obvious here, everything depends on you. If noise is not big problem for you, then you can choose reference card, especially there is 3 years warranty. Otherwise overpaying everything about 15-20$ you can choose R57Gaming OC.

strelly83

newbie

Activity: 20

Merit: 0

Hi everyone, at this moment what are the best cards in terms of price / performance based on RX 5700 for mining eth?
I plan to build a 4-card rig soon and I am undecided between the purchase of reference cards or some 2-fan variants (for example Sapphire Pulse RX 5700 or MSI RX 5700 MECH)...thank you!

bategojko74

member

Activity: 220

Merit: 12

Quote from: zenstrive on September 13, 2019, 04:26:03 AM

Quote from: bategojko74 on September 11, 2019, 12:57:51 PM

Try phoenix miner with the following settings: : -ethi 8 -tt 54 -cclock 1300 -cvddc 750 -mclock 910 -hstats 2
I make 52.65MH/s with 93 watts powerdraw and gpu hbm temperature 70 degree C. You can try with higher -tt to decrease the fan speed. 54 works for sure and is too safe.

what phoenix miner version?

Latest: 4.6c

zenstrive

jr. member

Activity: 46

Merit: 4

Quote from: bategojko74 on September 11, 2019, 12:57:51 PM

Try phoenix miner with the following settings: : -ethi 8 -tt 54 -cclock 1300 -cvddc 750 -mclock 910 -hstats 2
I make 52.65MH/s with 93 watts powerdraw and gpu hbm temperature 70 degree C. You can try with higher -tt to decrease the fan speed. 54 works for sure and is too safe.

what phoenix miner version?

bategojko74

member

Activity: 220

Merit: 12

Quote from: zenstrive on September 13, 2019, 01:35:47 AM

Theoritically, you can put Four Reference cards Navi in MSI X570 Godlike. My current tower cooler blocks the first GPU slot. Waiting on 3950X before actually putting the bonus MasterLiquid AIO from MSI X570's promotions.

Anyway, I tried with three GPUS (one custom AIO cooled, two reference coolers) and it works just fine. Their temps are not catastrophic either; With fan speeds less than 50% (therefore noise is tame), the GPU temps are around 65-75 oC, with memory junction temp around 90 oC.

https://youtu.be/IDsYGdhxayA

After 3950X is here, I will get one of that and another reference card 5700 and my goal of having four GPUs in one motherboard will be achieved.

For now the RX 5700 yiels around 48-49 Mh/S each with maximum GPU temps around 70 oC with memory temps around 90 oC, with fans running below 40 %.
This RX 5700 reference design is just great for me.

Can't see any reason not to make 52 mh/s with lower power consumption and gpu temps about 62 and mem temps about 72 and fans below 40. You will damage your memory in long term on 90 degree. Read my previous post and use PM with -tt 62 -cclock 1300 -mclock 900 -cvvdc 750
You can even use -mclock 910 for 52.7mh/s but you have to find lower -tt between 54 and 62. 54 works for sure. The fans will go between 40% and 60% depending on -tt .Currently testing ... -mclock 910 -tt 58 ..., seems very stable. Memory at 76 degree C. Fans at 46%. Power consumption 94 Watts.

zenstrive

jr. member

Activity: 46

Merit: 4

Theoritically, you can put Four Reference cards Navi in MSI X570 Godlike. My current tower cooler blocks the first GPU slot. Waiting on 3950X before actually putting the bonus MasterLiquid AIO from MSI X570's promotions.

Anyway, I tried with three GPUS (one custom AIO cooled, two reference coolers) and it works just fine. Their temps are not catastrophic either; With fan speeds less than 50% (therefore noise is tame), the GPU temps are around 65-75 oC, with memory junction temp around 90 oC.

https://youtu.be/IDsYGdhxayA

After 3950X is here, I will get one of that and another reference card 5700 and my goal of having four GPUs in one motherboard will be achieved.

For now the RX 5700 yiels around 48-49 Mh/S each with maximum GPU temps around 70 oC with memory temps around 90 oC, with fans running below 40 %.
This RX 5700 reference design is just great for me.

davidsheng

newbie

Activity: 5

Merit: 0

Quote from: adaseb on September 11, 2019, 04:27:58 PM

How much actual RAM do you have in your system? I've had this issue before years ago and when I added more GPUs it wouldn't work for some reason. I tried everything like increasing the virtual memory and still didn't work, then as a last resort I added more RAM and the issue was fixed. So try adding more ram and see if it goes away.

If you don't want to add RAM then remove a GPU or 2 and see if the problem is fixed. For some weird reason even though the DAG pagefile is identical, you need more RAM to load it on the GPUs.

