GPU FLOPS: Theory vs Reality

Message boards : Number crunching : GPU FLOPS: Theory vs Reality
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 13 · 14 · 15 · 16 · 17 · 18 · 19 . . . 20 · Next

AuthorMessage
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2018718 - Posted: 13 Nov 2019, 0:40:55 UTC
Last modified: 13 Nov 2019, 0:57:44 UTC

Excellent work Shaggie.

As expected the clear winners in credit/watt performance are the 1650Ti or not, 2060 super & 2070 as expected.

@Tom
You just need to remember this graphs are made with GPU who run stock builds, not the ones we use.
In our case we expect similar performances but that is to be proved.
ID: 2018718 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5126
Credit: 276,046,078
RAC: 462
Message 2018721 - Posted: 13 Nov 2019, 0:53:59 UTC - in response to Message 2018718.  

Excellent work Shaggie.

+1

As expected the clear winners if credit/watt performance are the 1650Ti or not, 2060 super & 2070 as expected.

@Tom
You just need to remember this graphs are made with GPU who run stock builds, not the ones we use.
In our case we expect similar performances but that is to be proved.


I understand that the graphs don't include the "beta test" apps like the petri/tBars special sauce. I am counting on a large enough sample that any extreme results will not be very noticeable.

Back when the gtx 750Ti was the king of the efficiency hill I got pretty good results with the "special sauce". So this is a relative scale of competitiveness rather than an absolute predictor of watts/performance.
I am comfortable with the results.

Tom
A proud member of the OFA (Old Farts Association).
ID: 2018721 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36811
Credit: 261,360,520
RAC: 489
Australia
Message 2018736 - Posted: 13 Nov 2019, 2:16:43 UTC

I understand that the graphs don't include the "beta test" apps like the petri/tBars special sauce. I am counting on a large enough sample that any extreme results will not be very noticeable.
Actually it doesn't count any of us who run under the "Anonymous Platform" banner (only those who run fully stock setups get counted). ;-)

Cheers.
ID: 2018736 · Report as offensive
catavalon21

Send message
Joined: 2 Nov 01
Posts: 13
Credit: 7,238,152
RAC: 48
United States
Message 2018862 - Posted: 14 Nov 2019, 0:03:54 UTC - in response to Message 2018703.  

Thanks for the update. I have been scouring the used market for a 1080, and am blown away that that 1660ti goes nearly toe to toe with it. Very impressive. Rethinking the path forward, that's for sure.

Thank you for this info - it's awesome!
ID: 2018862 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 2018913 - Posted: 14 Nov 2019, 6:51:11 UTC - in response to Message 2018862.  

Thanks for the update. I have been scouring the used market for a 1080, and am blown away that that 1660ti goes nearly toe to toe with it. Very impressive. Rethinking the path forward, that's for sure.
And although it's not on the list yet (since it's just been released) the GTX 1660 Super is meant to be on par (or even slightly better) than the 1660Ti, and it's cheaper.
Grant
Darwin NT
ID: 2018913 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 2018914 - Posted: 14 Nov 2019, 6:51:57 UTC - in response to Message 2018703.  

The RTX 2060 and 2070 SUPERs are included now -- some excellent performance for both of them depending on what you're looking for.
Thanks again for your efforts.
Grant
Darwin NT
ID: 2018914 · Report as offensive
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 2018929 - Posted: 14 Nov 2019, 12:43:55 UTC - in response to Message 2018913.  

Thanks for the update. I have been scouring the used market for a 1080, and am blown away that that 1660ti goes nearly toe to toe with it. Very impressive. Rethinking the path forward, that's for sure.
And although it's not on the list yet (since it's just been released) the GTX 1660 Super is meant to be on par (or even slightly better) than the 1660Ti, and it's cheaper.

I just checked and in the last scan there were four GTX 1660 SUPERs -- I'll give it a few weeks and scan again.
ID: 2018929 · Report as offensive
catavalon21

Send message
Joined: 2 Nov 01
Posts: 13
Credit: 7,238,152
RAC: 48
United States
Message 2018954 - Posted: 14 Nov 2019, 16:05:28 UTC - in response to Message 2018712.  

Agreed. Legacy data will always still be posted here. Go for it!
ID: 2018954 · Report as offensive
catavalon21

Send message
Joined: 2 Nov 01
Posts: 13
Credit: 7,238,152
RAC: 48
United States
Message 2018955 - Posted: 14 Nov 2019, 16:09:11 UTC - in response to Message 2018929.  

