Considering new Graphics card

Message boards : Number crunching : Considering new Graphics card
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 5 · Next

AuthorMessage
Profile Blurf
Volunteer tester

Send message
Joined: 2 Sep 06
Posts: 8962
Credit: 12,678,685
RAC: 0
United States
Message 1534594 - Posted: 1 Jul 2014, 23:42:32 UTC

As a gift to myself for finishing Relay for Life season in spectacular fashion, I am buying a new graphics card.

Currently have a NVIDIA GeForce GTX 460. Can't break the bank for over $250 or so.

Any suggestions?


ID: 1534594 · Report as offensive
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 11360
Credit: 29,581,041
RAC: 66
United States
Message 1534615 - Posted: 2 Jul 2014, 0:13:49 UTC - in response to Message 1534594.  

I'd look at a GTX 750, real cheap and very low power usage. You could buy 2 on your budget.
ID: 1534615 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1534619 - Posted: 2 Jul 2014, 0:19:14 UTC - in response to Message 1534594.  

Is $250 the max? If so then you could get a EVGA GTX 760 SC for that much over at Newegg.com If you don't want to spend that much, then could step down to several other cards. The GTX 750 are nice, run decent and cool. Theres a GTX 750 Ti Superclocked with ACX for $140 also at newegg. You will need a 6 pin connector for it. I'm sure you will get numerous suggestions. Those are just my 2 picks.

Happy Crunching...


Zalster
ID: 1534619 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1534651 - Posted: 2 Jul 2014, 1:16:06 UTC

A GTX 660Ti should fit that price or a GTX660 for under $200.

Cheers.
ID: 1534651 · Report as offensive
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 11360
Credit: 29,581,041
RAC: 66
United States
Message 1534665 - Posted: 2 Jul 2014, 2:31:59 UTC - in response to Message 1534651.  

A GTX 660Ti should fit that price or a GTX660 for under $200.

I really like mine, however I think 2 GTX 750s would be more better.
ID: 1534665 · Report as offensive
Profile Blurf
Volunteer tester

Send message
Joined: 2 Sep 06
Posts: 8962
Credit: 12,678,685
RAC: 0
United States
Message 1534671 - Posted: 2 Jul 2014, 3:11:45 UTC

Just buying one but I'll look at the 660. Thanks!


ID: 1534671 · Report as offensive
tbret
Volunteer tester
Avatar

Send message
Joined: 28 May 99
Posts: 3380
Credit: 296,162,071
RAC: 40
United States
Message 1534699 - Posted: 2 Jul 2014, 5:28:11 UTC - in response to Message 1534594.  
Last modified: 2 Jul 2014, 5:30:57 UTC



Currently have a NVIDIA GeForce GTX 460. Can't break the bank for over $250 or so.

Any suggestions?


I have a question, then maybe a suggestion.

Why are you going to replace the 460? Is there something wrong with it, or are you tired of looking at it, or do you want something that's better for games, or... what?

Recently I let one of my two 460s go with a machine I built for a friend, but the other one is chugging-along just fine, crunching like it always has.

Based on crunching one AP at a time, the 460 is taking about 2,200 seconds to crunch-through an AP task.

A brand new (I just installed it) AMD R9 270x seems to be doing a single AP task in 2,000 seconds.

Both of those measures are "more-or-less" but close enough.

What I'm about to tell you is patently unfair, but you might find it useful. My 660Tis seem to be running *two at a time* in about 4,500 seconds using the Lunatics' optimized applications or about 2,250/work unit. I honestly think it would be unfair to say they wouldn't run faster than that if they were doing one work unit at a time. On the other hand, the 660Ti never impressed me. It must not have impressed anyone else, either. They have gone away.

One of my 750Ti's is doing a single AP in about 5,000 seconds. The other one in a really old P4, but running XP instead of Windows 7, is doing a single AP work unit in about 4,500 seconds.

Okay, conclusion: If I were shopping for a card in that price range I would be looking at either a AMD R9 280 (to get the 384-bit memory) or a factory superclocked version of the GTX 760.