If this doesn't work then try another OS. Basically Win10 is a pain, especially on old CPUs but usually the drivers and miners work best in Win10. If Win10 doesn't work then try Ubuntu and if that doesn't work then I have no idea what the issue is.

thanks a lot for the suggestions! I have 8GB RAM with a 64GB pagefile. Will try adding more RAM tomorrow and see how it goes.

WhackOBill

jr. member

Activity: 73

Merit: 3

19.9.2 is out. I hate playing driver roulette but I have some stability issues in 14.7 so I might take it for a spin.

adaseb

legendary

Activity: 3808

Merit: 1723

Up to 300% + 200 FS deposit bonuses

Quote from: davidsheng on September 11, 2019, 12:20:06 PM

Bumping this. Still need a solution.

Quote from: davidsheng on September 08, 2019, 11:40:19 PM

Hi everyone. I need some help regarding mining with P106.

Rig specs:
- 7x Zotac P106, total sightly below 170 Mh/s
- Windows 7 64-bit
- nVidia driver version 382.53
- Claymore miner version 11.0 (any version newer than this will always crash for me!)
- MSI Afterburner version 4.4.2
- Virtual memory sufficiently high

Current situation:
- Been running this setup for past ~2 years with very few issues
- I've paid thousands of USD in fees to Claymore since the beginning
- Happy with performance, consistency and stability
- It's working fine right now

Problem:
- I am being forced to abandon v11.0 and install v15.0 once DAG #299 comes out
- I have tested the newer clients including v15.0, but they always crash to desktop immediately after detecting the GPUs
- This doesn't happen with v10 or v11
- Updating nVidia drivers to latest version / CUDA version makes no difference
- Phoenix miner is even worse for me (individual GPUs crash very frequently)

Anyone with similar experience / solution? Your comments will be hugely appreciated. Thanks!

How much actual RAM do you have in your system? I've had this issue before years ago and when I added more GPUs it wouldn't work for some reason. I tried everything like increasing the virtual memory and still didn't work, then as a last resort I added more RAM and the issue was fixed. So try adding more ram and see if it goes away.

If you don't want to add RAM then remove a GPU or 2 and see if the problem is fixed. For some weird reason even though the DAG pagefile is identical, you need more RAM to load it on the GPUs.

If this doesn't work then try another OS. Basically Win10 is a pain, especially on old CPUs but usually the drivers and miners work best in Win10. If Win10 doesn't work then try Ubuntu and if that doesn't work then I have no idea what the issue is.

bategojko74

member

Activity: 220

Merit: 12

Quote from: Bathmat on September 11, 2019, 11:03:19 AM

Quote from: bategojko74 on September 11, 2019, 08:05:16 AM

Quote from: Bathmat on September 11, 2019, 08:01:12 AM

Quote from: bategojko74 on September 11, 2019, 07:11:32 AM

Quote from: vuli1 on September 11, 2019, 05:09:39 AM

HBM ? 5700 have GDDR6 memory.

Quote from Martin - HWiNFO Author about GPU HBM temperature: 'I just clarified this with AMD and the situation is following.
This value on Navi GPUs is the memory temperature, but it's not the case temperature (T_case) that can be measured on the surface of the chip. It's the junction temperature (Tj) provided by the memory sensor.
Tj is always higher and the difference seen from Tcase can be up-to 18 C (according to measurements done by AMD). The maximum value of Tj here is 105 C.
So users that were concerned about seeing too high temperatures here should not worry about elevated values and not treats this as the surface temperature (which has a limit of 95 C).
I will adjust this value in the next build of HWiNFO so it provides correct description.'

I saw that site/link too when I did some searching about. So high temps are probably ok, but 90C+ still seems pretty high even if Tj Max is 105 C. However, there doesn't seem to be much I can do about it... Changing fan speeds, etc. doesn't lower the temp. It shoots up over 80C pretty much as soon as the miner starts. I guess the only thing would be to reduce mining intensity.

That being said, the miner ran all night (8 hours) without a hiccup or throttling. So as long as the temps aren't going to kill the mem long term, maybe they can handle it.

you can control GPU HBM temperature indirectly by passing very low -tt such as -tt 54 or -tt 53 and -fanmin 54 for an instance.

Tried those. Fans spin faster, core comes down a few deg, mem comes down a couple deg, but still 90C. Also tried -ethi 2, HBM temp showed 84C, but obviously speed was reduced too... 45 and 47Mh/s

So, perhaps it's a design flaw with the Sapphire Pulse heat sink/cooler.

Try phoenix miner with the following settings: : -ethi 8 -tt 54 -cclock 1300 -cvddc 750 -mclock 910 -hstats 2
I make 52.65MH/s with 93 watts powerdraw and gpu hbm temperature 70 degree C. You can try with higher -tt to decrease the fan speed. 54 works for sure and is too safe.

