Updated GPU Requirements

My event log says:
1/16/2015 4:28:31 PM | rosetta@home | Requesting new tasks for AMD/ATI GPU
1/16/2015 4:28:32 PM | | App version needs OpenCL but GPU doesn't support it
1/16/2015 4:28:32 PM | Milkyway@Home | [error] App version uses non-existent ATI GPU
1/16/2015 4:28:32 PM | Milkyway@Home | [error] Missing coprocessor for task de_80_DR8_Rev_8_5_00004_1413455402_44869832_0; aborting

So it seems my app is requesting tasks for a GPU I appear not to have? Why would that be?
I actually have a Radeon 4800, is that not supported?

According to the statistics, top gpu's it is #3 returning 0.828% of the total gpu work at the project. So my guess would be your drivers aren't the right ones to make it work yet.

The HD4800 series card supports OpenCL version 1.0. I think Milky Way requires version 1.1

The HD4800 series card supports OpenCL version 1.0. I think Milky Way requires version 1.1

Nope. I have several hosts with HD4850 GPUs (OpenCL 1.0) that have been crunching continuously here for a long time without problems. Here is one that joined up back in May 2009.

---

Cheers,
Gary.

Well, I think I finally figured out why MW@H isn't sending me any GPU WU's on my ATI machine. It see's my HD7950 and HD7850 as:

"Message from server: ATI GPU R600 (R38xx) does not support OpenCL"

I'm using the Catalyst 14.4 driver and E@H is going like gangbusters crunching multiple GPU WU's per card.

Basically, E@H has this machine all to itself right now and would like to thank MW@H for their generosity!

Regards,
Steve

Well, I think I finally figured out why MW@H isn't sending me any GPU WU's on my ATI machine. It see's my HD7950 and HD7850 as:

"Message from server: ATI GPU R600 (R38xx) does not support OpenCL"

I'm using the Catalyst 14.4 driver and E@H is going like gangbusters crunching multiple GPU WU's per card.

Basically, E@H has this machine all to itself right now and would like to thank MW@H for their generosity!

Regards,
Steve

See my reply in your thread.
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=3710&postid=63088#63088

---

Well, I think I finally figured out why MW@H isn't sending me any GPU WU's on my ATI machine. It see's my HD7950 and HD7850 as:

"Message from server: ATI GPU R600 (R38xx) does not support OpenCL"

I'm using the Catalyst 14.4 driver and E@H is going like gangbusters crunching multiple GPU WU's per card.

Basically, E@H has this machine all to itself right now and would like to thank MW@H for their generosity!

Regards,
Steve

See my reply in your thread.
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=3710&postid=63088#63088

Arkayn,

Thanks for the suggestion but I'm on Linux Mint (17.1), not Windows.

Regards,
Steve

Well, I think I finally figured out why MW@H isn't sending me any GPU WU's on my ATI machine. It see's my HD7950 and HD7850 as:

"Message from server: ATI GPU R600 (R38xx) does not support OpenCL"

I'm using the Catalyst 14.4 driver and E@H is going like gangbusters crunching multiple GPU WU's per card.

Basically, E@H has this machine all to itself right now and would like to thank MW@H for their generosity!

Regards,
Steve

See my reply in your thread.
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=3710&postid=63088#63088

Arkayn,

Thanks for the suggestion but I'm on Linux Mint (17.1), not Windows.

Regards,
Steve

Easily fixed with a new app_info.

Already posted in your thread.

---

Arkayn,

Thanks so much for the help. I'll get this onto my ATI machine right away.