Now the question is; Do I have a heat problem?

If yes, then I'd probably go with this one or the equivalent because it blows hot air out of its... exhaust:

Overclock 760

If no, then I'd give some serious thought to the R9 280. (notice there is no "X")

HOWEVER --- You know what AMD drivers are like. The AMD would probably be really good at some of the Einstein work. I don't have any idea what's best at the other projects you run.

Congratulations on spectacular finish.
ID: 1534699 · Report as offensive
Profile Blurf
Volunteer tester

Send message
Joined: 2 Sep 06
Posts: 8962
Credit: 12,678,685
RAC: 0
United States
Message 1534719 - Posted: 2 Jul 2014, 6:21:37 UTC

Thx Tberet---I like to have one machine AMD and one Nvidia.

As to the heat issue-yes the room gets warm so reducing the heat would be a great idea.

Thank you


ID: 1534719 · Report as offensive
tbret
Volunteer tester
Avatar

Send message
Joined: 28 May 99
Posts: 3380
Credit: 296,162,071
RAC: 40
United States
Message 1534765 - Posted: 2 Jul 2014, 7:57:42 UTC - in response to Message 1534719.  

Oh crud -

An Edit didn't take Blurf.

I discovered after my original post that the 750Ti on the P4 is running two at a time!!!

I also discovered that the one on the AMD has down-clocked. I'll have to deal with that, but the 750Ti that should be CPU-starved is moving right along!
ID: 1534765 · Report as offensive
BetelgeuseFive Project Donor
Volunteer tester

Send message
Joined: 6 Jul 99
Posts: 158
Credit: 17,117,787
RAC: 19
Netherlands
Message 1534860 - Posted: 2 Jul 2014, 12:44:51 UTC - in response to Message 1534699.  

This is weird. Your 750Ti should be faster than by (non-Ti) 750.
However, my 750 will do a single AP (no blanking) in 2700 seconds.
I really like the GTX-750. It is silent, does not run hot and it performs very well.

Just my 2 cents ...

Tom




Currently have a NVIDIA GeForce GTX 460. Can't break the bank for over $250 or so.

Any suggestions?


I have a question, then maybe a suggestion.

Why are you going to replace the 460? Is there something wrong with it, or are you tired of looking at it, or do you want something that's better for games, or... what?

Recently I let one of my two 460s go with a machine I built for a friend, but the other one is chugging-along just fine, crunching like it always has.

Based on crunching one AP at a time, the 460 is taking about 2,200 seconds to crunch-through an AP task.

A brand new (I just installed it) AMD R9 270x seems to be doing a single AP task in 2,000 seconds.

Both of those measures are "more-or-less" but close enough.

What I'm about to tell you is patently unfair, but you might find it useful. My 660Tis seem to be running *two at a time* in about 4,500 seconds using the Lunatics' optimized applications or about 2,250/work unit. I honestly think it would be unfair to say they wouldn't run faster than that if they were doing one work unit at a time. On the other hand, the 660Ti never impressed me. It must not have impressed anyone else, either. They have gone away.

One of my 750Ti's is doing a single AP in about 5,000 seconds. The other one in a really old P4, but running XP instead of Windows 7, is doing a single AP work unit in about 4,500 seconds.

Okay, conclusion: If I were shopping for a card in that price range I would be looking at either a AMD R9 280 (to get the 384-bit memory) or a factory superclocked version of the GTX 760.

Now the question is; Do I have a heat problem?

If yes, then I'd probably go with this one or the equivalent because it blows hot air out of its... exhaust:

Overclock 760

If no, then I'd give some serious thought to the R9 280. (notice there is no "X")

HOWEVER --- You know what AMD drivers are like. The AMD would probably be really good at some of the Einstein work. I don't have any idea what's best at the other projects you run.

Congratulations on spectacular finish.
ID: 1534860 · Report as offensive
Profile arkayn
Volunteer tester
Avatar

Send message
Joined: 14 May 99
Posts: 4438
Credit: 55,006,323
RAC: 0
United States
Message 1534926 - Posted: 2 Jul 2014, 15:56:44 UTC - in response to Message 1534619.  