NCarter84

jr. member

Activity: 195

Merit: 4

Quote from: davidsheng on September 11, 2019, 12:20:06 PM

Bumping this. Still need a solution.

Quote from: davidsheng on September 08, 2019, 11:40:19 PM

Hi everyone. I need some help regarding mining with P106.

Rig specs:
- 7x Zotac P106, total sightly below 170 Mh/s
- Windows 7 64-bit
- nVidia driver version 382.53
- Claymore miner version 11.0 (any version newer than this will always crash for me!)
- MSI Afterburner version 4.4.2
- Virtual memory sufficiently high

Current situation:
- Been running this setup for past ~2 years with very few issues
- I've paid thousands of USD in fees to Claymore since the beginning
- Happy with performance, consistency and stability
- It's working fine right now

Problem:
- I am being forced to abandon v11.0 and install v15.0 once DAG #299 comes out
- I have tested the newer clients including v15.0, but they always crash to desktop immediately after detecting the GPUs
- This doesn't happen with v10 or v11
- Updating nVidia drivers to latest version / CUDA version makes no difference
- Phoenix miner is even worse for me (individual GPUs crash very frequently)

Anyone with similar experience / solution? Your comments will be hugely appreciated. Thanks!

Windows 7 goes EOL in a few months... Maybe its time to move to Win10?

davidsheng

newbie

Activity: 5

Merit: 0

Bumping this. Still need a solution.

Quote from: davidsheng on September 08, 2019, 11:40:19 PM

Hi everyone. I need some help regarding mining with P106.

Rig specs:
- 7x Zotac P106, total sightly below 170 Mh/s
- Windows 7 64-bit
- nVidia driver version 382.53
- Claymore miner version 11.0 (any version newer than this will always crash for me!)
- MSI Afterburner version 4.4.2
- Virtual memory sufficiently high

Current situation:
- Been running this setup for past ~2 years with very few issues
- I've paid thousands of USD in fees to Claymore since the beginning
- Happy with performance, consistency and stability
- It's working fine right now

Problem:
- I am being forced to abandon v11.0 and install v15.0 once DAG #299 comes out
- I have tested the newer clients including v15.0, but they always crash to desktop immediately after detecting the GPUs
- This doesn't happen with v10 or v11
- Updating nVidia drivers to latest version / CUDA version makes no difference
- Phoenix miner is even worse for me (individual GPUs crash very frequently)

Anyone with similar experience / solution? Your comments will be hugely appreciated. Thanks!

jeca

newbie

Activity: 9

Merit: 0

Quote from: Claymore on September 05, 2019, 05:03:06 AM

Quote from: jstefanop on September 04, 2019, 08:28:00 PM

Quote from: wa11mine on September 04, 2019, 07:00:08 PM

I would expect the Non XT to run at around 48MHs (36 CU vs 40 CU), don't know much about power usage in that case.

Nah I’m running plain 5700...extra cores don’t do anything for eth, and it will matter even less when miners are running wave32 code.

Yes, both 5700XT and 5700 have the same hashrate.

HEY, Will the 40 computational units be available in the future? is this predictable?

Bathmat

jr. member

Activity: 78

Merit: 3

Quote from: bategojko74 on September 11, 2019, 08:05:16 AM

Quote from: Bathmat on September 11, 2019, 08:01:12 AM

Quote from: bategojko74 on September 11, 2019, 07:11:32 AM

Quote from: vuli1 on September 11, 2019, 05:09:39 AM

HBM ? 5700 have GDDR6 memory.

Quote from Martin - HWiNFO Author about GPU HBM temperature: 'I just clarified this with AMD and the situation is following.
This value on Navi GPUs is the memory temperature, but it's not the case temperature (T_case) that can be measured on the surface of the chip. It's the junction temperature (Tj) provided by the memory sensor.
Tj is always higher and the difference seen from Tcase can be up-to 18 C (according to measurements done by AMD). The maximum value of Tj here is 105 C.
So users that were concerned about seeing too high temperatures here should not worry about elevated values and not treats this as the surface temperature (which has a limit of 95 C).
I will adjust this value in the next build of HWiNFO so it provides correct description.'

I saw that site/link too when I did some searching about. So high temps are probably ok, but 90C+ still seems pretty high even if Tj Max is 105 C. However, there doesn't seem to be much I can do about it... Changing fan speeds, etc. doesn't lower the temp. It shoots up over 80C pretty much as soon as the miner starts. I guess the only thing would be to reduce mining intensity.

That being said, the miner ran all night (8 hours) without a hiccup or throttling. So as long as the temps aren't going to kill the mem long term, maybe they can handle it.

you can control GPU HBM temperature indirectly by passing very low -tt such as -tt 54 or -tt 53 and -fanmin 54 for an instance.

