New Intel Xeon processor available :

https://www.tomshardware.com/reviews/intel-cascade-lake-xeon-optane,6061.html

56 core and 112 Threads...is boinc able to manage more than 100 threads at once ?

---

56 core and 112 Threads

You'll need to sell both kidneys, an arm & both legs to afford just one (and they're aimed at 2 socket systems), and they are only available to OEMs- you have to order a complete system from them, and they then place the order with Intel.
Oh, and they're rated at 400W, per socket.

---

Grant
Darwin NT

New Intel Xeon processor available :

https://www.tomshardware.com/reviews/intel-cascade-lake-xeon-optane,6061.html

56 core and 112 Threads...is boinc able to manage more than 100 threads at once ?

Maybe not. Since the cpu limit seems to be 100 tasks, they would all be "running" (all 100) and as soon as one uploaded another could be downloaded. I know we are "spoofing" the Gpu's. I wonder if the 100 task cpu limit is at the client level or at the server level? I think I remember it being at the server level.

Another possibility would be to run two different Boinc Managers in two different directories which would effectively give you two "systems" living on one hardware platform. Or setup a couple of VM's to do the same trick.

Tom

---

A proud member of the OFA (Old Farts Association).

I know we are "spoofing" the Gpu's.

Interesting. I would like to learn how to do this on anonymous apps. I only know how to setup multiple clients but that's too cumbersome. Thanks.

I know we are "spoofing" the Gpu's.

Interesting. I would like to learn how to do this on anonymous apps. I only know how to setup multiple clients but that's too cumbersome. Thanks.

You have to know how to compile the source code unless you know someone who can do it for you.

The only advantage it provides is for high end gpu systems. It reduces or eliminates the lost gpu processing time on Tuesday by increasing the cache for gpu tasks. It is still possible to "run dry" when the Seti Servers experience extended outages or have other problems during the week.

One of the other things you can do is limit the number of tasks reported to say 100. Even if you need to report 500 tasks, this reduces the congestion and usually makes the task reporting work.

Tom

---

A proud member of the OFA (Old Farts Association).

Is the source code here: https://boinc.berkeley.edu/trac/wiki/CompileClient or here: https://boinc.berkeley.edu/trac/wiki/BuildSystem?

If it is C++ I may know someone but needs to know which routine or constants to update.

Thanks again.

Neither of your links - this is probably the best place to start, and be prepared for a serious nose bleed, the code is convoluted, and in places very poorly documented.
https://boinc.berkeley.edu/trac/wiki/SourceCodeGit

Some people do the GPU spoofing by changing the GPU BIOS or driver so that it miss-reports as a multi-gpu device, which of course cannot be done within BOINC, and may have some undesirable and expensive consequences.

Also remember the 100 tasks for the CPU is a server side limit, not a client side limit, so we can't do anything (reliable) about it, and people spoofing high GPU counts are doing so as a way of having enough tasks to chug through the Tuesday outage, not to run high numbers of concurrent tasks.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

New Intel Xeon processor available :

https://www.tomshardware.com/reviews/intel-cascade-lake-xeon-optane,6061.html

56 core and 112 Threads...is boinc able to manage more than 100 threads at once ?

Maybe someone should break down and buy one of those $250,000 "small" Super Computers ;)

When I googled "gpu server" it popped up something about a $25,000 box....

And rumor has it the next round of EPYC dual cpu servers will be something ridiculous like 128 cores/256threads?

Tom

---

A proud member of the OFA (Old Farts Association).

Some people do the GPU spoofing by changing the GPU BIOS or driver so that it miss-reports as a multi-gpu device, which of course cannot be done within BOINC, and may have some undesirable and expensive consequences.

it's actually neither of these.

the GPU spoofing is done by modification to the BOINC code.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

I cant even imagine the power bill to run that 24/7 let alone a 8 hours lol!

---

All of this has happened before and will happen again -Battlestar Galactica