Legion 5 which cpu would you chose?

[Shahrad]

I am considering to buy a laptop, the legion 5. I don't know which cpu is better for chess engines: AMD R7 4800H or i7-10750H. Any advice would be very appreciated.

[lithander]

I would go with the Ryzen. 8 cores instead of 6 and much more power efficient. Single core speed is about equal.

https://www.notebookcheck.net/R7-4800H- ... 596.0.html

That said... the BMI2 implementation of AMD's Zen2 cores isn't great, which matters for computer chess in so far that you'll want to use the AVX2 builds of Stockfish etc if you got a Zen2 core.

[Shahrad]

Thank you for your advice, I did go with Asus TUF Gaming A15 FX506IV R7 4800H.

[MikeB]

Thank you for your advice, I did go with Asus TUF Gaming A15 FX506IV R7 4800H.

You made a good choice. Enjoy!

[Wilson]

any good laptop with a fast Ryzen cpu like the 4800H but without pricey gpus ?

[Ras]

any good laptop with a fast Ryzen cpu like the 4800H but without pricey gpus ?

Maybe Schenker Via 15 Pro. Also available as Tuxedo Pulse 15 if you want Linux pre-installed. RAM and SSD can be configured individually.
https://www.schenker-tech.de/en/schenker-via-15-pro-en
https://www.tuxedocomputers.com/en/Linu ... en1.tuxedo

[Shahrad]

Many thanks for all the replies.
There is a weird problem: lc0 cudnn is 10kN/s faster than the cuda. As said it has a 2060 RTX.

[MMarco]

Many thanks for all the replies.
There is a weird problem: lc0 cudnn is 10kN/s faster than the cuda. As said it has a 2060 RTX.

That is the case for small networks (10x128) like the one included (703810) in Lc0 official release for many graphic card.

Since you have a 2060, try using a large 30x384 network like J94-100 here: https://github.com/jhorthos/lczero-trai ... a-Training

NPS will be much slower than for a small networks, but Lc0 will play stronger due to added "knowledge". For large networks, Cuda will be faster than CudNN. Also make sure you have the latest drivers installed for your card. If those are old, they might not use recent optimizations in the Lc0 binary.

[Shahrad]

Many thanks for all the replies.
There is a weird problem: lc0 cudnn is 10kN/s faster than the cuda. As said it has a 2060 RTX.

That is the case for small networks (10x128) like the one included (703810) in Lc0 official release for many graphic card.

Since you have a 2060, try using a large 30x384 network like J94-100 here: https://github.com/jhorthos/lczero-trai ... a-Training

NPS will be much slower than for a small networks, but Lc0 will play stronger due to added "knowledge". For large networks, Cuda will be faster than CudNN. Also make sure you have the latest drivers installed for your card. If those are old, they might not use recent optimizations in the Lc0 binary.

Good to know, I installed the J94-100, thank you. Now it is running at 4kn/s and the question is at which depth the evaluation is somewhat reliable? I am using it only for analysis and not for engine tournaments.

[Milos]

I am considering to buy a laptop, the legion 5. I don't know which cpu is better for chess engines: AMD R7 4800H or i7-10750H. Any advice would be very appreciated.

4800H is already outdated. Try finding a laptop with 5800H/HS/HX/U. There Intel doesn't have advantage even on single core.