Tried those. Fans spin faster, core comes down a few deg, mem comes down a couple deg, but still 90C. Also tried -ethi 2, HBM temp showed 84C, but obviously speed was reduced too... 45 and 47Mh/s

So, perhaps it's a design flaw with the Sapphire Pulse heat sink/cooler.

bategojko74

member

Activity: 220

Merit: 12

Quote from: Bathmat on September 11, 2019, 08:01:12 AM

Quote from: bategojko74 on September 11, 2019, 07:11:32 AM

Quote from: vuli1 on September 11, 2019, 05:09:39 AM

HBM ? 5700 have GDDR6 memory.

Quote from Martin - HWiNFO Author about GPU HBM temperature: 'I just clarified this with AMD and the situation is following.
This value on Navi GPUs is the memory temperature, but it's not the case temperature (T_case) that can be measured on the surface of the chip. It's the junction temperature (Tj) provided by the memory sensor.
Tj is always higher and the difference seen from Tcase can be up-to 18 C (according to measurements done by AMD). The maximum value of Tj here is 105 C.
So users that were concerned about seeing too high temperatures here should not worry about elevated values and not treats this as the surface temperature (which has a limit of 95 C).
I will adjust this value in the next build of HWiNFO so it provides correct description.'

I saw that site/link too when I did some searching about. So high temps are probably ok, but 90C+ still seems pretty high even if Tj Max is 105 C. However, there doesn't seem to be much I can do about it... Changing fan speeds, etc. doesn't lower the temp. It shoots up over 80C pretty much as soon as the miner starts. I guess the only thing would be to reduce mining intensity.

That being said, the miner ran all night (8 hours) without a hiccup or throttling. So as long as the temps aren't going to kill the mem long term, maybe they can handle it.

you can control GPU HBM temperature indirectly by passing very low -tt such as -tt 54 or -tt 53 and -fanmin 54 for an instance.

Bathmat

jr. member

Activity: 78

Merit: 3

Quote from: bategojko74 on September 11, 2019, 07:11:32 AM

Quote from: vuli1 on September 11, 2019, 05:09:39 AM

HBM ? 5700 have GDDR6 memory.

Quote from Martin - HWiNFO Author about GPU HBM temperature: 'I just clarified this with AMD and the situation is following.
This value on Navi GPUs is the memory temperature, but it's not the case temperature (T_case) that can be measured on the surface of the chip. It's the junction temperature (Tj) provided by the memory sensor.
Tj is always higher and the difference seen from Tcase can be up-to 18 C (according to measurements done by AMD). The maximum value of Tj here is 105 C.
So users that were concerned about seeing too high temperatures here should not worry about elevated values and not treats this as the surface temperature (which has a limit of 95 C).
I will adjust this value in the next build of HWiNFO so it provides correct description.'

I saw that site/link too when I did some searching about. So high temps are probably ok, but 90C+ still seems pretty high even if Tj Max is 105 C. However, there doesn't seem to be much I can do about it... Changing fan speeds, etc. doesn't lower the temp. It shoots up over 80C pretty much as soon as the miner starts. I guess the only thing would be to reduce mining intensity.

That being said, the miner ran all night (8 hours) without a hiccup or throttling. So as long as the temps aren't going to kill the mem long term, maybe they can handle it.

bategojko74

member

Activity: 220

Merit: 12

Quote from: vuli1 on September 11, 2019, 05:09:39 AM

HBM ? 5700 have GDDR6 memory.

Quote from Martin - HWiNFO Author about GPU HBM temperature: 'I just clarified this with AMD and the situation is following.
This value on Navi GPUs is the memory temperature, but it's not the case temperature (T_case) that can be measured on the surface of the chip. It's the junction temperature (Tj) provided by the memory sensor.
Tj is always higher and the difference seen from Tcase can be up-to 18 C (according to measurements done by AMD). The maximum value of Tj here is 105 C.
So users that were concerned about seeing too high temperatures here should not worry about elevated values and not treats this as the surface temperature (which has a limit of 95 C).
I will adjust this value in the next build of HWiNFO so it provides correct description.'

vuli1

jr. member

Activity: 238

Merit: 3

HBM ? 5700 have GDDR6 memory.

Bathmat

jr. member

Activity: 78

Merit: 3

Quote from: zenstrive on September 10, 2019, 09:50:45 PM

Thanks for the info.
Looks like Sapphire Pulse cooler is unable to cool down the memory chips under mining load, and the reference cooler does better job at it. Are you using custom fan speeds or left it on auto?
Also does The XT and non-XT performs the same regarding mining?

Non-XT is 0.5-0.8 Mh/s slower than XT given the same clock, etc. settings
I'm using MSI AB for fan curve, but it doesn't seem to be affecting it much (if it's even working). I had the fans up to 90%+ which might have cooled the core temp a little, but HBM temp remained the same.

Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 86. (Read 6590565 times)