I just checked and in the last scan there were four GTX 1660 SUPERs -- I'll give it a few weeks and scan again.


Much appreciated.
ID: 2018955 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 2019091 - Posted: 15 Nov 2019, 7:16:57 UTC - in response to Message 2018929.  

I just checked and in the last scan there were four GTX 1660 SUPERs -- I'll give it a few weeks and scan again.
People certainly aren't wasting any time putting those in their systems (it's also nice that they're actually available, and at a reasonable price).
Grant
Darwin NT
ID: 2019091 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2019108 - Posted: 15 Nov 2019, 13:25:49 UTC - in response to Message 2019091.  
Last modified: 15 Nov 2019, 13:39:44 UTC

I just checked and in the last scan there were four GTX 1660 SUPERs -- I'll give it a few weeks and scan again.
People certainly aren't wasting any time putting those in their systems (it's also nice that they're actually available, and at a reasonable price).

I just not imagine why somebody uses this new GPU's and choose to run stock apps.
This is IMHO something to think about. If you look the list of the top hosts most of them uses optimized apps (Linux, Windows, or whatever) , so you not sum to the numbers them, in theory, most optimized hosts. They rely reflects the real performance of the GPU's/CPU's combinations.
And there are another variable on the equation, the mining hosts who has huge amounts of GPU's driven normally by relatively small CPU's.
ID: 2019108 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22535
Credit: 416,307,556
RAC: 380
United Kingdom
Message 2019109 - Posted: 15 Nov 2019, 14:00:59 UTC

It is quite simple to understand, choose from any one of:
- Not everyone knows about the optimised application, -
- Not everyone wants to run Linux,
- Many people are in the "fit and forget" camp, and just let BOINC do what BOINC does,
- Not everyone is allowed to run Linux,
- Not everyone can run Linux.
I dare say there are more reasons, but those will do for starters.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 2019109 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2019114 - Posted: 15 Nov 2019, 14:22:40 UTC - in response to Message 2019109.  
Last modified: 15 Nov 2019, 14:26:14 UTC

I'm not talking only about Linux, there are Windows Optimized apps too. I'm sure you know that.

BTW I'm not a Linux fan, i only use Linux on my main cruncher because the apps are rely a lot faster.
The rest of my computers uses Windows.

<edit>
- Not everyone knows about the optimised application, -

IMHO that's a shame, if they exist and are faster than the stock apps and produce valid science, their use must be encouraged!

But this all is out topic. Sorry Shaggie76
ID: 2019114 · 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 2019115 - Posted: 15 Nov 2019, 14:22:55 UTC - in response to Message 2019109.  

It is quite simple to understand, choose from any one of:
- Not everyone knows about the optimised application, -
- Not everyone wants to run Linux,
- Many people are in the "fit and forget" camp, and just let BOINC do what BOINC does,
- Not everyone is allowed to run Linux,
- Not everyone can run Linux.
I dare say there are more reasons, but those will do for starters.


The optimized apps encompass much more than just the special app.

What Juan is talking about, is how the scans do not capture anything Anonymous Platform, that includes the people who have used the Lunatics installers.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2019115 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2019142 - Posted: 15 Nov 2019, 17:05:33 UTC - in response to Message 2019114.  

IMHO that's a shame, if they exist and are faster than the stock apps and produce valid science, their use must be encouraged!
Since we're talking about GPUs, the Windows stock apps are the same as the Lunatics installer apps. Lunatics supplied the optimised apps to the project for use as stock.

Only the CPU apps are truly optimised. But you can't half-optimise under Anonymous Platform - it has to be all or nothing. That's why the installer had to cover all apps.
ID: 2019142 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 2019175 - Posted: 15 Nov 2019, 21:12:22 UTC - in response to Message 2019108.  

I just not imagine why somebody uses this new GPU's and choose to run stock apps.
As Rob pointed out, in many cases people just don't know about optimised applications.
And if you look at many of the systems running stock, they also aren't aware of command line values either- their hardware is running with the default settings that were done so even the lowest end hardware could crunch without impacting on system performance.
These days, pretty much everything from the RTX 2060 & the GTX 1660Ti and up could run the SoG application with the most aggressive settings possible, and have little (if any) effect on system responsiveness. The only other thing the user would need to do would be to reserve a CPU core to support the GPU in order to not impact on their CPU processing times.
Grant
Darwin NT
ID: 2019175 · Report as offensive
wujj123456

Send message
Joined: 5 Sep 04
Posts: 40
Credit: 20,877,975
RAC: 219
China
Message 2019290 - Posted: 16 Nov 2019, 20:20:49 UTC - in response to Message 2019109.  
Last modified: 16 Nov 2019, 20:22:10 UTC

It is quite simple to understand, choose from any one of:
- Not everyone knows about the optimised application, -
- Not everyone wants to run Linux,
- Many people are in the "fit and forget" camp, and just let BOINC do what BOINC does,
- Not everyone is allowed to run Linux,
- Not everyone can run Linux.
I dare say there are more reasons, but those will do for starters.

