poor performance of DTSBV (via DPBTRS) on Zen 5

[hiokazaki]

Hello,

I noticed low performance of DPBTRS, profiling shows that a very large fraction of CPU time is spent in DTBSV. Using Intel MKL improves the situation so low memory bandwidth of desktop Zen 5 is probably not the root cause here.

Can you assist in enhancing performance of DTBSV on Zen 5?

[kvaragan]

[AMD Official Use Only - AMD Internal Distribution Only] Hi, We have not yet optimized these routines; we will add it to our to-do list. What's your case ? Thanks Get Outlook for iOS<https://aka.ms/o0ukef>

…

________________________________ From: hiokazaki ***@***.***> Sent: Sunday, January 12, 2025 6:40:25 AM To: amd/blis ***@***.***> Cc: Subscribed ***@***.***> Subject: [amd/blis] poor performance of DTSBV (via DPBTRS) on Zen 5 (Issue #26) Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding. Hello, I noticed low performance of DPBTRS, profiling shows that a very large fraction of CPU time is spent in DTBSV. Using Intel MKL improves the situation so low memory bandwidth of desktop Zen 5 is probably not the root cause here. Can you assist in enhancing performance of DTBSV on Zen 5? — Reply to this email directly, view it on GitHub<#26>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADAJQWAKAWWLX4HFYEZ6QTT2KG6IDAVCNFSM6AAAAABVAPPG6WVHI2DSMVQWIX3LMV43ASLTON2WKOZSG44DEMRRHAYTEMI>. You are receiving this because you are subscribed to this thread.

[hiokazaki]

Hi, We are using Nektar++ IncompressibleNavierStokes CFD code. If you need exact DPBTRS call parameters let me know when you are ready and I can send you a trace, or a test case if you want to reproducec with the original Nektar++ code.

…

On Mon, Jan 13, 2025 at 01:03:34AM -0800, Kiran wrote: [AMD Official Use Only - AMD Internal Distribution Only] Hi, We have not yet optimized these routines; we will add it to our to-do list. What's your case ? Thanks Get Outlook for iOS<https://aka.ms/o0ukef> ________________________________ From: hiokazaki ***@***.***> Sent: Sunday, January 12, 2025 6:40:25 AM To: amd/blis ***@***.***> Cc: Subscribed ***@***.***> Subject: [amd/blis] poor performance of DTSBV (via DPBTRS) on Zen 5 (Issue #26) Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding. Hello, I noticed low performance of DPBTRS, profiling shows that a very large fraction of CPU time is spent in DTBSV. Using Intel MKL improves the situation so low memory bandwidth of desktop Zen 5 is probably not the root cause here. Can you assist in enhancing performance of DTBSV on Zen 5? — Reply to this email directly, view it on GitHub<#26>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADAJQWAKAWWLX4HFYEZ6QTT2KG6IDAVCNFSM6AAAAABVAPPG6WVHI2DSMVQWIX3LMV43ASLTON2WKOZSG44DEMRRHAYTEMI>. You are receiving this because you are subscribed to this thread. -- Reply to this email directly or view it on GitHub: #26 (comment) You are receiving this because you authored the thread. Message ID: ***@***.***>

[kvaragan]

Hi,
Can you send mail about performance issue you are facing with Nektar++ to: toolchainsupport@amd.com

[hiokazaki]

sure

…

On Tue, Jan 14, 2025 at 08:59:40PM -0800, Kiran wrote: Hi, Can you send mail about performance issue you are facing with Nektar++ to: ***@***.*** -- Reply to this email directly or view it on GitHub: #26 (comment) You are receiving this because you authored the thread. Message ID: ***@***.***>