HowTo enable GPU for read mapping?

[claus-h-g]

How do I activate define the GPU for read mapping?
I did go though the working example and used nvidia-smi to monitor any potential activity on the GPU: I could not find any.

For the gem3-mapper I did find the --gpu switch to enable GPU acceleration. But to me it is unclear in which config file I set this switch.

[JakeLehle]

Hmmm good question. I'm guessing have already looked at the config file that is fed into the

gemBS prepare  

step correct? You could mess with the threads and cores in that config file. But I don't think it has any toggles for GPU stuff there.

Let me look into where that pulls the gem3-mapper settings from and get back to you.

[heathsc]

The GPU version of gem3 is not methylation aware so cannot be used with gemBS. Simon

…

On Wed, 17 Aug 2022, 21:02 Jake Lehle, ***@***.***> wrote: Hmmm good question. I'm guessing have already looked at the config file that is fed into the gemBS prepare step correct? You could mess with the threads and cores in that config file. But I don't think it has any toggles for GPU stuff there. Let me look into where that pulls the gem3-mapper settings from and get back to you. — Reply to this email directly, view it on GitHub <#93 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAY4657SZFL77OKSJKLSKATVZUZNJANCNFSM562DGEDQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[claus-h-g]

Many thanks for your fast replies.
I guess this explains why I could not see any error, progress or a result when I tried to use gem3 on methylation sequencing data.
Are you aware of a different mapper for Bisultfite sequencing data supporting the use of GPUs. I expect the need to analyze a multitude of different bisulfite sequencing data. The mapper should be kind of straightforward to use. I am aware of Airoc, but here the reference genome has to be split into pieces of a size less than 1 GB. I expect this might raise any issues for any diagnostic setting and I would prefer to work with a reference genome in one piece to minimize the number of steps that might go wrong.

[achacond]

Simon, what is missing in GEM-GPU to be compatible in gemBS? Because this is the first I'm hearing of it, theoretically all the GPU primitives and internal infra that I developed for GEM should be transparent to the workflow running on top. El mié, 17 ago 2022 a las 22:15, Simon Heath ***@***.***>) escribió:

…

The GPU version of gem3 is not methylation aware so cannot be used with gemBS. Simon On Wed, 17 Aug 2022, 21:02 Jake Lehle, ***@***.***> wrote: > Hmmm good question. I'm guessing have already looked at the config file > that is fed into the > > gemBS prepare > > step correct? You could mess with the threads and cores in that config > file. But I don't think it has any toggles for GPU stuff there. > > Let me look into where that pulls the gem3-mapper settings from and get > back to you. > > — > Reply to this email directly, view it on GitHub > <#93 (comment)>, or > unsubscribe > < https://github.com/notifications/unsubscribe-auth/AAY4657SZFL77OKSJKLSKATVZUZNJANCNFSM562DGEDQ > > . > You are receiving this because you are subscribed to this thread.Message > ID: ***@***.***> > — Reply to this email directly, view it on GitHub <#93 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCPCJ3CRESUNUSZYK4G5VLVZVI77ANCNFSM562DGEDQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[heathsc]

Hi Alex, There should be no major problem getting this working. The methylation part has a few changes w.r.t. the classic gem and I don't think it would be that difficult to get it working. At the moment I don't have many test platforms with GPUs so I haven't put too much time into it. Simon On Tue, Oct 11, 2022 at 9:37 AM Alejando (Alex) Chacon < ***@***.***> wrote:

…

Simon, what is missing in GEM-GPU to be compatible in gemBS? Because this is the first I'm hearing of it, theoretically all the GPU primitives and internal infra that I developed for GEM should be transparent to the workflow running on top. El mié, 17 ago 2022 a las 22:15, Simon Heath ***@***.***>) escribió: > The GPU version of gem3 is not methylation aware so cannot be used with > gemBS. > > > Simon > > On Wed, 17 Aug 2022, 21:02 Jake Lehle, ***@***.***> wrote: > > > Hmmm good question. I'm guessing have already looked at the config file > > that is fed into the > > > > gemBS prepare > > > > step correct? You could mess with the threads and cores in that config > > file. But I don't think it has any toggles for GPU stuff there. > > > > Let me look into where that pulls the gem3-mapper settings from and get > > back to you. > > > > — > > Reply to this email directly, view it on GitHub > > <#93 (comment)>, or > > unsubscribe > > < > https://github.com/notifications/unsubscribe-auth/AAY4657SZFL77OKSJKLSKATVZUZNJANCNFSM562DGEDQ > > > > . > > You are receiving this because you are subscribed to this thread.Message > > ID: ***@***.***> > > > > — > Reply to this email directly, view it on GitHub > <#93 (comment)>, or > unsubscribe > < https://github.com/notifications/unsubscribe-auth/ABCPCJ3CRESUNUSZYK4G5VLVZVI77ANCNFSM562DGEDQ > > . > You are receiving this because you are subscribed to this thread.Message > ID: ***@***.***> > — Reply to this email directly, view it on GitHub <#93 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAY4657CTJPMXT6VJVCKKB3WCUKLXANCNFSM562DGEDQ> . You are receiving this because you commented.Message ID: ***@***.***>

[achacond]

Thanks Simon, I have reached you by email regarding this topic. Best, Alex El mar, 11 oct 2022 a las 8:46, Simon Heath ***@***.***>) escribió:

…

Hi Alex, There should be no major problem getting this working. The methylation part has a few changes w.r.t. the classic gem and I don't think it would be that difficult to get it working. At the moment I don't have many test platforms with GPUs so I haven't put too much time into it. Simon On Tue, Oct 11, 2022 at 9:37 AM Alejando (Alex) Chacon < ***@***.***> wrote: > Simon, what is missing in GEM-GPU to be compatible in gemBS? > Because this is the first I'm hearing of it, theoretically all the GPU > primitives and internal infra that I developed for GEM should be > transparent to the workflow running on top. > > El mié, 17 ago 2022 a las 22:15, Simon Heath ***@***.***>) > escribió: > > > The GPU version of gem3 is not methylation aware so cannot be used with > > gemBS. > > > > > > Simon > > > > On Wed, 17 Aug 2022, 21:02 Jake Lehle, ***@***.***> wrote: > > > > > Hmmm good question. I'm guessing have already looked at the config file > > > that is fed into the > > > > > > gemBS prepare > > > > > > step correct? You could mess with the threads and cores in that config > > > file. But I don't think it has any toggles for GPU stuff there. > > > > > > Let me look into where that pulls the gem3-mapper settings from and get > > > back to you. > > > > > > — > > > Reply to this email directly, view it on GitHub > > > <#93 (comment) >, > or > > > unsubscribe > > > < > > > https://github.com/notifications/unsubscribe-auth/AAY4657SZFL77OKSJKLSKATVZUZNJANCNFSM562DGEDQ > > > > > > . > > > You are receiving this because you are subscribed to this > thread.Message > > > ID: ***@***.***> > > > > > > > — > > Reply to this email directly, view it on GitHub > > <#93 (comment)>, or > > unsubscribe > > < > https://github.com/notifications/unsubscribe-auth/ABCPCJ3CRESUNUSZYK4G5VLVZVI77ANCNFSM562DGEDQ > > > > . > > You are receiving this because you are subscribed to this thread.Message > > ID: ***@***.***> > > > > — > Reply to this email directly, view it on GitHub > <#93 (comment)>, or > unsubscribe > < https://github.com/notifications/unsubscribe-auth/AAY4657CTJPMXT6VJVCKKB3WCUKLXANCNFSM562DGEDQ > > . > You are receiving this because you commented.Message ID: > ***@***.***> > — Reply to this email directly, view it on GitHub <#93 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCPCJ77BGBYADKFQ7OSIB3WCULLXANCNFSM562DGEDQ> . You are receiving this because you commented.Message ID: ***@***.***>

[claus-h-g]

Sorry for the late reply.
I would be happy to contribute and step in for testing. I should be to run test on X86 xUbuntu Systems equipped with RTX 20820 or 3060.
If possible and or interest for you, I would interested to test on system on chip such as Nvidia Jetson AGX - a hybrid of aarch64 and GPUs.
Do you have a repository for developing the code?