As someone in "fit and forget" camp, I just want to contribute free compute time to science, not keep messing with configurations or profiles. I don't care all that much about credits either. I never bothered to look up optimizations until I noticed my WUs are failing consistently last weekend. Then I learnt the optimized apps thanks to all helpful people here to both solve the failure and get a boost in efficiency with same power.

I probably spent at most half an hour to set up a new Ubuntu install and those optimized apps on my secondary PC. Linux is both my job and work environment so I have no learning curve. However, I spent another few hours messing with all other apps like Steam, controllers, browsers, etc to ensure I can do what I normally do in Windows on this secondary PC. My initial try was actually a diskless boot as well just so I have no chance of messing up Windows install. I still run Windows exclusively on my main gaming PC since I don't want to bother with managing two installs and frequently reboot between the two OSes.

Use it normally is why most people bought their computers I suppose. I probably bought more powerful CPUs than I would otherwise need for BOINC, but that's about it. That's why I think it's very important such optimizations are provided by project by default and stop providing WUs to computers with faulty drivers. I am sure the admins running the projects is constantly weighing between spending more time in optimization and other stuff they do.
ID: 2019290 · 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 2019299 - Posted: 16 Nov 2019, 21:58:50 UTC

To prevent harmful interaction between an existing Windows OS installation, I alway recommend purchasing a cheap 128GB SSD for $25 and installing Linux onto that. Then just switch between the two OS' by using the BIOS boot choice. That way you can run your normal Windows OS without it ever knowing about the Linux installation and vice versa.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2019299 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36811
Credit: 261,360,520
RAC: 489
Australia
Message 2019303 - Posted: 16 Nov 2019, 22:09:56 UTC - in response to Message 2019299.  

To prevent harmful interaction between an existing Windows OS installation, I alway recommend purchasing a cheap 128GB SSD for $25 and installing Linux onto that. Then just switch between the two OS' by using the BIOS boot choice. That way you can run your normal Windows OS without it ever knowing about the Linux installation and vice versa.
I used a larger SSD, but yes that is how mine 2 rigs work. Connect SSD, disconnect Windows drive/s, install Linux, reconnect Windows drive/s, go into BIOS and select your preferred boot drive, then use whichever F key (F12 in my case) to boot into the other OS when desired.

Cheers.
ID: 2019303 · Report as offensive
wujj123456

Send message
Joined: 5 Sep 04
Posts: 40
Credit: 20,877,975
RAC: 219
China
Message 2019385 - Posted: 17 Nov 2019, 7:57:06 UTC - in response to Message 2019303.  

To prevent harmful interaction between an existing Windows OS installation, I alway recommend purchasing a cheap 128GB SSD for $25 and installing Linux onto that. Then just switch between the two OS' by using the BIOS boot choice. That way you can run your normal Windows OS without it ever knowing about the Linux installation and vice versa.
I used a larger SSD, but yes that is how mine 2 rigs work. Connect SSD, disconnect Windows drive/s, install Linux, reconnect Windows drive/s, go into BIOS and select your preferred boot drive, then use whichever F key (F12 in my case) to boot into the other OS when desired.

Cheers.

Usually I run update-grub when in Linux distro and it will find my Windows installation and add to grub menu. Then I can always boot from Linux and select Windows in menu if needed, without having to press a key at the right time. However, I couldn't get it working with my Windows 10, with either ntldr or chainloader command. None of them can find my Windows boot partition or efi file even though I am fairly confident I passed in the right uuid. I wonder if it has anything to do with bitlocker, but AFAIK it should only encrypt the partitions I specified, not including the boot partition.

Has anyone got grub to recognize Windows under UEFI when bitlocker is turned on? Mind sharing the grub menuentry?
ID: 2019385 · Report as offensive
Previous · 1 . . . 13 · 14 · 15 · 16 · 17 · 18 · 19 . . . 20 · Next

Message boards : Number crunching : GPU FLOPS: Theory vs Reality


 
©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.