Is $250 the max? If so then you could get a EVGA GTX 760 SC for that much over at Newegg.com If you don't want to spend that much, then could step down to several other cards. The GTX 750 are nice, run decent and cool. Theres a GTX 750 Ti Superclocked with ACX for $140 also at newegg. You will need a 6 pin connector for it. I'm sure you will get numerous suggestions. Those are just my 2 picks.

Happy Crunching...


Zalster


My 760 ran AP (0% blanking) in 2900 seconds, it was also running an MB at the same time.

It does a Primegrid task in 718 seconds. The 660 takes 927 seconds.
The 670 takes 637 seconds.

ID: 1534926 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1534959 - Posted: 2 Jul 2014, 17:09:48 UTC - in response to Message 1534926.  

serprising my 650 runs units in 20mins and does 4 at a time so the 750 isn't much faster if at all than a 650 open cl units in aprox 5hrs and astropulse in 9 hrs I'd go 650 and buy 2 and then crunch 8 units at a time
ID: 1534959 · Report as offensive
CZ

Send message
Joined: 18 May 06
Posts: 11
Credit: 1,278,969
RAC: 0
Australia
Message 1535000 - Posted: 2 Jul 2014, 19:11:28 UTC - in response to Message 1534594.  
Last modified: 2 Jul 2014, 19:13:56 UTC

The following lists show the most productive GPU models on different platforms. Relative speeds are shown in parentheses http://setiathome.berkeley.edu/gpu_list.php

NVIDIA


Total
1.(1.000) GeForce GTX 580
2.(0.867) GeForce GTX 770
3.(0.741) GeForce GTX 780
4.(0.710) Quadro K4000
5.(0.699) GeForce GTX 680MX
6.(0.601) GeForce GTX 670
7.(0.587) GeForce GTX 470
8.(0.583) GeForce GTX 680
9.(0.558) GeForce GTX 560 Ti
10.(0.556) GeForce GTX 560
11.(0.542) GeForce GTX 570
12.(0.526) GeForce GTX 660 Ti
13.(0.515) GeForce GTX 650 Ti BOOST
14.(0.472) GeForce GTX 260
15.(0.458) GeForce GTX 660
16.(0.443) Quadro 4000
17.(0.397) GeForce GTX 460
18.(0.396) GeForce GTX 550 Ti
19.(0.321) GeForce GTX 750 Ti
20.(0.312) GeForce GTX 760
21.(0.307) GeForce GTX 650 Ti
22.(0.293) Quadro 2000
23.(0.283) GeForce GTX 650
24.(0.260) GeForce GTS 250
25.(0.236) GeForce GT 640
26.(0.214) GeForce GT 650M
27.(0.210) GeForce GT 440
28.(0.205) GeForce 9800 GT
29.(0.173) GeForce 9600 GT
30.(0.157) GeForce GTS 450
31.(0.150) Quadro 600
32.(0.145) GeForce GT 430
33.(0.109) GeForce GTX 275
34.(0.047) GeForce 210

Windows
1.(1.000) GeForce GTX 580
2.(0.867) GeForce GTX 770
3.(0.741) GeForce GTX 780
4.(0.710) Quadro K4000
5.(0.699) GeForce GTX 680MX
6.(0.652) GeForce GTX 670
7.(0.587) GeForce GTX 470
8.(0.583) GeForce GTX 680
9.(0.558) GeForce GTX 560 Ti
10.(0.556) GeForce GTX 560
11.(0.542) GeForce GTX 570
12.(0.536) GeForce GTX 660
13.(0.526) GeForce GTX 660 Ti
14.(0.515) GeForce GTX 650 Ti BOOST
15.(0.472) GeForce GTX 260
16.(0.443) Quadro 4000
17.(0.431) GeForce GTX 650 Ti
18.(0.399) GeForce GTX 550 Ti
19.(0.397) GeForce GTX 460
20.(0.321) GeForce GTX 750 Ti
21.(0.312) GeForce GTX 760
22.(0.293) Quadro 2000
23.(0.283) GeForce GTX 650
24.(0.260) GeForce GTS 250
25.(0.236) GeForce GT 640
26.(0.214) GeForce GT 650M
27.(0.210) GeForce GT 440
28.(0.205) GeForce 9800 GT
29.(0.173) GeForce 9600 GT
30.(0.167) GeForce GTS 450
31.(0.150) Quadro 600
32.(0.145) GeForce GT 430
33.(0.109) GeForce GTX 275
34.(0.047) GeForce 210