Thanks again,
Steve

Could MilkyWay please clearly identify that or those application(s) on the MilkyWay@Home preferences page or the server status page which are GPU apps and which are not?
All that's shown is
http://milkyway.cs.rpi.edu/milkyway/prefs.php?subset=project

MilkyWay@Home: yes/no
MilkyWay@Home N-Body Simulation: yes/no
Milkyway@Home Separation: yes/no
Milkyway@Home Separation (Modified Fit): yes/no

http://milkyway.cs.rpi.edu/milkyway/server_status.php
application unsent

MilkyWay@Home _______________________ 275
MilkyWay@Home N-Body Simulation _______ 499
Milkyway@Home Separation ______________ 0
Milkyway@Home Separation (Modified Fit) ___ 343

Also, Please show on (one of) these page(s) which use multiple CPUs.
Thanks

---

I think, therefore I THINK I am.
My thinking's neither the source of my being--
not proves it to others.
God Is Love, Jesus proves it

Please take a look at the Applications page ...
http://milkyway.cs.rpi.edu/milkyway/apps.php

If the version includes ati or nvidia then it is GPU
If the version included mt then it is multi-threaded (cpu)

So I'm in the mist of doing upgrades to the pc's I'm running work on and need some direction here from you experts, so I have some questions and hope you can help answer them.

In terms of doing work on Boinc projects which is better to use: Nvidia or AMD? According to the first message on this board either will work but I'm looking to get more bang for my buck if you know what I mean.

If Nvidia, what should I be looking at when choosing a videocard, CUDA cores or CORE clock?

If AMD, what should I be looking at when choosing a videocard, CORE clock or STREAM processors.

And does video memory play a role with either card with projects?

Like I said, I'm no expert but I'm learning :-)

So I'm in the mist of doing upgrades to the pc's I'm running work on and need some direction here from you experts, so I have some questions and hope you can help answer them.

In terms of doing work on Boinc projects which is better to use: Nvidia or AMD? According to the first message on this board either will work but I'm looking to get more bang for my buck if you know what I mean.

If Nvidia, what should I be looking at when choosing a videocard, CUDA cores or CORE clock?

If AMD, what should I be looking at when choosing a videocard, CORE clock or STREAM processors.

And does video memory play a role with either card with projects?

Like I said, I'm no expert but I'm learning :-)

Both types work well, take a look at the top gpu's and users and you will get an idea of which model is the very best, be prepared to see some expensive ones though!! You can also check out your teammates pc's and see what they are using.

As for cuda cores or stream processors, yes they are very important, the more the better. As for on board memory, again the more the better.

Now we come to the hardest part the speed of the actual cuda cores or stream processors, yes it's important, but so is the thruput of the card as in 128 bit, 384 bit, 512 bit etc, that can make a card with a slower memory speed perform just as fast as a card with a faster memory speed.

BTW cuda cores and stream processors are NOT directly comparable to each other, you can only compare the number of cuda cores to other gpu's with cuda cores, and number of stream processors to other cards with stream processors.

Thanks for the response. I will do as you said and look at other users computer to see which is mostly used and go in that direction.

Seeing as you posted in the Milkway forum, the best cards for Milkyway (& possibly Einstein@home) are some of AMD's card due to their Double Precision power, although it's more the older generation cards. So HD 7870XT,7950,7970, R9 280 & 280X (Tahiti chip), .

For more recent AMD cards the R9 290 & 290X, R9 390 & 390X (Hawaii chip) are pretty decent, but only similar in DP power to the HD 7950 (single precision is obviously way ahead ;) ).
The R9 Fury & Nano would appear on paper to be ok on DP power too (yet less than the HD 7950) but I haven't seen any MW benchmarks for those yet.

The best single GPU card (at stock speeds) for MW@H is the 7970 GHz edition, with the R9 280X very close.

Whilst their are a few (top end) NVidia cards with more DP power they don't seem to do so well at MW (unless things have changed very recently), see my benchmarking thread for more info :) (go to the end of the thread for the latest table).

So what projects do you run & are likely to run?

---

Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG

Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit
2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7

I have both the 7970 and the R9 280x cards. Both are in computers with PCI-e v3 slots. There are differences as to the cpus and windows versions. But I found that if I set the core clock and memory clock the exact same, I cannot distinguish the speed between them. Both cards are capable of 41-42 sec running two tasks concurrently, as long as both tasks are not synchronize. So each day I check and make sure this is the case, otherwise when the tasks are synchronized, computation time increases to 44-45sec.

