I use the INNO3D 2 slot 4070Ti. I bought it specifically because it's a 2 slot card. The 4070Ti Super is the same. It's a great card, I've had no issues although I'm struggling to get any good results out of it with E@H currently.
There's not many 2 slot cards these days.
Hi Chooka!
I agree with you that there are not many 2-slot cards available these days. I've read your posts with interest, and since you mentioned INNO3D (which I never heard of until you mentioned it) I did some checking on them because I also would love to have 2-slot GPUs in my systems - as long as they perform well.
The 4070 Ti (and other INNO3D GPUs) would appear to be an okay GPU for gaming and for normal usage, but not for over clocking as their VRMs are subpar and tend to breakdown when pushed.
We in GPU Users Group, Tom included, tend to push our GPUs harder than most people. If I may ask:
How long have you owned your INNO3D 4070 Ti ?
Do you over clock your 4070 Ti ?
Do you notice any Coil whine?
Does the 4070 Ti run cool enough for you?
Have you had any problems setting up the 4070 Ti for running BOINC projects?
All-in-all, I'm interested in getting an NVIDIA 40XX series card, and may eventually do so with all three of my computers because of the newer NVIDIA's Ada Lovelace generation of GPU. The power requirements have dropped considerably. Plus, they are a bit faster.
BTW, I run Linux Ubuntu 22.04 and I see that you are on Windows 10 Pro which I felt was pretty good too, until Windows decided to continue doing constant upgrades, etc. It became a hassle for me so I made the switch to Linux at the urging of my teammates and I never want to go back.
Has anybody been using GRID or TESLA with multiple GPU chips on them?
I am interested how it runs the BOINC? Single thread per GPU chip or single thread per GPU board? ????♂️
Also, interested to find accurate 3D model (can be STL or STEP) for nVidia TESLA / GRID cards. As I plan to design my own cooler adapters for those cards (to be done on 3Dprinter). Can anybody help out?
Has anybody been using GRID or TESLA with multiple GPU chips on them?
I am interested how it runs the BOINC? Single thread per GPU chip or single thread per GPU board? ????♂️
Also, interested to find accurate 3D model (can be STL or STEP) for nVidia TESLA / GRID cards. As I plan to design my own cooler adapters for those cards (to be done on 3Dprinter). Can anybody help out?
Boinc is not setup to multi-thread gpu tasks so one task per gpu chip, gaming can use multiple gpu's for one game but Boinc can not.
Has anybody been using GRID or TESLA with multiple GPU chips on them?
I am interested how it runs the BOINC? Single thread per GPU chip or single thread per GPU board? ????♂️
Also, interested to find accurate 3D model (can be STL or STEP) for nVidia TESLA / GRID cards. As I plan to design my own cooler adapters for those cards (to be done on 3Dprinter). Can anybody help out?
I think at one time we had a professor running Radeon VII Pro cards that were two gpus per physical card. In his case Boinc and his Linux OS treated them as two cards. He was running either an apparent 6 or 8 gpus. I don't remember which.
As long as the drivers/OS recognize them as separate cards so will Boinc.
Because we don't have the coding for them, I doubt multiple GPU cards/chips can be treated as a single GPU. If the OS could present those multiple GPU cards/chips as a single GPU then it should be possible.
In many cases you can run multiple threads/tasks on a single GPU. I believe someone is running 5x on a Radeon VII. I am running 4x on an rtx 3080 ti. And 3x on 3 Titan V's.
Just takes an "app_config.xml" file in the project folder. 0.5 gpu_usage is 2x, 0.333 is 3x, 0.25 is 4x, 0.19 is 5x. For some reason mine won't do 5x with 0.20
Has anybody been using GRID or TESLA with multiple GPU chips on them?
I am interested how it runs the BOINC? Single thread per GPU chip or single thread per GPU board? ????♂️
Also, interested to find accurate 3D model (can be STL or STEP) for nVidia TESLA / GRID cards. As I plan to design my own cooler adapters for those cards (to be done on 3Dprinter). Can anybody help out?
Boinc is not setup to multi-thread gpu tasks so one task per gpu chip, gaming can use multiple gpu's for one game but Boinc can not.
But you can have multiple tasks per GPU "chip"/card..
I think multi-threading has a very specific meaning and it can be confusing. 1 (CPU) task can multi-thread if written to and run 8, 16, 32 cpus/threads under a single task. Yes, in theory, that means it approaches 32 times faster if the parallelization of tasks is present.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Has anybody been using GRID or TESLA with multiple GPU chips on them?
I am interested how it runs the BOINC? Single thread per GPU chip or single thread per GPU board? ????♂️
Also, interested to find accurate 3D model (can be STL or STEP) for nVidia TESLA / GRID cards. As I plan to design my own cooler adapters for those cards (to be done on 3Dprinter). Can anybody help out?
Boinc is not setup to multi-thread gpu tasks so one task per gpu chip, gaming can use multiple gpu's for one game but Boinc can not.
First off, KLiK is talking about a muliti-chip GPU, not multi-threaded GPU. To my knowledge... NO?... Really not sure.
Secondly, yes you can run multiplethreads on a GPU in BOINC with the proper setup in your <app_config.xml) file in the Projects section. E@H, for example:
...where the <gpu usage> is set for running 3 tasks simultaneously.
Plus there is no limitation other than the host's hardware to running more than one GPU for BOINC, whether with the same project or multiple projects. Use the Client Configuration web page to do that.
