I have a GeForce TITAN, there is there any GPU units for my card?

Tank you for you reply.

@+
*_*

https://developer.nvidia.com/category/zone/cuda-zone

---

You can see by yourself.
Every project have a page 'Applications' (usually linked on the Home page)

You can see that Asteroids@home have no GPU app for now:
http://asteroidsathome.net/boinc/apps.php

And this is how it looks on other projects:
http://setiathome.berkeley.edu/apps.php
http://einstein.phys.uwm.edu/apps.php
http://milkyway.cs.rpi.edu/milkyway/apps.php

If you see words like 'cuda', 'nvidia', 'ati', 'opencl' only then those apps are for the relevant GPU

---

- ALF - "Find out what you don't do well ..... then don't do it!" :)

Will be released a version for OpenCL?

Yes, but it still take a while.

I can't imagine how quick the WU times will be seeing as it only takes my aging C2 quad ~39mins/WU ;)

Any idea roughly when a GPU app will be available?

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

I have alpha version of CUDA app. It's not ready even for beta testing. I will start new thread about CUDA in forum soon.

I don't understand why bother with CUDA (first) when there are tons of unused OpenCL-capable GPGPU's left after WCG - HCC1 project finished. I, for example, am waiting for some interesting OpenCL GPU projects to utilize my lower-range 7750 and 6770M and AAH would be one of them :-)

Good luck anyway and cheers :V:

---

For Science and Shrubberies! Since January 13 2000.

@branjo,

Today I heard Einstein offers an OpenCL app.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

Einstein, Albert, Poem, Seti ... there are multiple projects supporting AMD OpenCL cards.

I think the choice of CUDA over OpenCL may be about more the developers having existing experience with CUDA, more mature tools etc.

I think the choice of CUDA over OpenCL may be about more the developers having existing experience with CUDA, more mature tools etc.

Exactly.

2 Dagorath & Jan Vaclavik. Thanks guys, time-to-time I am crunching EAH and SAH as well as Collatz, PrimeGrid and MilkyWay. But for one reason or another, I am not willing to dedicate a bigger part of my electricity bills to them :-). AAH seems to be worth it :wink:

2 HA-SOFT, s.r.o.: it is really important reason, thanks for reply and good luck :-)

Cheers, NI!

---

For Science and Shrubberies! Since January 13 2000.

2 HA-SOFT, s.r.o.: it is really important reason, thanks for reply and good luck :-)

For now yes. I know CUDA so no problem to do CUDA. But for OpenCL we need to reorganize codebase to add all code to one .cl file (or with includes). CUDA has linker so I have the same c and header files as in CPU app (code inside c file is changed to better meet CUDA architecture, of course).

We would like to support both nVIDIA and ATI.

Happy crunching
Z

I have a spare Nvidia GPU, when you feel like releasing an alpha build. Well I am able to test, at least.

I have a spare Nvidia GPU, when you feel like releasing an alpha build. Well I am able to test, at least.

I'm preparing it now.

Also willing to alpha test, Linux or WinXP (dual-boot), with GTX 660ti.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

Also willing to alpha test, Linux or WinXP (dual-boot), with GTX 660ti.

Do not expect to much from 6xx cards due to their double precision slowness.

Also willing to alpha test, Linux or WinXP (dual-boot), with GTX 660ti.

Do not expect to much from 6xx cards due to their double precision slowness.

No problem.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

HA Soft

Good to hear things moving on the GPU front :)
What card are you alpha testing the GPU app on? What WU times are you getting?

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

Also willing to alpha test, Linux or WinXP (dual-boot), with GTX 660ti.

Do not expect to much from 6xx cards due to their double precision slowness.

No problem.

OK, problem, I just didn't realize it until now. Also, I intend to support A@H with a GPU one way or another, not bailing out here, just looking for advice and an affordable path to decent performance.

The 5xx cards have double precision (DP) crippled. They are said to have 1/8 single precision performance, in other words 1/8 SP.

The 6xx cards also have double precision (DP) crippled but even more crippled than 5xx. They have DP = 1/24 SP. (not sure if the = is appropriate, perhaps -> is more correct?)

The 7xx cards are pretty much just 6xx cards with faster clocks so DP performance isn't much better than 6xx. The fastest 7xx for DP is the recently released 780Ti but it's DP is still crippled.

The best DP performer at a price the average cruncher can afford is Titan. Actually it's a 7xx too but they didn't cripple DP on Titan so it's sort of in a class separate from the other 7xx cards.

Here's a chart I derived from info I found. I'm not a GPU guru so don't take the chart as gospel, it's open to corrections, additions, updates, whatever.

================================================================
| model | price |  peak DP  |  TDP    | GFlops/$  | GFlops/watt |
|       |($ CDN)| (GFlops)  | (watts) |           |             |
=================================================================
| 560Ti | 189   | 104 - 162 | 170-210 | .55-.86   | .61 - .77   |
-----------------------------------------------------------------
| 660Ti | 265   |  103      | 150     |  .39      |  .69        |
-----------------------------------------------------------------
| 780Ti | 760   | 230       | 250     |  .30      |   .92       |
-----------------------------------------------------------------
| Titan | 1,030 | 1300-1500 | 250     | 1.26-1.46 |  4.12-6.0   |
-----------------------------------------------------------------
| Tesla | 3,499 | 1173      | 225     | .34       |   5.2       |
-----------------------------------------------------------------

(Prices are from newegg.ca, performance figures are from http://en.wikipedia.org/wiki/Comparison_of_Nvidia_graphics_processing_units)


See the problem? Everything sucks for this project except Titan and Titan costs $1,030 CDN. The suck factor is ~4 for 780Ti and ~6 for the other models except Tesla which is ~$3,500. Below Titan, the 5xx cards give the best performance/dollar but the worst performance per watt.

Again, I'm a newbie at the GPU crunching game. Does anybody have some encouraging news? I mean news other than "you can afford a Titan if you go dumpster diving for bottles and cans".

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

HA Soft

Good to hear things moving on the GPU front :)
What card are you alpha testing the GPU app on? What WU times are you getting?

My development environment is cheap msi notebook with w7 and gtx660m. Yes mobile gpu. It is horible slow, but enought for profiling and debuging for now. I'm using vs2010 + cuda 5.0 + last nvidia nSight.

I plan to test app first on my titan on linux for speed. I think 5xx should be ok and gk110 7xx with dp enabled.

PS:
When I did optimized app for Distrrtgen I had used titan on my work pc and then I had problems with slow cards (which were all others except Tesla). So now I have decided to use slow card for development.