Linux
1.(1.000) GeForce GTX 660
2.(0.698) GeForce GTX 650 Ti
3.(0.629) GeForce GTS 450

Mac
No GPU tasks reported

ATI/AMD


Total
1.(1.000) AMD Radeon HD 7870/7950/7970/R9 280X series (Tahiti)
2.(0.778) AMD Radeon HD 7850/7870 series (Pitcairn)
3.(0.661) AMD Radeon HD 6900 series (Cayman)
4.(0.591) AMD Radeon HD 7870/7950/7970 series (Tahiti)
5.(0.454) AMD Radeon HD 7700 series (Capeverde)
6.(0.404) AMD Radeon HD 6790/6850/6870 series (Barts)
7.(0.293) ATI Radeon HD 5800/5900 series (Cypress/Hemlock)
8.(0.205) AMD Radeon HD 7500/7600/8500/8600 series (Devastator)
9.(0.196) AMD Radeon HD 6570/6670/7570/7670 series (Turks)
10.(0.182) ATI Radeon HD 5700/6750/6770 series (Juniper)
11.(0.180) ATI Radeon HD 5500/5600 series (Redwood)
12.(0.060) AMD Radeon HD 6350/6450/7450/7470 series (Caicos)
13.(0.041) ATI Radeon HD 5400 series (Cedar)
14.(0.031) ATI Radeon HD 2300/2400/3200/4200 (RV610)

Windows
1.(1.000) AMD Radeon HD 7870/7950/7970/R9 280X series (Tahiti)
2.(0.770) AMD Radeon HD 7850/7870 series (Pitcairn)
3.(0.648) AMD Radeon HD 6900 series (Cayman)
4.(0.450) AMD Radeon HD 7700 series (Capeverde)
5.(0.400) AMD Radeon HD 6790/6850/6870 series (Barts)
6.(0.289) ATI Radeon HD 5800/5900 series (Cypress/Hemlock)
7.(0.203) AMD Radeon HD 7500/7600/8500/8600 series (Devastator)
8.(0.194) AMD Radeon HD 6570/6670/7570/7670 series (Turks)
9.(0.181) ATI Radeon HD 5700/6750/6770 series (Juniper)
10.(0.178) ATI Radeon HD 5500/5600 series (Redwood)
11.(0.059) AMD Radeon HD 6350/6450/7450/7470 series (Caicos)
12.(0.041) ATI Radeon HD 5400 series (Cedar)
13.(0.031) ATI Radeon HD 2300/2400/3200/4200 (RV610)

Linux
1.(1.000) AMD Radeon HD 6900 series (Cayman)
2.(0.682) AMD Radeon HD 7870/7950/7970/R9 280X series (Tahiti)
3.(0.353) ATI Radeon HD 5800/5900 series (Cypress/Hemlock)
4.(0.000) AMD Radeon HD 7870/7950/7970 series (Tahiti)

Mac
No GPU tasks reported

Intel


Windows
1.(1.000) Intel(R) HD Graphics 4600
2.(0.877) Intel(R) HD Graphics 4400
3.(0.768) Intel(R) HD Graphics 4000
4.(0.622) Intel(R) HD Graphics 2500
5.(0.426) Intel(R) HD Graphics

Linux
No GPU tasks reported
Mac
No GPU tasks reported
ID: 1535000 · 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 1535001 - Posted: 2 Jul 2014, 19:17:19 UTC

Becareful with this list, it only apears GPUs who run stock apps, so anyone who is optimized is out of the list and most of the heavy users actualy runs optimized.
ID: 1535001 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1535015 - Posted: 2 Jul 2014, 19:47:44 UTC

Yes, that list is very misleading by not including anonymous platforms it's also based on the numbers of those cards being used.