Performance of gpus is easy to check by looking at the reviews of AMD cards on Anandtech. They publish the specs of the chips in each card. The spec to look for is the SP to DP ratio. There are not many AMD cards that have a ration of 1/4. The graphic card vendors have been crippling their chips for a while now, as they make the high DP ratio available in the multi thousand $ cards. The nVidia cards have pretty pathetic SP/DP rations except for Titan cards, but those are $1000 or higher.

After I learned of this information last year, I went to look for refurbished or used 7970/280x cards. People were dumping them because they no longer made any money mining bitcoins. My local Microcenter store had several refurbished ones at less then $180, so I picked them up. You may still be able to find them on ebay for a good price, but without the return policy I had at the local microcenter store.

Yea that's because the 7970 & the R9 280X use the same GPU chip ;), but they do use different PCBs & bioses with slightly different stock clock speeds.

I picked up this HD 7950 for £82 including postage this September :), it was slightly cheaper being the reference design with a not very quiet cooler, but it's not that bad :).

AnandTech do indeed do good graphic cards reviews, but for a quick over view of DP power I use wiki ;) (you can always check it's figures afterwards if wanted).

---

Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG

Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit
2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7

Hi there. So, what are the current requirements for GPU crunching to work? I'm also getting the message from MW@h, saying ""Message from server: ATI GPU R600 (R38xx) does not support OpenCL"", but I have Radeon HD 8400 (Southern islands, or R3, or something)

fglrx version: 15.200
BOINC version: 7.6.22
Host link

BOINC says:

Sat  9 Jan 00:15:35 2016 |  | OpenCL: AMD/ATI GPU 0: Kalindi (driver version 1729.3 (VM), device version OpenCL 2.0 AMD-APP (1729.3), 512MB, 512MB available, 96 GFLOPS peak)
Sat  9 Jan 00:15:35 2016 |  | OpenCL CPU: AMD Athlon(tm) 5350 APU with Radeon(tm) R3 (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 1729.3 (sse2,avx), device version OpenCL 1.2 AMD-APP (1729.3))

clinfo says:

