GTX is dead long live RTX !!

Message boards : Number crunching : GTX is dead long live RTX !!
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 9 · 10 · 11 · 12

AuthorMessage
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1969120 - Posted: 7 Dec 2018, 18:33:38 UTC - in response to Message 1969110.  

I want a Hybrid one! Any idea when they will be available?

I get the feeling that EVGA is not going to be offering pre-built Hybrid or Hydro Copper cards anymore. They do have the DIY AIO cooler kits now for the Turing cards.
https://www.evga.com/products/productlist.aspx?type=18&family=Cooling&chipset=GPU+HYBRID+Cooler
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1969120 · Report as offensive
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 1969121 - Posted: 7 Dec 2018, 18:36:19 UTC - in response to Message 1969120.  
Last modified: 7 Dec 2018, 18:38:23 UTC

I want a Hybrid one! Any idea when they will be available?

I get the feeling that EVGA is not going to be offering pre-built Hybrid or Hydro Copper cards anymore. They do have the DIY AIO cooler kits now for the Turing cards.
https://www.evga.com/products/productlist.aspx?type=18&family=Cooling&chipset=GPU+HYBRID+Cooler


yeah i was just going to post this. add the 2070 hybrid cooler to notify you, then make it yourself. the word on EVGA forums claims they should be available next week.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 1969121 · Report as offensive
Profile zoom3+1=4
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 65746
Credit: 55,293,173
RAC: 49
United States
Message 1969123 - Posted: 7 Dec 2018, 18:45:58 UTC

Me I'd love to get a 2070 blower card, maybe one day in 2 years, Asus, MSI, and Zotac all make one or should if I read right.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1969123 · Report as offensive
Bruce
Volunteer tester

Send message
Joined: 15 Mar 02
Posts: 123
Credit: 124,955,234
RAC: 11
United States
Message 1969125 - Posted: 7 Dec 2018, 18:51:01 UTC - in response to Message 1969109.  

Bruce, can you tell me what you see typically for your core clock frequency while running the SoG app. Also what is the memory on the card clocking under compute load?
Hi Keith,
According to GPU-Z 2.13 my GPU Core Clock runs between 1905 and 1920 MHz on both.
The GPU Memory Clock runs at 1700 MHz on both.
GPU Temps average about 150F to 155F with fans at 50%. (room temp 75F)
Memory used is about 4735 MB.
Of coarse this can very a little bit depending on the different WU make up.

Thanks Bruce, that aligns with what I see. I run Linux so I see my memory clock different than Windows7. If I do 1700 X 8 that equals 13600Mhz which is where my card runs under compute load. Did you notice the power state the card runs under? The power states are defined differently in Windows and in Linux. What surprised me was that Turing has four discrete power states compared to only three in Pascal cards.
Keith,
It looks like SIV is reporting P2 state on the cards.
SIV is also showing 5 states - P0, P2, P3, P5, P8. Not sure what all those mean.
Hope info helps.
Bruce
ID: 1969125 · Report as offensive
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 1969126 - Posted: 7 Dec 2018, 18:52:23 UTC - in response to Message 1969123.  

EVGA has a model listed for that also, but i dont think i've actually seen it for sale.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 1969126 · Report as offensive
RuthlessRufus

Send message
Joined: 18 Oct 07
Posts: 11
Credit: 70,386,101
RAC: 28
United States
Message 1969132 - Posted: 7 Dec 2018, 19:31:04 UTC

Sup guys,

Just got my 2080Ti running Seti. Currently running the latest Lunatics optimizer I could find, and running 4 Cuda instances simultaneously for the card. Way too early to tell how fast it's going, but wanted to find out what other users configurations are like for the GPU.
ID: 1969132 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1969135 - Posted: 7 Dec 2018, 19:56:06 UTC - in response to Message 1969125.  
Last modified: 7 Dec 2018, 19:57:52 UTC


Keith,
It looks like SIV is reporting P2 state on the cards.
SIV is also showing 5 states - P0, P2, P3, P5, P8. Not sure what all those mean.
Hope info helps.

Yes, that is the traditional power states on Nvidia cards that SIV has always published. On Linux, the power states are labeled differently to confuse matters. P0 power state is the highest in Windows and where the cards usually run with a standard graphics load. They drop to P2 state when the drivers detect a compute load. That usually knocks considerable memory clock from the normal P0 state.

On Linux, P3 is the highest mode called Performance Mode 3 and what the card runs under a graphics load. It drops to Performance Mode 2 when the driver detects a compute load.

I also have discovered that my 2080 won't respond to any of the normal overclock commands that my Pascal cards do. The settings just get ignored. So the reason why I was asking what you typically see under compute load. My card runs the core at 1980Mhz and the memory at 13600Mhz, so that is the P2 power state downclock for detected compute loads.

Need to investigate how to get the memory back to full 14000Mhz or add a bit more which is usually possible. It has been with all my Pascal cards.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1969135 · Report as offensive
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 1969139 - Posted: 7 Dec 2018, 20:15:56 UTC - in response to Message 1969135.  

Keith, I was able to overclock my 2070 (both core and mem) with the nvidia-settigs GUI. I haven’t tried the command line commands however.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 1969139 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1969140 - Posted: 7 Dec 2018, 20:16:13 UTC - in response to Message 1969132.  