So that that page with a pinch of salt.

Cheers.
ID: 1535015 · Report as offensive
CZ

Send message
Joined: 18 May 06
Posts: 11
Credit: 1,278,969
RAC: 0
Australia
Message 1535024 - Posted: 2 Jul 2014, 20:00:01 UTC - in response to Message 1535015.  
Last modified: 2 Jul 2014, 20:07:20 UTC

The list is NOT based on the number of cards used (but less popular cards can provide statistically less significant results).

While the fact that it may or may not include all platforms/optimised BOINC versions/etc is a factor that should be considered by those "power" users.

For the "average user" though the list is the best reference because it is the official ranking of card performance in performing SETI (GPU based) tasks and is based on hundreds of millions of results returned by all the GPUs over a very long time.

It is a fact though that the list is not going to reflect the performance of the card in other BOINC projects, gaming, or other uses it may be put to. Heavy GPU users/overclockers/etc also have their favourite cards and often there is an emotional or personal connection to a brand or model.

The list is also an "average performance" of the card - the cards will perform differently based on the motherboard/chipset/CPU/OS/Memory and a great number of other factors. So you will get as many different opinions as there are users.

So while far from perfect, this list is at least unbiased by peoples' personal preferences, and based on real statistical data.
ID: 1535024 · 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 1535034 - Posted: 2 Jul 2014, 20:24:37 UTC
Last modified: 2 Jul 2014, 20:27:11 UTC

The table is based on a failed statistical data so i can´t agree it is useful.

Look few examples the table say: 770 is faster than the 780, 670 is faster than the 680, the 560 is faster than the 570 etc.

Something we all know is totaly wrong.
ID: 1535034 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1535041 - Posted: 2 Jul 2014, 20:39:23 UTC

Believe as you like Chris, but that would be the last page that I would use to point out performance differences of cards to anyone. ;-)

Cheers.
ID: 1535041 · Report as offensive
tbret
Volunteer tester
Avatar

Send message
Joined: 28 May 99
Posts: 3380
Credit: 296,162,071
RAC: 40
United States
Message 1535048 - Posted: 2 Jul 2014, 20:54:11 UTC - in response to Message 1534860.  

This is weird. Your 750Ti should be faster than by (non-Ti) 750.
However, my 750 will do a single AP (no blanking) in 2700 seconds.
I really like the GTX-750. It is silent, does not run hot and it performs very well.



You may not have seen the edit, Tom. The 750Ti in question is doing two-at-a-time and I had forgotten setting it up that way. (lots of playing "musical computers" lately)

That means the 750Ti, which has a mild factory overclock on it, is taking 4,500 seconds to do two on a P4-based machine. An average of 2,250 per work unit.

The other 750Ti was found to be down-clocked (driver crash) due to someone switching users on it.

So my 750Ti numbers are low at this point.

I really should get a set-up, run stock on it, and put each card on that setup for a decent comparison. One of the reasons I don't is that it would be of most interest to people running the expensive cards and I'm just not very likely to buy expensive cards for SETI crunching.
ID: 1535048 · Report as offensive
Cruncher-American Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor

Send message
Joined: 25 Mar 02
Posts: 1513
Credit: 370,893,186
RAC: 340
United States
Message 1535054 - Posted: 2 Jul 2014, 21:23:19 UTC - in response to Message 1535000.  

If you believe that a GTX 470 is faster than a GTX 680, I have a card I will gladly trade you, if you have a spare 680.

I think that list is based on number of WUs done by that class of card, not the amount of work done per unit time or any other kind of performance measure.

When I switched both of my machines from 460s to 660tis/670s/680s, I got an immediate more-than-doubling of the GPU work done.
ID: 1535054 · Report as offensive
1 · 2 · 3 · 4 . . . 5 · Next

Message boards : Number crunching : Considering new Graphics card


 
©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.