  Device Type:                                   CL_DEVICE_TYPE_GPU
  Vendor ID:                                     1002h
  Board name:
  Device Topology:                               PCI[ B#0, D#1, F#0 ]
  Max compute units:                             2
  Max work items dimensions:                     3
    Max work items[0]:                           256
    Max work items[1]:                           256
    Max work items[2]:                           256
  Max work group size:                           256
  Preferred vector width char:                   4
  Preferred vector width short:                  2
  Preferred vector width int:                    1
  Preferred vector width long:                   1
  Preferred vector width float:                  1
  Preferred vector width double:                 1
  Native vector width char:                      4
  Native vector width short:                     2
  Native vector width int:                       1
  Native vector width long:                      1
  Native vector width float:                     1
  Native vector width double:                    1
  Max clock frequency:                           600Mhz
  Address bits:                                  64
  Max memory allocation:                         134217728
  Image support:                                 Yes
  Max number of images read arguments:           128
  Max number of images write arguments:          64
  Max image 2D width:                            16384
  Max image 2D height:                           16384
  Max image 3D width:                            2048
  Max image 3D height:                           2048
  Max image 3D depth:                            2048
  Max samplers within kernel:                    16
  Image formats supported:
   2D Read-Write/Read-Only/Write-Only
    CL_R: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 half_f
loat float
    CL_A: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 half_f
loat float
    CL_RG: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 half_
float float
    CL_RGB: unorm-int_101010
    CL_RGBA: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 hal
f_float float
    CL_BGRA: snorm-int8 unorm-int8 signed-int8 unsigned-int8
    CL_ARGB: snorm-int8 unorm-int8 signed-int8 unsigned-int8
    CL_INTENSITY: snorm-int8 snorm-int16 unorm-int8 unorm-int16 half_float float
    CL_LUMINANCE: snorm-int8 snorm-int16 unorm-int8 unorm-int16 half_float float
   3D Read-Write/Read-Only/Write-Only
    CL_R: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 half_f
loat float
    CL_A: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 half_f
loat float
    CL_RG: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 half_
float float
    CL_RGB: unorm-int_101010
    CL_RGBA: snorm-int8 snorm-int16 unorm-int8 unorm-int16 signed-int8 signed-int16 signed-int32 unsigned-int8 unsigned-int16 unsigned-int32 hal
f_float float
    CL_BGRA: snorm-int8 unorm-int8 signed-int8 unsigned-int8
    CL_ARGB: snorm-int8 unorm-int8 signed-int8 unsigned-int8
    CL_INTENSITY: snorm-int8 snorm-int16 unorm-int8 unorm-int16 half_float float
    CL_LUMINANCE: snorm-int8 snorm-int16 unorm-int8 unorm-int16 half_float float
  Max size of kernel argument:                   1024
  Alignment (bits) of base address:              2048
  Minimum alignment (bytes) for any datatype:    128
  Single precision floating point capability
    Denorms:                                     No
    Quiet NaNs:                                  Yes
    Round to nearest even:                       Yes
    Round to zero:                               Yes
    Round to +ve and infinity:                   Yes
    IEEE754-2008 fused multiply-add:             Yes
  Cache type:                                    Read/Write
  Cache line size:                               64
  Cache size:                                    16384
  Global memory size:                            536870912
  Constant buffer size:                          65536
  Max number of constant args:                   8
  Local memory type:                             Scratchpad
  Local memory size:                             32768
  Max pipe arguments:                            16
  Max pipe active reservations:                  16
  Max pipe packet size:                          134217728
  Max global variable size:                      120795904
  Max global variable preferred total size:      536870912
  Max read/write image args:                     64
  Max on device events:                          1024
  Queue on device max size:                      524288
  Max on device queues:                          1
  Queue on device preferred size:                262144
  SVM capabilities:
    Coarse grain buffer:                         Yes
    Fine grain buffer:                           Yes
    Fine grain system:                           No
    Atomics:                                     No
  Preferred platform atomic alignment:           0
  Preferred global atomic alignment:             0
  Preferred local atomic alignment:              0
  Kernel Preferred work group size multiple:     64
  Error correction support:                      0
  Unified memory for Host and Device:            1
  Profiling timer resolution:                    1
  Device endianess:                              Little
  Available:                                     Yes
  Compiler available:                            Yes
  Execution capabilities:
    Execute OpenCL kernels:                      Yes
    Execute native function:                     No
  Queue on Host properties:
    Out-of-Order:                                No
    Profiling :                                  Yes
  Queue on Device properties:
    Out-of-Order:                                Yes
    Profiling :                                  Yes
  Platform ID:                                   0x7ffad48038f0
  Name:                                          Kalindi
  Vendor:                                        Advanced Micro Devices, Inc.
  Device OpenCL C version:                       OpenCL C 2.0
  Driver version:                                1729.3 (VM)
  Profile:                                       FULL_PROFILE
  Version:                                       OpenCL 2.0 AMD-APP (1729.3)
  Extensions:                                    cl_khr_fp64 cl_amd_fp64 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_image2d_from_buffer cl_khr_spir cl_khr_subgroups cl_khr_gl_event cl_khr_depth_images

The AMD Radeon 7850 also works fine.

So I noticed that a lot of people go with the Radeon R8 280x series for doing work units, seems to be the choice graphics card. I was wondering will a Radeon R9 290x series work just as well if not better and does the amount of video memory play a factor on the card for doing work units for milkyway@home?

Thank you in advance for any help in direction you can point me in.

Disregard this post as it was already answered which I did not pay attention to reading further down the thread. My apologies.