Run 3 SoG tasks ... Cuda on Windows is just SLOWWWWW.
It is for very old cards.
Make sure you reserve 3 or 4 threads for the SoG tasks.
ID: 1969140 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1969141 - Posted: 7 Dec 2018, 20:19:12 UTC - in response to Message 1969132.  

running 4 Cuda instances simultaneously

You're running SoG, which is the best application to run. But why on Earth would you run 4 WUs at a time???
1 or 2 at a time using some command line settings would produce considerably more work per hour.
Grant
Darwin NT
ID: 1969141 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1969143 - Posted: 7 Dec 2018, 20:28:59 UTC - in response to Message 1969141.  
Last modified: 7 Dec 2018, 20:29:30 UTC

People have found 3 SoG tasks on a 1080Ti to be the "Sweet Spot" for them.
I assume it would be very similar for the 2080Ti.
EDIT: With aggressive tuning as well.
ID: 1969143 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1969146 - Posted: 7 Dec 2018, 20:35:16 UTC - in response to Message 1969139.  

Keith, I was able to overclock my 2070 (both core and mem) with the nvidia-settigs GUI. I haven’t tried the command line commands however.

I hadn't tried with the GUI. I always used the command line in a script. Just added 400Mhz to the 13600Mhz memory clock to put it stock 14000Mhz.

But I just crashed the computer when I tried to add 40Mhz to the 1980Mhz core clock. Looks like the offset stays 0 for the 2080 card. GPU Boost3.0 does well enough in keeping the core clocks up under compute load.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1969146 · Report as offensive
Profile zoom3+1=4
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 65746
Credit: 55,293,173
RAC: 49
United States
Message 1969167 - Posted: 7 Dec 2018, 22:28:26 UTC - in response to Message 1969126.  

EVGA has a model listed for that also, but i dont think i've actually seen it for sale.

Yeah I found that 2070, if My cards were paid down, I could buy 2 right now, but I need to do work on My home, Zillow says My home might be worth $82,691.00, as hard as that is to believe.
EVGA certainly likes transparency, of course one can tell at a glance if the card needs cleaning or not.

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1969167 · Report as offensive
Profile Freewill Project Donor
Avatar

Send message
Joined: 19 May 99
Posts: 766
Credit: 354,398,348
RAC: 11,693
United States
Message 1969379 - Posted: 8 Dec 2018, 19:50:13 UTC - in response to Message 1961173.  

So i'm curious. I just run a Nvidia Geforce Gt 730 in the system here. It does one unit every two hours or so.

What would one of these expensive cards do? Do they have more cores and such? How much faster would they turn units back in?


Yes.
My AMD R9 380 returns a unit every 8 to 10 minutes.
A Nvidia 1080TI in less than 5 minutes.
On Linux with the special app 2 minutes.


A gtx 1060 3GB running about $150 on eBay will run a SOG gpu task in about 7-8 minutes under Windows 10 and about 3-4 minutes under Linux/Cuda91 (Linux/special app).

Tom


That speed increase under Linux/Cuda91 is impressive. I'm willing to convert a machine to Linux. Is this special app part of the Lunatics package or something separate? How much effort to configure beyond the basic install of Boinc?

Roger
ID: 1969379 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1969380 - Posted: 8 Dec 2018, 20:08:07 UTC - in response to Message 1969379.  

So i'm curious. I just run a Nvidia Geforce Gt 730 in the system here. It does one unit every two hours or so.

What would one of these expensive cards do? Do they have more cores and such? How much faster would they turn units back in?


Yes.
My AMD R9 380 returns a unit every 8 to 10 minutes.
A Nvidia 1080TI in less than 5 minutes.
On Linux with the special app 2 minutes.


A gtx 1060 3GB running about $150 on eBay will run a SOG gpu task in about 7-8 minutes under Windows 10 and about 3-4 minutes under Linux/Cuda91 (Linux/special app).

Tom


That speed increase under Linux/Cuda91 is impressive. I'm willing to convert a machine to Linux. Is this special app part of the Lunatics package or something separate? How much effort to configure beyond the basic install of Boinc?

Roger

No it isn't part of the old Lunatics package. It is another package developed by Petri33 and TBar. The package is located at the Crunchers Anonymous site. But I just discovered that all the posts and links for the special app have been removed. Need to investigate why.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1969380 · Report as offensive
Profile Freewill Project Donor
Avatar

Send message
Joined: 19 May 99
Posts: 766
Credit: 354,398,348
RAC: 11,693
United States
Message 1969385 - Posted: 8 Dec 2018, 20:53:14 UTC - in response to Message 1969380.  

Keith,

Thanks for the reply. If you re-discover how to access this special app package, I would appreciate hearing about it. I haven't built up a linux (well, it was unix at the time) system in quite a while. Might be interesting.

Roger
ID: 1969385 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5124
Credit: 276,046,078
RAC: 462
Message 1969449 - Posted: 9 Dec 2018, 5:03:47 UTC - in response to Message 1969380.  
Last modified: 9 Dec 2018, 5:04:16 UTC


No it isn't part of the old Lunatics package. It is another package developed by Petri33 and TBar. The package is located at the Crunchers Anonymous site. But I just discovered that all the posts and links for the special app have been removed. Need to investigate why.


Keith,
Tbar mentioned he was going to take a vacation and go hang with Jason on another thread.

It is possible that the missing links are due to that.

Tom
A proud member of the OFA (Old Farts Association).
ID: 1969449 · Report as offensive
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 11361
Credit: 29,581,041
RAC: 66
United States
Message 1969452 - Posted: 9 Dec 2018, 5:23:30 UTC - in response to Message 1969449.  

Tbar mentioned he was going to take a vacation and go hang with Jason on another thread.

What thread is that?
ID: 1969452 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1969454 - Posted: 9 Dec 2018, 5:44:48 UTC - in response to Message 1969452.  

Tbar mentioned he was going to take a vacation and go hang with Jason on another thread.

What thread is that?

https://setiathome.berkeley.edu/forum_thread.php?id=83307&postid=1968742
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1969454 · Report as offensive
Previous · 1 . . . 9 · 10 · 11 · 12

Message boards : Number crunching : GTX is dead long live RTX !!


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.