You’re talking about Gaurav. To my knowledge he had regular Radeon VIIs not the Pro version. And the Radeon Pro VII isn’t a dual GPU board. It’s basically the same single GPU as the normal RVII, but with more FP64 capability.
what he had was a “Pro” version of some other older AMD GPU (like Fiji era or something) that was a dual GPU board. (4x cards, BOINC reported 8)
You’re talking about Gaurav. To my knowledge he had regular Radeon VIIs not the Pro version. And the Radeon Pro VII isn’t a dual GPU board. It’s basically the same single GPU as the normal RVII, but with more FP64 capability.
what he had was a “Pro” version of some other older AMD GPU (like Fiji era or something) that was a dual GPU board. (4x cards, BOINC reported 8)
Chooka wrote: I use the
)
Hi Chooka!
I agree with you that there are not many 2-slot cards available these days. I've read your posts with interest, and since you mentioned INNO3D (which I never heard of until you mentioned it) I did some checking on them because I also would love to have 2-slot GPUs in my systems - as long as they perform well.
The 4070 Ti (and other INNO3D GPUs) would appear to be an okay GPU for gaming and for normal usage, but not for over clocking as their VRMs are subpar and tend to breakdown when pushed.
We in GPU Users Group, Tom included, tend to push our GPUs harder than most people. If I may ask:
All-in-all, I'm interested in getting an NVIDIA 40XX series card, and may eventually do so with all three of my computers because of the newer NVIDIA's Ada Lovelace generation of GPU. The power requirements have dropped considerably. Plus, they are a bit faster.
BTW, I run Linux Ubuntu 22.04 and I see that you are on Windows 10 Pro which I felt was pretty good too, until Windows decided to continue doing constant upgrades, etc. It became a hassle for me so I made the switch to Linux at the urging of my teammates and I never want to go back.
Hope all is well with you and yours!
Proud member of the Old Farts Association
Has anybody been using GRID
)
Has anybody been using GRID or TESLA with multiple GPU chips on them?
I am interested how it runs the BOINC? Single thread per GPU chip or single thread per GPU board? ????♂️
Also, interested to find accurate 3D model (can be STL or STEP) for nVidia TESLA / GRID cards. As I plan to design my own cooler adapters for those cards (to be done on 3Dprinter). Can anybody help out?
non-profit org. Play4Life in Zagreb, Croatia, EU
KLiK wrote: Has anybody been
)
Boinc is not setup to multi-thread gpu tasks so one task per gpu chip, gaming can use multiple gpu's for one game but Boinc can not.
KLiK wrote:Has anybody been
)
I think at one time we had a professor running Radeon VII Pro cards that were two gpus per physical card. In his case Boinc and his Linux OS treated them as two cards. He was running either an apparent 6 or 8 gpus. I don't remember which.
As long as the drivers/OS recognize them as separate cards so will Boinc.
Because we don't have the coding for them, I doubt multiple GPU cards/chips can be treated as a single GPU. If the OS could present those multiple GPU cards/chips as a single GPU then it should be possible.
In many cases you can run multiple threads/tasks on a single GPU. I believe someone is running 5x on a Radeon VII. I am running 4x on an rtx 3080 ti. And 3x on 3 Titan V's.
Just takes an "app_config.xml" file in the project folder. 0.5 gpu_usage is 2x, 0.333 is 3x, 0.25 is 4x, 0.19 is 5x. For some reason mine won't do 5x with 0.20
Example is this:
=====
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
mikey wrote: KLiK
)
But you can have multiple tasks per GPU "chip"/card..
I think multi-threading has a very specific meaning and it can be confusing. 1 (CPU) task can multi-thread if written to and run 8, 16, 32 cpus/threads under a single task. Yes, in theory, that means it approaches 32 times faster if the parallelization of tasks is present.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
mikey wrote: KLiK
)
First off, KLiK is talking about a muliti-chip GPU, not multi-threaded GPU. To my knowledge... NO?... Really not sure.
Secondly, yes you can run multiple threads on a GPU in BOINC with the proper setup in your <app_config.xml) file in the Projects section. E@H, for example:
<app>
<name>einstein_O3AS</name>
<gpu_versions>
<cpu_usage>1.0</cpu_usage>
<gpu_usage>0.333</gpu_usage>
</gpu_versions>
</app>
...where the <gpu usage> is set for running 3 tasks simultaneously.
Plus there is no limitation other than the host's hardware to running more than one GPU for BOINC, whether with the same project or multiple projects. Use the Client Configuration web page to do that.
https://boinc.berkeley.edu/wiki/Client_configuration
Proud member of the Old Farts Association
You’re talking about Gaurav.
)
You’re talking about Gaurav. To my knowledge he had regular Radeon VIIs not the Pro version. And the Radeon Pro VII isn’t a dual GPU board. It’s basically the same single GPU as the normal RVII, but with more FP64 capability.
what he had was a “Pro” version of some other older AMD GPU (like Fiji era or something) that was a dual GPU board. (4x cards, BOINC reported 8)
_________________________________________________________________________
Ian&Steve C. wrote: You’re
)
I did not know that. Thanks Ian!
Proud member of the Old Farts Association
I believe there is a
)
I believe there is a difference between a multi threaded application. And a single threaded application.
And apparently I wasn't clear enough in my exposition.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
To be more clear, I am
)
To be more clear, I am talking about multi-GPU chips per physical card.
Examples are:
- nVidia GRIDs: https://www.techpowerup.com/gpu-specs/?generation=GRID&sort=generation
- some nVidia TESLAs:
a) https://www.techpowerup.com/gpu-specs/tesla-k80.c2616
b) https://www.techpowerup.com/gpu-specs/tesla-m60.c2760
etc.
Just wondering if anybody is using those? & how does BOINC sees those multi-GPU chips? ????♂️
non-profit org. Play4Life in Zagreb, Croatia, EU